Adult Depression Clinical Practice Guideline

Transcription

1 NATIONAL CLINICAL PRACTICE GUIDELINE Adult Depression Clinical Practice Guideline This guideline is informational only. It is not intended or designed as a substitute for the reasonable exercise of independent clinical judgment by practitioners, considering each patient s needs on an individual basis. Guideline recommendations apply to populations of patients. Clinical judgment is necessary to design treatment plans for individual patients. Approved by the National Guideline Directors February 2010

2 Table of Contents Introduction... 1 Guideline Summary First-Line Treatment of Major Depressive Disorder (MDD) Hypericum (St. John s Wort) For Treatment of MDD Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, or Plan Second-Line Treatment of MDD Length of Treatment With Antidepressants In Patients With MDD Follow-Up For Patients In The Acute Phase of Treatment For MDD Follow-Up For Patients In The Continuation Phase of Treatment For MDD Follow-Up For Patients In The Maintenance Phase of Treatment For MDD Discontinuation of Antidepressants In Patients With MDD Treatment Preferences For MDD In Different Ethnic Groups Patient Self-Management Strategies For Improving Symptoms of MDD Behavioral Health Education Classes For Adults With MDD Antidepressants To Avoid During Pregnancy or Breastfeeding Rationale Statements First-Line Treatment of Major Depressive Disorder (MDD) Hypericum (St. John s Wort) for MDD Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, Or Plan Second-Line Treatment Of MDD Length Of Treatment With Antidepressants In Patients With MDD Follow-Up For Patients In The Acute Phase of Treatment For MDD Follow-Up For Patients In The Continuation Phase of Treatment of MDD Follow-Up For Patients In Maintenance Phase of Treatment of MDD Discontinuation of Antidepressants in Patients with MDD Treatment Preferences For MDD In Different Ethnic Groups Patient Self-Management Strategies for Improving Symptoms of MDD Behavioral Health Education Classes For Adults With MDD Antidepressants To Avoid During Pregnancy or Breastfeeding Appendix A: Criteria for Grading the Evidence Appendix B: Supporting Documentation First-Line Treatment of Major Depressive Disorder (MDD) Problem Formulation Search Strategy Evidence Tables Hypericum (St. John s Wort) For Treatment of MDD Problem Formulation Search Strategy Evidence Tables i National Adult Depression Clincial Practice Guideline

3 3. Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, or Plan Problem Formulation Search Strategy Evidence Tables Second-Line Treatement of MDD Problem Formulation Search Strategy Evidence Tables Length of Treatment With Antidepressants In Patients With MDD Problem Formulation Search Strategy Evidence Tables Follow-Up For Patients In The Acute Phase of Treatment For MDD Problem Formulation Search Strategy Evidence Table Follow-Up For Patients In The Continuation Phase of Treatment For MDD Problem Formulation Search Strategy Evidence Table Follow-Up For Patients In Maintenance Phase Treatment Of MDD Problem Formulation Search Strategy Evidence Table Discontinuation of Antidepressants In Patients With MDD Problem Formulation Search Strategy Evidence Tables Treatment Preferences For MDD In Different Ethnic Groups Problem Formulation Search Strategy Evidence Table Patient Self-Management Strategies For Improving Symptoms of MDD Problem Formulation Search Strategy Evidence Tables ii National Adult Depression Clincial Practice Guideline

4 12. Behavioral Health Education Classes For Adults With MDD Problem Formulation Search Strategy Evidence Tables Antidepressants To Avoid During Pregnancy or Breastfeeding Problem Formulation Search Strategy Evidence Tables References iii National Adult Depression Clincial Practice Guideline

5 Introduction Kaiser Permanente s National Guideline Program The National Guideline Program (NGP) supports the development of a core set of explicit, scientifically-based clinical practice guidelines, practice resources, and evidence synopses to assist Kaiser Permanente (KP) physicians, administrators, and other health care professionals in determining the most effective medical practices. This core set of evidence-based resources will: Create Programwide economies of scale, Support ongoing performance improvement activities, Consistently provide high quality resources for use in care delivery tools and systems, and Increase KP regions abilities to leverage clinical guidelines to improve clinical outcomes. Clinical practice guidance, based on scientific evidence, is essential for providing high quality care and continuously improving on it. Such guidance needs to be integrated into the electronic medical record and other decision support tools to be accessible to clinicians at the point of care. In addition, engaging our members in collaborative, shared decision-making conversations regarding their personal preferences is an essential component of patient-centered quality care. Furthermore, cost-effectiveness of various evidence-based interventions and resource limitations are important considerations. This involves addressing health problems in ways that maximize the health of the population given the available resources. Who are the National Guideline Directors? The National Guideline Directors (NGD) are a group of experts and advocates of evidence-based medicine who provide direction and oversight to the National Guideline Program (NGP). In this role, the NGD selects and approves topics for evidence-based knowledge products, owns Kaiser Permanente s Common Methodology, and is responsible for quality assurance review. This group is composed of representatives from the Care Management Institute (CMI) and all eight regions. What Is the Guideline Quality Committee? The Guideline Quality (GQ) Committee is a subcommittee of the NGD consisting of a group of evidence experts from various KP regions and CMI who review and approve all the National Guidelines. This review ensures that the processes used to develop guideline content have adhered to KP evidence-based methods and that the labels applied to clinical recommendations therein are accurate (e.g., evidence-based or consensus-based ). 1 National Adult Depression Clincial Practice Guideline

6 How Are Guidelines Developed? Guidelines are developed with the use of an evidence-based methodology and involve a systematic literature search, critical appraisal of the research design and statistical results of relevant studies, and grading of the sufficiency (quantity, quality, consistency, and relevancy) of the evidence for drawing conclusions. An evidence search includes literature published in peerreviewed scientific journals, existing evidence-based guidelines, consensus-based statements from external professional societies and government health organizations, and clinical expert opinion of KP regional specialty groups. For additional information on evidence grading, see Table 1 in Appendix A. To develop a or revise a guideline, CMI consultants work with a multidisciplinary Guideline Development Team (GDT). Each GDT consists of a core group of physicians, representing primary care and the specialties most affected by the guideline topic, and, as appropriate, other content experts from disciplines such as pharmacy, nursing, and health education. The members of a GDT are nominated by the respective National Guideline Directors to represent their regions. The GDT reviews the appraisal of the evidence and develops or revises clinical recommendations based on the current evidence. Each regional representative then presents the draft guideline recommendations to key experts and champions in their regions for critical review and support to improve the likelihood of implementation once the guideline is published. How Often Are Guidelines Reviewed and Revised? To keep current with changing medical practices, all guidelines are reviewed, and, if appropriate, revised at least every two years. To develop the Adult Depression Guideline, released in February 2010, a multidisciplinary, interregional GDT first met in November 2009 to define the scope of the guideline. The Project Management Team then performed systematic reviews of the medical literature on each of the clinical questions identified by the GDT, assembled the evidence, and developed draft recommendations for review by the GDT. All of the recommendations and supporting evidence were reviewed in depth by the GDT in a series of meetings from November through January The GQ Committee reviewed and approved the guidelines in February All recommendations included in the guideline were approved by the NGD. What Does It Mean for a Guideline to Be Evidence-Based? Each clinical recommendation within a guideline is labeled as evidence-based or consensusbased. A recommendation is considered evidence-based if there has been a systematic review of the evidence, the evidence is sufficient, and the recommendation is consistent with the evidence. A recommendation can also be considered evidence-based if there is insufficient evidence but either no particular intervention is recommended or options are recommended without favoring one of the options over others. A recommendation is considered consensusbased if there has been a systematic review of the evidence, the evidence is insufficient to support an evidence-based recommendation, and the GDT decides to make a consensus recommendation. 2 National Adult Depression Clincial Practice Guideline

7 What Does It Mean for a Guideline to Be Approved and National? A recommendation that is consistent with the above policies is labeled as National Guideline Directors Approved. A recommendation that fails to satisfy those criteria is not approved and will be noted as such. A National Guideline Directors Approved guideline for which at least 90% of the recommendations are approved by at least six of the eight KP regions is a "National Guideline." On the topics for which they exist, National Guidelines are the preferred evidence source for KP HealthConnect content. Contact information: David Price, MD Adult Depression Clinical Lead Care Management Institute david.price@kp.org Devon McCabe, MA Care Management Consultant Care Management Institute devon.d.mccabe@kp.org Acknowledgments The Kaiser Permanente (KP) Adult Depression Clinical Practice Guideline is the result of the extensive clinical expertise, collaborative efforts, and outstanding personal contributions of the following participants: KP Adult Depression Guideline Project Management Team David Price, MD Clinical Lead Care Management Institute Devon McCabe, MA Project Manager Care Management Institute Christy N. Pham, MPH Lead Analyst KP-Southern California Erin G. Stone, MD, FACP EBM Methodologist Care Management Institute Tabitha Pousson Staff Assistant Care Management Institute 3 National Adult Depression Clincial Practice Guideline

8 KP Adult Depression Guideline Development Team There were no conflict of interests for any member of the Guideline Development Team (GDT). Colorado Jean E Milofsky MD Regional Department Chair, Psychiatry Kerri Gaughan, PharmD, BCPP Clinical Pharmacy Specialist, Mental Health David Price, MD Director of Medical Education, CPMG; CMI Depression Guideline & Education Lead; Medical Director, KP National CME Program Georgia Sam W. Moss, MD, MS Lead Physician, Adult Medicine; Assistant to the Chief of Medicine for Asthma/Depression Hawai i John Draeger, MD Chief, Behavioral Health Service Samuel V Gadam Geriatric psychiatrist Mid-Atlantic States Timothy M Sitts, MD Psychiatrist Northern California Mason Turner, MD Chief, Department of Psychiatry; Assistant Director, Regional Mental Health and Chemical Dependency Joyce O. Arango, DrPH Sr. Managerial Consultant, Northern California Guidelines Director John Guzman, PhD Subchief, SSF Behavioral Medicine, Regional Chair; Behavioral Medicine Subchiefs Gabrielle Beaubrun, MD Psychiatrist Assistant Chief of Psychiatry Steve Olson, MD Family Physician, Depression Champion; Co-Manager Behavior Medicine Services Northwest Jonathan Ebbing, MD Psychiatrist Ohio William S. Schwab, MD PhD AGSF Chief of Geriatrics Horia Craciun, MD Psychiatrist Larissa Elgudin, MD Regional Chief of Behavioral Health Services, Parma Program Offices Andrew Bertagnolli, PhD Senior Consultant, Behavioral Medicine & Pain Mgmt Care Management Institute Southern California Christy N. Pham, MPH Consultant, Technology Assessment & Guidelines Unit Erin G. Stone, MD, FACP Physician Lead, Clinical Content and Decision Support Debbie R Kubota, PharmD Pharmacist Evidence Analyst & Strategist Misha Askren, MD Family Medicine Mark Dreskin, MD Regional Depression Co-Lead; Physician in Charge; Same Day Acute Medical Services Kerri Gaughan, PharmD, BCPP Clinical Pharmacy Specialist, Mental Health 4 National Adult Depression Clincial Practice Guideline

9 Guideline Summary This guideline is informational only. It is not intended or designed as a substitute for the reasonable exercise of independent clinical judgment by practitioners, considering each patient s needs on an individual basis. Guideline recommendations apply to populations of patients. Clinical judgment is necessary to design treatment plans for individual patients. 1. First-Line Treatment of Major Depressive Disorder (MDD) 1A 1B 1C 1D 1E For patients with mild to moderate Major Depressive Disorder (MDD), use either antidepressant medication or psychotherapy * as first-line treatment. Evidence-based: B Given the lack of evidence on a clearly superior approach for mild to moderate MDD, base treatment decisions on patient and clinician preference, potential side effects, and cost. Consensus-based For patients with severe or chronic MDD, combine antidepressant use and psychotherapy * as first-line treatment. Evidence-based: B If antidepressants are used, any class of antidepressant (SSRI, TCA, SNRI, NRI, or DA) may be prescribed as first-line treatment of MDD. Evidence-based: B Given the equivalence of therapeutic effect, base the choice of antidepressant on patient s prior response, patient and clinician preference, potential side effects, and cost. Consensus-based * (Interpersonal Therapy, Cognitive Behavioral Therapy, or Problem-Solving Therapy) 5 National Adult Depression Clincial Practice Guideline

10 2. Hypericum (St. John s Wort) For Treatment of MDD 2A The GDT makes no recommendation for or against providing hypericum (St. John s wort) in patients with mild to moderate Major Depression. 2B There is fair evidence of effectiveness of hypericum in this population. However, due to lack of consistency of preparation oversight and dosage across trials, and concerns about lack of FDA oversight and consistency of hypericum preparations, the balance of benefits, harms, and costs compared with other treatments cannot be determined. Evidence-based: C * The GDT recommends against providing, hypericum (St. John s wort) to patients with severe Major Depression. Evidence-based 3. Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, or Plan 3A 3B For patients with Major Depression expressing suicidal intent or plan, the GDT recommends consultation with specialty behavioral health. Consensus-based For patients with suicidal ideation or who have made previous suicide attempts, the GDT recommends consultation or collaboration with a psychiatrist before prescribing TCAs or venlafaxine. Consensus-based * Please note that only recommendations approved since the adoption in 2006 of evidence grading will use letters (A, B, C, etc.) to specify the grade of the evidence. Recommendations approved prior to 2006 will not include a letter grade following the statement evidence-based. For additional information on evidence grading, see Appendix A. 6 National Adult Depression Clincial Practice Guideline

11 4. Second-Line Treatment of MDD 4A 4B 4C 4D For patients with MDD whose symptoms fail to remit after first-line treatment, the GDT recommends an assessment of the adherence to the initial treatment regimen. Consensus-based For patients with MDD whose symptoms fail to remit after adherence to first-line treatment, the GDT recommends that treatment options include: Combining antidepressants and psychotherapy. Evidence-based Increasing the dose of the initial antidepressant. Consensus-based Switching to a different antidepressant of the same or different class. Consensus-based Switching from psychotherapy to antidepressants or antidepressants to psychotherapy. Consensus-based Combined pharmacologic treatment (monitoring for toxicity, side effects and drug interactions) with SSRIs and: low-dose TCAs, or bupropion, or buspirone, or mirtazepine, or lithium, or liothyronine (T3). Consensus-based (all in this list) The GDT makes no recommendation for or against providing folate or inositol to patients whose MDD symptoms fail to remit after adhering to first-line treatment. Evidence-based: I The GDT makes no recommendation for or against providing atypical an antipsychotics to primary care patients with (nonpsychotic, nonbipolar) MDD whose symptoms fail to remit after adherence to first-line treatment. 4E There is fair evidence of short-term effectiveness for use of atypical antipsychotic agents to augment antidepressants in patients with nonpsychotic, nonbipolar MDD who fail to remit after initial treatment. However, due to lack of longer-term data, known cardiometabolic risks of treatment with these medications, and lack of comparison data against other strategies, the balance of benefits, harms and costs compared with other treatments cannot be determined. Evidence-based: I The GDT recommends against providing augmentation with pindolol for patients with MDD whose symptoms fail to remit after adherence to first-line treatment. Evidence-based 7 National Adult Depression Clincial Practice Guideline

12 5. Length of Treatment With Antidepressants In Patients With MDD Patients Who Achieve Symptom Remission 5A The GDT recommends that patients with MDD who achieve symptom remission with antidepressants should continue antidepressants at the same dose for at least an additional six to 12 months. Evidence-based Patients With One Lifetime Episode of MDD 5B Based on patient and provider preference, the GDT recommends that a trial of antidepressant discontinuation is optional for patients in their first lifetime episode of MDD, who are being treated with antidepressants, achieve remission, and remain asymptomatic for six to 12 months after acute phase treatment. Consensus-based Patients With Two or More Lifetime Episodes of MDD 5C The GDT recommends that patients with two or more lifetime episodes of MDD, who are being treated with antidepressants and remain asymptomatic after acute phase treatment, should be maintained on the medication and dose with which they achieved remission for at least an additional 15 months to five years after acute phase treatment. Consensus-based Patients With Chronic MDD or MDD With Concurrent Dysthymia 5D The GDT recommends that patients with chronic MDD (continual symptoms for more than two years) or Double Depression (MDD and dysthymia) who improve with antidepressants during acute phase treatment should continue antidepressants for at least an additional 15 to 28 months after acute phase treatment. Evidence-based 6. Follow-Up For Patients In The Acute Phase (First Three Months) of Treatment For MDD 6 For patients who are starting treatment with antidepressants for Major Depression, the GDT recommends that the minimum recommended follow-up frequency is one patient contact * within the first month, and at least one additional patient contact four to eight weeks after the first contact. Assess for adherence, side effects, suicidal ideation, and patient response during both these visits. Consensus-based * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. The use of between patients and providers is relatively new, and has not been a widely utilized means of communication to date. However, it is being increasingly advocated as part of a patient-centered, more efficient ( less visit dependent ) model of care. At least one member of the GDT uses this modality regularly and deems it effective for follow-up contacts. 8 National Adult Depression Clincial Practice Guideline

13 7. Follow-Up For Patients In The Continuation Phase (Months Four To 12) of Treatment For MDD 7 After achieving symptom remission, the GDT recommends at least one follow-up contact * during the fifth or sixth month of treatment in patients with Major Depression. Assess for continuing symptom remission and dosage/treatment adjustment during this contact. The GDT recommends additional patient follow-up to consider either continuing treatment beyond the continuation phase, or attempting a trial of treatment discontinuation. Consensus-based 8. Follow-Up For Patients In The Maintenance Phase (Beyond 12 Months) of Treatment For MDD 8A 8B For asymptomatic patients with Major Depression who are continuing on antidepressants beyond 12 months, the GDT recommends at least one annual follow-up contact * is recommended to assess for continuing symptom remission, the need for ongoing treatment, and dosage/treatment adjustment. Consensus-based The GDT recommends that additional follow-up for asymptomatic patients with Major Depression who are continuing on antidepressants beyond 12 months should be based on patient preference and response. Consensus-based 9. Discontinuation of Antidepressants In Patients With MDD 9A 9B Fluoxetine may be discontinued, without tapering, with a relatively low risk of adverse effects. Evidence-based The GDT recommends tapering other antidepressants (other SSRIs, TCAs, SNRIs, NRIs, and DAs) over a two to four week period. Consensus-based 10. Treatment Preferences For MDD In Different Ethnic Groups 10 Because patient preferences for treatment may vary based on their ethnicity and culture, the GDT recommends asking patients from different ethnic groups about treatment preference when discussing treatment options for MDD. Evidence-based * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. The use of between patients and providers is relatively new, and has not been a widely utilized means of communication to date. However, it is being increasingly advocated as part of a patient-centered, more efficient ( less visit dependent ) model of care. At least one member of the GDT uses this modality regularly and deems it effective for follow-up contacts. 9 National Adult Depression Clincial Practice Guideline

14 11. Patient Self-Management Strategies For Improving Symptoms of MDD Exercise 11A Exercise is an adjunctive strategy (in addition to antidepressants or psychotherapy) for treating MDD. Evidence-based :B Internet Resources 11B Selected internet-based patient self-help materials may be used as an optional adjunct strategy (in addition to antidepressants or psychotherapy) for treating MDD. Consensus-based Bibliotherapy 11C Selected bibliotherapy * may be used as an optional adjunct strategy (in addition to antidepressants or psychotherapy) for treating MDD. Consensus-based Befriending 11D Befriending is an optional adjunct to antidepressants or psychotherapy for treating MDD. Consensus-based Patient-Initiated Combined Phone/Computer Programs 11E There is insufficient evidence for or against using patient-initiated combined phone/computer programs in the treatment of MDD. Evidence-based: I Light Therapy 11F There is insufficient evidence for or against using light therapy as a primary or adjunctive treatment for nonseasonal forms of MDD. Evidence-based: I Music Therapy 11G There is insufficient evidence for or against using music therapy in the treatment of MDD. Evidence-based: I Life Review Therapy 11H There is insufficient evidence for or against using life review therapy in the treatment of MDD. Evidence-based: I 12. Behavioral Health Education Classes For Adults With MDD (Cognitive Behavioral Skills or Problem-Solving Classes) 12 For patients with mild to moderate MDD, the GDT recommends behavioral health education classes as an adjunctive treatment option. However, these classes should not be used in lieu of either antidepressant medication or psychotherapy. Evidence-based * Bibliotherapy: Advising people to read specific written material based on cognitive-behavioral approaches to depression treatment. Befriending: Consists of a designated befriender who meets the depressed person to talk and socialize with for at least one hour per week. 10 National Adult Depression Clincial Practice Guideline

15 13. Antidepressants To Avoid During Pregnancy or Breastfeeding Pregnancy 13A Do not start paroxetine in women who are pregnant. Evidence-based: D 13B Use caution in starting other selective serotonin re-uptake inhibitors (SSRIs) in women who are pregnant. Consensus-based Discuss risks to the mother and fetus of untreated maternal depression, as well as the risk of fetal adverse effects from antidepressants. 13C If drug therapy is a consideration for treatment of maternal MDD during pregnancy and/or breastfeeding, then: Individualize according to patient history and need for medication, and Discuss the benefits and harms of the various treatment options with the patient. Consensus-based 13D If MDD is in remission and a woman becomes pregnant while taking antidepressants during the continuation or maintenance phase of treatment, then: Discuss the risks to the mother and fetus of untreated maternal depression or depression relapse after antidepressant discontinuation, as well as the risk of fetal adverse effects from continuing antidepressants, and Monitor for first trimester fetal malformations if taking paroxetine. Consult OB/GYN for considerations on fetal malformation screening. Consensus-based Breastfeeding 13E Do not start fluoxetine and/or citalopram in breastfeeding women in most circumstances. If used, they should be used with caution, and only in patients who had good results with these medications during pregnancy or a previous depression episode. Consensus-based. 13F In women taking antidepressants during pregnancy whose depression is in remission and who desire to breastfeed: Discuss the risks to the mother and fetus of untreated maternal depression or depression relapse after antidepressant discontinuation and the risk of adverse effects in the nursing newborn of mothers continuing antidepressants, and Consider changing treatment for depression if the newborn shows potential antidepressant-related adverse effects (withdrawal symptoms) during the first few hours after birth. Consensus-based 11 National Adult Depression Clincial Practice Guideline

16 Rationale Statements 1. First-Line Treatment of Major Depressive Disorder (MDD) 1A 1B 1C 1D 1E For patients with mild to moderate Major Depressive Disorder (MDD), use either antidepressant medication or psychotherapy * as first-line treatment. Evidence-based: B Given the lack of evidence on a clearly superior approach for mild to moderate MDD, base treatment decisions on patient and clinician preference, potential side effects, and cost. Consensus-based For patients with severe or chronic MDD, combine antidepressant use and psychotherapy * as first-line treatment. Evidence-based: B If antidepressants are used, any class of antidepressant (SSRI, TCA, SNRI, NRI, or DA) may be prescribed as first-line treatment of MDD. Evidence-based: B Given the equivalence of therapeutic effect, base the choice of antidepressant on patient s prior response, patient and clinician preference, potential side effects, and cost. Consensus-based Evidence Grade Evidence for Recommendation 1A: Fair. Evidence for Recommendation 1C, D: Good. Rationale: 2010 Update New evidence was found, the recommendation remains unchanged. * (Interpersonal Therapy, Cognitive Behavioral Therapy, or Problem-Solving Therapy) The criteria for grading the strength of the evidence as either good, fair, or insufficient adheres to the KP National Guideline Program s Policies and Procedures documents entitled Label and Language of Recommendations and KP System for Grading the Strength of a Body of Evidence, which are located in Appendix A. 12 National Adult Depression Clincial Practice Guideline

17 Search Strategy Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. Casacalenda (1) study was indexed as a review article in PubMed and did not show up in the search results due to the way PubMed indexing is done. Another information source was the AHRQ Evidence report on the Treatment of Depression- Newer Pharmacotherapies. (2) See Appendix B for more information on the search strategy. The GDT determined that a limited review of the clinical question focusing on first-line medication treatment for MDD was most appropriate for the current 2010 update. Systematic reviews and meta-analyses were included in the current update. Original studies that demonstrated selective reporting using non-systematic searches or were industry-sponsored were excluded. Note: There is no universally used definition of mild, moderate, and severe Major Depression. To determine severity, clinicians can use either the depression score from a standardized, validated depression rating scale (such as the PHQ-9, Beck Depression Inventory, Zung Depression Scale, and others that are often used in Major Depression studies) or use the following commonly used clinical consensus rating system as a guide to determining symptom severity. Severity of Major Depression SYMPTOM SEVERITY NUMBER OF MDD SYMPTOMS (DSM-IV CRITERIA) PATIENT FAMILY, SOCIAL, OR OCCUPATIONAL IMPAIRMENT Mild 5 to 6 Minor Moderate 6 to 7 Moderate Severe 8 to 9 Severe; including suicidal intention or plan After reviewing this new evidence, the overall recommendation for first-line antidepressant treatment remains unchanged. The included studies are summarized below (please refer to Evidence Table 1.1 for study details). A limited review focusing on first-line medication choice for treatment of MDD was conducted. Systematic reviews and meta-analyses were included. Analyses using non-systematic searches and single antidepressant vs. placebo trials of established (previously reviewed) antidepressants were excluded. Eighteen relevant systematic reviews and meta-analyses were identified. Efficacy outcomes were measured by response rates, defined as a reduction of 50% of baseline Hamilton Depression Rating Scale (HAM-D) or the Montgomery-Asberg Depression Rating Scale (MADRS) or a score of much improved on the Clinical Global Impression score(cgi); and/or remission, defined frequently in studies by a preset cut-point of 7 to 9 on the HAM-D score. Tolerability or acceptability was measured by overall dropout rates, discontinuation rates due to adverse events, or rates of adverse events. 13 National Adult Depression Clincial Practice Guideline

18 Three studies (Mukai 2009, (3) Gartlehner 2008a, (4) Gartlehner 2008b (5) ) compared the efficacy and tolerability of different drug classes (e.g., TCA, SSRI, SNRI, other antidepressants) to each other; nine studies (Cipriani 2009b, (6) Cipriani 2009c, (7) Cipriani 2009a, (8) Nakagawa 2009, (9) Omori 2009, (10) Cipriani 2008, (11) Van den Broek 2009, (12) Weinmann 2008, (13) Watanabe 2008 (14) ) compared individual antidepressants to other antidepressants; and six studies (Arroll 2009, (15) Hansen 2008, (16) Barbui 2009, (17) Barbui 2008, (18) Deshauer 2008, (19) Nelson 2008 (20) ) evaluated antidepressants, individually or as a class, relative to placebo. The majority of the studies focused on the 12 newer, second-generation antidepressants that included bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, sertraline, trazodone and venlafaxine. Head-to-Head Comparisons of Antidepressants * Mukai et al. (2009) (21) reviewed 18 RCTs examining the efficacy of single- versus dualaction Antidepressants for the treatment of depression among patients 59 years. In this narrative review, authors conducted a systematic search of published studies; however, data were scarce and insufficient to conduct a meta-analysis of head-to-head comparisons. In addition, publication bias assessment was not reported. Overall, limited data suggest that dual-action agents such as TCAs and SNRIs do not appear to confer any additional efficacy benefits over single-action agents such as SSRIs in the treatment of depression in the elderly. Two trials found no significant difference in efficacy between TCAs vs. SSRIs; three studies found no significant difference between venlafaxine vs. SSRIs; one study of duloxetine vs. placebo (funding source not reported) found significant improvement in depression (p < 0.001), pain (p < 0.001), and cognition (p = 0.013) in favor of duloxetine; and five studies suggested no additional efficacy benefit for the SNRI venlafaxine compared with SSRIs or TCAs. Gartlehner et al. (2008a) (4) conducted a meta-analysis of 203 studies (including head-tohead RCTs; observational studies; placebo trials; systematic reviews, meta-analyses, and studies with pooled data) on the comparative benefits and harms of second-generation antidepressants. Overall, no substantial differences in comparative efficacy and effectiveness of second-generation antidepressants for treatment of MDD were detected; the evidence did not support the choice of one second-generation antidepressant over another. Nevertheless, authors noted some differences in adverse events. Some statistical differences in onset of action were noted, however these differences may not be clinically significant since most response rates were similar after four weeks of treatment and all seven studies that examined speed of response were funded by manufacturer. * The great majority of antidepressant studies included here were funded by pharmaceutical companies. In almost all cases, at least some results favored the drug manufactured by the funder. 14 National Adult Depression Clincial Practice Guideline

19 In a separate sub-analysis, Gartlehner et al. (2008b) (5) assessed the comparative harms of second-generation Antidepressants for MDD by reviewing 104 studies (83 head-to-head RCTs (N > 17,000) and 21 observational studies (N > )). Outcome measures were rates of adverse effects, specific adverse effect reported, and overall rate of adverse effect. Adverse effects profiles of second-generation antidepressants were similar; nausea, vomiting, diarrhea, dry mouth, sweating, headache, dizziness, sexual dysfunction and weight gain were commonly reported adverse effect. However, individual drugs differed in frequencies of specific adverse effects, which might be clinically relevant and influence the choice of treatment for individual patients (on a case-by-case basis). Roughly 63% of patients in efficacy trials experienced at least one adverse effect. No significant difference was detected between second-generation antidepressants and SSRIs, except for venlafaxine, which had higher discontinuation from adverse effects and higher rate of nausea and vomiting than SSRIs [(Relative Risk (RR) discontinuation from adverse effects: 1.42 (95% CI: 1.15 to 1.75); RR = from nausea and vomiting: RR = 1.53 (95% CI: 1.26 to 1.86); NNH = 9 (95% CI: 6 to 23)]. Compared with other second-generation antidepressants, paroxetine frequently was associated with higher rates of sexual dysfunction and bupropion with lower rates of sexual dysfunction; mirtazapine and paroxetine were associated with more weight gain (up to 3 kg, although not always specified); and sertraline was associated with higher rates of diarrhea. However, differences did not lead to significantly different discontinuation rates. Cipriani et al. (2009a) (8) evaluated the efficacy and acceptability of 12 second-generation antidepressants in a meta-analysis of 117 RCTs (N = 25,928). Only seven of these trials were in primary care settings. Included studies were of short duration. Fifteen unpublished studies from industry were included, it was not specified how many of these favored the funder s antidepressant (no funnel plots or assessment of publication bias was noted, and it was not clear how many of these studies were also included in the FDA trial database). Authors noted a small number included in each pair-wise comparison. Mirtazapine, escitalopram, venlafaxine, and sertraline were more efficacious treatments for depression response (remission was not assessed). Escitalopram, sertraline, bupropion, and citalopram were better tolerated than other antidepressants. Reboxetine, fluvoxamine, paroxetine, and duloxetine tended to have lower efficacy and tolerability than other antidepressants. Numbers needed to treat or harm were not calculated, making it difficult to assess the absolute differences in each analysis. Additionally, statistical significance (defined as p < 0.05 in this analysis) was not adjusted for multiple comparisons, so many differences could have been due to chance. 15 National Adult Depression Clincial Practice Guideline

20 Escitalopram vs. Other Antidepressants Cipriani et al. (2009b) (6) examined the efficacy and tolerability of escitalopram compared to other Antidepressants in 22 RCTs. Patients with medical comorbidity were excluded, limiting the potential applicability to primary care settings. In terms of efficacy, escitalopram was shown to be significantly more effective than citalopram in achieving acute response (OR = 0.67, 95% CI: 0.50 to 0.87) and remission (OR = 0.53, 95% CI: 0.30 to 0.93). In terms of tolerability, fewer patients allocated to escitalopram withdrew from trials due to any cause compared to duloxetine (OR = 0.62, 95% CI: 0.38 to 0.99). This analysis is limited by potential publication and sponsorship biases and potential selective reporting of results. Milnacipran vs. Other Antidepressants Nakagawa et al. (2009) (9) examined milnacipran versus other antidepressants in a metaanalysis of 16 RCTs (N = 2777). When compared to TCAs, patients taking milnacipran were associated with fewer dropouts due to adverse events (OR 0.55; 95% CI: 0.35 to 0.85) and there was a trend that suggested fewer adverse events of sleepiness/ drowsiness, dry mouth or constipation. No other significant differences in efficacy, acceptability and tolerability were detected when comparing milnacipran with other antidepressive agents. However, authors noted that there remained inadequate evidence to conclude whether milnacipran is superior, inferior or the same as other antidepressive agents in terms of efficacy, acceptability and tolerability in the acute phase treatment of major depression. Milnacipran is not currently available in the United States. Fluvoxamine vs. Other Antidepressants Omori et al. (2009) (10) reviewed the efficacy and tolerability of fluvoxamine and other antidepressants in 53 RCTs (N = 8,244 for efficacy; N = 6,440 for tolerability). In terms of efficacy, no significant differences were detected in response and remission rates between fluvoxamine and other antidepressants as a class (TCAs, heterocyclics, SNRIs, SSRIs) in early or end of acute phase of treatment. In addition, dropouts for any reason or for adverse effects were not significantly different between fluvoxamine and other antidepressants as a class or individually. Nausea or vomiting and weight loss or anorexia were more frequent with fluvoxamine than with TCAs and some other antidepressants (mianserin, milnacipran, and newer antidepressants); constipation and dry mouth were more common with TCAs than with fluvoxamine. The authors concluded that the initial selection of an antidepressant medication will and should largely be based on the anticipated side effect profile and patient s preference. 16 National Adult Depression Clincial Practice Guideline

21 Sertraline vs. Other Antidepressants Cipriani et al. (2009c) (22) assessed the efficacy and tolerability of sertraline compared to other antidepressants in 59 RCTs (N = 9,950). Evidence favoring sertraline over some individual antidepressants for the acute phase treatment of major depression was found, either in terms of treatment response (fluoxetine) or acceptability/tolerability (amitriptyline, imipramine, paroxetine, and mirtazapine). No statistically significant difference in remission between fluoxetine and sertraline was noted. However, some differences favoring other antidepressants over sertraline in terms of response (mirtazapine, amitriptyline) and acceptability (bupropion) were also noted. Sertraline was generally associated with a higher rate of participants experiencing diarrhea. Overall, the quality of included studies was low, not all pre-specified outcomes were reported in included studies, and outcomes of clear relevance to patients and clinicians were not reported in any of the included studies. No analysis of publication bias was conducted. Few studies reported remission rates, those that did were underpowered to detect clinical significance. Analysis of sertraline vs. other antidepressants as a class was not conducted. Cipriani et al. (2008) (11) evaluated sertraline compared to other antidepressants in a metaanalysis of 56 RCTs (N = 8,507 for efficacy; N = 8,387 for tolerability). There was substantial overlap of studies in this analysis and the analysis conducted in Cipriani et al. (2009a) (8), thus many of the same limitations (short duration of included trials, few trials conducted in primary care settings) apply. Trials including patients with medical disorders were excluded from this analysis, also limiting applicability to the primary care setting. FDA trial databases and similar European trial databases were searched for study inclusion in this analysis; most included studies were funded by the maker of sertraline. This analysis also did adjust the level of significance to p < 0.01 to account for multiple comparisons. Included studies used different dosing schedules, making it difficult to determine differences (or lack thereof) between equivalent effective doses of antidepressants. Sertraline was more efficacious than other SSRIs [RR = 0.88 (99% CI: 0.78 to 0.99), p = 0.009; NNT = 17], particularly fluoxetine [RR = 0.85 (99% CI: 0.74 to 0.98); p = 0.004; NNT = 12]. But sertraline s efficacy was not significantly different from TCAs [RR = 0.95 (99% CI: 0.83 to 1.09)] or any other antidepressants, For acceptability, no significant differences were detected between sertraline and TCAs [RR = 0.83 (99% CI: 0.66 to 1.04)], SSRI [RR = 0.9 (99% CI: 0.68 to 1.18)], or any other antidepressants. In the conclusions section, the authors site cardiovascular physicians belief and clinical care practices in treating depression in patients with cardiovascular disease as observational evidence supporting their conclusion of sertraline as a candidate for initial choice of antidepressant. Venlafaxine vs. Other Antidepressants Van den Broek et al. (2009) (12) compared the effectiveness of venlafaxine to TCAs in seven RCTs (N = 947). There were no significant differences for response [OR = 0.88 (95% CI: )] or withdrawal [OR = 1.21 (95% CI: )] of TCA vs. venlafaxine. However, authors noted that because of the heterogeneity of the odds ratios, one cannot conclude that they are of equal efficacy. 17 National Adult Depression Clincial Practice Guideline

22 Weinmann et al. (2008) (13) evaluated the efficacy and tolerability of venlafaxine compared with SSRI in 17 RCTs (N = 3,523 for response analysis; N = 3,142 for remission analysis). There were no statistically significant differences between venlafaxine and the SSRI group with respect to response [random-effect RR = 1.05, (95% CI: 1.00 to 1.10); NNT = 27] and remission [random-effect RR = 1.04, (95% CI: 0.96 to 1.13); NNT = 34]. Total rate of treatment discontinuation did not differ between venlafaxine and SSRIs (RR = 1.05, 95% CI: 0.93 to 1.2, NNH = 100), but there were significantly more dropouts due to adverse effects in the venlafaxine vs. SSRIs group [RR = 1.38, 95% CI: 1.08 to 1.77, NNH = 32]. Mirtazapine vs. Other Antidepressants Watanabe et al. (2008) (14) conducted a systematic review on 25 RCTs (N = 4,842) on the effectiveness and tolerability of mirtazapine compared with other antidepressants (TCAs, SSRIs, SNRIs, and other). Results were stratified according to treatment duration, including early phase (2 weeks), end of acute-phase (6 to 12 weeks), and end of continuation phase (4 to 6 months). Analysis of efficacy at early phase of treatment found no significant differences between mirtazapine and TCAs in response [RR = 0.9 (99% CI: 0.69 to 1.18)] and remission [RR = 0.87 (99% CI: 0.52 to1.47)]; mirtazpine was superior to SSRIs in both response [RR = 1.36 (99% CI: 1.13 to 1.64); NNT = 11] and remission [RR = 1.68 (99% CI: 1.2 to 2.36); NNT = 25], particularly, significantly better than paroxetine in response [RR = 2.02 (99% CI: 1.09 to 3.75); NNT = 8] and sertraline in remission [RR = 1.73 (99% CI: 1.01 to 2.98); NNT = 12]; and mirtazapine was significantly superior to SNRI in terms of response [RR = 1.77 (99% CI: 1.08 to 2.89); NNT = 6] but not in remission [RR = 2.21(99% CI: 0.93 to5.26)]. At the end of the acute phase (6 weeks), no statistical significant differences were detected, except for superior remission outcome in comparison of mirtazapine with paroxetine [RR = 1.34 (99% CI: 1.04 to 1.73); NNT = 10)]. At the end of the continuation phase (24 weeks), one study examined mirtazapine with paroxetine and no significant differences were detected. No significant differences in tolerability were identified between patients treated with mirtazapine and TCAs (RR = 0.87 (95% CI: 0.7 to 1.08)], SSRIs (RR = 1.07 (95% CI: 0.92 to 1.26)], SNRI (venlafaxine) [RR = 0.82 (95% CI: 0.58 to 1.16)], or other antidepressants (trazodone) [RR = 0.93 (95% CI: 0.58 to 1.5)]. Mirtazapine dropouts due to adverse effects were similar to SSRI [RR = 1.22 (95% CI: 0.87 to 1.73)], SNRI [RR = 0.59 (95% CI: 0.27 to 1.29)], and trazodone [RR = 0.66 (95% CI: 0.3 to 1.46)]. Subgroup analysis found mirtazapine had lower withdrawals due to adverse effects compared with sertraline [RR = 2.58 (95% CI: ); NNH = 11]. Based on findings authors concluded that although mirtazapine is highly likely to have better efficacy profile than paroxetine or venlafaxine in terms of early response, in view of similar efficacy of mirtazapine and other antidepressants, results suggest that clinicians should also focus on other practically or clinically relevant considerations such as differences in the side effect profiles, to tailor treatment to best fit an individual patient s needs. 18 National Adult Depression Clincial Practice Guideline

23 Antidepressants vs. Placebo * Arroll et al. (2009) (15) reviewed 14 RCTs examining the effectiveness of TCAs and/or SSRIs compared to placebo. Both TCAs and SSRIs were significantly more effective than placebo for depression reduction and symptoms [RR = 1.24 (95% CI: 1.11 to 1.38) and NNT = 7 to 16; and RR = 1.28 (95% CI: 1.15 to 1.43) and NNT = 7 to 8, respectively]. More adverse effects were associated with TCAs than with SSRIs [NNH = 4 to 30 vs. 20 to 90, respectively]. Hansen et al. (2008) (16) compared MDD relapse and recurrence rates during continuation and maintenance phase treatment with second-generation antidepressant compared with placebo in 27 RCTs. Results were stratified by duration (those <1 year were categorized as relapse prevention and trials 1 year were categorized as recurrence prevention). NNTs for relapse prevention over a (mean of eight months) and recurrence (mean of 16 months) were each five [RR = of relapse: 0.54 (95% CI: 0.46 to 0.62), RR = of recurrence: 0.56 (95% CI: 0.48 to 0.66)]. In addition, loss to follow-up because of adverse events was not significantly different between active treatment and placebo (RR = 1.42, 95% CI: 0.92 to 2.20). Nelson et al. (2008) (20) reviewed 10 RCTs (N = 2377) on the efficacy of second-generation antidepressants for depression in late-life (> 60 years) with respect to treatment response, remission, and tolerability. Those assigned to active drug treatment had significantly greater response [OR = 1.4 (95% CI: 1.24 to 1.57); NNT = 13] and remission [OR = 1.27 (95% CI: ); NNT = 20] compared with placebo. Response rates were higher in the longer trials compared to the shorter trials [10 to 12 weeks OR = 1.73 (95% CI: 1.42 to 2.09) vs. 6 to 8 weeks OR = 1.22 (95% CI: 1.05 to 1.42)]. However, compared with placebo, there were increased odds of overall medication discontinuation [OR = 1.22 (95% CI: 1.06 to 1.4); I2 = 48.2%] and discontinuation due to medication adverse effects [OR = 1.84 (95% CI: 1.51 to 2.24)]. Evidence did not suggest superiority of one class of medication over another. Authors concluded that for every 100 patients treated, eight would show a response and five a remission in excess of placebo and for every two patients who responded, one discontinued prematurely because of adverse effects. Deshauer et al. (2008) (23) pooled 6 RCTs (N = 1299) to evaluate SSRIs versus placebo for MDD. Treatment response at sic to eight months showed SSRIs to be superior to placebo [OR = 1.66 (95% CI: 1.12 to 2.48)], particularly in depressed patients without other comorbidities [OR = 2.13 (95% CI: 1.11 to 4.08)]. There were no statistically significant differences in remission [OR = 1.46 (95% CI: 0.92 to 2.32)] or drop-out rates [OR = 0.87 (95% CI: 0.67 to 1.14)] between SSRI and placebo. * The great majority of antidepressant studies included here were funded by pharmaceutical companies. In almost all cases, at least some results favored the drug manufactured by the funder. 19 National Adult Depression Clincial Practice Guideline

24 Paroxetine vs. Placebo Barbui et al. (2008) (18) compared paroxetine with placebo in a meta-analysis of 40 RCTs (N = 6391). The primary outcome was discontinuation and the secondary outcome was response. Significantly more patients assigned to receive paroxetine left the study because of side effects [random effect RR = 1.77 (95% CI: 1.44 to2.18); NNH = 17], reported any adverse effects [OR = 1.27, 95% CI: 0.88 to 1.83, NNH = 9], and experienced suicidal tendencies compared with patients given placebo OR = 2.55 (95% CI: 1.17 to 5.54); NNH = 142]. Patients who received paroxetine were more likely to experience an improvement in depressive symptoms compared to patients receiving placebo [random effect RR = 0.83 (99% CI: 0.77 to 0.90); NNT = 9]. Suicide Risk Antidepresants vs. Placebo Barbui et al. (2009) (17) conducted a meta-analysis of eight observational studies (> 200,000 patients) to assess the risk of suicide based on SSRI exposure. Overall, there was an increased risk of attempted or completed suicide among adolescents exposed to SSRIs [random-effect OR = 1.92 (95% CI: 1.51 to 2.44)]. In subgroup analysis, paroxetine and venlafaxine were associated with increased risk. However there was a decreased risk of attempted or completed suicide among adults and elderly individuals exposed to SSRIs ( 65 years) [OR = 0.57 (95% CI: 0.47 to 0.70); OR = 0.46 (95% CI: 0.27 to 0.79)]. No individual antidepressant was significantly associated with completed or attempted suicide in adults. Patients may have had additional comorbidities that may have influenced selection of treatment and have unaccounted for effects on suicide risk. Funnel plot suggested that small negative studies may not have been published and included in the analysis. Discussion: In aggregate, updated evidence suggests no significant differences in efficacy among the different classes of antidepressants (TCAs, SSRIs, SNRIs; N = 4 new studies (21), (4), (5), (24) ). Five newly identified meta-analyses/systematic reviews ( (20), (16), (17), (25), (23) ) support the proposition that antidepressant treatment for MDD is more efficacious than placebo, especially among patients without other comorbidities. (23) One study (Nelson 2008 (20) ) found that response to treatment is greater in longer trials (10 to 12 weeks) compared with shorter trials (6 to 8 weeks), supporting findings of the STAR*D trial (reviewed previously). Three studies ( (21), (4), (5) ) did not detect significant differences among antidepressant drug classes in discontinuation for any reason and/or from side effects. However, one study (Arroll 2009 (15) ) suggested that more adverse events were associated with TCAs than SSRIs and another study (Nelson 2008 (20) ) noted increased odds for discontinuation of medication treatment compared to placebo. One study (Barbui 2009 (17) ) reported a decreased risk of suicide attempts or completion in depressed adults treated with SSRIs, but supported previously found associations between antidepressant exposure and suicide risk in adolescents. Although significant differences in tolerability were not detected in several studies, there was some evidence that showed marginal differences in the adverse event profiles among individual antidpressants ( (4), (5), (13) ). 20 National Adult Depression Clincial Practice Guideline

25 A variety of results were reported among studies that assessed the effectiveness and tolerability of individual antidepressants relative to other antidepressants or placebo. In a head-to-head comparison study of 12 new second-generation antidepressants, Cipriani et al. (2009a) (8) reported that mirtazapine, escitalopram, venlafaxine and sertraline were among the most efficacious while reboxetine, fluvoxamine, and duloxetine were the least efficacious. One study (Watanabe 2008 (14) ) found that mirtazapine was equivalent to TCAs and more efficacious than SSRIs, (particularly paroxetine); no significant differences in tolerability were detected between mirtazapine and other antidepressants. Cipriani 2009b (6) showed statistically better response for escitalopram compared with citalopram and better escitalopram tolerability compared with duloxetine. Two studies demonstrated no significant differences in efficacy between venlafaxine and SSRI (13) or TCAs, (12) although one study found more dropouts due to adverse events in venlafaxine treated patients compared with SSRI patients (13). Cipriani (2009c (7) ) found better response for sertraline compared with fluoxetine, and a trend in favor of mirtazapine for efficacy and bupropion for tolerability over sertraline. Another Cipriani analysis (2008 (11) ) reported no differences in efficacy and tolerability between sertraline, TCAs, and other antidepressants while sertraline was superior to SSRI, especially fluoxetine, in efficacy. There were significant overlaps in the studies included in these two Ciprani analyses. One study (9) found better tolerability of milnacipran compared with TCAs; however, the authors noted that there was insufficient evidence to establish the overall effectiveness and acceptability of milnacipran compared to other antidepressants (milnacipran is not available in the United States). One study (10) reported no significant differences in tolerability between fluvoxamine and other classes of antidepressants. Finally, one study (18) concluded that paroxetine was superior to placebo in response, but more withdrawals due to adverse events were observed in patients allocated to paraoxetine than placebo. Altogether, the NNT for antidepressant treatment, either as a class or individually, ranged from 4 to 34 and the NNH ranged from 9 to 90. As a result, the benefits of antidepressant treatment for MDD appear to outweigh the identified harms. Several important limitations were present in most of the included systematic review and metaanalyses that warrant cautious interpretation of the reported results. The majority of included studies were funded by the pharmaceutical industry, which tends to favor the funded medication. Unpublished results were likely missing from many of the analyses. The same small group of researchers published many of the studies included, and in some cases, separate analyses were presented that included several common trials, making it appear that there is stronger evidence to support a conclusion than really exists. Only a minority of studies in some analyses was conducted in primary care settings, and several studies excluded patients with comorbidities, making generalization to primary care difficult. Data was unavailable for many head-to-head comparison studies of individual antidepressants, and selective reporting of results may have occurred in several of the included trials. Selective participation of symptomatic volunteers was observed in most included trials, which may potentially overstate the true effectiveness in real world settings. Most studies reported on symptom response rather than symptom remission. 21 National Adult Depression Clincial Practice Guideline

26 Many trials included multiple comparisons, increasing the risk of finding a statistically significant result that is truly a chance finding. Data on tolerability based on dropout rates due to adverse events were also difficult to interpret due to the variability among individual drug adverse effects profiles. Methodological concerns such as inadequate blinding and analytical methods (e.g., completer analyses and last observation carried forward methods of imputing missing data) may bias studies in favor of active treatment). Variability in dosing strategies can lead to higher effective doses of one medication being compared with lower effective doses of another. Although some studies found statistically significant results, whether these findings translate to significant clinical relevance is questionable. Conclusion: New evidence from systematic reviews of mostly lower quality studies did not suggest the need to change the previous guideline recommendation regarding antidepressants for treatment of severe MDD. Although individual studies suggest marginal benefit for specific antidepressants, in aggregate there is no consistent directionality amongst the studies. Furthermore, the absolute clinically important difference between antidepressants is uncertain, due to the limited quantity of the comparison studies. Therefore, the GDT continues to find that the balance of evidence supports the use of antidepressant medication as a first-line treatment for MDD. The GDT continues to recommend, consistent with the previous guideline recommendations, that any class of antidepressant (SSRI, TCA, SNRI, NRI, or DA) may be prescribed as first-line treatment of MDD. Similarly, the GDT maintains that clinicians base the choice of antidepressant on a patient s prior response, patient and clinician preference, potential side effects, and cost Guideline No new evidence was found, the recommendation remains unchanged. 22 National Adult Depression Clincial Practice Guideline

27 2006 Guideline New evidence was found that did not change the existing recommendation. Antidepressants Alone or vs. Placebo * Montgomery, et al. (2003) (26) reviewed four RCTs examining the effectiveness of reboxetine in the treatment of severe depression. In three of the four trials, the reboxetine group had a significantly greater decrease in Hamilton Depression Scale (HAMD) score when compared with placebo. In three of the four trials, antidepressant efficacy occurred significantly faster (at two weeks) in the reboxetine group when compared with the placebo group. In three of the four trials, the responder rate (50% improvement from baseline HAM-D score) was significantly higher in the reboxetine group (56 to 74%) compared with placebo (20 to 52%). Patients unresponsive to previous antidepressant treatment were excluded. Remission rates were not reported. Limitations of this analysis include: small number of studies, all studies included were funded by Pharmacia, the manufacturer of reboxetine, and a systematic review was not performed (unpublished studies may not have been included), the four studies were not similar in terms of duration or number of subjects, and one of the four studies had a higher baseline mean HAMD score than the rest (35.4 vs. 26.4, 27.0, and 27.2). One systematic review (27) found in the Cochrane Database found that TCAs, SSRIs, and MAOIs are all effective in treating older adult patients with depression and a severe medical illness. The results, using a fixed effects model, for the three groups respectively were: TCAs: OR = 0.32 (95% CI: 0.21 to 0.47); SSRIs: OR = 0.51 (95% CI: 0.36 to 0.72); MAOIs: OR = 0.17 (95% CI: 0.07 to 0.39). Another Cochrane review (28) analyzed the effectiveness of antidepressants compared with active placebos, based on growing concerns that differential rates of side effects between active antidepressant drugs and inactive (inert) placebos in RCTs may lead to unblinding of patients and clinicians to their treatment group and may potentially bias results in favor of active medication. All the studies in the review produced a pooled estimate of effect of 0.39 standard deviations (95% CI: 0.24 to 0.54) in favor of antidepressants. When one strongly positive trial was omitted, sensitivity analysis reduced the pooled effect size to 0.17 (95% CI: 0.00 to 0.34). The review concludes that differences between antidepressants and active placebos are small. In their analysis of five systematic reviews and seven subsequent RCTs, Clinical Evidence (Issue 14, January 2006) found a positive treatment effect of antidepressants over placebo for adults in primary and secondary care with MDD. They note, however, that there is relatively little information given about severe side effects when compared with placebo, and that publication bias has been found in RCTs of selective serotonin reuptake inhibitors. On average, 69% of people taking placebo had worse outcomes over a mean of six weeks than the average person taking antidepressant drugs (mean effect size = 0.5 for change in score with antidepressant drugs versus placebo). (29) * The great majority of antidepressant studies included here were funded by pharmaceutical companies. In almost all cases, at least some results favored the drug manufactured by the funder. 23 National Adult Depression Clincial Practice Guideline

28 Low-dose TCAs significantly increased the proportion of people who responded at four weeks and at three to 12 months (RR = 1.65, 95% CI: 1.36 to 2.00) compared with placebo (RR = 2.14, 95% CI: 1.41 to 3.26). (30) Newer antidepressants significantly increased the proportion of people who responded compared with placebo (51% vs. 31% respectively; RR = 1.6, 95% CI: 1.5 to 1.7). (31) Antidepressants significantly increased the proportion of people who responded to treatment over four to 12 weeks compared with placebo (RR = 1.9, 95% CI: 1.6 to 2.3). (32) Reboxetine significantly increased the proportion of people who responded at six weeks compared with placebo (74% vs. 20% respectively, p < 0.001). (33) Reboxetine at any dose significantly increased the proportion of people who responded over four weeks compared with placebo (60% vs. 35% respectively; p < 0.05). (34) Reboxetine significantly improved mean Social Adaptation Self-Evaluation Scale score at eight weeks compared with placebo (35.3 vs respectively, p < 0.05). (35) Escitalopram significantly improved depressive symptoms compared with placebo at eight weeks (change in MARDS score: -15 with escitalopram, -12 with placebo, p = 0.002). (36) Escitalopram 10 mg daily, escitalopram 20 mg daily, and citalopram 40 mg daily all significantly improved depressive symptoms compared with placebo at eight weeks (change in MADRS score from baseline: -12.8, -13.9, -12.0, -9.4 respectively; p 0.05 for all treatments versus placebo). (37) Duloxetine significantly improved depressive symptoms compared with placebo (change in HAM-D score from baseline: vs respectively, p < 0.001). (38) In a meta-analysis (nonindustry funded) examining the efficacy and tolerability of tricyclic antidepressants and SSRIs compared with placebo in primary care, Arroll, et al. (2005) (39) extracted pooled data showing that both classes of antidepressant were significantly more effective than placebo. Most studies were of short duration (less than eight weeks) and of lower methodologic quality. For TCAs, the pooled NNT for improvement was about four; the NNH for withdrawal due to side effects was 5 to 11. Low doses of tricyclic antidepressants (75 to 100 mg of amitriptyline or equivalent) were as effective as higher doses. For SSRIs, the NNT for improvement was six, the NNH for withdrawal due to side effects ranged from 21 to 94. This meta-analysis included many more studies comparing TCAs with placebo than those comparing SSRIs with placebo, and patients severity of depression was heterogeneous across studies. The authors conclude that prescribing antidepressants in primary care is a more effective clinical strategy than prescribing placebo. 24 National Adult Depression Clincial Practice Guideline

29 Head-to-Head Comparisons of Antidepressants* At the end of an eight week randomized trial with no placebo control subjects, Moore, et al. (2005) (40) found that patients treated with escitalopram had significantly greater adjusted (2.1 points, p < 0.05) decrease in MADRS (Montgomery-Asberg Depression Rating Scale) scores from baseline and higher responder (76.1% vs. 61.3%, p < 0.01) and remitter rates (56.1% vs. 46.3%, p < 0.05) compared with patients treated with citalopram. Adverse effects and treatment withdrawals were not significantly different between groups. Both the difference in mean score changes from baseline, and the calculated difference between mean MADRS scores at endpoint (13.9 vs points) are of marginal clinical significance. In a small Japanese study examining the efficacy and safety of fluvoxamine and nortriptyline, Otsubo, et al. (2005) (41) found no significant differences in total HAMD score change (-9 vs. -12, p = 0.14), CGI (Clinician rating of Global Improvement) scores (figures not reported, p = 0.055), responder rates (55.6% vs. 57.9% on HAMD, 52.8% vs. 44.7% on CGI) or HAMD remission rates (38.9% vs. 26.3%). The authors state that the incidence of adverse events was generally higher in nortriptyline-treated patients, but no statistical analysis of significance was reported. In a six week study with small sample sizes (n = 41 patients), Wehmeier, et al. (2005) (42) found that geriatric patients taking fluoxetine had the same HAMD treatment response (57.1% vs. 60.0%) and improvement in HAMD scores (11.9 vs points) as those taking trimipramine. The fluoxetine group reported fewer adverse events than the trimipramine group (54.5% vs. 66.7%) but no statistical analysis of significance was performed. The authors conclude that the two drugs are similar in terms of effectiveness and tolerability. In a two-phase study (eight week acute phase and six month continuation phase) Detke, et al. (2004) (43) tested the effectiveness of duloxetine (both 80 mg and 120 mg doses) and paroxetine against a placebo control. In the acute phase, the authors found both doses of duloxetine and paroxetine to be statistically significantly different from placebo on all of the following outcomes: HAMD score reduction (range -11 to points), response rates (70 to 82%), and remission rates (47 to 58%). In the continuation phase, of all the treatment groups, the higher dose of duloxetine had the longest time to loss of response when compared with placebo (p = 0.023). The adverse events ranged from 4 to 11.4%, statistical significance was not reported). Despite the implication of superiority of duloxetine, it should be noted that there was no statistical analysis comparing duloxetine with paroxetine, and duloxetine was used at doses higher than currently recommended by the manufacturer. Sechter, et al. (2004) (44) compared milnacipran and paroxetine and found no significant differences between the two drugs on any of the following measures: HAMD score reduction from baseline (-11.8 vs. -12 points), MADRS score reduction (-16.2 vs points), response rate (58.1 vs. 60.3%), and remission rate (33.1% vs. 35.1%). The paroxetine group had significantly greater post-treatment discontinuation symptoms (31.8% vs. 13.0%) (p = 0.032). The study was funded by the manufacturer of milnacipran. * The great majority of antidepressant studies included here were funded by pharmaceutical companies. In almost all cases, at least some results favored the drug manufactured by the funder. 25 National Adult Depression Clincial Practice Guideline

30 In a multicenter, 22 week study of 151 geriatric outpatients, Allard, et al. (2004) (45) examined the efficacy and tolerability of venlafaxine compared with citalopram, and found that both treatment groups showed comparable improvements in MADRS score (-18 vs points) over time. The incidence of spontaneously reported side effects/adverse effects was higher in the venlafaxine group than in the citalopram group (62% vs. 43% respectively), but no statistical analysis was performed to evaluate this difference. Treatment discontinuation due to side effects was rare (less than 10% in each group). In an eight week trial, Montgomery, Huusom, & Bothmer (2004) (46) compared escitalopram and venlafaxine and found no significant differences between groups on MADRS (-20.7 vs points) or HAMD (-14.4 vs. -14 points) scores. The escitalopram group achieved response and remission significantly faster than the venlafaxine group (4.6 days p < 0.05 and 6.6 days p < respectively). These authors found greater incidences of side effects in the venlafaxine group, specifically nausea, constipation and sweating (p < 0.05), but overall treatment discontinuation rates were similar (14.4% vs. 13.3%). Rapaport (2003) (47) studied the efficacy of paroxetine controlled release (CR) compared with paroxetine immediate release (IR) and placebo. A statistically significant adjusted difference compared with placebo was found for both active treatments in HAMD total score change (-12.1 and points vs points), CGI response rate (43% and 44% vs. 26%) and HAMD remission rate (72% and 65% vs. 52%) at 12 weeks. Post-hoc analysis noted that patients with chronic depression (duration of more than two years) responded as well as those with short-term depression. Statistical comparisons between active treatments were not reported. Withdrawal due to adverse events was highest in the paroxetine IR group, (16% vs. 12.5% vs. 8.3% for placebo) but statistical significance was not evaluated. Two authors were either employees or major stockholders of the company funding the study. Sauer, Uppertz-Helmhold & Dierkes (2003) (48) found noninferiority of venlafaxine ER compared with amitriptyline ER, based on equivalent changes in HAMD total scores (-10.5 vs points). Adverse effects (70.9% vs. 81.8%, p = NS) and adverse drug reactions (55.7% vs. 71.4%, p = 0.04) were high in both groups. In a 2003 study, Wade, et al. (49) found no significant difference between mirtazapine and paroxetine in HAMD score changes (-18.2 vs points), response (87% vs. 78%, p = 0.28) and remission rates (61% vs. 42%, p = 0.075) at 24 weeks. However, there were statistically significant two point differences in HAMD scores in favor of mirtazapine at two and four weeks (p = 0.012, and p = respectively). Overall adverse effects were high and similar in each group 79% vs. 85%); the mirtazapine group had higher incidences of fatigue while the paroxetine group had significantly more sweating, headache and nausea. More than fifty percent of patients in each group discontinued treatment. In a comparison of venlafaxine and nortriptyline in older adult patients, Gasto, et al. (2003) (50) found no significant differences in remission rates (approximately 70% overall for each drug), regardless of the severity of the depressive episode. Both drugs were similarly well tolerated; side effects were frequent (73.5% vs. 82.3%) but generally mild. Venlafaxine scored lower on the autonomic side effects subscale. 26 National Adult Depression Clincial Practice Guideline

31 Hansen, et al. (2005) (51) conducted a systematic review of RCTs, meta-analyses, and observational studies examining the efficacy and safety of second-generation antidepressants, and found similar outcomes among different SSRIs, and comparing SSRIs with other second generation antidepressants on all of the following outcome measures: efficacy, speed of response, quality of life, and tolerability/adverse effects. The authors concluded that overall, second generation antidepressants probably to not differ substantially for treatment of Major Depressive Disorder. The majority of systematic reviews and RCTs reported in Clinical Evidence (Issue 14, January 2006) found no significant differences in outcomes among different classes of antidepressants (tricyclic antidepressants, selective serotonin reuptake inhibitors, or monoamine oxidase inhibitors). The two most relevant reviews are reported here. One review suggested that TCAs were more effective than MAO inhibitors, but may be less effective in depression with atypical biological features (increased sleep, increased appetite). Specific data were not cited. (52) Another review cited in Clinical Evidence found that about twice as many people taking TCAs compared with SSRIs had dry mouth, constipation, and dizziness; but that slightly more people taking SSRIs had nausea, diarrhea, anxiety, agitation, insomnia, nervous-ness, and headache. It concluded differences in overall side effects were small, but didn t quantify the magnitude of difference. (53) One Cochrane systematic review (54) found that SSRIs show only a modest advantage in tolerability over TCAs. Specifically, in 136 trials, participants taking SSRIs had significantly lower drop-out rates than those taking tricyclic/heterocyclic antidepressants (OR = 1.21, 95% CI: 1.12 to 1.30). The authors note that this difference from short duration, RCTs may not have clinical significance over the longer term. Another Cochrane review (55) found no clinically significant differences in effectiveness between SSRIs and TCAs: the standardized effect size was (95% CI: to 0.09). Given the apparent equal efficacy, the authors advise that treatment decisions should be based on patient acceptability, toxicity, and cost. Another Cochrane review (56) found amitriptyline as effective as other TCAs or newer compounds. Compared with SSRIs, however, amitriptyline was not as well-tolerated: OR = 0.84 (95% CI: 0.75 to 0.95). The evidence-based NICE (National Institute for Clinical Excellence) (57) guideline found strong evidence showing a statistically significant difference favoring SSRIs over TCAs on reducing the likelihood of patients leaving treatment early due to side effects (RR = 0.69, 95% CI: 0.62 to 0.67). However, other reviews, such as the Cochrane review cited above, show that all cause discontinuation rates do not vary significantly among antidepressant classes. 27 National Adult Depression Clincial Practice Guideline

32 Psychotherapy vs. Antidepressants In a study comparing cognitive therapy with antidepressants for treatment of moderate to severe depression, DeRubeis, et al. (2005) (58) found both active treatments were significantly more effective than placebo at eight weeks in improving depression symptoms (50% antidepressants, 43% cognitive therapy, 25% placebo), with no significant difference between active treatments. At 16 weeks the response rate was still similar for both active treatments (58%), although the remitter rate was significantly different, in favor of anti-depressant medication (46% vs. 40%, p = 0.04). These results should be interpreted with caution, since the antidepressant group had twice as many subjects as the cognitive therapy group (120 vs. 60 respectively) and a significant interaction effect was noted for site of intervention (p = 0.02), suggesting that therapists experience levels were not equal between the two sites used in the study. Combination Therapy In a complex study with multiple outcome measures and several limitations de Jonghe (2004) (59) found that patients with mild to moderate MDD initially treated with psychotherapy alone and patients treated with combination psychotherapy and antidepressants each experienced improvement over six months (primary outcome change in HAMD score: points in psychotherapy group vs points in combined therapy group, p = 0.083). Combination therapy did not lead to significantly greater improvements in depressive symptoms in the intent to treat sample compared with psychotherapy alone. In the per-protocol sample, there were minor differences favoring combined therapy on some measures, but not others. The authors conclude that the benefits of combination therapy in this population are equivocal. In a study of specialty mental health outpatients, Kool, et al. (2003) (60) found that antidepressant treatment (initially fluoxetine, changed to amitriptyline if initially intolerant, then moclobemide [a MAOI] if still intolerant) combined with 16 sessions of short psychodynamic supportive psychotherapy (SPSP), starting two weeks after the initiation of medication, was more effective than pharmacotherapy alone for depressed patients with personality disorders (change in HAMD scores vs , p = 0.04), but not in depressed patients without personality disorders (change in HAMD scores vs. -9.1, p = 0.74). The generalizability of this study to primary care is uncertain; most patients with depression and personality disorders are referred to specialty mental health for treatment. A search of Clinical Evidence (Issue 14, January 2006) yielded two systematic reviews and two subsequent RCTs: The first review found that combination therapy significantly improved depressive symptoms compared with drug treatment alone (OR = 1.86, 95% CI: 1.38 to 2.52). A greater effect was found in longer duration (12-week) studies with shorter treatment times (OR = 2.20, 95% CI: 1.22 to 4.03). (61) The second review found a small yet significant effect of combination pharmacotherapy and psychotherapy over either intervention alone (effect size: Cohen s d = 0.34 with BDI and 0.18 with HRSD; further details not reported). (62) 28 National Adult Depression Clincial Practice Guideline

33 Clinical Evidence also cited the studies by de Jonghe (2004) (59) and Kool (2003) (60) which were reviewed by the GDT and discussed above. Another RCT (63) of 102 older patients (older than 60 years) with MDD compared desipramine plus cognitive behavioral therapy, desipramine alone, and/or cognitive behavioral therapy alone. All three groups showed a significant reduction in symptoms from baseline as assessed using the HAM-D after 16 to 20 weeks of treatment. It found that combined treatment significantly improved symptoms compared with desipramine alone (p < 0.05), but there was no significant difference between combined treatment and cognitive behavioral therapy alone. Other Interventions Proudfoot, et al. (2004) (64) found that, compared with usual care (with or without antidepressants), a computerized CBT intervention yielded a significantly greater improvement in depressive symptoms, assessed primarily by BDI scores at three months (6.5 point difference favoring intervention group); differences continued to be present at eight months (5.8 point difference favoring the intervention group). Slightly over half of each patient group received pharmacotherapy. It is unclear how many patients in the usual care group received pharmacotherapy if many usual care patients received no treatment, it would bias results in favor of the intervention group. Both authors have commercial interests in the computerized CBT intervention. In a poorly reported study, Sirey, Bruce, and Alexopoulous (2005) (65) examined the effectiveness of The Treatment Initiation Program (TIP), an individualized, early intervention to address older adults attitudes about depression and barriers to care. The control group was given pharmacotherapy as usual; the intervention group was given both pharmacotherapy and TIP. The intervention group showed greater (71% vs. 42%) remission (defined with a low threshold as HAMD scores < 10) after 24 weeks. The intervention group had significantly lower HAMD scores at baseline, suggesting poor randomization; and overall changes in HAMD scores were not reported (lower pretreatment scores in the intervention group might bias remission results in favor of the intervention group, when in fact both groups could have equal changes in HAMD scores). The authors claim, but do not quantify, that patients in the intervention group were more likely to remain in treatment. Discussion: There has been increasing discussion about the true efficacy of antidepressants compared with placebo. Concerns have been raised about publication bias, especially since analyses show that published studies funded by the pharmaceutical industry overwhelmingly tend to favor the funder s medication. Some studies have found statistically significant differences on some outcomes (differences between changes in scores on depression rating scales) whose clinical relevance is debatable. The methodological quality of studies, especially as it relates to blinding and analytic methods (completer analyses and last observation carried forward methods of imputing missing data may bias studies in favor of active treatment). Interestingly, a recent analysis (66) notes that over the past 20 years, overall antidepressant response and placebo responses have both increased This suggests that study design factors and other external factors (patient awareness, direct to consumer and other antidepressant marketing) might be influencing trial results. 29 National Adult Depression Clincial Practice Guideline

34 Almost all clinical trials include some degree of attention for patients receiving placebo, and patients who enroll in clinical trials may be more responsive to placebo than patients at large; therefore, the placebo response in trials may be higher than the placebo response that would be found in the real world, that is, uncontrolled clinical conditions with a wider range of patients. While trials may overstate the magnitude of benefit of antidepressants, the NNT for antidepressant treatment still would likely remain more favorable than the NNH in most analyses. Considering all this information, the GDT concludes that the balance of evidence supports the use of antidepressant medication for treatment of MDD, but also underscores the need (given studies showing it s effectiveness) to consider structured psychotherapeutic treatment options (Interpersonal Therapy, Cognitive Behavioral Therapy, or Problem-Solving Therapy) and the use of shared decision-making with patients. New evidence did not surface to suggest the need to change the previous guideline recommendation regarding combined antidepressant and psychotherapy for treatment of severe MDD. Although this combination may provide marginal benefit in patients with milder forms of depression, the benefit is small in light of the additional resources required, so the GDT recommends, consistent with the previous guideline recommendation, limiting combination therapy in this group of patients to patients who fail to respond to initial treatment, which by definition includes patients with chronic depression (see the recommendations for second-line treatment). Patients with personality disorders and depression are usually treated in specialty behavioral health settings. After review of the new evidence summarized here, the GDT believes that the previous recommendations are still valid Guideline Supporting Evidence for Antidepressants versus Placebo Since the previous revision of the Depression Guidelines, three RCTs and one systematic review were found which looked at the efficacy of antidepressants versus placebo. Detke (38) included adult patients with MDD that had a baseline score of 15 on the 17-item Hamilton Rating Scale for Depression (HAM-D-17). The study was a randomized, doubleblind, placebo controlled trial. The study compared a SNRI (duloxetine), 60 mg /day (n = 121) and inert placebo (n = 115) for efficacy. This study also compared duloxetine (n = 123) and inert placebo (n = 122) for safety analysis. The duration of study was nine weeks. Duloxetine was significantly superior to placebo (4.8 point difference on the HAM- D-17 scale (p < 0.001), NNT to achieve one additional remission = 4). Significant side effects associated with duloxetine versus placebo included: nausea (p < 0.001), dry mouth (p < 0.001), somnolence (p < 0.001). Other statistically significant side effects of duloxetine were dizziness (p = 0.010), diarrhea (p = 0.006), insomnia (p = 0.021), and constipation (p = 0.001). 30 National Adult Depression Clincial Practice Guideline

35 Goldstein (67) included adult patients with MDD that had a baseline score of at least 15 on the HAM-D-17 scale for depression. The study was a randomized, double-blind, placebo controlled trial comparing a SNRI (duloxetine), 40 mg/day to 120 mg/day (n = 70) and inert placebo (n = 70) for a period of eight weeks. An SSRI (fluoxetine), 20 mg/day (n = 33) was included as an internal control to be sure patients were antidepressant responsive. Duloxetine was statistically superior to placebo in reducing the symptoms of MDD, with a 3.1 point difference between treatments on the HAM-D-17 scale (p = 0.009), and a NNT = 4 to achieve one additional remission compared with placebo. Insomnia was significantly more common with SNRI treated patients (p = 0.046, NNH = 13 to 14). There was a mean increase in standing diastolic blood pressure for SNRI treated patients of 2.8 mm Hg (p = 0.041). The study also showed a small but statistically significant (p = 0.005) reduction in body weight (0.59 kg) in SNRI treated patients. The clinical significance of the latter two findings is questionable. Wade (68) included adult patients between ages of 18 and 65. There was a 3:1 ratio of women to men, with more than 97% being Caucasian. All patients met the DSM-IV criteria for MDD and had a baseline Montgomery-Asberg Depression Rating Scale (MADRS) total score between 22 and 40. The study was a randomized, double-blind, placebo controlled trial comparing a SSRI (escitalopram) (n = 191) and inert placebo (n = 189) for a period of eight weeks. Escitalopram (10 mg/day) was statistically superior to placebo, with a 2.7 point greater adjusted mean change in MADRS total score from baseline to week eight (p = 0.002) (NNT not given). Headache and nausea were the most frequent adverse events, but nausea was the only adverse effect that was statistically significant in SSRI treated patients (p 0.05, NNH = 20). Moncrieff (69) reviewed six randomized and quasi-randomized, double-blinded studies of men and women of all age groups with a primary diagnosis of depressive disorder. These studies compared TCA drugs and an active placebo containing atropine. The duration of studies varied from three to 12 weeks. Change in mood at the end of treatment was the outcome of interest in these studies. The majority of trials found only small differences (p not stated) between TCA antidepressants and active placebos. The Cochrane systematic review, Wilson, (27) included 17 trials that assessed the efficacy of different antidepressants in older adult patients (over 60 years) with MDD. All trials compared SSRIs, TCAs, or MAOIs (monoamine oxidase inhibitors) with placebo, no headto-head active drug comparisons were reported. The authors concluded that TCAs, SSRIs, and MAOIs were effective in the treatment of older community patients and they recommended that at least six weeks of antidepressant treatment to achieve optimal therapeutic response. 31 National Adult Depression Clincial Practice Guideline

36 Clinical Evidence, Vol. 9, was also searched and our recommendations are in line with their conclusions. They note that systematic reviews in people aged 16 years or greater have found that antidepressant drugs (monoamine oxidase inhibitors, selective serotonin reuptake inhibitors, or tricyclic antidepressants) versus placebo improve symptoms in acute treatment of all grades of depressive disorder. One systematic review in people aged 55 years or greater with all grades of depressive disorder has found that tricyclic antidepressants, selective serotonin reuptake inhibitors, or monoamine oxidase inhibitors versus placebo significantly reduce the proportion of people who fail to recover over 26 to 49 days. They found no specific evidence on adverse effects in older adults. (70) Overall Conclusion: Since the previous revision of these guidelines, three RCTs and one meta-analysis were found that supported the use of antidepressants compared with inert placebo. One systematic review found only a small difference between treatment and active placebo. Although antidepressants have side effects, they are superior to placebo in treating MDD. Supporting Evidence for Psychotherapy vs. Placebo Volume 9 of Clinical Evidence, (70) most recent search date November 2003, was searched and one systematic review was found showing benefit of cognitive therapy vs. placebo in younger and older adults with mild to moderate depression. RCTs in younger and older adults with mild to moderate depression found that problemsolving treatment or interpersonal psychotherapy versus placebo significantly improved depressive symptoms in the short term. One systematic review in people aged 55 years with mild to moderate depression found five of six comparisons of cognitive or cognitivebehavioral therapy vs. no treatment lead to significant mean changes of -7.3 points on the HAM-D rating scale (95% CI: to -4.4); however, there were no significant differences found between active treatments and nonspecific attention control groups. This review was based on a small number of studies, the populations varied (although most were community samples), and many of the studies were short term. RCTs found limited evidence about the effects of psychological treatments in severe depression. This evidence of the efficacy of psychotherapy versus placebo is summarized in Table 1.26, Effects of Specific Psychological Treatments for Depressive Disorders, in Appendix B. No additional systematic reviews or RCTs since the Clinical Evidence review were found that addressed this question. Other Considerations: There is insufficient evidence across trials to determine the optimal number of sessions for each type of psychotherapy, and some concern that rigorous study protocols may not be delivered in real world settings, potentially affecting the achieved effectiveness of psychotherapy in different settings. Overall Conclusion: The evidence supports the efficacy of structured forms of psychotherapy (cognitive, cognitivebehavioral, interpersonal, or problem-solving therapy) delivered by behavioral health professionals as a first-line treatment option for patients with Major Depression. 32 National Adult Depression Clincial Practice Guideline

37 Supporting Evidence for Antidepressants versus Psychotherapy or Combination of Antidepressants and Psychotherapy One systematic review that addressed this issue was found. Six multiple-cell, randomized, controlled, double-blind trials were included in the Casacalenda (1) systematic review. The studies included adults with nonpsychotic Major Depression. These studies compared medication (five studies used tricyclics and one used phenelzine, a MAOI) (n = 261) and psychotherapy (three studies used cognitive behavior therapy, three used interpersonal therapy, one used problem-solving) (n = 352) and placebo (n = 270). Intention-to-treat analyses indicated that pharmacotherapy and psychotherapy were significantly more efficacious than control conditions (p < ) but were not significantly different from each other when treating mildly to moderately depressed patients. Structured forms of psychotherapy have established efficacy in treating the symptoms of Major Depression. Mynors-Wallis (1, 71) compared the following: Problem-solving therapy delivered by a general practitioner Problem-solving therapy delivered by a nurse Antidepressant medication (either fluvoxamine or paroxetine) Combination of psychotherapy and medication. Patients receiving psychotherapy met with the doctor or nurse for one hour on the first visit and for 30 minutes on subsequent visits. All patients received 12 weeks of treatment and at 52 weeks had follow-up interviews. A total of 151 patients participated in the study. All patients had "probable" or "definite" acute Major Depression according to Diagnostic Research Criteria. Patients receiving psychotherapy had lower completion rates (completed all 52 weeks of treatment) at the end of the study (64% and 68% respectively) as compared with medication alone or combination therapy (83% and 86% respectively). At 52 weeks, there were no statistically significant differences between patients who received problem-solving therapy, medication, or combination therapy (as determined by a 50% reduction in depression symptom score on the Hamilton Depression Rating Scale and the Beck Depression Inventory). Therefore, the authors concluded that problem-solving therapy and medication were effective treatments for Major Depression. However, patients receiving problem-solving therapy had higher dropout rates than those in the medication or combination group. The Mynors-Wallis study showed that patients assigned to medication experienced higher rates of symptom resolution at 12 weeks (67% versus 54%, ARR = 13, NNT = 8), therefore psychotherapy had slower onset of therapeutic effect. 33 National Adult Depression Clincial Practice Guideline

38 Schulberg (72) compared usual care, nortriptyline, and Interpersonal Therapy (IPT) in 276 primary care patients. Patients were aged 18 to 64 years and had to meet DSM-III-R (73) criteria for MDD. Patients were followed for eight months. The authors concluded that the severity of depressive symptoms was reduced more rapidly and more effectively among patients randomized to pharmacotherapy or psychotherapy than among patients assigned to a physician's usual care. Among treatment completers, approximately 70% of patients participating in the full pharmacotherapy or psychotherapy protocol achieved resolution of symptoms but only 20% of usual care patients were judged as recovered at eight months. However, this intervention is very resource intensive. Clinical Evidence (70) was also searched and our recommendations are in line with their conclusions. The following are their findings: One systematic review in people aged over 18 years with recent onset psychological problems, including depression, found that brief, nondirective counseling versus usual care by a physician significantly reduced symptom scores in the short term (less than six months), but found no significant difference in scores in the long term (more than six months). They found one nonsystematic review showing that the combination of psychotherapy and antidepressant medication in patients age 18 to 80 with severe MDD is more beneficial than either treatment alone. One RCT was also found showing that combination antidepressant treatment is more effective than either treatment alone for patients with mild to moderate Major Depression. Overall Conclusion: New evidence since the previous Depression Guideline revision shows that combining antidepressant medication and psychotherapy may be more beneficial than either treatment alone for patients with severe, and perhaps even mild to moderate, Major Depression. However, many patients with mild to moderate depression will respond to monotherapy (either antidepressants or psycho-therapy used alone), and extra resources are required to provide combined treatment to all depressed patients. Therefore, the GDT recommends that combining antidepressants and psychotherapy be reserved as an option for those patients with mild to moderate depression who do not initially respond to either treatment alone. However, for patients with severe MDD, who are most adversely affected, most significantly impaired, and are at higher risk for suicide, the added cost of initially starting with both antidepressant medication and psychotherapy in combination may be worth the incremental cost (in time and resources). Therefore, the GDT believes that combined antidepressantpsychotherapy is appropriate first-line treatment for patients with severe MDD. 34 National Adult Depression Clincial Practice Guideline

39 Supporting Evidence For Head-To-Head Comparison of Antidepressants Since the previous revision of the Depression Guidelines, one RCT and one nonsystematic review were found that compared the efficacy of antidepressant against each other. Thase, (74) a nonsystematic review, included eight randomized, double-blinds studies comparing a SNRI (venlafaxine) and SSRIs (fluoxetine, paroxetine, and fluvoxamine) and placebo for six to eight weeks (only four of eight trials were placebo controlled). Studies included adults (18 and older) who met the DSM-III-R or DSM-IV criteria for MDD for at least one month. All studies were funded by the SNRI manufacturer. Remission rates (using Hamilton Rating Scale for Depression, HRSD) were significantly higher (10%) with SNRI (venlafaxine) than with an SSRI (fluoxetine, paroxetine, and fluvoxamine). Odds ratio indicated that venlafaxine-treated patients had a 50% greater chance of attaining remission than patients treated with an SSRI (NNT = 10). However, the reported SSRI remission rates in this analysis (35%) are noticeably lower than the SSRI remission rates seen in other trials, raising the possibility of publication bias. Mulsant (75) included outpatients and inpatients age 60 and older with MDD and a HAM-D-17 score of 15 or above. The population was 71.6% female and 86.2% white. The study was a 12 week randomized double-blind trial comparing a TCA (nortriptyline, 25 to 50 mg) and a SSRI (paroxetine, dose range of 10 to 20 mg). There was no statistical difference in efficacy based on HAM-D-17 scores (p = 0.16). Patients were twice as likely to discontinue TCA (33%) as SSRI (16%) due to a significant side effect (p = 0.04), but the overall discontinuation rates did not differ significantly, p = AHRQ s systematic review, Mulrow (76) included 81 trials that addressed efficacy of antidepressant medications. The majority of trials were conducted in outpatient setting. All trials were in patients with MDD. All trials were double-blind and were six to eight weeks in duration. All studies were sponsored by pharmaceutical companies. The review included 80 studies that demonstrated that multiple newer antidepressants (SSRIs, SNRIs, and NRIs) were equally as effective as older tricyclic antidepressants. In 55 studies reporting response rates, SSRIs were compared with TCA1 (n = 38), TCA2 (n = 5), triazolopyridine (trazodone) (n = 4), tetracyclic antidepressants (n = 7), and MAOI (n = 1). SSRIs were equally effective compared with TCA1 (RR = 1.0, 95% CI: 0.9 to 1.1), TCA2 (RR = 1.1, 95% CI: 0.9 to 1.3), triazolopyridine (RR = 1.1, 95% CI: 0.7 to 1.6), tetracyclic (RR = 1.1, 95% CI: 0.9 to 1.3), and MAOI antidepressants (RR = 0.9, 95% CI: 0.7 to 1.3). There is no evidence that any one medication is more efficacious than any other. SSRIs have a small, short-term advantage in six week continuation rates over TCAs and heterocyclic medications (NNT = 20 to 33). 35 National Adult Depression Clincial Practice Guideline

40 The two subsequent systematic reviews in the Cochrane database included Geddes and Barbui. Geddes (77) compared SSRIs with a variety of antidepressants, including TCAs and heterocyclics. This systematic review included 98 studies conducted in primary and secondary care in 13,336 inpatients and outpatients with MDD. All the trials were double-blind and were six weeks in duration. The analysis of efficacy was based on 5,044 patients treated with an SSRI or related drug and 4,510 treated with an alternative antidepressant. The standardized effect size for SSRIs and related drugs together versus alternative drugs using a fixed effects model was (95% CI: to 0.076; Q = , df = 97, p < 0.001). The standardized effect size for SSRIs alone versus TCAs was (95% CI: to 0.107). There was no evidence of statistically or clinically significant differences between the drugs. In AHRQ s systematic review (76) less than 10% of the trials addressed adherence. These trials demonstrated that overall dropouts did not differ significantly between active treatments. In the Geddes (77) review, 27.7% of patients treated with SSRIs dropped out, compared with 32.7% of patients treated with TCAs (ARR = 5%, NNT = 20). Barbui s systematic review (78) focused on adherence and included 136 trials, 58% of which were conducted in the primary care setting, the total N was not reported. The majority of trials were six weeks in duration (84%) and nearly all studies were in patients with Major Depression. The estimate of overall dropout rates showed significantly fewer dropouts in the SSRI group compared with TCAs. (OR = 1.21, 95% CI: 1.12 to 1.30) On average, 27 out of every 100 patients taking SSRIs will dropout compared with 30 of every 100 taking TCAs. (ARR = 3%, NNT = 33) The authors concluded that there was a small difference in dropout rates between SSRIs and TCAs (amitriptyline, imipramine, clomipramine, desipramine, dothiepin, doxepin, lofepramine, and nortriptyline). Eight studies were found since the Cochrane systematic review that addressed treatment of MDD in older patients. Of these, Finkel (79) and Newhouse (80) compared sertraline with fluoxetine in 75 (university center) and 236 (outpatients) respectively. The trials were 12 weeks in duration. Finkel reported results that favored sertraline; however, this study was sponsored by Pfizer and there was evidence of selection bias. Newhouse concluded that both drugs were equally efficacious in this population; however, the authors didn't report any head-to-head comparisons and not all comparisons were reported. In addition, Newhouse used a one week placebo run-in, which doesn't control for the difference in half-life between sertraline and fluoxetine. 36 National Adult Depression Clincial Practice Guideline

41 Bondareff (81) and Oslin (82) compared sertraline with nortriptyline in specialty care (n = 210 and 76) and nursing home patients (n = 97). Trials ranged from 10 to 12 weeks in duration. The studies report conflicting results; Bondareff and Finkel favor sertraline while Oslin favors nortriptyline. Oslin compared sertraline open-label patients with double-blind treated nortriptyline (regular and low-dose) patients. Bondareff also reported that nortriptyline patients over 70 had poorer prognosis. Bondareff excluded placebo responders. Schweizer (83) compared imipramine and buspirone with placebo in 177 primary care outpatients. Results indicated that imipramine was superior to placebo but no head-to-head comparisons were reported. More patients taking imipramine also used over the counter medications (such as painkillers, etc.). Mulsant (84) compared nortriptyline with paroxetine in 80 mixed setting patients (inpatients and outpatients). The trial was six weeks in duration. The authors concluded that both drugs were equally efficacious. Forlenza (85) compared sertraline with imipramine in 55 specialty care outpatients. The trial was six weeks in duration. The authors concluded that both drugs were equally efficacious; while there was no significant difference in overall dropout rates, the majority of patients taking imipramine dropped out in the first two weeks. The trials were short in duration; had selection and publication bias; and small N s. Clinical Evidence, Vol. 9, (70) was searched and it was found that our recommendations are in line with their conclusions. The followings are their findings: Two systematic reviews have found no clinically significant difference in outcomes with different kinds of antidepressant drug, although one systematic review found that monoamine oxidase inhibitors were less effective than tricyclic antidepressants in people with severe depressive disorders, but may be more effective in atypical depressive disorders, for example increased sleep, increased appetite, mood reactivity, and rejection sensitivity. Systematic reviews have found that antidepressant drugs differ in their adverse event profiles. One systematic review has found that, on average, people seem to tolerate selective serotonin reuptake inhibitors a little better than tricyclic antidepressants, but the difference was small. Another systematic review and one retrospective cohort study found no strong evidence that fluoxetine was associated with increased risk of suicide. One RCT and observational data suggest that abrupt withdrawal of selective serotonin reuptake inhibitors is associated with symptoms including dizziness and rhinitis, and that these symptoms are more likely with drugs with a short half-life, such as paroxetine. 37 National Adult Depression Clincial Practice Guideline

42 Other Considerations Group Health Cooperative s 2003 Adult Major Depression Guideline included one additional study on venlafaxine vs. SSRIs identified from a Medline search using similar mesh terms that did not appear in our PubMed search (Smith (86) ). Group Health concluded: This study (Smith) found that venlafaxine was more effective than other antidepressants at treating depression. The difference in effect size was equal to about 1.2 points on the HAM- D which, although statistically significant, may not be a clinically significant difference. The meta-analysis was funded by Wyeth Laboratories, the manufacturer of venlafaxine, and this may have introduced bias in the study design, analysis, or reporting. The existing evidence is not sufficient to recommend the use of venlafaxine as first-line medication rather than well-established antidepressants such as SSRIs. Overall Conclusion The preponderance of evidence suggests that generally, all antidepressants are equally effective. While some evidence suggests that venlafaxine may be more effective that SSRIs in achieving remission in the first few weeks of treatment, due to concerns about publication bias and the lack of longer-term outcomes data, the GDT believes that currently there is insufficient evidence to preferentially recommend venlafaxine over other antidepressants for first-line treatment of MDD in the primary care setting. 2. Hypericum (St. John s Wort) for MDD 2A The GDT makes no recommendation for or against providing hypericum (St. John s wort) in patients with mild-to-moderate Major Depression. There is fair evidence of effectiveness of hypericum in this population. However, due to lack of consistency of preparation and dosage across trials, and concerns about lack of FDA oversight and consistency of hypericum preparations, the balance of benefits, harms, and costs compared with other treatments cannot be determined. Evidence-based: C 2B The GDT recommends against providing hypericum (St. John s wort) to patients with severe Major Depression. Evidence-based Evidence Grade Evidence for Recommendation 2A: Fair Evidence for Recommendation 2B: Evidence-based 38 National Adult Depression Clincial Practice Guideline

43 Rationale: 2008 Guideline New evidence was found that did not change the existing recommendations. Search Strategy We found four RCTs comparing hypericum with placebo and/or standard antidepressants: Kasper et al. (2006) (87) found that hypericum decreased depressive symptoms significantly more than placebo in patients with mild-to-moderate Major Depressive Disorder. The percentages of responders (NNT = 3) and remitters (NNT = 5) were higher in patients receiving hypericum than in patients receiving placebo as well. Gastpar et al. (2005) (88) found that hypericum was not inferior to sertraline in reducing depressive symptoms in moderately depressed patients (score 20 to 24 on the Hamilton Depression Scale) meeting the ICD-10 diagnosis for moderate depression, with at least four or more DSM-IV typical depression symptoms. Although no placebo group was studied, the percentage of responders did not differ between treatment groups either. Due to the inclusion criteria, it is possible that not all patients would meet the DSM-IV classification for Major Depressive Disorder. Gastpar et al. (2006) (89) found that hypericum was not inferior to citalopram in reducing depressive symptoms in moderately depressed patients (score 20 to 24 on the Hamilton Depression Scale and diagnosis of Major Depressive Disorder), whereas both hypericum and citalopram were superior to placebo. The percentage of responders was higher in both treatment groups than in the placebo group as well. Fava et al. (2005) (90) found that patients with mild-to-moderate Major Depressive Disorder diagnosed by Structured Clinical Interview for DSM-IV who were taking hypericum showed a significant improvement in depressive symptoms over patients taking fluoxetine (NNT = 12), and reported a trend toward superiority for hypericum over placebo (NNT = 6). Overall Conclusion We found no new evidence on the use of hypericum for patients with severe Major Depressive Disorder. Therefore, there is no change in the recommendation on use of hypericum in these patients. Three of the four studies we reviewed studied patients with Major Depressive Disorder, addressing one of the GDT s concerns about the heterogeneity of study populations in previous studies. However, the GDT has persistent concerns regarding the lack of standardization of dose and preparation of hypericum across trials. Given the lack of FDA oversight, it is uncertain whether trial preparations and doses of hypericum would be consistently available to patients for routine use. The GDT is, therefore, unable to determine the balance of benefit and risk of hypericum for mild-to-moderate Major Depression compared with other existing evidence-based therapies. 39 National Adult Depression Clincial Practice Guideline

44 For the 2008 Guideline, the health outcome of change in symptoms was the only relevant outcome for which evidence was found. In addition, all three health interventions prescription antidepressants, hypericum, and no treatment - were addressed in this 2008 Guideline Guideline We found four RCTs comparing hypericum with placebo and/or standard antidepressants, with mixed results: Gelenberg, et al. (2004) (91) found that the lack of hypericum efficacy in previous studies was unlikely to be attributable to treatment-resistant subjects. Uebelhack, et al. (2004) (92) found that moderately depressed patients taking hypericum showed a significant improvement in depressive symptoms over placebo control subjects. Szegedi, et al. (2005) (93) showed decreases in depressive symptoms in both patients taking hypericum and patients taking paroxetine, and concluded that hypericum is not inferior to paroxetine. Bjerkenstedt, et al. (2004) (94) found a significantly higher remission rate among mild to moderately depressed subjects who had taken hypericum versus those who had taken placebo, but study duration was short (four weeks) and there was no comparison of remission rates between hypericum and fluoxetine subjects. The Cochrane systematic review (95) included 37 trials comparing hypericum with placebo or antidepressant. Larger placebo-controlled trials restricted to patients with Major Depression showed only minor effects of hypericum (RR = 1.15, 95% CI: 1.02 to 1.29); older, smaller trials including patients with minor depression showed more marked effects (RR = 2.06, 95% CI: 1.65 to 2.59). Trials comparing hypericum extracts and standard antidepressants were heterogeneous. Patients given hypericum extracts had fewer adverse effects than those given SSRIs specifically, but the difference was not statistically significant (OR = 0.60, 95% CI: 0.31 to 1.15). The authors conclude that current evidence regarding hypericum is inconsistent and confusing. Some trials involving patients with Major Depression suggest hypericum has a minimal beneficial effect, while others suggest the effect is comparable with standard antidepressants. Furthermore, the pharmaceutical quality of hypericum preparations varies considerably, and trial results are limited and specific to preparations used in the study. Clinical Evidence (Issue 14, January 2006) also cited the Linde & Mulrow systematic review and the authors made similar conclusions about the quality of the trials and the concerns about pharmaceutical quality of hypericum. They emphasize the need to interpret RCTs on the efficacy of hypericum cautiously due to the lack of standardization of preparations used across trials and the varying doses of comparator standard antidepressants. We found no trials comparing hypericum with psychotherapy. 40 National Adult Depression Clincial Practice Guideline

45 Overall Conclusion The statement that the evidence is "insufficient and confusing" is an appropriate summary of the studies we reviewed. Based on this finding and other concerns expressed in previous reviews (varying definitions and inclusion criteria for depression, concerns over adequacy of blinding in the studies, and short trial duration), the GDT believes that there is insufficiently consistent evidence to recommend hypericum instead of prescription antidepressants for the treatment of MDD Guideline Supporting Evidence for Hypericum (St. John s Wort) versus Placebo Two RCTs examining the efficacy of hypericum versus placebo for treatment of MDD were found. The hypericum Depression Trial Study Group (96) was a randomized, double-blind, controlled trial including adults (at least 18 years old) with MDD and a minimum score of 20 on the HAM-D scale for depression. The study compared hypericum (900 to 1,500 mg/dl) (n = 113) and placebo (n = 116) for eight weeks, with sertraline (50 to 100 mg/day) (n = 111) as an active comparator to evaluate the study s sensitivity. The study found no evidence that hypericum is more effective than placebo in treating moderately severe Major Depression (p = 0.59). The only significant adverse event with hypericum was anorgasmia (p = 0.04). Kalb (97) included adult outpatients between 18 and 65 years old with mild to moderate single episode or recurrent MDD and a total score of greater than 16 on HAMD-17 scale. The study was a randomized, double-blind, controlled trial comparing hypericum extract WS 5572 (3 x 300 mg/day) (n = 37) and placebo (n = 35) for six weeks. The study concluded that the standardized hypericum extract WS 5572 has superior efficacy compared with placebo and very good tolerability in the acute treatment of mildly to moderately depressed patients. There was a significant decrease in HAM-D scores (p < 0.001) in hypericum (54.8%, or 10.8 points) compared with placebo (29.2%, or 5.7 points). There was no significant difference in the percentage of patients with at least a 50% improvement in HAM-D scores (a commonly used measure of antidepressant response) 62.2% for hypericum vs. 42.9% for placebo, p = 0.10, but the study may have been underpowered to detect this difference. There was, however, a significant difference in percentage of patients with at least a 60% improvement in HAM-D scores (51.4% vs. 17.1%, p = 0.002); however, this measure is not commonly used in other antidepressant efficacy trials. 41 National Adult Depression Clincial Practice Guideline

46 Clinical Evidence (70) was also searched and our recommendations are in line with their conclusions. They concluded: Two systematic reviews in people with mild to moderate depressive disorders have found that St. John's wort (Hypericum perforatum) versus placebo significantly improves depressive symptoms over four to 12 weeks, and have found no significant difference in symptoms with St. John's wort versus prescription antidepressant drugs. However, these findings have not yet been repeated in people with all grades of depression using standardized preparations of St. John's wort. The evidence cited must be interpreted cautiously because it is unclear how closely people in these trials match people in clinical practice, and the preparations and doses of H. perforatum and types and doses of standard antidepressants varied. More studies are needed on clearly defined, clinically representative people using standardized preparations. Interactions with other drugs are possible and should be considered. Shelton (98) compared St. John's wort with placebo in 167 outpatients attending academic medical centers for the treatment of severe Major Depression. The authors concluded that St. John's wort was no more effective than placebo in treating severe Major Depression. Since severe Major Depression is associated with morbidity and an increased risk for suicide, also since there are other effective available treatments, and the only study of St. John's wort in severely depressed patients showed no efficacy, the GDT recommends that St. John's wort not be used in patients with severe Major Depression. Supporting Evidence for Hypericum (St. John s Wort) Versus Antidepressants One RCT examining the efficacy of hypericum versus antidepressants for treatment of MDD was found since the previous revision of the Depression Guidelines. Behnke (99) included adult patients (18 to 73 years old) with mild to moderate Major Depression and HAM-D scores between 16 and 24. The study was a randomized, doubleblind, controlled trial comparing hypericum (150 mg twice daily) (n = 35) and SSRI (fluoxetine, 20 mg twice daily) (n = 35) for six weeks. There was no placebo control group. The study concluded that Hypericum perforatum is therapeutically equivalent to fluoxetine and therefore a rational alternative to synthetic antidepressants. There was a significant decrease in HAM-D score (p < 0.001) in both hypericum (50%) and fluoxetine (58%), but it was not significantly different (p = 0.23) between the two groups. Five RCTs of patients with milder forms of depression were reviewed. All were six to eight weeks in duration. Harrer, (100) Philipp, (101) Schrader, (102) and Woelk (103) studied patients with mild to moderate depression, while Brenner (104) looked at 30 patients with MDD (number of patients with severe MDD not specified), dysthymia, adjustment disorder, and depression - not otherwise specified. Harrer (100) and Schrader (102) compared St. John's wort with fluoxetine in 161 older adult and 238 outpatients with mild to moderate depression. The authors concluded that both medications are equally efficacious in treating mild to moderate depression. However, the Harrier study used low doses of fluoxetine (11 mg/day) that in many instances may be insufficient to achieve symptom remission. Comparisons with higher yet commonly used fluoxetine dosage regimens were not included in the Harrier study, and no placebo group was included to control for possible no effect of low-dose fluoxetine in this setting. 42 National Adult Depression Clincial Practice Guideline

47 Both Philipp (101) and Woelk (103) compared St. John's wort with imipramine in 263 and 288 outpatients with mild to moderate depression. The authors concluded that both medications are equally efficacious but patients tend to tolerate St. John's wort better than imipramine. (103) Philipp's study used low doses of imipramine that are often not sufficient to induce remission of depressive symptoms. Woelk's study was conducted in primary and specialty care patients. Brenner's (104) study compared St. John's wort with sertraline in 30 patients with Major Depression (number of patients with severe MDD not specified), dysthymia, adjustment disorder, and depression - not otherwise specified. The authors concluded that both medications were equally efficacious in treating mild to moderate depression. This study may be underpowered to detect significant differences, and no placebo control group was included. The GDT expressed several concerns about St. John's wort including: Wide range of dosages and different preparations of St. John's wort used across trials (some preparations may have other, undetermined active ingredients). Short duration of the trials (longer-term results are not clear). The definitions and classification of depression varied in trials, making it difficult to translate the results consistently into practice. Many of the patients in the reviewed trials with mild depression had adjustment disorders or minor/subsyndromal depression. There is no current evidence that suggests these types of depression need treatment with medication. Therefore, without a placebo control group in these trials, it is possible that while St. John's wort was equal to a comparator medication, neither medication may have been superior to placebo. Potential for bias due to difficulty in patient blinding (due to the characteristic taste and smell of St. John's wort). The products used in trials may not be widely available commercially. Dietary supplements are not subject to evaluation for safety and efficacy by the US Food and Drug Administration (FDA) nor is their manufacture held to compliance with the FDA s Good Manufacturing Practices. As a result of the lack of standardization and quality control, herbal content and efficacy vary, and contamination and misidentification of plant species may occur. St. John's wort is not approved by the FDA for treatment of depression. Traditional treatment alternatives (regulated antidepressant medication, structured psychotherapy) are readily available, well-tested, known to be effective, and subject to oversight. St. John's wort can cause potentially serious drug interactions, which were not adequately addressed in the reviewed studies. 43 National Adult Depression Clincial Practice Guideline

48 Overall Conclusion There is no evidence that St. John s wort is effective for severe MDD. There is lower quality evidence of short-term effectiveness in milder forms of depression. These trials often lacked placebo controls and included patients with subsyndromal forms of depression, who often remit spontaneously and therefore, it is unclear that any medication is indicated. Due to concerns about the quality of the evidence, absence of data on longer-term outcomes, and reports of adverse drug-hypericum interactions that are not yet routinely documented in prescribing decision-support software, the GDT believes there is insufficient evidence for or against St. John's wort as an alternative to other effective, regulated treatments for mild to moderate Major Depression. The GDT recognizes that some patients may initiate discussions about St. John's wort, or may already be taking St. John's wort at the time they consult a clinician. In these instances, the GDT suggests that clinicians discuss the evidence and concerns with patients, using a shared decisionmaking approach. 3. Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, Or Plan 3A 3B For patients with Major Depression expressing suicidal intent or plan, the GDT recommends consultation with specialty behavioral health. Consensus-based For patients with suicidal ideation or who have made previous suicide attempts, the GDT recommends consultation or collaboration with a psychiatrist before prescribing TCAs or venlafaxine. Consensus-based Rationale: 2008 Guideline No new evidence was found, the recommendations remain unchanged Guideline We found four studies that examined the association of antidepressants with suicidal intent or behavior. Yerevanian, et al. (2004) (105) found no difference in rate of suicidal behavior between patients taking tricyclic antidepressants (TCAs) versus those taking selective serotonin reuptake inhibitors (SSRIs). They did, however, find a significant increase in suicidal behavior in the discontinuation period of both classes of antidepressant. Martinez, et al. (2005) (106) showed no difference in rate of suicidal behavior in patients prescribed SSRIs versus those prescribed TCAs. There was a similar risk for different SSRIs, as well as for different TCAs. Fergusson, et al. (2005) (107) found no difference in odds of suicide attempts in patients receiving SSRIs versus those receiving TCAs. 44 National Adult Depression Clincial Practice Guideline

49 Gibbons, et al. (2005) (108) did not find a significant association between all prescribed antidepressants and suicide rate. The authors did find a significant positive association between TCAs specifically and suicide rate, but since this was an ecological study and individual patient data was not collected, it cannot be concluded that a causal relationship exists. A Clinical Evidence (Issue 14, January 2006) systematic review, one retrospective cohort study, and two case control studies specifically examined the risk of suicide between different classes of antidepressants. The systematic review found no difference in risk of suicide between SSRIs and TCAs (OR = 0.88, 95% CI: 0.54 to 1.42). It should be noted that RCTs were included in this review regardless of treatment condition and 59% of the RCTs used were in people who had a diagnosis other than MDD. (109) The retrospective cohort study found that the risk of suicide was higher in people who received fluoxetine (19/10,000 person years, 95% CI: 9/10,000 person years to 34/10,000 person years) than in those receiving dosulepin (dothiepin; RR = of suicide vs. dosulepin 2.1, 95% CI: 1.1 to 4.1). However, a subgroup analysis of people with no history of suicidal behavior or previous antidepressant prescription broadened the confidence interval to make (110, 111) the result nonsignificant (RR = 2.1, 95% CI: 0.6 to 7.9). One case control study reported no significant increase in suicide risk in people prescribed SSRIs compared with TCAs (OR = for suicide 0.57, 95% CI: 0.26 to 1.25). (106) Another case control study found no significant increase in suicide risk with individual SSRIs (amitriptyline, fluoxetine, paroxetine) compared with dothiepin (amitriptyline: OR = 0.83, 95% CI: 0.61 to 1.13; fluoxetine: OR = 1.16, 95% CI: 0.90 to 1.50; paroxetine: OR = 1.29, 95% CI: 0.97 to 1.70). (110) Overall Conclusion There is little consistent evidence of differences in all-cause suicide between classes of antidepressants. However, the consensus of the GDT remains that due to potential toxicity in overdose, TCAs and venlafaxine should be prescribed to suicidal patients only after consultation or collaboration with a psychiatrist Guideline No studies were found with a design that directly addressed our question. However, we did find two studies on safety of antidepressants in overdose situations. Buckley (112) calculated the fatal toxicity index expressed as death per million prescriptions for antidepressants. The serotoninergic drug class overall had a much lower index (1.6) than the tricyclic antidepressants (34.8) and monoamine oxidase inhibitors (20.0). Venlafaxine had a higher index (13.2) than the individual and combined results of other serotoninergic drugs. Shah (113) examined trends in suicide from drug overdose in older adults (65 years old and older) between 1993 to Antidepressants were the third (15%) most commonly used drugs in overdose leading to suicide. Of these deaths, 95% were due to tricyclic antidepressants. Death rates increased with age, with highest rates in men over National Adult Depression Clincial Practice Guideline

50 Volume 9 of Clinical Evidence, (70) most recent search date November 2003, was also searched, and our recommendations are in line with their conclusions. They found: One systematic review (search date not stated, which included RCTs completed by December 1989) pooled data from 17 double-blind RCTs in people with depressive disorders aged 12 to 90 years comparing a TCA (731 people) versus fluoxetine (1,765 people) or versus placebo (569 people). It found no significant difference in the rate of suicidal acts [attempts] between the groups (TCAs 0.4%, fluoxetine 0.3%, and placebo 0.2%), but development of suicidal ideation was less frequent in the fluoxetine group (1% fluoxetine vs. 3% placebo, p = 0.04; and vs. 4% TCAs, p = 0.001). One historical cohort study followed 172,598 people who had at least one prescription for one out of ten antidepressants during the study period in general practice in the UK. The risk of suicide was higher in people who received fluoxetine (19/10,000 person years, 95% CI: 9 to 34) than those receiving the TCA dosulepin (RR = of suicide vs. dosulepin 2.1, 95% CI: 1.1 to 4.1). In a nested case controlled, subanalysis in people with no history of suicidal behavior or previous antidepressant prescription, the risk remained the same, although the confidence interval broadened to make the result nonsignificant (RR = 2.1, 95% CI: 0.6 to 7.9). Although the apparent association may be because of residual confounding, there remains uncertainty about the possible association between fluoxetine and suicide. However, any absolute increase in risk is unlikely to be large. One systematic review was found that addressed suicide attempts or completions. AHRQ's review (2) included 15 studies, two cohort studies, and 13 meta-analyses. Only one of these studies was conducted in primary care. Nine trials compared fluoxetine with placebo or TCA, two trials compared fluvoxamine with placebo or TCA, three trials compared paroxetine with placebo, and one trial compared fluoxetine with placebo or a TCA. There is no evidence that any one medication is associated with higher rates of attempted or completed suicides, when all methods of suicide are considered in the primary care setting. Evidence suggests that there is a higher rate of completed suicide by overdose with TCA. Evidence also suggests that there is a lower rate of suicide by overdose with SSRIs than expected. On closer examination of the studies, Henry (114) looked at the relative mortality of antidepressants in the UK. The authors calculated the number of deaths per million prescriptions during the six years for all drugs taken together, for each of the four groups of antidepress-ants, and for each drug individually. The chi-square test was applied to the groups of antidepressants. The expected numbers of deaths were given for the individual drugs, with the Fisher s exact probability applied to the data. The authors concluded that TCAs had a higher ratio of observed to expected rates of suicide by overdose as compared with all antidepressants taken together. SSRIs has a lower ratio of observed to expected rates of suicide by overdose. A head-to-head TCA to SSRI comparison was not done. 46 National Adult Depression Clincial Practice Guideline

51 The Jick (111) study had small N s underscoring the low frequency of suicide in primary care settings, and was probably not sufficient to draw conclusions. Studies or reviews might not have large enough N's to account for differences in rare events. There may be unreported differences in trials because patients with suicidal ideation or severe forms of depression are often excluded from studies. These trials include not only patients with suicidal ideation, intent, or plan; but also patients without these characteristics. However, the GDT believes that the evidence above could be extrapolated to patients with suicide ideation, intent, or plan. Other Considerations and Overall Conclusion There is no evidence to support referring patients with suicidal intent or plan to specialty behavioral health. However, retrospective studies and expert opinion note that patients who express intent to commit suicide or have a specific plan to commit suicide are at higher risk for completed suicide. Therefore, consistent with current practice and common sense, and for clarity, the GDT recommends patients who endorse suicidal intent or plan be referred to specialty behavioral health. In contrast, many patients with depression will endorse occasional thoughts of suicide but deny intent or plan, and will agree not to commit suicide. These patients can be and often are managed successfully in primary care. However, observational evidence suggests that certain classes of antidepressants (TCAs and venlafaxine) are associated with higher incidences of death by overdose than other anti-depressants. But there are some patients who might benefit from receiving these antidepressants instead of other available alternatives. For example, patients may have other medical comorbidities where TCAs are useful in treatment (for example, chronic pain). In other instances, patients may have responded to these antidepressants in the past, and are an established patient well known to their provider. In these instances, the GDT would not automatically exclude these patients from receiving these medications, but recommends that the provider discuss the potential risks and benefits with a psychiatrist. (Note: MAOIs are excluded from this guideline, as this class of antidepressants is not recommended for use by primary care clinicians in treating depression). 47 National Adult Depression Clincial Practice Guideline

52 4. Second-Line Treatment Of MDD 4A 4B 4C 4D 4E For patients with MDD whose symptoms fail to remit after first-line treatment, the GDT recommends an assessment of the adherence to the initial treatment regimen. Consensus-based For patients with MDD whose symptoms fail to remit after adherence to first-line treatment, the GDT recommends that treatment options include: Combining antidepressants and psychotherapy. Evidence-based Increasing the dose of the initial antidepressant. Consensus-based Switching to a different antidepressant of the same or different class. Consensus-based Switching from psychotherapy to antidepressants or from antidepressants to psychotherapy. Consensus-based Combined pharmacologic treatment (monitoring for toxicity, side effects, and drug interactions) with SSRIs and low-dose TCAs, or bupropion, or buspirone, or Consensus-based (all in this list) mirtazepine, or lithium, or liothyronine (T3). The GDT makes no recommendation for or against providing folate or inositol to patients whose MDD symptoms fail to remit after adhering to first-line treatment. Evidence-based: I The GDT makes no recommendation for or against providing atypical antipsychotics to primary care patients with (nonpsychotic, nonbipolar) MDD whose symptoms fail to remit after adherence to first-line treatment. Evidence-based: I The GDT recommends against providing augmentation with pinodol for patients with MDD whose symptoms fail to remit after adherence to first-line treatment. Evidence-based Rationale: Evidence for Recommendation 4C: Insufficient Evidence for Recommendation 4D: Insufficient Evidence for Recommendation 4E: Evidence-based Note: It is the expert opinion of the GDT, based on criteria used in trials of Major Depression treatment, that partial response is usually seen within four weeks of initiating treatment for MDD, and that remission, if it is going to occur, usually occurs within six to 12 weeks of initiation of treatment. For purposes of this guideline, patients whose symptoms have not improved by four weeks or resolved by six to 12 weeks after initiation of treatment are candidates for second-line treatment. 48 National Adult Depression Clincial Practice Guideline

53 2008 Guideline New evidence was found that did change the existing recommendations. Five RCTs and one new meta-analysis were identified that address a second line of treatment for patients whose symptoms did not resolve after initial treatment. Mahmoud et al. conducted a six-week, multicenter, double-blind, placebo-controlled trial of augmentation of antidepressant therapy with risperidone (Risperdal) (up to 2 mg per day) in patients with Major Depressive Disorder whose symptoms had not resolved after a minimum of four weeks. (115) In the intention-to-treat analysis, remission of depression was seen in significantly more patients in the active treatment group than in the placebo group (24.5% vs 10.7%, p = 0.004, NNT = 7). The intervention group also had a higher percentage of patients with response (50% improvement in HAM-D scores; 46.2% vs. 29.5%, p = 0.004, NNT = 6). The 2.8 point mean change in HAM-D scores between groups, although statistically significant, is of marginal clinical significance. Patients who received risperidone gained an average of 2.8 lb in six weeks vs. an average of 0.3 lb in the placebo group (p < 0.001). One meta-analysis and one RCT were identified that addressed the efficacy of augmentation of antidepressant therapy with atypical antipsychotic agents. Papakostos et al. (2007) (116) published a meta-analysis that pooled the results of ten randomized placebo-controlled trials of augmentation with three atypical antipsychotic agents, olanzapine, quetiapine, and risperidone. Of these trials, four were published in the peer-reviewed literature and six were data from unpublished scientific reports (the quality of the latter studies cannot be explicitly determined). Data from a total of 1500 patients were pooled. These authors found that remission rate and symptom scores were significantly improved in the actively treated population (NNT for response and remission, ~ 4). There was no difference in overall discontinuation rates, but more patients in active treatment discontinued medication due to side effects (NNH not determined). In a randomized, double-blind, placebo-controlled trial (Berman et al., 2007) (117) the efficacy and safety of apiprazole as augmentation therapy for patients with treatment-resistant depression. Both remission rates and improvement in depression scores were reported to be improved in actively treated patients. While the overall rate of adverse effects was high, the discontinuation rate due to adverse effects was only 2.2% in the apiprazole group. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial reported the results of comparing treatment modalities as second-step therapy after initial treatment failure with citalopram monotherapy. Both augmentation and substitution strategies were examined. Comparable outcomes were seen with cognitive therapy and pharmacotherapy with several different medications, although fewer than one-third of participants consented to randomization strata that permitted comparison of pharmacotherapy with cognitive therapy. Switch to cognitive therapy seemed to be associated with fewer side effects than switch to medication, but because of the small sample size, the differences were not statistically significant (NNT undetermined). When it was used as augmentation of citalopram, the effect of pharmacotherapy was seen more rapidly than that of cognitive therapy, with mean time to first remission (for those who did remit) 40 days in the pharmacotherapy group compared with 55 days in the cognitive therapy group. (118) 49 National Adult Depression Clincial Practice Guideline

54 In a comparison of sustained-release bupropion, sertraline, and extended-release venflaxine, Rush et al. studied the substitution of these drugs as monotherapy for patients whose depression had not responded adequately to SSRIs. (119) All three strategies resulted in the same degree of improvement; no significant difference in side effects was seen among the three treatment groups. Trivedi et al. compared augmentation of monotherapy with bupropion or buspirone. Both groups had similar rates of remission, but a greater reduction in the severity of symptoms and fewer side effects (NNT = 8 for discontinuation due to side effects) were seen with bupropion treatment. (120) Two RCTs were identified that compared third-line treatment for patients who had not achieved remission in two separate therapeutic trials. Fava et al. (121) compared mirtazapine and nortriptyline as sole medications for patients who had not achieved remission with, or had been unable to tolerate, two previous therapies in the STAR*D trial. Both groups achieved similar results, with remission rates lower than 20%. In another study in the third level of the STAR*D trial, Nierenberg et al. (122) compared lithium with triiodothyronine as augmentation therapy for patients who did not have remission of symptoms after two previous trials. Outcomes were similar in the two groups, but triiodothyronine use was associated with fewer side effects (NNT = 14). Overall Conclusion: The STAR*D trial is a national, multicenter, academic and community, private and public sector equipoise RCT. Patients not responding to initial treatment with citalopram (an SSRI) at mean doses of 41 mg were randomized to different, multiple follow-up treatment options, but patients could be excluded from consideration of randomization to treatments they found unacceptable. (This design mimics real-world discussion of shared decision making.) Patients who failed to adequately respond to second-phase treatment proceeded to third-stage treatment with similar patient opt out possibilities; continued refractory patients proceeded to stage 4 treatment. The STAR*D and other trials reinforce, or in some cases add, options for primary care clinicians treating patients with Major Depression who fail to remit with initial treatment. Because of the lack of placebo in these trials, the GDT did not feel these recommendations deserved an evidence-based rating, unless previous placebo RCTs supported specific recommendations. (It is possible that some patients would have responded to placebo in these trials as a result of the nonspecific attention effect of being in a clinical trial.) Also of note, the STAR*D population might not exactly reflect the Kaiser Permanente population (STAR*D patients tended to be more socioeconomically disadvantaged, a high percentage were uninsured, and 15% to 18% expressed suicidal ideation). 50 National Adult Depression Clincial Practice Guideline

55 Contrary to previously reviewed evidence on atypical antipsychotic augmentation for Major Depression, the Mahmoud and Berman studies and the Papakostis meta-analysis suggest that short-term augmentation with risperidone may be beneficial. All of these trials were drug company sponsored and of short duration. Some of the findings (i.e., the Mahmoud study) were of questionable clinical significance, whereas the Papakostis meta-analysis suggests some potentially clinically significant findings. All of these studies compared atypical antipsychotic augmentation with placebo, not with other proven or established augmentation or second-line strategies. Additionally, concerns have been raised about cardiometabolic side effects and safety of atypical antipsychotics. Patients taking these medications require laboratory, as well as mood, monitoring (consensus statement: Diabetes Care 2004: 27: ). Finally, atypical antipsychotics are currently significantly more costly than other available strategies. There is fair evidence of short-term effectiveness for use of atypical antipsychotic agents to augment antidepressants in patients with nonpsychotic, nonbipolar MDD who fail to remit with initial treatment. However, due to the lack of longer-term data, the known cardiometabolic risks of treatment with these medications, and the lack of comparison data against other strategies, the balance of benefits, harms, and costs compared with other treatments cannot be determined. The GDT sought evidence of both health outcomes and health interventions for the clinical question listed above. The only relevant health outcome found in the previous guideline pertains to change in depression symptoms among participants in the various trials listed above. Health interventions were sought in: changing antidepressant medication, increasing existing antidepressant dose, switching to psychotherapy, adding psychotherapy, adding another antidepressant to existing antidepressant, augmentation to existing antidepressant, rtms, and vagus nerve stimulation. Relevant information on all listed health interventions were found in the studies listed above. However, to avoid duplication of efforts, the GDT has decided to defer the recommendations for vagus nerve stimulation and rtms to those made by KP s Interregional New Technology Committee (INTC). Excerpts from Kaiser Permanente s INTC recommendations on vagus nerve stimulation and repetitive transcranial magnetic stimulation (rtms) are summarized below: Vagus Nerve Stimulation The INTC maintains its prior recommendation: There is insufficient evidence to determine whether vagus nerve stimulation is a medically appropriate treatment for any patient with treatment-resistant depression. The existing evidence is of insufficient quantity and quality. 51 National Adult Depression Clincial Practice Guideline

56 Repetitive Transcranial Nerve Stimulation Based upon a review conducted by the INCT in Oct. 2003, there is insufficient evidence to determine whether repetitive transcranial magnetic stimulation (rtms) is a medically appropriate treatment for any patient. For more information on vagus nerve stimulation and Repetitive Transcranial Magnetic Stimulation (rtms) for the treatment of Depression, please refer to the following, updated statements made by (INTC): Vagus Nerve Stimulation: rtms: Guideline We found several studies that address a second-line of treatment for patients whose symptoms did not resolve after initial treatment. Schatzberg (2005) (123) found that patients with Major Depression who failed to respond to initial treatment with nefazodone or CBASP (cognitive behavioral therapy) both showed statistically significant improvement when switched to the alternate therapy. Patients who switched from nefazodone to CBASP showed a significantly greater improvement in depressive symptoms than those who switched from CBASP to nefazodone, possibly due to a decrease in medication-related side effects when switching to CBASP. There was no nonresponder control group continuing in the initial treatment. Parker, Brotchie & Parker (2005) (124) found that the addition of olanzapine to antidepressants did not significantly improve symptoms in patients with nonpsychotic Major Depression when compared with control subjects who were treated with antidepressants exclusively. Perry, et al. (2004) (125) found that, when compared with control, pindolol augmentation did not have a significant effect on depressive symptoms of patients who previously did not respond to SSRIs. Kauffmann (2004) (126) found a significantly positive effect of right prefrontal transcranial magnetic stimulation (rtms) over ten days in a small group of patients who failed to respond to at least two standard antidepressants given at adequate doses for at least eight weeks. Rumi (2005) (127) studied the effect of rtms on patients with severe nonpsychotic Major Depression and found that rtms had a significantly better impact on depressive symptoms and remission rates when compared with control. Clinical Evidence reports several RCTs within systematic reviews that found positive improvement in depressive symptoms after the addition of an antidepressant to psychotherapy, or the addition of psychotherapy to an antidepressant. These findings are consistent across all severities of depression mild, moderate, and severe. For more details, see the rationale statements under First-line treatment recommendations. Clinical Evidence found one systematic review and one subsequent RCT examining augmentation of prescription antidepressant drug treatment with lithium or pindolol versus placebo in adults with treatment-resistant depression. 52 National Adult Depression Clincial Practice Guideline

57 One systematic review found no significant difference in the proportion of people who responded over one to eight weeks between pindolol augmentation and placebo; responders: 10/53 [19%] with pindolol augmentation vs. 6/53 [11%] with placebo; absolute risk difference +8%, 95% CI: -6 to +21%; RR = not reported). The same review found that lithium augmentation significantly increased the proportion of people who responded over two weeks compared with placebo; 11/26 [42%] with lithium vs. 4/24 [17%] with placebo; absolute risk difference 25%, 95% CI: 2% to 49%; RR = not reported. (128) The subsequent RCT found no significant difference between lithium augmentation and placebo in the proportion of people who responded over six weeks was 2/18 [11%] with lithium vs. 3/17 [18%] with placebo; reported as nonsignificant, CI not reported. (129) A Cochrane review (130) yielded two RCTs (151 patients) that examined the use of folate in addition to other treatments for Major Depression. It found that patients treated with folate in addition to other treatments had an additional average decrease in Hamilton Depression Scale ratings of 2.65 points (95% CI: 0.38 to 4.93); fewer folate patients experienced an inadequate response to treatment (less than 50% decrease in depression scores from baseline) with a NNT of 5. The trials did not reveal any problems with the safety or acceptability of folate. The Cochrane review suggests that limited evidence to suggest folate may be of benefit as a supplement to other treatments for depression, but it is unclear if this benefit depends on baseline serum folate levels (or presence of baseline folate deficiency). Another Cochrane systematic review (131) examined the effectiveness of inositol as an adjunct to antidepressants. Combining continuous measures using Standardized Weighted Mean Difference (SMD), no statistically significant overall heterogeneity of effect between trials was observed (Chi-square = 3.57, df = 3, p = 0.31). The authors concluded that there was no clear therapeutic benefit and that further research would be needed before a recommendation for this intervention could be made. Cochrane also examined the therapeutic efficacy and safety of rtms for depression. (132) Five studies compared rtms with a sham rtms intervention and showed statistical homogeneity. The relative risk, using a fixed effects model was 0.81 (95% CI: 0.36 to 1.83; p = 0.6). The authors concluded that there is no strong evidence for benefit from using transcranial magnetic stimulation to treat depression, although the small sample sizes do not exclude the possibility of benefit. In their systematic review of the literature, the National Institute for Clinical Excellence (NICE) (57) found strong evidence suggesting there is a clinically significant difference favoring combined CBT and antidepressants over antidepressants alone on increasing the likelihood of remission in people with chronic depression (which, by definition, is refractory to initial treatment). (N = 1, n = 454, RR = 0.73, 95% CI: 0.62 to 0.84). Overall Conclusion The evidence continues to support our previous recommendations for treatment of patients with MDD whose symptoms do not resolve after the initial treatment. Additional wording was included to clarify that these recommendations assume that patients were adherent to the initial treatment but resistant to it (differentiating lack of adherence from true treatment nonresponse). 53 National Adult Depression Clincial Practice Guideline

58 Psychotherapy We found one new study that supports changing from antidepressants to psychotherapy or vice versa in patients who fail to respond to initial treatment for Major Depression after 12 weeks. There was no control group (continuing on previous therapy) in this study, therefore it is unclear how many patients would have responded to the initial therapy given more time. However, clinically, a change in therapy for nonresponse is usually initiated before 12 weeks. Given that this is the only study we identified that suggests that switching is effective for nonresponse to initial treatment for Major Depression, and the presence of more evidence for other treatment options, the GDT does not believe that it can make an evidence-based recommendation for this strategy at this time. However, structured psychotherapy (CBT, IPT, and PST, as outlined in the problem formulation) and antidepressants have been shown to be equivalent treatment options for initial treatment for many patients with Major Depression. Extrapolating from this, the GDT believes that offering the alternative treatment as one of several strategies (albeit a strategy without as much direct evidence as others) for patients who fail to respond to initial MDD treatment to consider is reasonable, and includes this as a consensus option. Lithium Augmentation Evidence remains mixed on lithium augmentation. The Fava RCT (133) reviewed by the GDT in 2004 was not included in the systematic reviews covered above. We therefore have one systematic review and one RCT (although the latter has some limitations as discussed in the 2004 review) favoring lithium augmentation, and one RCT finding no significant difference. The weight of the evidence seems to favor use of lithium as a potential augmentation strategy, with a consensus-based (rather than evidence-based) recommendation. Primary care clinicians (the main audience for this guideline) seldom initiate lithium therapy, although they commonly see patients taking lithium prescribed by a psychiatrist. Therefore, the GDT recommends consultation with psychiatry for patients in whom this option is being considered. Transcranial Magnetic Stimulation Since our previous iteration of the guideline, we found two studies on transcranial magnetic stimulation. Although these studies showed benefit, the studies were small and of short duration, the technology is not readily available, no comparisons have been made with other augmentation strategies, and the added value has not been quantified. Therefore, the GDT agrees with the Cochrane review that there is insufficient evidence at this time to recommend this intervention for the treatment of MDD. Folate The GDT discussed the Cochrane reviews on folate in combination with antidepressants for treatment of Major Depression. The GDT believes that more studies are needed, particularly examining differences in patients with normal vs. below-normal serum folate levels, before this strategy can be recommended. Inositol The GDT concurs with the Cochrane review on inositol. The authors concluded that there was no clear therapeutic benefit and that further research would be needed before a recommendation for this intervention could be made. 54 National Adult Depression Clincial Practice Guideline

59 Vagus Nerve Stimulation The following rationale statement on Vagus Nerve Stimulation for depression comes from the Kaiser Permanente Interregional New Technologies Committee (INTC) meeting of November 14, 2005: The INTC reviewed this topic in July of 2004 and had an update March At that time, the INTC found there is insufficient evidence to determine whether vagus nerve stimulation (VNS) is a medically appropriate treatment option for any patient with Major Depressive Disorder. The existing evidence regarding how VNS effectively treats Major Depressive Disorder (MDD) is of insufficient quantity and quality. The published studies are not blinded, report incomplete data sets of short-term results, and fail to follow intent-to-treat principles, which can cause results to be overestimated. Recent publications and requests from TPMG, NWPMG, and TSPMG are bringing this topic back to the INTC. Marc Meisner, MD, TPMG Chief of Psychiatry, presented this topic to the committee. Recent publications, an ECRI Target 10/05, Hayes 10/05, and BCBSA TEC 08/05 assessment served as the basis for the discussion. The vagus nerve stimulation (VNS) Therapy System, developed by Cyberonics, Inc., Houston, TX, consists of an implanted pacemaker-like pulse generator and nerve stimulation and delivers intermittent stimulation to the patient's left vagus nerve. The device costs $15,500. In 1997, the FDA approved this Cyberonics technology for treatment-resistant seizures in epilepsy patients. The mechanism of action of VNS for treatmentresistant depression (TRD) remains unknown. However, the vagus nerve projects into areas of the brain associated with neuropsychiatric disorders, neuroimaging studies have shown changes in areas of the brain linked to mood regulation with VNS, and evidence of mood improvement was seen in VNS for epilepsy studies irrespective of seizure control. In July 2005, the FDA approved Cyberonics' VNS pre-market approval application for "the adjunctive long-term treatment of long term treatment of chronic or recurrent depression for patients 18 years of age or older who are experiencing a major depressive episode and have not had an adequate response to four or more adequate antidepressant treatments." With its approval, the FDA required two post-approval studies to examine optimal stimulation dosing with one-year follow-up for 450 patients in a randomized double-blind study and five-year patient outcomes in 1,000 implanted patients. Of the 18 million people in the United States who experience a major depressive episode in one year, four million suffer chronic or recurrent pharmacoresistant depression. Some of these latter patients have TRD per the July 2005 FDAlabeled indication for VNS. 55 National Adult Depression Clincial Practice Guideline

60 Four published studies, all sponsored by Cyberonics, compose the scientific evidence for VNS for TRD. They involved a total of 422 patients with TRD. Dr. Meisner provided an overview on the pilot study (D-01), the Acute and Long-Term Pivotal Study (D-02), and the Comparative Study (D-04). Recent publications from Rush and George 2005, focused on D-02 and D-04. An open label study of 60 patients (pilot study D-01) who received VNS implant showed a response, defined as greater or equal to a 50% reduction on the Hamilton Rating Scale for Depression (HRSD), of 31% at 12 weeks, 44% at one year, and 42% at two years. Remission, defined as less than ten on the HRSD, was 15%, 27% and 22% for the same periods, respectively. However, differences in response rates between periods were not statistically significant. Rush, 2005, reported that in the D week, pivotal double-blind, randomized controlled trial, active VNS therapy (n = 112) was no more effective than sham VNS (n = 110) in alleviating symptoms of depression among a population of adults diagnosed with Major Depressive Disorder or bipolar disorder. Active and sham treatment groups were well balanced with respect to baseline characteristics. At week 12, there was no significant difference between active and sham VNS in treatment response rates (15.2% versus 10.0%, respectively), nor were there significant differences between active and sham VNS groups for 4 of 5 scales used as secondary measures of efficacy. The only endpoint to show a significant difference between the two study arms was the self-administered Inventory of Depressive Symptomatology-Self-Report (IDS-SR) (17% vs. 7%). A two-year follow-up active treatment study (long-term pivotal study D-02, Rush 2005) consisted of patients from both active and sham arms of the acute pivotal study. This open label study demonstrated statistically significant improvement in depression scores measured by the 24-item HRSD and the 30- item IDS-SR. The study found an HRSD response of 27% at one year and 21% at two years as well as HRSD remission in 16% of patients at one year follow-up. Dr. Meisner discussed various study limitations. Comparisons of long-term outcomes must also be interpreted with caution, as patients were not randomized or blinded to treatment in longer-term trials. Patients treated with long-term VNS in clinical trials were allowed to receive additional concomitant treatments, raising the possibility that the observed response was not a result of VNS. Stimulation parameters varied among patients, which may have affected treatment outcomes. The manufacturer supported these studies, and a number of investigators listed as authors in these studies have disclosed individual financial relationships with the manufacturer, therefore, the potential for bias cannot be excluded. 56 National Adult Depression Clincial Practice Guideline

61 A fourth study (comparative study D-04) examined outcomes for the more than 200 patients in the long term pivotal study relative to those of more than 120 patients with TRD from a separately recruited control group who had not received VNS implants and were being treated with standard of care therapies. This trial resulted in a statistically significant HRSD response rate at one year of 27% for the VNS group, which could also use standard of care remedies, versus 13% HRSD response for the group only allowed standard of care treatment. Study limitations were also noted with the D-04 comparative study. The patients from the comparison control group were recruited separately, after results from D-02 were known, so do not represent adequate controls for the long-term effect of VNS. These studies, together with prior experience with VNS therapy in seizures, confirmed that surgical complications for VNS implant, including infection and vagus nerve damage, are low. More than half of implanted patients report stimulation-related voice alteration; much smaller percentages note increased cough, dyspnea, neck pain and other VNS-induced discomforts which tended to lessen with time. Further, VNS therapy does not appear to pose a greater risk for mania, suicide, and worsening depression. Overall, the evidence is limited by the studies' designs and the data quality suffers from inadequate controls and blinding, placebo effect, research site differences, heterogeneous stimulation parameters, various concomitant treatments allowed, and possible investigator bias. The updated Hayes Assessment from 10/05 concluded that the currently available evidence is insufficient to permit conclusions regarding the efficacy and safety of VNS as an adjunct therapy in treatment-resistant Major Depression and bipolar disorder. A Hayes Rating of C is assigned for VNS as an adjunctive therapy in adults with severe Major Depression or bipolar disorder when symptoms associated with a major depressive episode are refractory to multiple regimens of standard medication and other therapies, including electroconvulsive therapy (ECT) and psychotherapy; D is assigned for VNS in patients with other types of depression and in patients with Major Depression or bipolar disorder who respond to medical treatment, psychotherapy, and/or ECT; and D is assigned for VNS in patients with contraindications to VNS. These Ratings reflect the paucity or lack of evidence regarding the safety and/or efficacy of VNS in these patient populations. 57 National Adult Depression Clincial Practice Guideline

62 The updated ECRI Assessment also concludes there is limited ability to draw strong conclusions about the efficacy of VNS to treat treatment-resistant depression (TRD) using the available evidence. "All studies thus far have been manufacturer-sponsored with notable design weaknesses. The largest study fails to follow intent-to-treat principles, potentially overestimating results by not reporting outcomes of all patients entered in the study. Although VNS therapy lacks the severe side effects (e.g., long-lasting cognitive impairments, renal and thyroid toxicity, hypertensive crises, anticholinergic effects) of the treatments typically offered to patients with TRD, conclusions cannot be drawn regarding adverse events associated with long-term use of VNS therapy for depression due to incomplete adverse event reporting in the larger of the two trials used to write this report." Apart from the KP Northwest, which intends to use VNS for highly select patients with TRD, no other region expressed support for its use at this time as no definitive benefit of VNS for TRD has been yet shown in a large, prospective, randomized trial with adequate follow-up. There is insufficient evidence to determine whether vagus nerve stimulation is a medically appropriate treatment for any patient with treatment-resistant depression. The existing evidence is of insufficient quantity and quality. Pindolol Augmentation We found one additional negative study on pindolol augmentation. Combining this with our previous findings on pindolol, the GDT recommends against the use of pindolol as an augmenting agent for Major Depression at this time. Assessing Adherence Finally, we found no evidence examining the benefit of assessing adherence to initial treatment before attempting other treatment alternatives. However, the studies on alternative treatments we reviewed included patients who were adherent to initial treatment regimens. Furthermore, assessing patient adherence is consistent with patient-centered practices designed to promote self-care, and is a generally accepted tenant of practice. Therefore, the GDT elected to include this consensus recommendation in order to be thorough. 58 National Adult Depression Clincial Practice Guideline

63 2004 Guideline Supporting Evidence For Antidepressant or Psychotherapy Alone Versus Combined Treatment of Antidepressant and Psychotherapy One RCT looked at efficacy of combined psychotherapy and antidepressant treatments versus psychotherapy or antidepressant medication alone in patients with chronic MDD. Hirschfeld (134) (n = 681) included adult patients with chronic Major Depression (who by definition have failed initial treatment) between the ages of 18 and 75. The population was 65% female, 91% white. The study was a randomized, open-label, controlled trial comparing a SNRI (nefazodone 200 to 600 mg/day) and Cognitive Behavioral Analysis System of Psychotherapy (CBASP) and the combined nefazodone/cbasp treatment for 12 weeks. There was no placebo control group. HAM-D score analysis (taken from the original publication of the study by Keller (135) ) indicated that that the combined treatment was statistically superior to either treatment alone; p for nefazodone vs. combined treatment, p for CBASP vs. combined treatment. SF-36 general health indicated the same results, p = and p = 0.02 respectively. There was also greater improvement in psychosocial functioning when combined treatment was used (p = 0.02 and p = 0.02). We also searched Volume 9 of Clinical Evidence, (70) most recent search date November 2003, and found that our recommendations are in line with their conclusions. They found: One nonsystematic review showing that the combination of psychotherapy and antidepressant medication in patients age 18 to 80 with severe MDD is more beneficial than either treatment alone. Two RCTs showing that combination antidepressant-psychotherapy treatment is more effective than either treatment alone for patients with mild to moderate Major Depression. One of these studies specifically examined patients with chronic, refractory, or depression recurring on treatment (who by definition have failed first-line treatments). Overall Conclusion Combining antidepressant treatment with psychotherapy is more effective than either treatment alone for patients who have failed first-line treatment or for patients with severe MDD. See discussion in the rationale for first-line treatment of MDD for discussion of the role of initially combining antidepressant-psychotherapy for patients with mild to moderate MDD. 59 National Adult Depression Clincial Practice Guideline

64 Supporting Evidence For Increasing Existing Antidepressant Dose, Switching Antidepressants, Adding Another Antidepressant to Existing Antidepressant or Adding An Augmenting Agent to The Existing Antidepressant Four RCTs looked at second-line treatment strategies for the treatment of MDD in adults whose symptoms did not resolve after the first treatment. Fava (n = 101) included outpatients with MDD (initial HAM-D-17 score of 16) between ages of 18 and 65 who were either partial responders or nonresponders to (133) eight weeks of treatment with SSRI (fluoxetine, 20 mg/day). The study was a randomized, double-blind, controlled trial comparing increasing the SSRI (fluoxetine) dose (40 to 60 mg/day), adding low dose desipramine (25 to 50 mg/day) to existing 20 mg/day fluoxetine, and adding an augmenting agent (lithium, 300 to 600 mg/day) to existing 20 mg/day fluoxetine for a period of four weeks. Mean change in HAM-D-17 score (from visit one to endpoint) showed no statistically significant differences among these patients (p = 0.4, for all comparisons). The difference in response rates (patients who showed at least a 50% reduction in HAM-D scores) across these three treatment groups was also not significantly significant (p = 0.5, for all comparisons). There was also no significant differences in dropout rates across the three groups (p = not stated). The study did not include a placebo group, so placebo response to augmentation or an eventual response to continued treatment at the same does cannot be excluded, and the study may not have had sufficient power to detect true differences between groups. Peet (136) included patients aged 18 to 70 with HAM-D-17 score of 15 or more despite ongoing treatment with a standard antidepressant at an adequate dose. The study was a small (n = 70), randomized, double-blind, controlled study comparing augmenting with ethyleicosapentaenoate (E-EPA) at 1 g/dl, 2 g/dl, 3 g/dl; liquid paraffin placebo for a period of 12 weeks. Change in 17-item Hamilton Depression Rating Scale (HDRS) score (p = 0.02), Montgomery-Asberg Depression Rating Scale (MADRS) score (p = 0.006) and Beck Depression Inventory (BDI) score (p = 0.007) for the intention-to-treat population showed that augmentation with ethyl-eicosapentaenoate at a dosage of 1 g/dl was statistically significant in treatment of Major Depression in patients who remain depressed despite adequate standard therapy. Nemets (137) included adults 18 to 75 years old with a current diagnosis of MDD (24-item Hamilton Depression rating Scale of 18 or higher). The study was an extremely small (n = 20), randomized, double-blind, controlled study comparing ethyl-eicosapentaenoate (E-EPA at 2 g/day) and placebo for a period of four weeks. The author concluded that the study showed significant benefits in adding E-EPA to the existing treatment (p = not stated). 60 National Adult Depression Clincial Practice Guideline

65 Perez (138) included adults 18 to 65 years old with single or recurrent MDD, with the current episode resistant to pharmacological treatment. The study was a small (n = 80), randomized, double-blind, controlled study comparing augmenting with pindolol and placebo for a period of ten days. There was no significant difference in the change in HAM-D score from day 0 to day 10 between pindolol and the placebo groups (p not given). The study does not support the hypothesis that the addition of pindolol results in a rapid augmentation of the effects of SSRIs in depressed patients resistant to treatment. Other pindolol studies (of mixed results) were found, but those studies addressed the effect of adding pindolol to antidepressants at the onset of treatment for Major Depression, rather than augmenting antidepressants with pindolol after initial nonresponse to antidepressants. As these studies examine a different clinical question (improving speed of initial response to antidepressants rather than augmentation in cases of nonresponse), these studies were excluded from formal review. Other Considerations E-EPA may not be widely available in a standardized preparation, and the two E-EPA trials were small. Therefore, the GDT considers the E-EPA findings preliminary, and pending further study, does not currently recommend E-EPA be used regularly by primary care physicians as an augmentation strategy. No evidence was found on the efficacy or safety of combining antidepressants from the same class in the treatment of MDD that has failed to respond to initial treatment. There are theoretic safety concerns about the increased risk of serotonin syndrome when combining SSRIs or using SSRIs with other highly serotonergic antidepressants (SNRIs). MAOIs are generally not recommended for use by primary care, given their potential toxicity and drug interactions, and have been excluded from this guideline. Combining MAOIs and SSRIs have been associated with adverse patient outcomes and are contraindicated. Only one RCT was found on combining low-dose TCAs and SSRIs for MDD that failed to respond to initial treatment and it used desipramine, a later generation TCA with fewer side effects and more noradrenergic action than most other TCAs. There are theoretic reasons that augmenting serotonergic agents with norepinephrinergic agents would lead to improved outcomes (by affecting two different neurotransmitter systems). However, we did not find good evidence that so-called dual action antidepressants (enhancing both serotonin and norepinephrine) are superior to either SSRIs or TCAs in the treatment of depression. 61 National Adult Depression Clincial Practice Guideline

66 There was also no consensus among the GDT that the TCA augmentation effect seen in this study can be extended to other TCAs as a class-effect. Therefore, the GDT elected to limit the recommendation to desipramine because it had been specifically studied. The recommendation is labeled consensus because of the limitations in the Fava study discussed above. Combined use of SSRIs and higher doses of TCAs increases the risk of TCA toxicity, primarily manifested as cardiac arrhythmia. Therefore, only low doses (< 50 mg) of TCAs should be used, and the GDT recommends careful monitoring for symptoms of TCA toxicity. Most primary care physicians do not prescribe lithium, due to complexities in dosage and laboratory monitoring. Some primary care physicians may be comfortable prescribing lithium, but in general the GDT believes that consultation with psychiatry should occur when this treatment option is being considered. Group Health Cooperative s 2003 Adult Major Depression Guidelines included three additional studies from a Medline search using similar mesh terms that did not appear in our PubMed search: Bauer, (139) Appelberg, (140) and Thase. (141) Group Health concluded: Bauer found that lithium augmentation was superior to placebo for nonresponders to antidepressants. The majority of the studies included initial treatment with TCAs and findings may be less applicable to SSRI treatment. Appelberg did not find a significant difference between augmentation with buspirone or placebo among patients initially treated with SSRIs. In the Appelberg study, 33% of patients responded to buspirone augmentation and 31% responded to placebo augmentation the Appelberg RCT may have been underpowered a substantial proportion of patients responded to placebo augmentation, so it is difficult to draw conclusions from studies that do not include a placebo group. Thase conducted an RCT in which patients who did not respond to initial treatment with SSRIs or TCAs switched to the other class of medication. There was no significant difference in outcomes between groups after switching. A substantial proportion of patients in each group experienced a remission after switching, 32% in the imipramine to sertraline group and 55% in the sertraline to imipramine group. Despite the presence of observational studies and case reports, we did not find any RCTs or quasi-experimental trials evaluating the use of Cytomel, carbamazepine, valproic acid, or methylphenidate for augmentation of antidepressants. Clinical Evidence did not examine this question. Overall Conclusion Increasing the dose of the initial antidepressant; adding low-dose desipramine to an SSRI; switching antidepressants; or adding lithium 300 to 600 mg/day are potential treatment options for patients with MDD whose symptoms do not remit after initial treatment. Due to complexities in management, most primary care clinicians will not prescribe lithium without psychiatric consultation or assistance, so the GDT added consultation with psychiatry to this consensus option to reflect real world practice. Due to methodological limitations, including lack of control groups simultaneously maintained on pre-existing treatment, the evidence is not as strong as the evidence for combining psychotherapy and antidepressants, so the GDT has labeled these recommendations consensusbased. 62 National Adult Depression Clincial Practice Guideline

67 5. Length Of Treatment With Antidepressants In Patients With MDD Patients Who Achieve Symptom Remission 5A The GDT recommends that patients with MDD who achieve symptom remission with antidepressants should continue antidepressants at the same dose for at least an additional six to 12 months. Evidence-based Patients With One Lifetime Episode of MDD 5B Based on patient and provider preference, the GDT recommends that a trial of antidepressant discontinuation is optional for patients in their first lifetime episode of MDD, who are being treated with antidepressants, achieve remission, and remain asymptomatic for six to 12 months after acute phase treatment. Consensus-based Patients with Two or More Lifetime Episodes of MDD 5C The GDT recommends that patients with two or more lifetime episodes of MDD, who are being treated with antidepressants and remain asymptomatic after acute phase treatment should be maintained on the medication and dose with which they achieved remission for at least an additional 15 months to five years after acute phase treatment. Consensus-based Patients with Chronic MDD or MDD with Concurrent Dysthymia* 5D The GDT recommends that patients with chronic MDD (continual symptoms for more than two years) or Double Depression (MDD and dysthymia) who improve with antidepressants during acute phase treatment should continue antidepressants for at least an additional 15 to 28 months after acute phase treatment. Evidence-based Rationale: 2008 Guideline No new evidence was found, the recommendations remain unchanged Guideline Our search yielded three RCTs that examined depression relapse or recurrence rates in an extended continuation phase with antidepressants: Keller, et al. (2005) (143) found that over a 44 week continuation phase, there was a significantly lower percentage of relapse in gepirone ER treated patients compared with those treated with placebo. Montgomery, et al. (2004) (144) found that over a 52 week continuation phase, the cumulative probability of recurrence was significantly lower in patients treated with venlafaxine versus those treated with placebo. * Dysthymia = Depressed mood plus at least two additional DSM-IV (142) symptoms present more days than not for at least two years. 63 National Adult Depression Clincial Practice Guideline

68 Rapaport, Bose & Zheng (2004) (145) found that over a 36 week continuation phase, the cumulative rate of relapse was significantly lower in patients treated with escitalopram versus those treated with placebo. Clinical Evidence found one systematic review that underscored the importance of the continuation phase. Patients continuing on a course of antidepressants after recovery had a reduced risk of relapse over one to three years. This effect was independent of duration of initial treatment, duration of previous antidepressant treatment, or underlying risk of relapse. The review found that, overall, continuing antidepressant drugs in people who had responded to them significantly reduced the proportion of people who relapsed compared with placebo (31 RCTs, 4,410 people with first episode or recurrent depression; numbers relapsing: 465/2,527 [18%] with continuing antidepressants vs. 1,031/2,505 [41%] with placebo; OR = 0.30, 95% CI: 0.22 to 0.38). (146) In their systematic review of RCTs, the National Institute for Clinical Excellence (NICE) (57) depression guideline makes an evidence-based recommendation that patients with moderate to severe depression should continue antidepressants for at least six months after remission. This conclusion is based primarily on the findings of two studies in which a greater percentage of patients on SSRIs were able to complete six months of treatment than those taking TCAs. An RCT conducted in the USA randomized 536 adults to receive desipramine, imipramine, or fluoxetine (Simon, et al., 1996). (147) Sixty percent of the fluoxetine patients completed six months of treatment compared with less than 40% of the TCA patients. Those who discontinued one antidepressant were offered another. There were no differences in overall completers or response rates at endpoint suggesting that initial drug choice did not affect outcome. However, outside of clinical trials, patients may not return to their general practitioner to have their treatment changed and outcome may be less positive. For example, a Swedish study of 949 patients found that 35% only ever received one prescription irrespective of whether it was for a TCA or a SSRI (Isacsson, et al., 1999). (148) After six months, 42% of SSRI patients were still receiving prescriptions compared with 27% of TCA patients. There is some evidence from this study that the relapse rate may have been higher in the TCA group: 28% of TCA treated patients received a subsequent prescription for an antidepressant after a nine-month treatment-free gap compared with 10% of SSRI patients. Overall Conclusion Evidence consistently shows that continuing antidepressants after remission of symptoms of MDD reduces the risk of relapse. This new evidence, in combination with some of the evidence reviewed previously, suggests that a minimum of six months of antidepressant continuation after acute phase treatment and symptom remission may be more appropriate than the previously recommended four month minimum continuation phase. Consistent with all the previously reviewed evidence, these new studies support continuation at the same antidepressant dose that was used to achieve remission. Therefore, the GDT recommends continuation phase treatment should last at least six to 12 months following acute phase antidepressant treatment of MDD. 64 National Adult Depression Clincial Practice Guideline

69 2004 Guideline How Long Should Patients With MDD Who Have Responded To Antidepressant Medication Continue Taking Medications Beyond The Acute Phase Of Treatment? Acute Phase: Up to three months after starting treatment (period to assess for response to treatment/remission.) Continuation Phase: Next four to 12 months. Maintenance Phase: Treatment beyond continuation phase. Supporting Evidence for Minimal Length of Treatment With Antidepressants in Patients With MDD after Acute Phase Symptom Remission Two new RCTs were found that addressed this issue. Weihs (149) studied men and women at least 18 years old with moderate to severe, recurrent Major Depression based on Diagnostic and Statistical Manual-IV (DSM-IV) criteria who had a minimum score of 18 on the 21-item Hamilton Depression Scale (HAMD). The current depressive episode must have been preceded by at least one other episode within the last 60 months. All patients (n = 816) were received open label treatment for eight weeks (acute phase), with bupropion SR (150 to 300 mg/day). Patients who responded to the treatment and continued to meet the selection criteria entered a 44-week (11 months) randomized, double-blinded, placebo controlled continuation phase with bupropion SR, 300 mg/day (n = 207) or placebo (n = 210). The study showed that treatment with bupropion SR for up to 44 weeks decreases the risk of depression relapse. Time to depression relapse (time from randomization to intervention) was 44 weeks for bupropion SR treated patients and 24 weeks for the placebo group (p = 0.003, favoring continuation treatment). Survival estimates indicated 52% of placebo treated patients and 37% of bupropion patients would have become depressed by the end of the study (p = favoring bupropion). By the end of one year treatment, the odds of placebo-treated patients requiring treatment intervention for a relapse of depression were 1.83 times greater that those of bupropion SR-treated patients (confidence interval not specified). 65 National Adult Depression Clincial Practice Guideline

70 Reimherr (150) studied male and female outpatients 18 to 65 years of age that met DSM-III-R criteria for Major Depression with a duration of at least one month and had a modified 17-item Hamilton Depression Rating Scale score of at least 16. After a five to nine day medication-free baseline phase, all patients (n = 839), received 12 to 14 weeks of open-label acute therapy phase with fluoxetine, 20 mg/day. The 395 remaining patients were then randomized into one of four double-blind continuation treatment groups: 14 weeks of continuation therapy with fluoxetine followed by 38 weeks of placebo (n = 97), 38 weeks of continuation therapy with fluoxetine followed by 14 weeks of placebo (n = 100), 50 weeks of continuation therapy with fluoxetine (n = 102), or 50 weeks of placebo (n = 96). Relapse rates were calculated during 12 week periods after each double-blind transfer from fluoxetine to placebo (weeks 12, 26, and 50). Relapse rate (Kaplan-Meier estimates) were 26.4% for the fluoxetine-treated group and 48.6 for the placebo group after 12 weeks of continuation treatment (p < 0.001, favoring continuation treatment), and 9.0% for the fluoxetine-treated group and 23.2% for the placebo group after 26 weeks of continuation treatment (p < 0.001, favoring continuation treatment). There was no statistical difference (p = 0.54) in the relapse rate between the fluoxetine group (10.7%) and placebo (16.2%) after 50 weeks of continuation treatment (possibly due to lack of statistical power due to patient attrition). The British Medical Journal (151) review consisted of six RCTs with a total sample size of 312 patients on current treatment with antidepressant medication. Findings indicated that continuation of antidepressant medication four to six months after acute phase treatment reduced the relapse rate by nearly half. The authors also pointed out that several more RCTs confirm the reduction in risk of early relapse with continuing medication for six to 12 months after acute phase treatment. Based on this evidence, the GDT recommends that depressed patients treated with antidepressants who are asymptomatic at the end of the acute phase (three months) of MDD treatment be maintained on antidepressants for an additional four to 12 months, to decrease the risk of early symptom relapse. Clinical Evidence, (70) Volume 9, included the above studies in their review. Their conclusions are consistent with ours. 66 National Adult Depression Clincial Practice Guideline

71 Supporting Evidence For A Trial Discontinuation For Patients With One Lifetime Episode of MDD No new evidence was found. A published review (152) notes that patients in their first episode of MDD have a lifetime risk of relapse of approximately 50%. Viguera (153) examined data from 27 randomized trials to determine the effects of discontinuation on relapse rates. Of these, two trials included only patients in their first episode of depression. In both of these trials, amitriptyline was the anti-depressant used, one trial included combination amitriptyline and lithium. Both trials were published before This review did not separately analyze the data in these two trials from five other trials whose inclusion criteria included one or more lifetime episodes of depression. Among patients with one or more past episodes, the corresponding two year survival rates (symptom-free at two years) were not statistically different: 64.1% with treatment (95% CI: 58.7% to 69.4%) versus 52.7% without (95% CI: 42.9% to 62.6%). The relative weighting of patients with only one lifetime episode vs. one or more lifetime episodes in this data is unknown. No additional trials were identified that specifically examined the risk of relapse in patients with one lifetime episode of MDD. Based on the paucity of data, and recognizing that some patients might choose, based upon their individual circumstances, to continue antidepressant treatment if made aware of the 50% lifetime risk of recurrence, the GDT has labeled this an option for patients. Because the GDT was unable to identify clean data in the systematic review that adequately addresses this question, this recommendation has been labeled as consensus-based. Supporting Evidence For Maintenance of Medication For Patients With Two or More Lifetime Episodes of MDD Two new RCTs were found that addressed this issue. Hochstrasser (154) studied mental health in- and out-patients 18 to 65 years of age with recurrent unipolar Major Depression (DSM-IV), a Montgomery Asberg Depression Rating Scale (MADRS) score of 22 and two or more previous depressive episodes, one within the past five years. Four hundred, twenty-seven patients entered acute phase treatment with citalopram, 20 to 60 mg/day. Patients who responded after six to nine weeks (n = 327) continued citalopram for an additional 16 weeks (continuation phase). Patients whose depressive symptoms remained in remission (MADRS score < 11, n = 269) were then randomized in a double-blind fashion to maintenance treatment with citalopram (n = 132) versus placebo (n = 132) for an additional 48 to 77 weeks. Data from five patients initially assigned to the placebo group was excluded due to major protocol violations. Time to recurrence was longer in patients taking citalopram than in patients taking placebo (p < 0.001), with crude rates of recurrence of 0.22 recurrences /person-year at risk in the citalopram vs recurrences/person-year at risk in the placebo group. 67 National Adult Depression Clincial Practice Guideline

72 Gilaberte (155) studied male and female outpatients, 18 to 65 years of age, who met DSM-III-R criteria for unipolar Major Depression, had at least one previous major depressive episode in the last five years, a score of at least 18 on the 17-item Hamilton Rating Scale for depression (HAM-D-17) and at least four on the Clinical Global Impression (CGI) severity scale. All patients (n = 253) completed 32 weeks of open-label treatment phase consisting of both an acute (eight weeks) and a continuation period (six months) before being randomized to a double-blind maintenance phase with fluoxetine, 20 mg/day (n = 70) or placebo (n = 70) for 48 weeks. The recurrence rate was 20% for the fluoxetine-treated group compared with 40% for placebo (p = 0.010, favoring maintenance fluoxetine treatment). The symptom-free period was significantly longer for patients treated with fluoxetine vs. placebo (295 days vs. 192 days; Kaplan-Meier estimates, log-rank test, p = 0.002). In the Viguera (153) analysis, a total of 3,037 depressed patients were treated for 5.78 (0 to 48) months and then followed for a mean of 16.6 (5 to 66) months with antidepressants either continued or discontinued. Compared with patients whose antidepressants were discontinued, those with continued treatment showed much lower relapse rates (1.85 versus 6.24% per month), longer time to 50% relapse (48.0 vs months), and lower 12 month relapse risk (19.5 versus 44.8%) (all p < 0.001). A survival analysis presented in the Viguera article shows a statistically significant difference in symptom remission between patients continuing medication and patients not continuing medication, which persisted through 54 months of follow-up. Among patients with more than two or more past episodes, the corresponding one year survival rates (symptom-free at one year) were: 79.5% with treatment (95% CI: 73.2% to 85.8%) versus 56.5% without treatment (95% CI: 48.9% to 64.2%). Viguera did not differentiate between patients in their second lifetime episode of MDD and those with more than two lifetime episodes. In patients with at least three past episodes or a chronic course, the two year survival rate (no recurrence of depressive symptoms) was 71.7% (95% CI: 64.6% to 78.9%) with antidepressant versus 14.7% (95% CI: 7.10% to 22.2%) without antidepressant - a highly significant 4.88 fold difference. Viguera did not differentiate between patients with and without psychiatric comorbidities or history of suicide attempt. Frank (156) found that active imipramine at a dose of 200 mg per day is effective at maintaining remission for three years in patients with three or more lifetime episodes of MDD. Kupfer (157) followed a smaller subset of patients from the Frank study for an additional two years and found that imipramine 200 mg per day continued to effectively prevent recurrence of symptoms for up to five years. There is no patient outcomes data available to demonstrate the effect on patients maintained on medications for more than five years. The studies examined all include maintenance therapy with antidepressants at the same or higher dose that achieved symptom remission or resolution. No evidence was found examining the effect of different dosage strategies, or changing medication, on disease-free interval. 68 National Adult Depression Clincial Practice Guideline

73 A recently published review (152) notes that patients with three or more episodes of MDD have a lifetime risk of relapse of approximately 90%. Patients with two prior lifetime episodes of depression have a lifetime risk of relapse of 70%, intermediate to those with one prior episode and those with three or more prior episodes. No additional studies were found that examined outcomes of maintenance therapy specifically for patients with two past episodes of Major Depression. Conclusion Our systematic review did not identify any RCTs or meta-analyses on length of antidepressant treatment that clearly distinguished between patients with a history of two lifetime episodes vs. three or more lifetime episodes of MDD. There is insufficient clean evidence to make different recommendations for patients with two lifetime episodes of MDD and patients with three or more lifetime episodes of MDD. The data, in aggregate, suggest benefit to long-term continuation of antidepressants in patients with at least two lifetime episodes of MDD who respond to antidepressant treatment. The optimal duration of treatment is not clear from available data, but it appears that a minimum of 15 months and in some cases as long as five years of treatment after the acute phase response demonstrates benefit. It is possible that longerterm treatment (beyond five years) also provides continued benefit, but studies have not been conducted to examine the benefits of longer-term antidepressant treatment compared with discontinuation after five years. Because of these limitations, the GDT has labeled this recommendation consensus-based. Supporting Evidence For Patients With Chronic MDD or Double Depression to Continue Medication Two new RCTs were found that this addressed this issue. Kocsis (158) studied out patients who met DSM-III diagnostic criteria for pure dysthymia (40%), chronic Major Depression (10%), and dysthymia with current Major Depression ( double depression ) (50%). All patients were treated with desipramine 50 to 325 mg/d during the acute phase (ten weeks) and continuation phase (16 weeks = four months). Patients who responded to treatment were then randomized to double-blind maintenance treatment with desipramine (n = 28) or placebo (n = 25) for 24 months. 15% of patients receiving long-term maintenance treatment with desipramine experienced a relapse, compared with 52% of patients receiving placebo (ARR = 37%, NNT = 3, p = 0.01). Fourty percent of patients studied did not have MDD (they had dysthymia only), although response in each patient subgroup did not differ in either the acute, continuation, or maintenance phases. Keller (159) studied outpatients meeting a DSM-III-R structured clinical interview diagnosis of chronic Major Depression (symptoms for at least two years) or dysthymic disorder with Major Depression (double depression) and a minimum baseline severity of 18 on the 24-item Hamilton Depression Scale (HAM-D). All patients (n = 253) completed 12 weeks of treatment (acute phase) and four months of continuation phase treatment with sertraline 50 to 200 mg/day. Responders were then randomized to 48 weeks of double-blind maintenance phase with sertraline (n = 77) or placebo (n = 84). Sertraline treated patients had significantly fewer recurrences than placebo treated patients (6% vs. 23%), p = for the log-rank test of time-to-recurrence distributions. Clinically significant depressive symptoms reemerged in 26% of patients treated with sertraline versus 50% in the placebo group (p = 0.001). 69 National Adult Depression Clincial Practice Guideline

74 The available evidence suggests benefit to long-term continuation of antidepressants in patients with chronic MDD or double depression (MDD with concurrent dysthymia) who respond to initial acute phase antidepressant treatment. The optimal duration of treatment is not clear from available data, but it appears that a minimum of 15 to 28 months of treatment after acute phase response demonstrates benefit. It is possible that longer-term treatment also provides continued benefit, but studies have not been conducted to examine the benefits of longer-term antidepressant treatment compared with discontinuation after 28 months. 6. Follow-Up For Patients In The Acute Phase (First Three Months) of Treatment For MDD 6 For patients who are starting treatment with antidepressants for Major Depression, the GDT recommends that the minimum recommended follow-up frequency be one patient contact within the first month, and at least one additional patient contact four to eight weeks after the first contact. Assess for adherence, side effects, suicidal ideation, and patient response during both these visits. Consensus-based Rationale: Note: HEDIS requires three follow-up contacts in the first 12 weeks of treatment with antidepressants, one of which must be with a prescribing clinician. HEDIS allows one of the three follow-up contacts to be by telephone; currently the other two contacts must be in-person visits. The HEDIS requirements are not evidence-based Guideline No new evidence was found, the recommendation remains unchanged Guideline We found one RCT examining an intervention for patients in the first three months (acute phase) of treatment for MDD: Swindle, et al. (2003) (160) tested whether the services of a Clinical Nurse Specialist (CNS) would improve symptoms of primary care patients receiving treatment for depression or dysthymia. The CNS coordinated a treatment plan including medication, cognitive behavioral therapy, referral to Mental Health, and monitoring through telephone and/or in-person visits. Monitoring occurred at two weeks, one month, and two months after the initial visit. The authors found no significant difference in depression symptoms, patient satisfaction, or rates of antidepressant prescribing or adequate dosing between patients with CNS services versus those without. Patients in the intervention group did have significantly more recorded depression diagnoses and referrals to mental health in the medical record. Specific data on the follow-up frequency of patients in the usual care group were not provided. However, CNS s often chose to use watchful waiting with patients in the intervention group (disagreeing clinically with depression scale diagnoses), thus affecting the fidelity of the intervention. 70 National Adult Depression Clincial Practice Guideline

75 Overall Conclusion The Swindle, et al. study, due to reasons outlined above, adds no new information that helps determine a follow-up schedule for newly diagnosed Major Depression patients. Given the lack of additional information, the previous consensus-based recommendations still seem valid. However, recognizing the discrepancy between this recommendation and the HEDIS guidelines (which, albeit not evidence-based, are a standard on which health plans are judged) the GDT suggests clarifying the previous recommendation, by adding that the MINIMUM recommended follow-up frequency for patients who are starting treatment with antidepressants for Major Depression is one patient contact within the first month, and at least one additional patient contact four to eight weeks after the first contact. Research on this topic does not differentiate between modalities for follow-up (in-person visits, phone calls, messages). The GDT believes that the active ingredient is the follow-up per se, and not the modality of follow-up; therefore we recommend that any of these types of contacts are acceptable Guideline Supporting Evidence For Follow-Up Frequency in the Acute Phase Treatment of MDD There is no evidence that determines the optimal number or frequency of follow-up for adults in the acute phase (first 12 weeks of treatment) of Major Depression. There is no evidence that suggests that in-person visits during acute phase follow-up are superior to phone contacts. Despite the lack of evidence, HEDIS requires three follow-up contacts in the first 12 weeks of treatment with antidepressants, one of which must be with a prescribing clinician. HEDIS allows one of the three follow-up contacts to be by telephone, but currently the other two contacts are defined as in-person visits. The GDT recommends targeting at least one contact within the first month and at least one additional contact four to eight weeks thereafter to evaluate treatment adherence, symptom response, potential side effects, suicidal ideation, and the need for treatment adjustment. This recommendation is based on medication discontinuation patterns seen in short-term trials of antidepressants (where patient dropouts tend to occur within the first month or between the second and third month of treatment), and the clinical experience and consensus of the GDT. Additional contacts may be needed for patients who are not tolerating or responding to treatment. There is no evidence that patients who are tolerating and responding to treatment will achieve additional benefit with three contacts (as HEDIS suggests) as opposed to two contacts as outlined above (meaning extra patient and clinician time and resources may be utilized in some cases for no appreciable added benefit). Since there is no evidence that directly addresses this question, the GDT decided to base its recommendation on indirect evidence from the literature, recognizing that this recommendation would also be consensus-based. 71 National Adult Depression Clincial Practice Guideline

76 The GDT recognizes that this recommendation may be viewed by some as an opposition to the current HEDIS standard (which is based solely on expert opinion without reference [that we found] to any systematic literature review to support it). However, by specifying a minimum number of visits in the Depression Guidelines; clinicians, teams, and regions still have the option to base performance benchmarks on the HEDIS criteria. The GDT believes also that antidepressant continuation rates (also a HEDIS measure and also addressed in this guideline) are founded on better evidence and are better proxies for processes and outcomes of care. The recommendation for follow-up, with at least some basis in evidence (however indirect), provides flexibility for clinicians and teams to focus on and address other aspects of quality depression care, while also providing a prompt for teams to examine their care processes if the number of contacts falls below this minimum recommendation. In March, 2004, the FDA issued a directive that the ten most commonly used antidepressants (fluoxetine, paroxetine, sertraline, bupropion, citalopram, fluvoxamine, mirtazapine, nefazodone, escitalopram, and venlafaxine) carry a warning that thoughts about suicide sometimes occur in people taking antidepressants. The findings have been noted in antidepressant studies of children and adolescents. Even though to date there have not been specific concerns raised about depressed adults being treated with antidepressants in primary care settings, the FDA chose to issue the directive for all patients as a precautionary measure. The essence of the FDA direction is to advise patients of the possibility that thoughts about suicide may arise, particularly at the onset of treatment and with changes in treatment (dosage and/or medication changes), and to encourage patients to inform clinicians immediately if these thoughts occur. Regardless of the treatment option selected, the GDT believes that primary care patients should be screened for suicidal ideation, intent, or plan at the diagnosis of Major Depression, with each follow-up visit for depression, and with any adjustments in treatment. Consultation with mental health should occur as indicated. 7. Follow-Up For Patients In The Continuation Phase (Months Four To 12) of Treatment of MDD 7 After achieving symptom remission, the GDT recommends at least one follow-up contact * during the fifth or sixth month of treatment in patients with Major Depression. Assess for continuing symptom remission and dosage/treatment adjustment during this contact. The GDT recommends additional patient follow-up to consider either continuing treatment beyond the continuation phase, or attempting a trial of treatment discontinuation. Consensus-based * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. The use of between patients and providers is relatively new, and has not been a widely utilized means of communication to date. However, it is being increasingly advocated as part of a patient-centered, more efficient ( less visit dependent ) model of care. At least one member of the GDT uses this modality regularly and deems it effective for follow-up contacts. 72 National Adult Depression Clincial Practice Guideline

77 Rationale: 2008 Guideline No new evidence was found, the recommendation remains unchanged Guideline No new evidence was found. Research on this topic does not differentiate between modalities for follow-up (in-person visits, phone calls, messages). The GDT believes that the active ingredient is the follow-up per se, and not the modality of follow-up; therefore recommends that any of these types of contacts are acceptable Guideline Supporting Evidence for Follow-Up Frequency In The Continuation Phase Treatment of MDD Clinical Evidence (70) identified one RCT that addressed patient follow-up during the continuation phase of MDD treatment. Three hundred eighty-six people aged 19 and older, with recurrent Major Depression or dysthymia, who had largely recovered after eight weeks of antidepressant treatment were randomized to a relapse prevention program (two primary care visits and three telephone calls) versus usual care for one year. Patients in relapse prevention had significantly improved depressive symptoms over one year (p = 0.04), but no significant difference in relapse rates (35% in both groups). While this study supports follow-up after symptom remission, we found no studies comparing different numbers or appropriate timing of follow-up contacts for patients with MDD. Furthermore, this study also included patients with dysthymia, and did not provide separate outcome data for patients with Major Depression only. The Depression Outcomes Report V (161) reports show that adherence with antidepressant treatment continues to decline between the third and sixth month of treatment. Premature discontinuation of antidepressants is associated with sub-optimal patient outcomes (see recommendations and rationale on length of treatment with antidepressants). Conclusion We found only one RCT, conducted in a mixed population (MDD or dysthymia), that addressed follow-up during continuation phase treatment of MDD. We found no studies addressing different numbers or timing of follow-up contacts for patients in the continuation phase of MDD. The optimal number and timing of follow-up contacts is therefore undetermined. Adopting a strategy of five follow-up contacts, as suggested by the Clinical Evidence review, would have major impact on patient convenience, patient expense, clinician time, and access to care. 73 National Adult Depression Clincial Practice Guideline

78 Nevertheless, the Clinical Evidence study, our review of evidence on duration of treatment and our own observational evidence from the Depression Outcomes report suggest that some level of follow-up is beneficial. Due to these considerations, the GDT believes (by consensus) that a minimum of one follow-up contact in the fifth or sixth month of treatment should be made to monitor for symptom relapse, to adjust treatment, and to stress ongoing treatment adherence. This would potentially capture most of the patients who have discontinued antidepressants prematurely. Some patients will be eligible for, or desire, discontinuation of antidepressant treatment after successful continuation phase treatment. Therefore, additional patient follow-up should occur as the decision is made to either continue treatment past the continuation phase, or attempt a trial of treatment discontinuation. There is no evidence to support this recommendation, which is based upon usual clinical care and consensus of the GDT. 8. Follow-Up For Patients In Maintenance Phase (Beyond 12 Months) of Treatment of MDD 8A 8B Rationale: For asymptomatic patients with Major Depression who are continuing on antidepressants beyond 12 months, the GDT recommends at least one annual follow-up contact to assess for continuing symptom remission, the need for ongoing treatment, and dosage/treatment adjustment. Consensus-based The GDT recommends that additional follow-up for asymptomatic patients with Major Depression who are continuing on antidepressants beyond 12 months should be based on patient preference and response. Consensus-based 2008 Guideline No new evidence was found, the recommendations remain unchanged Guideline No new evidence was found. Research on this topic does not differentiate between modalities for follow-up (in-person visits, phone calls, messages). The GDT believes that the active ingredient is the follow-up per se, and not the modality of follow-up; and therefore recommends that any of these types of contacts are acceptable Guideline Supporting Evidence for Follow-Up Frequency in the Maintenance Phase Treatment of MDD There is no evidence that determines optimal number or frequency of follow-up for asymptomatic patients on long-term antidepressant treatment for Major Depression. The GDT believes that a minimum of one annual follow-up contact should be made with patients on maintenance phase treatment, to assess for ongoing symptom control and potential treatment adjustment. 74 National Adult Depression Clincial Practice Guideline

79 9. Discontinuation of Antidepressants in Patients with MDD 9A 9B Fluoxetine may be discontinued, without tapering, with a relatively low risk of adverse effects. Evidence-based The GDT recommends tapering other antidepressants (other SSRIs, TCAs, SNRIs, NRIs, and DAs) over a two to four week period. Consensus-based Rationale: 2008 Guideline No new evidence was found, the recommendations remain unchanged Guideline Our search yielded one RCT addressing the issue of discontinuing antidepressants: Sunder, et al. (2004) (162) designed a study to differentiate between symptoms of antidepressant discontinuation vs. depression relapse in women treated with sertraline or placebo for postpartum depression. Both sertraline and placebo were tapered over three weeks, starting at week 18. The authors found that a three week taper did not lead to emergence of symptoms typical of SSRI antidepressant withdrawal. The study was not designed to compare a three week taper with a more rapid discontinuation schedule. We found no further studies addressing methods of discontinuing fluoxetine. Overall Conclusion Although the Sunder study is limited to postpartum women, it is consistent with earlier studies suggesting that antidepressants with short half lives (such as sertraline) should be tapered. Therefore, the previous recommendation that fluoxetine may be discontinued without tapering with a relatively low risk of adverse effects, while other antidepressants should be tapered over a two to four week period, is still valid Guideline One new study was found since the previous revision of the Depression Guidelines that addressed this issue. Judge (163) studied outpatients aged 18 with a diagnosis of unipolar depressive disorder successfully treated with fluoxetine or paroxetine whose Montgomery-Asberg Depression Rating Scale (MADRS) score was 12 or less (indicating symptom remission). Patients (n = 150) received a blinded drug equivalent to their current daily maintenance dose (fluoxetine 20 to 60 mg or paroxetine 20 to 50 mg) at visit one. Patients were then randomized to two different treatment groups comparing fluoxetine (n = 75) and paroxetine (n = 75) with periods of treatment interruptions lasting three to five days in each group. Mean change following treatment interruption for Discontinuation Emergent Signs and Symptoms (DESS) was 0.2 for the fluoxetine group and 2.2 for the paroxetine group (p = favoring fluoxetine-treated patients). Mean change in the MADRS was 0.2 for fluoxetine patients and 1.8 for paroxetine patients (p = 0.021). 75 National Adult Depression Clincial Practice Guideline

80 Paroxetine-treated patients also had significantly greater increases in Clinical Global Impressions Severity (CGI-Severity scores) than did fluoxetine-treated patients following treatment interruption. The deterioration from baseline within the paroxetine treatment group was also significant for both MADRS (p = 0.010) and CGI scores (p = 0.001). This study did not evaluate for discontinuation side effects beginning after five days of treatment interruption, when side effects might be expected to appear with a long half-life antidepressant such as fluoxetine. Two studies were found both of which looked at the effects of discontinuing SSRIs. Zajecka (164) compared fluoxetine (varying treatment lengths) with placebo in 395 patients (274 women, 121 men, mean age ± SD = 40 ± 10 years) who were randomly assigned to continuation treatment with fluoxetine (n = 299) or placebo (n = 96). Subjects were men and women who initially met DSM-III-R (73) criteria for Major Depression and had a 17-item HAM-D score 16 and whose depressive symptoms significantly improved during the Acute Phase of a multicenter examination of fluoxetine in the maintenance treatment of depression. (Improvement was defined as a HAM-D 17 score 7 after 12 weeks of acute treatment with fluoxetine 20 mg daily). Upon completion of this acute phase of treatment, subjects were assigned by random allocation to double-blind placebo (n = 96) or to one of three arms of ongoing active treatment with fluoxetine 20 mg daily (n = 299) for various periods of time. (These three arms were pooled for analysis purposes in this study.) Fluoxetine treatment was discontinued without a tapering-off period in patients assigned to placebo. Patients were seen at weeks one, two, four, and six after randomization. Reports of new or worsened adverse events were similar for both groups at each visit after randomization. The authors concluded that abrupt discontinuation of fluoxetine was well-tolerated. Rosenbaum (165) compared discontinuation syndromes in patients with a history of unipolar (Major) Depressive Disorder on sertraline, paroxetine, and fluoxetine as compared with placebo in 231 patients. Fluoxetine had the least problems when discontinued, followed by sertraline and paroxetine. No studies were found since the previous revision of the Depression Guidelines comparing discontinuation of higher doses of fluoxetine (40 to 60 mg/day) with lower doses (10 to 20 mg/day). All studies found included lower and higher dose fluoxetine patients in the same treatment group. It takes approximately 5.5 half lives for medications to be cleared from the bloodstream. Medications which short half lives (and thus more rapid clearance) are more prone to produce symptoms upon discontinuation than medications with longer half-lives. The half-life of fluoxetine is 96 hours, sertraline is 24 hours, and paroxetine is 17 to 22 hours. 76 National Adult Depression Clincial Practice Guideline

81 No evidence was found for tapering TCAs. The two to four week time period recommended by the GDT is consistent with clearance of these agents, based on their half-lives. Conclusion There is some evidence that fluoxetine can be discontinued without a high risk of many adverse effects. This is consistent with its known long half-life. Extrapolating from the Rosenberg study and principals of pharmacology, SSRIs with shorter half-lives should be tapered, rather than discontinued. There is no evidence that determines the optimal tapering regimen. The two to four week time period recommended by the GDT is consistent with clearance of these agents, based on their half-lives. Pharmacotherapy of Antidepressant Medications: Half-life (Active Metabolite) (166) Medication Half-Life in Hours (Active Metabolite) SSRI Fluoxetine (Prozac) (146) Sertraline (Zoloft) 26 (66) Paroxetine (Paxil) 24 Citalopram (Celexa) 33 TCAs Desipramine (Norpramin) 12 to 28 Nortriptyline (Pamelor) 18 to 56 Amitriptyline (Elavil) 9-46 (18-56) Imipramine (Toframil) 6-28 (12-28) Doxepin (Sinequan) 11 to 23 Others Bupropion (Wellbutrin) 10 to 21 Venlafaxine (Effexor) 4 (10) Nefazodone (Serzone) 4-5 (4-18) Mirtazapine (Remeron) 20 to Treatment Preferences For MDD In Different Ethnic Groups 10 Because patient preferences for treatment may vary based on their ethnicity and culture, the GDT recommends asking patients from different ethnic groups about treatment preference, when discussing treatment options for MDD. Evidence-based Rationale: 2008 Guideline No new evidence was found, the recommendation remains unchanged. 77 National Adult Depression Clincial Practice Guideline

82 2006 Guideline We found one study that addressed the preferences of one particular ethnic group. Dwight-Johnson, et al. (2004) (167) used conjoint analyses to assess the preferences of low income Latinos. They found that this population preferred individual over group treatment, combination therapy over counseling or medication alone, and all treatment settings equally. Overall Conclusion The GDT believes that the Dwight-Johnson study is consistent with our previous analysis and recommendation, which remains that patients from different ethnic groups should be asked about their treatment preferences for MDD Guideline Supporting Evidence for Treatment Preferences for MDD in Different Ethnic Groups PubMed (1/1/2001 through 4/1/2003) and PsychInfo (1985 through 4/1/2003) databases were searched and no new evidence on this topic was found since the previous revision of the Depression Guidelines. One article was found from another search that did not show up in the above search due to indexing issues in Pub-Med. Cooper, et al. (168) conducted a telephone survey of 829 adult patients (659 non-hispanic whites, 97 African Americans, 73 Hispanics) recruited from primary care offices across the United States who reported one week or more of depressed mood or loss of interest within the past month and who met criteria for major depressive episode in the past year. Within this cohort, they examined differences among African Americans, Hispanics, and non-hispanic whites in acceptability of antidepressant medication and acceptability of individual counseling. African Americans (adjusted OR = 0.30; 95% CI: 0.19 to 0.48) and Hispanics (adjusted OR = 0.44; 95% CI: 0.26 to 0.76) had lower acceptance of antidepressant medications compared with Whites. African Americans had somewhat lower odds (adjusted OR = 0.63; 95% CI: 0.35 to 1.12), and Hispanics had higher odds (adjusted OR = 3.26; 95% CI: 1.08 to 9.89) of finding counseling acceptable than non-hispanic whites. Racial and ethnic differences on preferences for treatment modalities were found. Clinicians should consider patients' cultural and social context when negotiating treatment decisions for depression. The AHRQ (169) report did not find any data on patient preference and ethnic background. 78 National Adult Depression Clincial Practice Guideline

83 Two studies were found previously on this topic. Dwight-Johnson's (167) longitudinal study surveyed 1,187 English and Spanish speaking patients regarding their treatment preferences for depression. Nearly all had MDD. Patients were given five treatment options to choose from; including free medication for six months; individual or group counseling; and watchful waiting. Eighty-three percent of the total sample reported wanting active treatment for depression; of these, 27% preferred antidepressant medication, 29% individual counseling, and 26% group counseling. Patient characteristics of those who chose counseling included being African American and having a greater knowledge of counseling. It is important to note that treatment preference may have been influenced by cost (total cost for medication was $480, individual counseling $300, and group counseling $75). Patients were not necessarily actively diagnosed with depression at the time of the survey but had to have experienced a two-week period of depression over the prior two-year period. Chilvers (170) examined treatment preference in 323 general practice MDD patients, patients could either choose counseling or medication, or agree to be randomized into either treatment arm. The study did not analyze results by ethnic group. The study concluded that patients who chose their care did better than those who were randomized into treatment. Also, patients taking antidepressant medication got better faster than patients receiving counseling. Consult the National Diversity Handbooks on culturally competent care approaches in Latinos; Asian/Pacific Islanders; African American; and Lesbian, Bisexual, Gay, and Transgender groups. To order, please contact Kaiser Permanente s National Diversity Institute for Culturally Competent Care. All orders for the products must be submitted in writing. Download the order form at this website: and fax to (510) Overall Conclusion Patients from different cultural/ethnic backgrounds may have different understandings of and frames of reference on depression. Therefore, they may express different treatment preferences for Major Depression. Many primary care clinicians are not aware of these cultural considerations. Since patient outcomes or satisfaction may be related to choice of treatment, the GDT recommends approaching the treatment of Major Depression in adult patients from different cultural and ethnic backgrounds in the context of the patient s cultural beliefs. 79 National Adult Depression Clincial Practice Guideline

84 11. Patient Self-Management Strategies for Improving Symptoms of MDD Internet Resources 11A The GDT recommends scientifically validated internet-based patient self-help materials as an optional adjunct strategy (in addition to antidepressants or psychotherapy) for treating symptoms of MDD. Evidence-based: B Exercise 11B The GDT recommends exercise as an adjunctive strategy (in addition to antidepressants or psychotherapy) for treating the symptoms of MDD. Evidence-based :B Bibliotherapy 11C The GDT recommends scientifically validated bibliotherapy as an optional adjunct strategy (in addition to antidepressants or psychotherapy) for treating the symptoms of MDD. * Consensus-based Befriending 11D The GDT recommends befriending as an optional adjunct to antidepressants or psychotherapy for treating the symptoms of MDD. Consensus-based Patient-Initiated Combined Phone/Computer Programs 11.E. There is insufficient evidence for or against using patient-initiated combined phone/computer programs in the treatment of MDD. Evidence-based: I Light Therapy 11F The GDT makes no recommendation for or against light therapy as a primary or adjunctive treatment for nonseasonal forms of MDD. Evidence-based: I Music Therapy 11G The GDT makes no recommendation for or against music therapy as an adjunct to antidepressants or psychotherapy for treating the symptoms of MDD. Evidence-based: I Life Review Therapy 11H The GDT makes no recommendation for or against life review therapy as an optional adjunctive depression management strategy for depressed older adult patients who are concurrently receiving regular social services care. Evidence-based: I * Bibliotherapy: Advising people to read specific written material based on cognitive-behavioral approaches to depression treatment. Befriending: Consists of a designated befriender who meets the depressed person to talk and socialize with for at least one hour per week. 80 National Adult Depression Clincial Practice Guideline

85 Evidence Grade Evidence for Recommendations 11A: Fair; F, G, and H: Insufficient. Search Strategy Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. The Clarke (2005) (171) study was not identified by the search due to PubMed Indexing, but was identified by a GDT member and reviewed due to relevance. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded.see Appendix B for more information on the search strategy Update New evidence was found, recommendations have been changed based on expert/consensus opinion. Our recent search yielded two relevant RCTs (Craft, 2007 (172) ; Meyer, 2009 (173) ) that address the different categories of self-management strategies. The studies included patients with symptoms of depression, which consisted of participants with minor depression and dysthymia as well. Studies provided little detailed information about concurrent use of antidepressants or psychotherapy in the intervention groups. One RCTs (Meyer, 2009 (173) ) evaluated the effectiveness of an internet-based therapy and one RCT (Craft, 2007 (172) ) assessed an exercise intervention. Please refer to Evidence Tables 11.1 and 11.2 for study details. Exercise Craft et al. (2007) (172) conducted a randomized pilot study to compare the effectiveness between a home-based versus a clinic-based exercise intervention for depressive symptoms in low-income, minority women. A total of 32 women were assigned to home-based (N = 16) or clinic-based (n = 16) exercise program for 3 months. Overall, both interventions were associated with improvements in depressive symptoms (effect size: and -1.3 for home and clinic based, respectively) and time spent in physical activity. However, at 3-month follow-up, there were no significant between-group effects for depressive symptoms, selfreported physical activity, body composition, or cardiovascular fitness. It is important to note that the study is limited by lack of control group, (it is essentially a pre-post study with post comparisons of the two groups) small sample size, lack of power, short duration, and low generalizability. Therefore, interpretation of results should be done with caution. Please refer to Evidence Table 11.1 for study details. Overall Conclusion Results from one low-quality study identified does not warrant changes to the 2008 recommendation. Although the study found modest short-term improvement in depressive symptoms with exercise, no long-term effect was observed and it was not designed to substitute exercise for antidepressant or psychotherapy treatments; thus, it can not be definitively concluded that exercise may be used effectively instead of antidepressant medication or structured psychotherapy. 81 National Adult Depression Clincial Practice Guideline

86 Internet/Computer Meyer et al. (2009) (173) randomized 396 participants from an internet depression forum in Germany to internet-based psychotherapy intervention plus treatment-as-usual (n = 320) or to wait-list control plus treatment-as-usual (n = 76) for nine weeks and followed-up at 18 weeks and six months. Continuation rates were reported at 54.5% at nine weeks, 36.9% at 18 weeks, and 25% at six months. About three quarter of the sample were women, many were unemployed, and 58% currently received treatment for depression. At nine weeks, intent-to-treat (ITT) analysis found significant reduction in depression symptoms and social functioning in the internet group compared to no change in wait-list controls (effect size for depression symptoms: d = 0.3; effect size for social functioning: d = 0.36). Changes were also maintained at six-months follow-up among completers (d = 0.74). In addition, 25.4% of those in the internet group vs. 1.9% of the wait-list control reached the criteria of clinical significant improvement/ recovered (change of at least 8.46 points on BDI, with post test score of < 14.29, p < 0.001). Based on findings, authors suggested that the program could serve as an adjunctive or stand-alone treatment tool for patients suffering from symptoms of depression. However, limitations such as no placebo-controlled group, participants not clinically diagnosed with MDD, high attrition rate, selection bias, and low generalizability warrant cautious interpretation of results. Please refer to Table 11.1 for study details. Overall Conclusion Only one trial on internet-based self-study intervention was identified and outcomes from the trial did not warrant changes to the 2008 recommendation. The trial found positive results among participants assigned to the internet intervention in terms of social functioning and reduction in depressive symptoms compared to wait-listed control. Authors stated some participants were concurrently under other treatments for depression; however, it was unclear whether all participants were clinically diagnosed with MDD, which makes it difficult to draw a definitive conclusion on the effectiveness of the internet program exclusively. Editorial note: no new evidence was found on bibliotherapy. The wording of the recommendation was modified by consensus approval of the GDT to be consistent with the wording of the internet-resources recommendation. In addition, no new evidence was found for combined phone/computer programs; however, the language for this recommendation was also modified by consensus approval to reflect the insufficient evidence based on only one previously identified study Guideline New evidence was found, the recommendation remains unchanged. One relevant RCT (174) was identified that studied the effect of providing patients with personalized workbooks that addressed various aspects of depression in addition to providing them with usual care. No differences in Beck Depression Inventory scores were seen between groups, although the patients in the group given workbooks were more likely to report that they felt well educated about depression. Please refer to Table 11.1 for study details. 82 National Adult Depression Clincial Practice Guideline

87 Overall Conclusion We found only one trial on the above list of patient self-management strategies in our 2008 update. This negative trial on adjunct bibliotherapy is the first trial identified since the original guideline recommendations. The overall evidence is mixed. However, even in this new trial, patients gained some benefit from understanding their condition. Since there is little apparent harm in providing patient education materials as an adjunct to other clinician-directed therapy, the GDT retained its belief in the use of bibliotherapy as an optional adjunct strategy. Bibliotherapy was the only health intervention for which new evidence was found. A change in symptoms as a result of bibliotherapy is the only health outcome addressed in this new evidence. The overall evidence on bibliotherapy remains mixed, so our recommendation remains unchanged Guideline Our recent search yielded a number of studies that address the different categories of selfmanagement strategies. Most studies were designed to compare these self-management strategies with treatment or with different intensities or types of the same general selfmanagement intervention. Many studies included patients with symptoms of depression, but were not designed to specifically study (nor did they specify the number of patients with) MDD. Few compared self-management strategies with traditional first-line treatment approaches; most studies provided little if any detailed information about concurrent use of antidepressants or psychotherapy in the intervention groups. Please refer to Table 11.1 for study details. In reviewing the evidence as noted below, the GDT concluded that due to limitations in the current body of evidence, there is insufficient evidence to recommend for or against these selfmanagement strategies as first-line monotherapy for patients with MDD. However, when appropriate (as discussed below), pending appropriately designed studies, the GDT extrapolated the findings to make consensus recommendations for considering selected self-management strategies as an adjunct to other evidence-based treatment for MDD. Exercise Dunn (2004) (175) found that mild to moderately depressed subjects partaking in a public health dose of exercise (17 kcal/kg/week) for at least three days a week had a significantly greater reduction in depression symptoms when compared with those partaking in a low dose (7 kcal/kg/week) or no exercise. None of the subjects were involved in another treatment for depression. Clinical Evidence (Issue 14, January 2006) found one systematic review examining the effectiveness of exercise as a treatment option for depression. The review found limited evidence that exercise may improve symptoms compared with no treatment [standardized mean difference in effect size -1.1, 95% CI: -1.5 to -0.6; weighted mean difference in Beck depression inventory -7.3, 95% CI: to -4.6, and that exercise may be as effective as cognitive therapy (standardized mean difference -0.3, (95% CI: -0.7 to 0.1)]. However, according to the authors, the following methodological problems rendered the results inconclusive: randomization was adequately concealed in only three RCTs, intention to treat analysis was undertaken in only two, and assessment of outcome was blinded in only one RCT. (176) 83 National Adult Depression Clincial Practice Guideline

88 One RCT involving older adults and included in the systematic review above found no significant difference among groups (aerobic exercise, sertraline, and combination treatment) in the proportion of people who recovered (60% with exercise, 69% with sertraline, and 66% with combination treatment). A ten month follow-up of this RCT found significantly lower relapse rates with exercise than with antidepressant drugs (30% with exercise, 52% with sertraline, and 55% with combination treatment; p = 0.28 for exercise vs. either treatment). However, about half of the people in the medication group engaged in exercise during follow-up, making it difficult to draw firm conclusions. The authors conclude that the clinical importance of the observed difference at ten months remains unclear. (177) Overall Conclusion The evidence comparing exercise with antidepressants or psychotherapy as first-line strategy for treatment of adults with Major Depression is relatively sparse. The study populations, inclusion criteria, and outcome measures are heterogeneous. The GDT does not believe that there is sufficient evidence at this time to recommend for or against exercise as a sole treatment option for MDD. However, exercise is often recommended for other health reasons, and is (in appropriately selected patients) a low risk self-management strategy. Extrapolating from the available evidence, the GDT believes that exercise is an adjunctive strategy (in addition to antidepressants or psychotherapy) for treating the symptoms of MDD. Bibliotherapy No new evidence was found. Overall Conclusion Many of the limitations noted with studies of exercise above also apply to studies of bibliotherapy found in earlier iterations of this guideline. In light of these past findings, the GDT believes that the previous recommendation should stand: Bibliotherapy is an optional adjunct strategy (in addition to antidepressants or psychotherapy) for treating the symptoms of MDD. Computer/Internet Christensen, Griffiths, & Jorm (2004) (178) placed subjects with depression (and not being treated by a psychologist or psychiatrist) into one of three groups: 1) Blue Pages, and internet program to improve depression literacy, 2) MoodGYM an internet-based cognitive behavioral therapy, or 3) attention control. The researchers found that both Blue Pages and MoodGYM were significantly more effective than control in reducing symptoms of depression; neither internet intervention was more effective than the other. There was no information on how many patients in each study arm were being treated with antidepressants. 84 National Adult Depression Clincial Practice Guideline

89 Clarke, et al. (2005) (171) conducted a follow-up to their 2002 study examining the effectiveness of the internet-based intervention Overcoming Depression in the InterNet (ODIN), available at The authors concluded that the 2002 trial found no significant differences between the internet and control interventions because participant usage of the internet site was low. In the 2005 update they attempted to increase participant adherence by using postcard and telephone reminders. The authors did in fact find that the addition of the reminders yielded significant results: those in the either intervention group (postcard or telephone reminders) showed greater decreases in CES-D depression scores than those in the treatment-as-usual control group who did not have access to the site at all. This difference was even greater in those who had more severe depression at baseline. However, enrollment rates in the study were very low, and more patients dropped out of the treatment groups than the control group. Overall Conclusion The specific web-based interventions studied by Christensen, Griffiths, and Jorm, and the 2005 Clarke intervention were found to be effective for reducing symptoms of depression, although the data do not suggest that these sites can be used in lieu of antidepressants or psychotherapy. It should also be noted that the interventions examined in the previously reviewed studies by Clarke (179) and Patten (180) in the 2004 update were ineffective. The effectiveness of web-based interventions may depend on a number of factors, potentially including patient motivation to utilize the resources, the specific content utilized and the usability of the website. The GDT believes that some internet-based interventions may be beneficial adjuncts in reducing symptoms of depression, but based on the currently available (heterogeneous) evidence, the recommendation should be limited to specific interventions (Blue Pages, Mood GYM, and ODIN). Note: The GDT acknowledges that internet sites are dynamic and continually evolving. Prior to referral to the aforementioned sites, care should be taken to ensure the information provided is still accurate and up-to-date. Befriending Clinical Evidence Volume 14 found one small RCT that provided insufficient evidence to assess befriending in people with mild to moderate depression. It found that befriending significantly increased the proportion of women with remission of mild depression symptoms at 13 months compared with waiting list control (86 women with chronic depression, aged 18 years and older, primarily aged 25 to 40 years; 65% with befriending vs. 39% with control; p < 0.05; NNT = 4, 95% CI: 2 to 18). (181) Overall Conclusion The Clinical Evidence update above is consistent with previously reviewed evidence suggesting some benefit of befriending in patients with depression. None of the studies specified how many patients were receiving other forms of depression treatment, nor did they differentiate between patients with Major and Minor Depression (the latter often resolves spontaneously). However, given the potential for some effect, and the low risk of befriending as a strategy, the GDT makes a consensus extrapolation from the available evidence to consider befriending as an option for patients as an adjunct to (but not a substitute for) treatment of Major Depression. 85 National Adult Depression Clincial Practice Guideline

90 Automated Telephone Programs We found no new evidence to support for or against the use of automated telephone programs. Overall Conclusion The GDT previous recommendation stands; use of automated telephone programs is not currently recommended as adjunctive therapy for MDD. Light Therapy Note: These analyses address the use of light therapy for nonseasonal forms of MDD. The use of light therapy for Seasonal Affective Disorder (SAD) is currently being systematically reviewed by the Cochrane Collaborative. Seasonal Affective Disorder is beyond the scope of these current recommendations. Therefore, conclusions are limited to nonseasonal Major Depression. Leppamaki, et al. (2004) (182) examined the effect of three interventions on mood improvement: 1) aerobics in bright light, 2) aerobics in normal light, and 3) relaxation and stretching in bright light. They found that a significant number of subjects who responded to one of these treatments were given light therapy. This study should be interpreted with caution, however, since it is unclear how many participants had Major Depression or Seasonal Affective Disorder (SAD - a subtype of Major Depression where light therapy is considered to be effective); nor did the authors provide information on the concurrent use of other treatments. In a study examining the effectiveness of adjunctive bright light for nonseasonal Major Depression, Martiny, et al. (2005) (183) treated all subjects with sertraline and concomitantly had them exposed to either dim light or bright light. At the end of the five week study period, the authors found a significant reduction in HAMD scores in the bright light group (p < 0.01). This group also had significantly higher response (p = 0.006) and remission (p = 0.015) rates when compared with the dim light group. While these results are promising, it should be noted that the two groups had unequal durations of treatment (30 minutes for dim light vs. 60 minutes for bright light), and it is unclear how this factor might have affected the overall results. The Cochrane Collaboration has reviewed the evidence on light therapy for treatment of nonseasonal depression. (184) Twenty studies were included in the review. Seventeen studies reported on participants mostly suffering from MDD. Ten studies had both unipolar and bipolar patients in their sample. Inclusion and exclusion criteria varied among studies. Almost all studies took place in the hospital or long-term care facility. Only two studies assessed outpatients. None of the studies were multicenter. Only five studies reported on the time of the year when the study was performed. 86 National Adult Depression Clincial Practice Guideline

91 Bright light therapy was administered in a wide range of intensities. The duration of active treatment varied between 30 minutes and nine hours. Eleven studies administered bright light in the morning. Most of the studies applied bright light as adjunctive treatment to drug therapy, sleep deprivation, or both. Light was administered adjunctive to sleep deprivation in nine studies, and in two additional studies the participants were awakened before their usual wake up time for light treatment. Standardized adjunctive pharmacotherapy was applied in seven studies, and in ten studies, concomitant drug treatment of the participants was kept unchanged. All included studies described themselves as randomized, but presented little methodological detail to elaborate on the truly random nature of allocation. Blinding of assessment in administration of light therapy is more difficult than in studies with drug intervention, since the active treatment due to its brightness looks dissimilar to the control treatment. Subjects cannot fail to perceive the treatment and cannot be literally blind to treatment, though they may not know which is intended as the active treatment. In general, the quality of reporting was poor. All but two trials reported the randomization procedure without adequate information on allocation concealment. Blinding procedures were also generally inadequately described. Many studies did not report the number of dropouts and did not specify reasons for drop-out. The trials did not report if intention-to-treat analysis was performed. According to the criterion of 50% decrease in the HDRS score, there was no difference between groups: 20 out of 39 patients (51%) in the bright light group and 17 out of 32 patients (53%) in the control treatment group were not improved (three studies, 71 patients, relative risk (RR) 0.94, CI: 0.61 to 1.46). The treatment response in the bright light group was better than in the control treatment group, but did not reach statistical significance in random effects models. The result was mainly based on studies of less than eight days of treatment. In studies using a light box, a fixed effect model approach showed that bright light was more effective than the control treatment (12 studies, 275 patients, standardized mean difference (SMD) = -0.50, CI: to -0.25), and the statistical significance remained even when a random effects model was applied (SMD = -0.47, CI: to -0.08). The response to bright light was significantly better than to control treatment in high-quality studies (SMD = -0.90, 95% CI: to -0.31), in studies applying morning light treatment (SMD = -0.38, CI: to -0.14), and in sleep deprivation responders (SMD = -1.02, CI to -0.45). Evaluation of the effect of combination of concomitant drug and morning bright light showed that there was no difference between the two morning light conditions with or without pharmacotherapy. With a fixed effect model approach, both conditions were statistically significantly in favor of bright light over control treatment (combination treatment: nine studies, 243 patients, SMD = -0.32, CI: to -0.08; light only: two studies, 54 patients, SMD = -0.57, CI: to -0.01), but with a more conservative random effects model the statistical significance was lost (combination treatment: SMD = -0.36, CI: to 0.07; light only: SMD = -1.03, CI: to 0.58). 87 National Adult Depression Clincial Practice Guideline

92 Many reviews did not report adverse effects systematically. Hypomania was more common in the bright light group compared with the control treatment group (RR = 4.91, CI: 1.66 to 14.46; NNH = 8, CI: 5 to 20). The reviewers noted that categorizing the drop-out subjects as "failures" might overestimate the number of subjects with this adverse effect as well as with other poor outcomes. The reviewers conclude that in patients suffering from nonseasonal depression, light therapy may offer modest antidepressive efficacy when administered in the morning, during the first week of treatment and as an adjunctive treatment to sleep deprivation responders. These benefits need to be balanced against the potential for hypomania. Finally, the reviewers note that due to limited data and heterogeneity of studies these results need to be interpreted with caution. Overall Conclusion In discussing the results of the above studies and in particular, the details of the Cochrane review, the GDT believes that, based on the information available, it would be difficult to formulate specific recommendations for patients with nonseasonal MDD on how to specifically utilize light therapy as a self-help strategy. Due to the heterogeneity of the studies, and pending further clarification, the GDT concluded that there is insufficient evidence at the current time to recommend for or against the use of light therapy as a self-help strategy for patients with nonseasonal forms of MDD. Music We found no new evidence to support for or against the use of music therapy. Overall Conclusion The GDT s previous recommendation stands: There is currently insufficient evidence to recommend for or against music therapy as an adjunct to antidepressants or psychotherapy for treating the symptoms of MDD. Life Review Serrano & Latore (2004) (185) tested life review therapy on a population of older adults (65 to 93) with clinically significant symptoms of depression and receiving social services regularly. The life review consisted of autobiographical retrieval practice that entailed focusing on a particular life period each week childhood, adolescence, adulthood, and summary. When compared with controls, the life review therapy group had significantly greater improvements in depressive symptoms. This study was conducted in Spain; it is unclear if it is generalizable to populations in other countries. Overall Conclusion As with other self-management strategies, this study was not designed to test the relative efficacy of this strategy vs. standard (psychotherapy or antidepressant) treatment for MDD. Due to the limited population studied (and the unknown generalizability), the GDT believes that currently there is insufficient evidence to recommend for or against life review therapy an optional adjunctive depression management strategy. 88 National Adult Depression Clincial Practice Guideline

93 2004 Guideline Supporting Evidence for Self-Management Strategies for Improving Depressive Symptoms in Patients with MDD Effect of Exercise Three RCTs and one systematic review were found that addressed this issue. Singh (186) studied patients age 60 and older with Beck Depression Inventory (BDI) score of greater than 12 and a diagnosis of either unipolar major or minor depression or dysthymia according to DSM-IV diagnostic criteria. The patient population was 37% male and 63% female. In phase one, patients (n = 32) were randomized to two treatment groups comparing ten weeks of supervised exercise in a laboratory setting (intervention group), with health education lectures (control group). In phase two (weeks 11 to 20), (n = 15) intervention patients received unsupervised exercise at home, laboratory or health club setting (n = 15, 18 ±2 weight lifting exercise sessions), while control patients (n = 14) received no further intervention. The BDI score at 20 weeks decreased significantly from 21 ±2.0 to 9.2 ±2.8 for the exercise group (p = 0.036), while the score for the control group decreased from 18.4 ±1.7 to only 14.4 ±2.2. At the 26 month follow-up, the BDI score for the exercise group decreased from 21 ±2.0 to 13 ±2.2, and the score for the control group decreased from 18.4 ±1.7 to 14.4 ±2.2, p = (favoring exercise group). The exercise group showed a statistically significantly reduced depression severity compared with control group at both 20 weeks and 26-month follow-up. The difference between BDI scores between groups may not be clinically significant. The effect may not be evident if exercise is not continued. Over time, the severity of depression also improved in the control group. McNeil (187) studied older adult outpatients with a mean age of 72.5 years and a BDI score of 12 to 24. Patients (n = 30) were randomized to three treatment groups comparing exercise (walking between 20 to 40 minutes, three times a week) with social contact (two home visits (20 to 40 minutes each week by an undergraduate psychology student) and a wait-list (control group) for a period of six weeks. The total BDI score decreased from 16.6 to 11.1 for the exercise group, 16.0 to 11.8 for the social contact group, and 15.2 to 14.7 for the control group (p < 0.05, favoring exercise or social contact). Exercise and social contact both resulted in significant reductions in both total and the psycho-social subscale of the Beck Depression Inventory. The exercise condition, however, resulted in decreased somatic symptoms of the BDI (p < 0.05). Mather (188) studied outpatients 53 years and older with a diagnosis of depression and a score of at least ten on the Geriatric Depression Scale (GDS). The patient population was 16% male and 84% female. Patients (n = 86) were randomized to two treatment groups comparing exercise (n = 43, 45 minutes of predominantly weight-bearing exercise; two sessions per week) with a nonexercise social control group (health education talks, n = 43) for ten weeks. 89 National Adult Depression Clincial Practice Guideline

94 Change in 17-item Hamilton Rating Scale for Depression (HRSD) from baseline at ten weeks for the exercise group was 55% and for the control group was 33% (p = 0.05, favoring exercise group). Response was defined as a 30% reduction in HRSD score from baseline, a threshold that is lower than the usual 50% decrease in symptoms used in most other studies (thus setting the bar lower for a positive effect). The study found that ten weeks of twice-weekly exercise was associated with a modest reduction in depression symptoms in a group of older people with depression. There was preponderance of women in the exercise group. North (189) systematically reviewed 80 studies on the effect of exercise on depression. Patients ranged from 11 to 55 years with a mean age of 31.8 ±12.4 years. Twenty studies included males only and 16 studies included all females. Studies included looked at the effect of different variables on depression (sixty three studies looked at age effect on depression and thirty six studies were included for the effect of gender.) Studies varied from four weeks to 24 weeks with the majority of studies being between five to 12 weeks. The studies compared exercise (aerobic endurance and/or muscular strength, varying in number of sessions) with comparison groups (no intervention control, leisure activity, or psychotherapy). The overall Effect Size (ES) of the studies included, was ±0.85, indicating that patients in exercise groups decreased their scores an average of 0.53 standard deviations more than subjects in the comparison groups (p < 0.001). A significant, negative corelational relationship was found between mean age and ES (p < 0.05), suggesting that the older the subjects, the greater the decrease in depression with exercise. The ES for males and females were not significantly different (p > 0.05) suggesting equal benefit of exercise on depression in both genders. The overall results indicate that exercise has a beneficial effect on improving symptoms of depression. There are multiple differences in the designs of the studies included in this meta-analysis, including mode of exercise, length of the exercise programs and types of control groups. This meta-analysis also excluded people older than 55. No heterogeneity data was reported. Results should be interpreted with caution due to internal validity issues. Our search and Clinical Evidence (70) identified one systematic review (search date 1999, 14 RCTs, 851 people). This review found limited evidence that exercise versus no treatment may improve depressive symptoms, and that exercise may be as effective as cognitive therapy. However, it suggested that these results were inconclusive because of methodological problems in all of the RCTs; randomization was adequately concealed in only three of the RCTs, intention-totreat-analysis was undertaken in only two, and assessment of outcome was blinded in only one of the RCTs. 90 National Adult Depression Clincial Practice Guideline

95 Older Adults The systematic review identified one RCT (156 people with Major Depression, mean age 57 years) comparing aerobic exercise, sertraline hydrochloride (a selective serotonin reuptake inhibitor), and combined treatment for 16 weeks. It found that the proportion of people who recovered (those no longer meeting criteria for depression or with a Hamilton Depression Rating Scale score < 8) was not significantly different across the treatment groups (60% with exercise vs. 69% with sertraline vs. 66% with combined treatments). A ten-month follow-up of this RCT found lower rates of relapse with exercise versus medication (30% with exercise vs. 52% with sertraline vs. 55% with combined treatment). However, about half of the people in the medication group engaged in exercise during follow-up, making it difficult to draw firm conclusions about effects of exercise. The clinical importance of the observed difference at ten months remains unclear. (Note that this RCT did not include a control group.) Conclusion There is evidence that exercise, at least in the short term, may be helpful in reducing symptoms of depression. Most studies were not designed to substitute exercise for antidepressant or psychotherapy treatments, so we can not conclude that exercise may be used effectively instead of antidepressant medication or structured psychotherapy. Most of these studies do not describe how many patients in the exercise group were receiving other forms of depression treatment, and many did not differentiate between patients with Major Depression and other forms of depression. Given the beneficial effect of exercise in general and on many often comorbid medical conditions, the GDT believes that exercise can be recommended as an adjunctive strategy for patients with MDD. The type of exercise will depend in part on the presence or absence of other medical conditions. Effect of Bibliotherapy (Patient Handouts and Other Reading Material) Two RCTs that addressed this issue were found. Jamison (190) studied patients with mean age of 40 years and a score of ten or greater on the Hamilton Rating Scale for Depression 21-item version (HRSD) and score of ten or greater on the 21-item Beck Depression Inventory (BDI). All patients met the DSM-III-R criteria for mild to moderate Major Depression. Patients (n = 80) were randomized to two treatment groups comparing cognitive bibliotherapy (n = 33) with control group (wait list, n = 39) at four weeks (treatment phase) and three months (follow-up phase). In treatment phase, the HRSD score decreased from 20.2 to 9.9 for the bibliotherapy group and decreased from 19.6 to 19.0 for the control group (p < 0.05, favoring bibliotherapy). BDI scores showed comparable results with p < 0.05 (favoring bibliotherapy). In the follow-up phase, HRSD score further decreased, from 10.0 to 9.2 for the treatment group and decreased from 18.7 to 10.0 (p < 0.05, favoring treatment when starting from baseline). (Note that the control group experienced a large decrease in depression scores in the follow-up phase, to within a point of the intervention group). BDI score again showed similar outcomes (p < 0.05, favoring treatment when starting from baseline). 91 National Adult Depression Clincial Practice Guideline

96 This study suggests that cognitive bibliotherapy for depression is an effective treatment for depression, with participants experiencing more rapid symptom improvement compared with nonparticipants. Treatment gains in the intervention group were maintained over a three-year follow-up (Smith (191) ) period, as participants (n = 50) were less depressed than when they began the program. There was a significant decrease in HRSD and BDI scores from pretreatment to follow-up (p < 0.05 each), with no changes from post treatment to follow-up (p > 0.05 each). This study suggests that there maybe long-range benefits to participants in structured, minimalcontact bibliotherapy programs for depression. There was no control group in the three-year follow-up period. Sensitivity analysis was not included to account for patients who completed the initial study but did not participate in the follow-up (some of whom may have relapsed). Scogin (192) studied 57 female and ten male patients 60 years and older with score of ten or higher on the Hamilton Rating Scale for Depression (HRSD) and score of eight or higher on the Mental Status Questionnaire. Patients (n = 67) were randomized to three treatment groups comparing behavioral bibliotherapy (n = 23, reading a copy of Control Your Depression (Lewinsohn (193) ), cognitive bibliotherapy (n = 22, reading a copy of Feeling Good (Burns, 1980 (194) )) and delayed treatment (control group) (n = 22), for an initial study period of four weeks and a six month follow-up. The two treatment groups were assessed before treatment (first time), immediately following treatment (second time) and six months later (third time). The control group participants were assessed at the first time and at one month following the second time, at which point their own treatment began, and a third time immediately following treatment. HRSD Score (first time, second time and third time) for the behavioral bibliotherapy group were 17.8 to 9.7 to 9.1, and were 16.3 to 7.5 to 8.9 for the cognitive bibliotherapy group. Control group scores were 16.4 to 15.9 to 7.2. There were statistically significant differences between cognitive bibliotherapy and control group (p < 0.05, favoring cognitive bibliotherapy), and between behavioral bibliotherapy and control group (p < 0.05, favoring behavioral bibliotherapy). There were no significant differences between the two bibliotherapy groups. This study supports the efficacy of self-paced bibliotherapy for mild to moderate depression in older adults. Bibliotherapy s effect for mild to moderate depression in older adults was maintained after two years. (195) HRSD scores post-treatment and at a two-year follow-up for the treatment groups (behavioral and cognitive bibliotherapy, n = 22) decreased from 8.1 to 7.4, p < 1.00, indicating no significant effect for time. There were no male participants and no control group in the two-year follow-up study. Sensitivity analysis not included to account for patients who completed the initial study but did not participate in the follow-up (some of whom may have relapsed). 92 National Adult Depression Clincial Practice Guideline

97 Our search and Clinical Evidence (70) identified two systematic reviews (search date not stated and search date 1999). The first review identified six small, short-term RCTs of bibliotherapy versus waiting list control in 273 people (described as adults in four RCTs and older adults in two RCTs; no age range provided) recruited by advertisement through the media and probably with only mild depression. The mean effect size of bibliotherapy was 0.82 (95% CI: 0.50 to 1.15). Seventy nine percent of the people in the waiting list control group had a worse outcome than the average person in the bibliotherapy group. The second systematic review identified eight randomized and nonrandomized trials in younger and older people, but only one of them included people with depression. It found that, in people with combined anxiety and depression, anxiety, or chronic fatigue, bibliotherapy may improve symptoms over two to six months compared with standard care. The RCT identified by the second review that included people with depression found that bibliotherapy versus standard care significantly improved symptoms of anxiety over four weeks as assessed using the Hamilton Depression Rating Scale, but found no significant difference in symptoms of depression at four or 12 weeks. Clinical Evidence found no systematic review or RCTs specifically in older adults. Conclusion There is evidence, albeit not unanimous (or as strong as the evidence for exercise), that bibliotherapy may be helpful in reducing symptoms of depression. Most of these studies do not describe how many patients were receiving other forms of depression treatment, and many did not differentiate between patients with Major Depression and other forms of depression. But the majority of studies support the use of bibliotherapy and the GDT believes that bibliotherapy is an optional adjunctive strategy for patients with MDD. Effect of Befriending The McNeil study (187) discussed earlier noted a statistically significant improvement in depressive symptoms in patients randomized to the social contact group. Our search and Clinical Evidence identified one small additional RCT of befriending versus waiting list control. This trial had 86 women with chronic depression, aged > 18 years, primarily aged 25 to 40 years, based in London, UK. Initial identification was by postal screening of women registered with, but not attending, primary care. It found that befriending versus waiting list control significantly increased the proportion of women with remission of symptoms at 13 months (65% with befriending vs. 39% with control; p < 0.05; NNT 4, 95% CI: 2 to 18). 93 National Adult Depression Clincial Practice Guideline

98 Older Adults We found no systematic review or RCTs specifically in older adults. Conclusion These two small studies provide some evidence suggesting that befriending may be helpful in reducing symptoms of depression. The studies did not describe how many patients were receiving other forms of depression treatment, and it did not differentiate between patients with Major and Minor Depression (the latter often resolves spontaneously). Due to the available evidence the GDT believes that befriending is an option for patients as an adjunct to (but not a substitute for) treatment of Major Depression. Effect of Music Therapy We found one RCT that addressed this issue. Hanser (196) studied older adults (61 to 86 years old, 77% female) with diagnosis of Major or Minor Depressive Disorder, based on a structured interview using the Schedule of Affective Disorders and Schizophrenia. Patients (n = 30) were randomized to three treatment groups comparing home based structured music therapy (n = 10), with self-administered music therapy (n = 10) and a wait-list control group (n = 10) for eight weeks. Geriatric Depression Scale (GDS) scores (pre-test to post-test) was to 7.70 for the home based music therapy, to for the self-administered music therapy, and to for the control group (p < 0.05, favoring the two music therapy groups). Follow-up results at nine months revealed no significant differences on the GDS score from post-test results in all three groups. This study supports the use of music therapy strategies with older adults experiencing symptoms of depression. Conclusion This one small study suggests that music therapy may be helpful in reducing symptoms of depression. The study was not large enough to differentiate between patients with Major and Minor Depression (the latter often resolves spontaneously). While there seems to be little apparent harm in recommending music therapy as adjunctive treatment (particularly if patients enjoy music), formal music therapy is associated with added cost without clear proven benefit. Pending more and larger studies, the GDT believes there is insufficient evidence to recommend for or against formal music therapy as an adjunctive strategy for patients with MDD. Effect of Depression Internet Sites or Automated Telephone Programs Two RCTs were found that addressed this issue. Clarke (171) studied a combination of depressed and nondepressed participants (n = 299) who were randomized to two treatment groups comparing access to a Depression Internet site (n = 144, depression cases n = 107) with usual care (n = 155, depression cases n = 116), for a period of 32 weeks. Much of the Internet site content was adapted from group CBT psychotherapy manuals. The mean (SD) age for the internet group was 43.3 (12.2) years with 73.6% females. 94 National Adult Depression Clincial Practice Guideline

99 The mean age for the control group was 44.4 (12.4) years with 77.4% females. In depressed patients, mean (SD) scores on the Center for Epidemiologic Studies Depression scale (CES-D) (baseline to week 32) were 30.7 (12.9) to 22.2 (12.8) in the internet group and 31.3 (11.5) to 25.2 (14.2) in the control group (p = 0.12). Stratified analyses by high vs. low baseline CES-D scores, gender, and age greater or less than 45 all showed no statistically significant between group differences. The study found no differences between the control and experimental conditions on self-reported depression (CES-D) over the study period, indicating a lack of treatment effect. Patten (180) studied patients with Major Depression diagnosed by the Composite International Diagnostic Interview (95% CIDI) who had a mean age 45.2 ±11.9 years. The population was 90.1% female, and 9.9% male. Patients (n = 786) were randomized to two treatment groups comparing use of web and telephony-based program (n = 420) with control group (n = 366) for three months. The intervention consisted of an interactive computer program accessible through the internet or by interactive voice telephone. The percentage of patients exceeding Center for Epidemiological Studies Depression rating scale (CES-D) score of more than 15 in the active group was 37.4% (baseline of 35.7%) and 34.1% (baseline of 33.3%) for the control group (p = not stated). The study did not find any significant differences between the groups. Conclusion Based on the available evidence, the GDT does not recommend Internet or automated telephonebased programs for the treatment of MDD. 12. Behavioral Health Education Classes For Adults With MDD (Cognitive Behavioral Skills or Problem-Solving Classes) 12 For patients with mild to moderate MDD, the GDT recommends behavioral health education classes as an adjunctive treatment option. However, these classes should not be used in lieu of either antidepressant medication or psychotherapy. Evidence-based Rationale: 2008 Update No new evidence was found, the recommendation remains unchanged Update No new evidence was found, the recommendation remains unchanged. It should be pointed out that there are several elements of behavioral health education classes that could have an impact on depressive symptoms, including but not limited to the content of the curriculum (which may be cognitive, behavioral, or problem-solving), the class setting, or the social aspects of a class setting. The curriculum was different in the studies cited in the 2004 update, and may be heterogeneous from setting to setting. It is therefore not currently possible to define which of the variables in of behavioral health education classes are the active ingredients of treatment, or to which degree each variable contributes to patient outcome. 95 National Adult Depression Clincial Practice Guideline

100 2004 Guideline Supporting Evidence for Behavioral Health Education Classes in Improving Depressive Symptoms in Patients with MDD Brown (197) studied adults 16 to 65 years old (70% female) with a Beck Depression Inventory (BDI) score of 22.2 (consistent with mild to moderate Major Depression) and a diagnosis of unipolar depression according to Research Diagnostic Criteria (RDC). Patients receiving other forms of treatment concurrently for depression were included. The number of patients with severe Major Depression was not disclosed, although severe depression was not mentioned as an exclusion criterion. Patients with concurrent substance abuse were excluded. The study did not specify if patients with chronic Major Depression were excluded. Patients (n = 63) were randomized to three treatment groups and a control group. The Coping with Depression course curriculum was standardized across the three treatment conditions, and consisted of 12 sessions over an eight week period plus a visit or phone contact at one and six months post-treatment. The study compared group class format (n = 25), individual tutoring (n = 13) and phone contact (n = 14) against a wait-list control group (n = 11). Participants in the three treatment groups improved more during the eight-week period than did the wait-list control group (p < 0.05). There were no statistically significant differences among the three treatment groups, when compared at eight weeks (treatment duration), one and six months follow-up sessions. Among the participants in the three treatment groups, only 25% still met the criteria for depression at the six-month follow-up session. From an economic point of view, the in-person class condition was by far the most cost effective. The difference in response between those receiving concurrent treatment and those not receiving other treatment for depression was not specified. 34 of 122 individuals chose not to participate in the study, with differences between enrolled and not enrolled group not specified. Among patients participated in the study, dropout rate was 4.6%. Dowrick (198) studied adults, ages 18 to 65, with DSM-IV diagnoses of MDD (52% single episode, 19% recurrent) dysthymia (16%), adjustment disorder (4%) or other depressive disorders (9%). Suicidal patients and patients with concurrent substance abuse were excluded from the study; it is unclear if patients with severe Major Depression who were not suicidal were included. Patients (n = 452) were randomized to three treatment groups comparing six individual sessions of problem-solving treatment delivered in a setting convenient for the patient (usually the patient s home) by behavioral health, allied health, or nursing personnel (N = 128), eight group sessions (2.5 hours each) of a course on prevention of depression (n = 108), and a control group (n = 189). Outcomes were assessed at six and 12 months using intention-to-treat-analysis. Mean difference in Beck Depression Inventory (BDI) score for the problem-solving group at six months was (95% CI: to -0.32), p = At 12 months, the BDI score was (95% CI: to 1.31), p = National Adult Depression Clincial Practice Guideline

101 The prevention of depression group had a BDI mean difference of (95% CI: to 1.17), p = at six months and 1.11 (95% CI: to 3.52), p = at 12 months (not significant). There were significant improvements in SF36 scores (mental role, social function, and mental health) at six months for patients in the depression prevention group compared with controls at six months (p = 0.042, 0.048, and respectively). There were also significant improvements in SF36 scores (mental role, social function, and mental health) at six months for patients in the problem-solving therapy group compared with controls at six months (p = 0.030, 0.012, and respectively). Neither treatment group had statistically significant improvement in any of these SF36 domains at 12 months, compared with controls. Statistical comparison of SF 36 scores comparing the two treatment groups with each other was not assessed. At six months 17% fewer patients in the problem-solving group were depressed compared with controls (NNT = 6) and 14% fewer patients assigned to prevention of depression were depressed compared with controls (NNT = 7). At 12 months there were no differences between either study condition and controls. More patients completed problem-solving therapy than the course on prevention of depression (the authors attribute this difference to better acceptance of the former treatment). This study concludes that both problem-solving treatment and the course on prevention of depression reduced the severity and duration of depressive disorders over the short term, and improves subjective mental and social functioning. Results in this mixed population (patients with different types of depression, not all MDD) were not stratified by condition. The number of different individuals delivering the problem-solving therapy was not specified, the problem-solving therapy protocol was not specified, and the inter-provider variability of adherence to the therapy protocol was not assessed. Therefore, this may not be directly comparable to the problem-solving therapy delivered by trained mental health personnel in traditional behavioral health settings. Other Considerations Behavioral health education experts on the GDT report that these classes in KP target patients with acute depressive symptoms. KP Behavioral Health education classes for Major Depression exclude patients with severe depression symptoms, and like the Brown and Dowrick studies, exclude patients with concurrent substance abuse. From the available literature and KP experience, therefore, the role of behavioral health education classes for patients with severe Major Depression or substance abuse is not clear. Controlled studies are needed to verify and more accurately quantify the initial benefit seen in pre-post analyses of Behavioral Health Education classes for MDD in KP Northern California and KP Northwest. 97 National Adult Depression Clincial Practice Guideline

102 Overall Conclusion There is limited evidence that behavioral health education classes may be effective in reducing symptoms and improving functioning for some patients with MDD. There is insufficient evidence that health education classes should be used in lieu of antidepressants or structured psychotherapy for treatment of MDD. Therefore, the GDT believes that health education classes (if available) are an optional adjunctive treatment for patients with mild to moderate MDD, but does not recommend that health education classes be used instead of antidepressants or psychotherapy at this time. 13. Antidepressants To Avoid During Pregnancy or Breastfeeding Pregnancy 13A Do not start paroxetine in women who are pregnant. Evidence-based: D 13B Use caution in starting other selective serotonin re-uptake inhibitors (SSRIs) in women who are pregnant. Consensus-based Discuss risks to the mother and fetus of untreated maternal depression, as well as the risk of fetal adverse effects from antidepressants. 13C If drug therapy is a consideration for treatment of maternal MDD during pregnancy and/or breastfeeding, then: Individualize according to patient history and need for medication, and Discuss the benefits and harms of the various treatment options with the patient. Consensus-based 13D If MDD is in remission and a woman becomes pregnant while taking antidepressants during the continuation or maintenance phase of treatment, then: Discuss the risks to the mother and fetus of untreated maternal depression or depression relapse after antidepressant discontinuation, as well as the risk of fetal adverse effects from continuing antidepressants, and Monitor for first trimester fetal malformations if taking paroxetine. Consult OB/GYN for considerations on fetal malformation screening. Consensus-based Breastfeeding 13E Do not start fluoxetine and/or citalopram in breastfeeding women in most circumstances. If used, they should be used with caution, and only in patients who had good results with these medications during pregnancy or a previous depression episode. Consensus-based. 13F In women taking antidepressants during pregnancy whose depression is in remission and who desire to breastfeed: Discuss the risks to the mother and fetus of untreated maternal depression or depression relapse after antidepressant discontinuation and the risk of adverse effects in the nursing newborn of mothers continuing antidepressants, and Consider changing treatment for depression if the newborn shows potential antidepressant-related adverse effects (withdrawal symptoms) during the first few hours after birth. Consensus-based 98 National Adult Depression Clincial Practice Guideline

103 Evidence Grade Evidence for Recommendation 13A: Fair Search Strategy Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. The Pedersen (2009) (199) study was not identified by the search due to PubMed Indexing, but was identified by a GDT member and reviewed due to relevance. See Appendix B for more information on the search strategy. Rationale: 2010 Update New evidence has been identified, and recommendations have been changed based on both new evidence and expert/consensus opinion. Our recent search yielded 11 relevant studies, including six cohort studies (Pedersen 2009 (199) ; Wichman 2009 (200) ; Jordan 2008 (201) ; Diav-Citrin 2008 (202) ; Kallen and Olausson 2008 (203) ; Oberlander 2008 (204) ) and 5 case-control studies (Andrade 2009 (205) ; Toh 2009 (206) ; Ramos 2008 (207) ; Salkeld 2008 (208) ; Maschi 2008 (209) ) that addressed adverse effects associated with the use of antidepressant medication during pregnancy. Outcomes examined included congenital malformation (n = 5), neonatal hypertension (n = 2), neonatal withdrawal syndrome (n = 2) and maternal adverse events (n = 2). No new studies were identified that evaluated potential risks associated with antidepressant use during lactation. Please refer to 13.2 for study details. Congenital Malformations Pedersen et al. (2009) (199) conducted a retrospective cohort of 493,113 children born in Denmark between to assess the association between SSRI during the 1 st trimester of pregnancy and congenital malformations. Results indicated that filled prescription for SSRIs was not associated with overall major malformation, but was significantly correlated with septal heart defects [OR = 1.99 (95% CI: )]. In particular, septal heart defects were highly associated with Sertraline [OR = 3.25 (95% CI: 1.21 to 8.75)] and Citalopram [OR = 2.52 (95% CI: 1.04 to 6.1)]. In women with one or more redemptions for antidepressants (mostly multiple SSRIs), there was a two-fold increase in risk septal heart defects [OR = 4.7 (95% CI: 1.74 to 12.7)]. However, the absolute increase in prevalence of septal heart defects was low [unexposed: 0.5% vs. any SSRI exposure: 0.9% vs. more than one SSRI: 2.1%]. Wichman et al. (2009) (200) examined whether the use of SSRIs during pregnancy was associated with congenital heart disease (CHD), in particular, ventricular septal defects (VSDs), and PPHN among 25,214 deliveries at the Mayo Clinic. Of the 25,214 deliveries, 808 of study participants took a SSRI at some point during pregnancy and results found three cases of CHD and no cases of persistent pulmonary hypertension (PPHN) or VSD in this group. Sertraline was the most frequently prescribed SSRI (36.6%). The prevalence of CHD was similar between the exposed (4%) and unexposed (8%) neonates, p = None of the newborns from the 134 mothers who took paroxetine were diagnosed with VSDs. Thus, authors concluded that there was no association between CHD or PPHN and SSRI use during pregnancy. However, the study was underpowered to detect a small association and other major methodological concerns warrant cautious interpretation of results. 99 National Adult Depression Clincial Practice Guideline

104 Ramos et al. (2008) (207) conducted a case-control study of pregnant women (N = 2,339) with psychiatric disorder prior to pregnancy to evaluate the effect of duration of antidepressant use during the 1 st trimester of pregnancy and the risk of major congenital malformation (MCM). Results found no significant association between antidepressant use during 1 st trimester and MCM [adjusted OR = 1.1 (95% CI: )]. Subgroup analysis also indicated nonsignificant difference between duration of antidepressant used and risk for MCM [adjusted OR = for 1 to 30 days vs. 0 day: 1.23 (95% CI: ); 31 to 60 days vs. 0 day: 1.03 (95% CI: ); 561 days vs. 0 day: 0.92 (95% CI: )]. Additionally, the class of antidepressant used was also not significantly associated with major birth defects. Oberlander et al. (2008) (204) investigated the incidence of congenital anomalies following prenatal exposure to SSRI monotherapy and/or in combination with benzodiazepines (BZ) using maternal and neonatal data linked to prenatal prescription records for all live births in British Columbia from 1998 to 2001 (N = 119,547). After adjusting for confounders, no significant risk differences were detected for major congenital anomalies for SSRI only, BZ only, or SSRI + BZ compared to no exposure. However, SSRI + BZ was significantly associated with increased risk for cardiovascular congenital defects [regression-adjusted incidence risk difference: 1.18 (95% CI: )], whereas SSRI alone was significantly associated with increased in risk for ASD [regression-adjusted incidence risk difference 0.21 (95% CI: )]. Among the SSRI used, citalopram was the only SSRI significantly associated with cardiovascular congenital defects [RR = 2.28 (95% CI: )] while fluoxetine + BZ significantly increased the risk for major congenital anomalies [RR = 5.18 (95% CI: )], although numbers were very small (n = 7 and n = 3, respectively). In addition, no association was detected between risk and 1 st trimester medication dose/day. Diav-Citrin et al. (2008) (202) reported on the risk of paroxetine and fluoxetine compared to nonteratogens exposure during pregnancy and major congenital anomalies. A total of 2,276 women who contacted the Israeli, Italy, and Germany Teratology Information Services (TIS) were followed-up (rates ranged: 44 to 91%). Among participants, 463 were exposed to paroxetine, 346 were exposed to fluoxetine and 1467 were exposed to nonteragens (controls). Rates of major anomalies were significantly greater in the paroxetine and fluoxetine group compared to controls (5.2% vs. 6.3% vs. 2.9%, p < 0.05), mostly from cardiovascular anomalies. No significant differences in rates of noncardiovascular anomalies were detected among the three groups (3.4% for paroxetine vs. 3.2% for fluoxetine vs. 2.4% for control). After adjusting for confounders, cardiovascular risk remained significant for fluoxetine [OR = 4.47 (95% CI: 1.31 to 15.27)] and smoking 10 cigarettes/day [OR = 5.4 (95% CI: 1.76 to 16.54)]. Neonatal Pulmonary Hypertension Andrade et al. (2009) (205) reported on a retrospective case-control study of 1104 women exposed to antidepressants (SSRIs, tricyclics, an miscellaneous) during the 3 rd trimester and matched controls (n = 1104) that found no significant association between SSRIs use during late pregnancy and PPHN in newborn. PPHN prevalence was 2.14 per 1000 [95% CI: 0.26 to 7.74] vs per 1000 [95% CI: 0.56 to 7.93] for infants whose mothers were exposed and not exposed to SSRI in the 3 rd trimester. However, the study was underpowered to detect a difference and other major methodological concerns limit the generalizability of results. 100 National Adult Depression Clincial Practice Guideline

105 Kallen and Olausson (2008) (203) assessed the effect of SSRI use during pregnancy and PPHN in newborns using data from the Swedish Medical Birth Register. A total of 506 infants were born between 1997 and 2005 with PPHN and after adjusting for maternal age, BMI, smoking status and parity, a significant association between maternal use of SSRI and PPHN in births after 34 weeks was detected in women who reported drug use in early pregnancy [RR = 2.38 (95% CI: ]. For women who reported SSRI use in early and later in pregnancy, the risk increased to 3.57 with wide confidence interval (95% CI: ). SSRI use among the women included citalopram, sertraline, fluoxetine, and paroxetine; however, the study was underpowered to evaluate the effect of specific SSRI on neonatal PPHN. Neonatal Behavioral Syndrome Maschi et al. (2008) (209) conducted a case-control study of 200 cases and 1,200 controls to investigate the association between antidepressant use during pregnancy and adverse effects, particularly, poor neonatal adaptation. After adjusting for possible confounders, results found no statistically significant difference in the incidence of poor neonatal adaptation in the exposed group compared to the unexposed control, although there was a trend for an increased risk in newborns exposed during the 2 nd and 3 rd trimester. However, the study was underpowered to detect small differences between the two groups and misclassification error could have occurred due to the conglomeration of conditions that constitute poor neonatal adaptation. Jordan et al. (2008) (201) investigated the severity of neonatal behavioral syndrome (NBS, e.g., irritability, jitteriness, hypo- or hypertonia, hyperreflexia, apnea, hypoglycemia, etc.) among infants of mothers with psychiatric illness taking SSRI during pregnancy (n = 46) compared to infants born to women with psychiatric illness not treated with medication (n = 59). No significant differences in the overall number of NBS (28% of SSRI-treated pregnancies vs. 17% in non-treated pregnancies), need for transfer to a higher level nursery for NBS (11% vs. 10%), or length of hospital stay (2 vs. 6 days) was detected between the two groups. However, study was underpowered with relatively small sample size. Summary of Findings for Neonatal Outcomes For infant adverse events, evidence of potential cardiovascular congenital defects was reported in three studies. One large cohort study found that multiple antidepressant redemptions (mostly SSRIs) and the combination of SSRIs and benzodiazepines increase the risk for cardiovascular congenital defects. Additionally, citalopram (n = 2), sertraline (n = 1) and fluoxetine (n = 1) were shown to be significantly associated with cardiovascular congenital defects, although the studies had small sample sizes which make it difficult to generalize results. The trend in recent studies suggests a possible SSRI class effect on associations with cardiovascular congenital defects. Conversely, two studies found no significant differences between antidepressant exposure and non-exposure for major congenital and congenital cardiovascular malformations. Two studies found conflicting results on the effect of antidepressant use and persistent pulmonary hypertension in newborns, with the one study suggesting an SSRI class effect (but being underpowered to detect if the effect was related to only specific SSRIs). 101 National Adult Depression Clincial Practice Guideline

106 Maternal Adverse Events Toh et al. (2009) (210) evaluated a subgroup of 199 women who participated in the multicentered case-control Slone Epidemiology Center Birth Defects Study, who received SSRI during pregnancy, and who gave birth to non-malformed and non-hypertensive newborns. Gestation hypertension was detected in 19.1% vs. 9% and of women with and without exposure [RR = 1.9 (95% CI: )] to SSRI and hypertension with preeclampsia was detected in 9% vs. 2.4% [3.16 (95% CI: )] in women with and without exposure. Adjusted relative risk for hypertension with preeclampsia for women who discontinued SSRI before the end of the 1 st trimester was 1.37 (95% CI: ) and 4.86 (95% CI: ) for those who continued SSRI. However, the study was underpowered to assess the attributable risk associated with specific SSRIs. Salkeld et al. (2008) (208) conducted a case-control to investigate the effect of antidepressant use during 3 rd trimester of pregnancy on postpartum hemorrhage among 2,460 cases and 23,943 matched controls. Results did not suggest a significant association between risk for postpartum hemorrhage and late-term exposure to SSRIs relative to non-ssris [adjusted OR = of SSRIs within 90 days before index date: 1.30 (95% CI: ) vs (95% CI: ) for non-ssris; p = 0.065]. Sensitivity analysis at 30, 60, and 180 days before index date also found consistent findings. However, cautious interpretation of outcomes is warranted due to unclear definition of postpartum hemorrhage and difficulty with drug adherence measurement. For maternal health, one study suggested no significant association between antidepressant use during pregnancy and postpartum hemorrhage while the other found increased in risk for hypertension with preeclampsia for women who continue on SSRI through late-pregnancy. There were several important concerns that warrant cautious interpretation of the results. First, most studies extracted data related to drug intake from pharmacy registries, interviews, or physician prescriptions, which are not indicative of actual drug adherence. Second, specific SSRIs could not be evaluated individually in some studies due to power limitation and were often lumped together by class of SSRI, tricyclics or antidepressant drugs in general. Third, major congenital malformations were defined differently among studies to either include or exclude congenital cardiovascular defects. Fourth, several outcomes such as PPHN and NBS do not have explicit criteria or diagnosis and frequently require expert evaluations. The combinations of these limitations plus heterogeneous study methodologies make it difficult to draw definitive conclusion. Overall Conclusion Overall, updated evidence combined with previously reviewed evidence suggests no significant associations between major neonatal non-cardiovascular congenital malformations (n = 5 new studies) and neonatal behavioral syndrome (n = 2 new studies) with antidepressant use during pregnancy. In addition, two new studies did not observe significant increases in risk associated with treatment dose or duration (Oberlander 2008 (204) ; Ramos 2009 (207) ). 102 National Adult Depression Clincial Practice Guideline

107 Newer studies provided conflicting evidence on neonatal cardiovascular congenital anomalies and neonatal pulmonary hypertension. Overall (including previously reviewed evidence) evidence continues to suggest an association of paroxetine use in pregnancy with congenital cardiovascular malformations; the newer studies reviewed raise the possibility of a potential SSRI class effect. Maternal adverse events were reported in two studies, one of which found no significant association between antidepressant use during pregnancy and postpartum hemorrhage while the other found increased in risk for hypertension with preeclampsia for women who continue on SSRI through late-pregnancy. The new available evidence is based on heterogeneous poor- to fair-quality studies, with limited assessment on attributable risk associated with specific SSRIs due to small sample size. Studies showed conflicting results. Combined with previously reviewed evidence, there still appears to be an association of paroxetine use during pregnancy and fetal cardiovascular malformations, therefore the 2008 recommendation against new starts of paroxetine in pregnancy remains unchanged. The newer studies suggest an association with pre-natal use of citalopram (two studies), sertraline and fluoxetine (one study each) and congenital cardiovascular malformations. Other new studies suggest (albeit not consistently) associations between SSRI use and fetal pulmonary hypertension and maternal pre-eclampsia. Consequently, the GDT has added a new recommendation recommending caution in starting other selective serotonin re-uptake inhibitors (SSRIs) in pregnant women. A single study, with detection bias limitations, suggests a synergistic association between prenatal use of SSRIs + benzodiazepines with fetal cardiovascular malformations compared with benzodiazepines alone. The GDT continues to recommend that the balance of risks and benefits of medication treatment and the risks of untreated depression be discussed with the mother. The recommendations for infant adverse effects due to antidepressant exposure from breastfeeding remain unchanged from 2008 due to no new evidence identified at this time Guideline New evidence was found, the recommendation remains unchanged. In our search exploring antidepressant efficacy for treatment of Major Depression in adults, we did not distinguish between adults in general and women who were pregnant or breastfeeding. In some antidepressant efficacy studies (particularly of newer medications) pregnant or breastfeeding women are excluded; however, they are not excluded from all studies. Our evidence search for this problem formulation did not identify any studies specifically relating to efficacy of antidepressants in pregnant or breastfeeding women. Thus, we have extrapolated from the general efficacy evidence that the efficacy of first-line treatment options for adults with Major Depression would be similar in pregnant or breastfeeding women. 103 National Adult Depression Clincial Practice Guideline

108 Five systematic reviews with meta-analyses of pooled data and a single case-control cohort study were selected for analysis. These publications addressed the adverse effects associated with the use of antidepressant medication on gestational outcomes. These outcomes included spontaneous abortion, congenital malformation, neonatal withdrawal syndrome or other neonatal central nervous system abnormality. There was some overlap of included studies in these metaanalyses; however, different outcomes were being evaluated. Please refer to Evidence Tables 13.1 and 13.2 for details. A meta-analysis of six cohort studies of women taking antidepressants in pregnancy (211) reported a rate of spontaneous abortion of 12.4% compared with 8.7% in women not taking antidepressants. The risk ratio was calculated to be (95% CI: 1.19 to 1.77). No significant difference in risk was seen between classes of antidepressants. (In this study, eight different antidepressants results were blended together). Lattimore et al. published a meta-analysis of data from nine prospective cohort studies of the effect of maternal SSRI use during pregnancy. (212) This meta-analysis did not analyze results by different medications in this class. It also included some studies in which intervention group women were using other psychotropic medications (primarily benzodiazepines); the extent to which this inclusion biases the results is unknown. Nonsignificant increases were seen in the rates of prematurity, low birth weight, special-care nursery admissions, and the diagnosis of poor neonatal adaptation. None of the effects cited in this study reached statistical significance at the p = 0.01 level. Moses-Kolko et al. reported that a meta-analysis of data from five cohort studies showed that compared with early or no SSRI exposure, late gestational exposure was associated with an overall risk ratio of 3.0 (95% CI: 2.0 to 4.4) for neonatal behavior syndrome. (213) This selflimiting syndrome has been characterized as including neonatal irritability, agitation, tremors, increased muscle tone and respiratory and digestive disturbances. 104 National Adult Depression Clincial Practice Guideline

109 A meta-analysis of nine studies of SSRI exposure in the first trimester and throughout pregnancy addressed outcomes including spontaneous abortion, major malformation, cardiovascular malformation, and minor malformation. (214) The authors reported an association between maternal use of any SSRIs and spontaneous abortion (12.5% vs. 7.6% OR = 1.7 (95% CI: 1.28 to 2.24)), but no statistically significant increased risk of major or minor malformations were detected. Bar-Oz, et al. (215) reported the results of a meta-analysis of seven studies that compared the pregnancy outcomes of women using paroxetine in the first trimester with those of women using other potentially teratogenic drugs or other antidepressants. They reported a 74% relative increase in the risk for cardiovascular malformation in the infants of women exposed to paroxetine in the first trimester compared with women who were not taking antidepressants (95% CI: 1.22 to 2.42). This was similar to the increase in risk of 70% for paroxetine exposure when compared with other antidepressants. The increase in risk for other abnormalities was not statistically significant. However, detection bias may have influenced the results, as significantly more women reported using paroxetine than other SSRIs for anxiety disorders, and women using SSRIs had approximately twice as many prenatal ultrasounds compared with women who took no antidepressants. This study also used teratogens and other anti-depressants as controls, no true controls. In response to earlier reports of persistent pulmonary hypertension of the newborn (PPHN) associated with maternal use of antidepressants, Chambers et al. (216) studied the occurrence of this diagnosis in the infants of women who used antidepressants during pregnancy. Data collected from 377 women whose infants were diagnosed with PPHN was compared with data collected from 836 controls. The two groups were comparable in terms of maternal education, ethnicity, age, body mass index, maternal diabetes, smoking status, and NSAID use. Fourteen infants with PPHN were born to mothers who had used SSRIs after the 20th week of gestation, compared with six infants in the control group (OR = 6.1, 95% CI: 2.2 to 16.8). Please refer to Table Two meta-analyses were identified in the peer-reviewed medical literature that addressed the potential risks associated with antidepressant use during lactation. Please refer to Table 13.3 for details. Rubin et al. (217) collected 129 publications on breastfeeding and maternal use of CNS-acting medications. Their findings indicated that the majority of CNS-acting drugs, including all antidepressants do not reach a concentration of greater than 10% of the maternal dosage when ingested via breast milk. 105 National Adult Depression Clincial Practice Guideline

110 Weissman et al. (218) pooled the data in 57 studies of medication levels in the breast milk of nursing mothers using antidepressant medication. The infant plasma level of most medications (an intermediate, biological outcome) was found to be less than 10% of that of the mother. However, 22% of infants exposed to fluoxetine and 17% of the infants exposed to citalopram demonstrated plasma drug levels that exceeded 10% of the maternal level (a level that researchers have adopted by consensus as defining an elevated level * ). In addition to weaknesses in the Bar-Oz study noted above, recommendations on paroxetine are based on meta-analyses of studies that have not been evaluated for quality. The evidence about fluoxetine and citalopram in breastfeeding is insufficient because the outcome variable is intermediate, not clinical. The studies showing increased risk to the fetus with SSRIs are insufficient because they are somewhat inconsistent, mostly lump all SSRIs together, and the ORs are < 2.0, which is often taken as a threshold for inferring causation from an observational study. There is significant heterogeneity for several studies (Hemels, Lattimore, Rahimi), and the model used for analysis was not known for the other two (Moses-Kolko, Bar-Oz). There were significant biases or potential biases cited for most of the meta-analyses. Therefore, all of the recommendations in this problem formulation are consensus-based, because of the overall insufficient nature of the evidence. Other Considerations The baseline rate of all fetal malformations is 3%; for fetal cardiac malformations the baseline rate is approximately 1%. Ten to 25% of pregnant women and 10 to 20% of postpartum women experience an episode of MDD. (219, 220) Untreated or relapsing depression during this time is associated with adverse maternal outcomes (including suicide), increased risk of obstetrical complications (212) low birth weight, premature birth, and fetal loss (221) and short- and long-term neurological and cognitive-behavioral issues in these children, including increased risks of (212, 222) depression later on in the child. One study (223) suggests that up to 68% of women who stop antidepressants during pregnancy will experience a relapse of depression, compared with 26% of women who continue treatment (NNH = 2.4). Therefore, the GDT believes that, when treating women who are pregnant or breastfeeding, the clinician should: Discuss pregnancy and what is known regarding antidepressant teratogenicity, perinatal, neonatal and long-term effects. Discuss what is known regarding effects of maternal depression on pregnancy and infant development. Advise that pregnancy does not protect against the onset or effects of depression. For patients already taking antidepressants, discuss risks of depression relapse. The Bar-Oz meta-analysis shows an increased risk of cardiac fetal abnormalities associated with paroxetine exposure. In addition, the FDA has required a black box warning for fetal abnormalities associated with the use of paroxetine in pregnant women. Therefore, the GDT has made specific recommendations against new paroxetine starts during pregnancy. * Hale TW. Medications and Mothers Milk, 12 th edition , Amarillo, TX: Hale Publishing National Adult Depression Clincial Practice Guideline

111 Overall Conclusion: Potential adverse affects of prenatal antidepressant exposure include increased risk of fetal malformations (for paroxetine), transient pulmonary complications, and transient neonatal behavioral syndromes. Long-term risks to the child from exposure in-utero and breast feeding seem to be relatively small, based on limited observational evidence. Based on available evidence, the risk of untreated depression seems to outweigh the small increase risk to the fetus from in-utero antidepressant exposure. Avoidance of new starts of paroxetine during early pregnancy seems prudent due to the increased risk in cardiac malformations compared with other antidepressants, especially in light of an FDA black box warning on paroxetine for use in pregnancy. If the mother is already using paroxetine when pregnancy is detected, the risk of teratogenicity due to exposure may have already occurred; hence increased fetal surveillance for malformations seems clinically indicated. The potential risk of depression relapse for the mother seems to outweigh the small increased fetal risk of continued antidepressant exposure; however, the GDT recommends that the balance of risks and benefits be discussed with the mother. Evidence for actual infant adverse effects due to antidepressant exposure from breastfeeding is limited at this time. In theory, higher levels of infant antidepressant exposure could increase the risk for adverse effects. Therefore, given the availability of other antidepressant options, in breastfeeding women, avoiding new starts of medications with long half-lives (fluoxetine) or increased neonatal plasma levels (fluoxetine, citalopram) seems prudent, although the concerns may mostly be theoretical, based on small numbers of infants studied, and are of unknown clinical significance. However, for women who wish to breastfeed who are continuing antidepressants taken during pregnancy, most fetal exposure will have already occurred in-utero. In the absence of neonatal behavioral syndromes or pulmonary symptoms, the risk of antidepressant discontinuation for the mother and the effect on maternal-fetal bonding is probably greater than neonatal risks of continued exposure during breastfeeding. However, the GDT also recommends a discussion of the balance of risks and benefits with mothers in this situation. 107 National Adult Depression Clincial Practice Guideline

112 Appendix A: Criteria for Grading the Evidence 108 National Adult Depression Clincial Practice Guideline

113 109 National Adult Depression Clincial Practice Guideline

114 Appendix B: Supporting Documentation 1. First-Line Treatment of Major Depressive Disorder (MDD) Problem Formulation 1 Clinical Question: Intended Use of the Guideline: Population: Health Problem: What first-line treatment methods should be used to treat adults with MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients with Major Depression MDD Health Intervention: Antidepressant treatment (SSRIs, TCAs, DAs, SNRIs, NRIs) * Psychotherapy (Interpersonal Therapy, Cognitive Behavioral Therapy, Problem-Solving Therapy) Combination of antidepressants and psychotherapy No treatment Practitioners: KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Setting: Outpatient office visit, emergency department, urgent care clinics Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention: Change in symptoms Quality of life Missed school/work days Sexual problems Drowsiness Headache Nausea Insomnia Agitation/ nervousness Dry mouth Seizures Elevated blood pressure Constipation Diarrhea Office/UCC/ER visits Hospitalizations Mortality Abdominal pain Dizziness Blurred vision Weight gain GI bleeding Fall * SSRI = Selective Serotonin Reuptake Inhibitor TCA = Tricyclic Antidepressant DA = Dopamine Agonist SNRI = Serotonin Norepinephrine Reuptake Inhibitor NRI = Norepinephrine Reuptake Inhibitor Cognitive Behavioral Therapy = Brief structured treatment, incorporating elements of cognitive therapy and behavioral therapy. Behavioral therapy is based on learning theory and concentrates on changing behavior. Interpersonal Therapy = Standardized form of brief psychotherapy primarily intended for outpatients with unipolar nonpsychotic depressive disorders. It focuses on improving the person s interpersonal functioning and identifying the problems associated with the onset of the depressive episode. Problem-Solving Therapy = Consists of three stages: (1) identifying the main problems for the person; (2) generating solutions; and (3) trying out the solutions. 110 National Adult Depression Clincial Practice Guideline

115 Search Strategy Database: PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Search Date 8/2005 to 9/2007 9/2007 to 10/2009 No. Included / Total Retrieved * 0/129 7/138 PubMed National Institute for Clinical Excellence (NICE) Cochrane Clinical Evidence "Antidepressive Agents/therapeutic use"[ All Fields] OR "Antidepressive Agents/therapeutic use"[mesh] OR "psychotherapy"[mesh] OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms]) OR "interpersonal therapy"[all Fields] OR ("Problemsolving"[MeSH Terms] AND "therapy"[mesh Subheading])) Depression Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Only items with abstracts, Humans, Metaanalyses, English, All Adult: 19+ years Systematic Review, Meta-analyses and RCTs Systematic Review, Meta-analyses Systematic 8/2005 to 9/2007 9/2009 to 10/2009 Dec. 6, /222 0/70 N/A 6/ N/A Q4, /131 Reviews / Systematic Jan N/A Reviews and RCTs /2 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 111 National Adult Depression Clincial Practice Guideline

116 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * Hand Searches N/A Systematic Reviews and Meta-analysis 2007 to /1 PubMed ("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "Major Depressive Disorder" [All Fields]) AND ("Antidepressive Agents/adverse effects"[mesh] OR "Antidepressive Agents/therapeutic use"[all Fields] OR "Antidepressive Agents/therapeutic use"[mesh] OR "psychotherapy"[mesh] OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms]) OR "interpersonal therapy"[all Fields] OR ("Problemsolving"[MeSH Terms] AND "therapy"[mesh Subheading])) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Randomized, controlled trials, All adults: 19+ years, English, Human 01/01/03-08/01/05 17/236 Cochrane Depression Systematic reviews 03/05/03 1/295 Clinical Evidence Depression Systematic reviews and RCTs 03/05/03 1/ National Adult Depression Clincial Practice Guideline

117 No. Database: Search Terms: Article Type and Other Limits: Search Date Included / Total Retrieved * PubMed (((((((((((depression/drug Clinical trials, /222 therapy[mesh] OR depression/therapy[mesh]) OR ("depressive disorder/drug therapy"[mesh] OR "depressive disorder/therapy"[mesh])) OR ("dysthymic disorder/drug therapy"[mesh]or "dysthymic disorder/therapy"[mesh])) AND ("Serotonin Uptake Inhibitors"[MESH] OR "Antidepressive Agents, Tricyclic"[MESH])) AND notpubref[sb]) AND Randomized Controlled Trial[ptyp]) AND English[Lang]) AND notpubref[sb]) AND "adult"[mesh Terms]) All adults 19+ years, English, Human - 03/2001 ("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR Meta-analysis, All Adult: 19+ years English, Human 01/01/01-04/01/03 1/5 "Major Depressive Disorder"[All Controlled 01/01/01 4/124 Fields]) AND ("Antidepressive Trials, All - Agents/adverse effects"[mesh] OR Adult: 19+ years 04/01/03 "Antidepressive Agents/therapeutic English, Human use"[mesh] OR "psychotherapy"[mesh] OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms]) OR "interpersonal therapy"[all Fields] OR ("Problemsolving"[MeSH Terms] AND "therapy"[mesh Subheading])) Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. Casacalenda (1) study was indexed as a review article in PubMed and did not show up in the search results due to the way PubMed indexing is done. The GDTdecided to include this study, since it met the inclusion criteria. Other Information Source: Evidence report on the Treatment of Depression-Newer Pharmacotherapies. AHRQ (169) 113 National Adult Depression Clincial Practice Guideline

118 Evidence Tables First-Line Treatment of MDD Author & Title Mukai et al. (2009) Treatment of depression in the elderly: A review of the recent literature on the efficacy of singleversus dualaction antidepressants. Table 1.1: Summary of Systematic Reviews and Meta-Analyses Search Database / Method Databases: MEDLINE, PsycINFO, and PubMed (January 2003 January 2009). Search terms: Antidepressant, SSRI, SNRI, TCA, depression, randomized controlled trials, human trials, and individual antidepressant names. Outcome measures: Response (reduction of at least 50% from baseline scores on HDRS, MADRS, or 1-2 on CGI); remission (MADRS <9 to <12 or HAMD<7 to <10); and tolerability (withdrawals due to AEs) Method: - Meta-analysis was not conducted; - no assessment of publication bias (e.g. funnel plot analysis); - no information on study funding was reported Study (N) Characteristics N: 18 RCTs (10 SSRIs head-to-head or placebo; 2 TCAs vs. SSRIs; and 6 SNRIs vs. placebo, TCA or SSRI) % F: Participants primarily female Duration: wks Sample size: range Inclusion criteria: published double-blind design; placebo control or active comparator group; aged 59 years; patients with a diagnosis of MDD; published meeting abstracts; publications in any language; studies conducted in the community, nursing, home, outpatient, and hospital settings; studies that enrolled patients with comorbid dementia or medical illnesses; and studies of maintenance therapy for depression. Published meeting abstracts that meet the criteria above were also considered. Results 2 comparative trials found no significant difference in efficacy between TCAs vs. SSRIs 3 studies found no significant difference between venlafaxine vs. SSRIs 1 study of duloxetine vs. placebo found significant improvement in depression (reductions in HAM-D and GDS, p<0.001) than placebo with addition reduction in pain (p<0.001) and cognition (p=0.013). Data from 5 studies using various measures (including changes in MADRS, HAM-D, or Geriatric Depression Scale [GDS] scores; response rates; and remission rates) suggested no additional efficacy benefit for the SNRI venlafaxine compared with SSRIs or TCAs. Conclusions/ Limitations Limited data suggest that dual-action agents such as TCAs and SNRIs do not appear to confer any additional efficacy benefits over singleaction agents such as SSRIs in the treatment of depression in the elderly. Limitations: heterogeneity among study sample, design, dosing and duration; inconsistent diagnosis of MDD or comorbididites, outcome definitions and measures; underpowered; small number of studies included; limited head-to-head comparison studies; low generalizability, no attempt to look for unpublished studies. 114 National Adult Depression Clincial Practice Guideline

119 Author & Title Gartlehner et al. (2008a) Comparative benefits and harms of secondgeneration antidepressants: background paper for the American College of Physicians. Search Database / Method Databases: MEDLINE, EMBASE, PsychLit, Cochrane, International Pharmaceutical Abstracts ( April 2007). Examined Center for Drug Evaluation and Research database (CDER) for unpublished research. Invited pharma to submit dossiers. Search strategy: AHRQ assisted in formulating key questions and data analysis. Search terms: ( adverse events, harms, drug reactions, toxicity, with major depressive disorder [MeSH] and 12 specific second-generation antidepressants (bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, sertraline, trazodone and venlafaxine). Humans and English language. Outcome measures: - Outcomes of Overall rate of adverse events (AEs) - Discontinuation because of AEs - Specific AEs (e.g. gastrointestinal symptoms, weight gain, dizziness and others) - Severe AEs (e.g. suicidality, hyponatremia, sexual dysfunction and others) Study (N) Characteristics N: 203 studies [including 105 head-tohead trials; 66 placebo trials; 6 SR; 23 observational studies (>740,000 subjects); 15 studies with pooled data. Inclusion criteria: - Population: Adult inpatients and outpatients with MDD. - Intervention: 12 AD listed previously - Study design: Experimental and observational head-tohead studies - Minimum study duration: 6 wks - Minimum sample size: none for experimental designs; n 100 for observational studies Results Of the 203 included studies, 140 (69.0%) were financially supported by pharmaceutical companies and 19 (9.3%) by governmental agencies or independent funds. For 44 (21.7%) studies, we could not determine the funding source. Note: the 7 studies that suggest some differences with respect to onset of action were all supported by manufacturer of mirtazapine and after 4 weeks of treatment, most response rates were similar. Conclusions/ Limitations Current evidence does not warrant the choice of one second-generation antidepressant over another on the basis of differences in efficacy and effectiveness. Nevertheless, there were some differences with respect to onset of action and adverse events that may be relevant for the choice of a medication. Limitations: low generalizability due to highly selected patient populations in efficacy studies; indirect comparisons due to lack of headto-head evidence resulted in low power and wide confidence intervals; residual publication bias; inconsistency and variability among studies in terms of drug dose and duration of treatment; small number of studies for individual comparisons. 115 National Adult Depression Clincial Practice Guideline

120 Author & Title Search Database / Method Method: Indirect comparison was calculated using metaregressions of placebocontrolled trials using individual drugs as covariates. Study (N) Characteristics Results Conclusions/ Limitations When number of trials was insufficient for metaregression, modified network meta-analysis was used. - meta-analysis of headto-head comparisons (both random and fixedeffects model) - test of heterogeneity (I 2) index) - adjusted direct comparison using metaregressions of placebo trials using individual drugs as covariates - when there s insufficient trials available for metaregressions, used modified network metaanalysis - publication bias assessed by funnel plots 116 National Adult Depression Clincial Practice Guideline

121 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 117 National Adult Depression Clincial Practice Guideline

122 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 118 National Adult Depression Clincial Practice Guideline

123 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 119 National Adult Depression Clincial Practice Guideline

124 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 120 National Adult Depression Clincial Practice Guideline

125 Author & Title Gartlehner et al. (2008b) Comparative risk for harms of secondgeneration antidepressants: A systematic review and metaanalysis. Search Database / Method Databases: MEDLINE, EMBASE, PsychLit, Cochrane, International Pharmaceutical Abstracts ( April 2007). Examined CDER for unpublished research. Invited pharma to submit dossiers. Search strategy: ( adverse events, harms, drug reactions, toxicity, with major depressive disorder [MeSH] and 12 specific second-generation antidepressants (bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, mirtazapine, nefazodone, paroxetine, sertraline, trazodone and venlafaxine). Humans and English language. Outcome measures: - Outcomes of Overall rate of adverse events (AEs) - Discontinuation because of AEs - Specific AEs (e.g. gastrointestinal symptoms, weight gain, dizziness and others) - Severe AEs (e.g. suicidality, hyponatremia, sexual dysfunction and others) Study (N) Characteristics N: 104 studies [81 head-to-head trials (>17,000 subjects) ; 21 observational studies (>740,000 subjects)] Inclusion criteria: - Population: Adult inpatients and outpatients with MDD. - Intervention: 12 AD listed previously - Study design: Experimental and observational head-tohead studies - Minimum study duration: 6 wks - Minimum sample size: none for experimental designs; n 100 for observational studies Results General tolerability AE profiles of second-generation antidepressants were similar with nausea, vomiting, diarrhea, dry mouth, sweating, headache, dizziness, sexual dysfunction and weight gain were commonly reported AEs. However, individual drugs differed in frequencies of specific AEs. About 63% of patients in efficacy trials experienced at least one AE. Discontinuation rates Overall, about 15% of patients treated with 2nd-gen antidepressant discontinue because of intolerable AEs. No significant difference was detected between 2nd-gen antidepressants and SSRIs (refer to Figure 2). However, pooled estimated indicated higher discontinuation from AEs for patients taking venlafaxine than SSRIs(RR: 1.42 (95% CI: )] Gastrointestinal AEs (refer to Figure 3) Venlafaxine had a statistically significantly higher rate of nausea and vomiting than SSRIs as a class [RR: 1.53 (95% CI: ); NNH = 9 (95% CI 6, 23)]. Other AEs Compared with other 2nd-gen antidepressants, paraoxetine frequently led to higher and bupropion to lower rates of sexual dysfunction; mirtazapine and paraoxetine led to higher weight gains; and sertraline to higher rates of diarrhea. However, differences did not lead to different discontinuation rates. Conclusions/ Limitations Overall, the spectrum of AEs was similar between different 2 nd - gen antidepressants; however, different frequencies of specific AEs might be clinically relevant and influence the choice of treatment. Limitations: comparative risk for rare AEs were insufficient to draw firm conclusion; heterogeneity between experimental and observational studies; low quality assessment and misclassification of AEs based on variation in definitions; poor reporting lack of objective scale; under-reporting in some studies; and evidence was too sparse to assess risk of harms for all 66 possible comparisons between 2 nd -gen antidepressants. Method: - when RCTs sufficient, meta-analyses (fixed and random effects) of RR = was conducted (only reported random effects); 121 National Adult Depression Clincial Practice Guideline

126 Author & Title Search Database / Method otherwise, evidence were synthesized qualitatively - test of heterogeneity (I 2 ) was applied to both random and fixed effects model - sensitivity analysis was conducted when I 2 >60% according to population characteristics, dosages or drug formulations - When pooled outcomes were significant, NNH was calculated - publication bias was assessed using funnel plots - due to lack of data on individual comparisons, assessments were made with SSRI as a class Study (N) Characteristics Results Conclusions/ Limitations Cipriani et al. (2009a) Comparative efficacy and acceptability of 12 newgeneration antidepressants: a multipletreatments metaanalysis Databases: Cochrane collaboration depression, anxiety and neurosis review group controlled trial registers (1991 Nov, 30, 2007). Asked pharmaceutical companies, regulatory agencies, and study investigators to supply all available information. Search terms: Not specified. Outcome measures: Response (>50% reduction of baseline N: 117 RCTs (25,928 subjects) % F: 64% Mean duration: 8.1 wks Mean sample size: (range: 9-357) 63% trials were done in Europe and North America. 53 studies included people <65 years. Inclusion criteria: only RCTs that compared any of the following 12 new-generation Only 7 of 117 trials were in primary care settings; 15 unpublished studies from industry were included, it was not specified how many of these favored the funder s antidepressant and it was not clear how many of these studies were also included in the FDA trial database No funnel plots or assessment of publication bias was noted Heterogeneity Overall, heterogeneity was moderate, although for most comparisons the 95% CI included values that showed very high or no heterogeneity, reflecting the small number of included studies for each pair-wise comparison. Efficacy Escitalopram, mirtazapine, sertraline, and venlafaxine were significantly more efficacious than duloxetine, fluoxetine, fluvoxamine, paroxetine, and reboxetine (refer Figure 3) Reboxetine was significantly less efficacious than all the other 11 antidepressants. Escitalopram and sertraline might be the best choice when starting a treatment for moderate to severe major depression because they have the best possible balance between efficacy and acceptability. Reboxetine, fluvoxamine, paroxetine, and duloxetine were the least efficacious and acceptable drugs, 122 National Adult Depression Clincial Practice Guideline

127 Author & Title Search Database / Method HAM-D, MADRS, or who scored much improved on CGI) and dropout rates during 8 wks of tx (dropouts). Comparative efficacy analyses for the 12 drugs were done using fluoxetine as reference drug. Method: - when dichotomous efficacy outcomes were not reported tx response at 8 wks was estimated - Responders to treatment were calculated on an intention-to-treat basis - pair-wise meta-analyses using random-effects model - multiple-treatments meta-analysis - heterogeneity was assessed using forest plots - sensitivity analyses based on dose and imputation were conducted - meta-regression analysis was conducted to examine sponsorship effect Study (N) Characteristics antidepressants (bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, milnacipran, mirtazapine, paroxetine, reboxetine, sertraline, and venlafaxine) as monotherapy in acutephase tx of adults with unipolar major depression. Study quality was rated: adequate, unclear, or inadequate, according to the adequacy of the random allocation concealment and blinding. Studies that scored adequate or unclear were included. Exclusion criteria: placebo groups present and RCTs of women with post-partum depression Results When using Fluoxetine as reference: Mirtazapine, escitalopram, venlafaxine, and sertraline were more efficacious than fluoxetine, and fluoxetine was more efficacious than reboxetine Acceptability duloxetine and paroxetine were less well tolerated than escitalopram and sertraline; fluvoxamine less well tolerated than citalopram, escitalopram, and sertraline; venlafaxine less well tolerated than escitalopram; reboxetine less well tolerated than many other antidepressants, such as bupropion, citalopram, escitalopram, fluoxetine, and sertraline; and escitalopram and sertraline were better tolerated than duloxetine, fluvoxamine, paroxetine, and reboxetine. fluoxetine was better than reboxetine Ranking of Efficacy and Acceptability (refer to Figure 4) Mirtazapine, escitalopram, venlafaxine, and sertraline were among the most efficacious treatments, and escitalopram, sertraline, bupropion, and citalopram were better tolerated than the other remaining antidepressants Cumulative probabilities of being among the four most efficacious treatments were: mirtazapine (24 4%), escitalopram (23 7%), venlafaxine (22 3%), sertraline (20 3%), citalopram (3 4%), milnacipran (2 7%), bupropion (2 0%), duloxetine (0 9%), fluvoxamine (0 7%), paroxetine (0 1%), fluoxetine (0 0%), and reboxetine (0 0%). Cumulative probabilities of being among the four best treatments in terms of acceptability were: escitalopram (27 6%), sertraline (21 3%), bupropion (19 3%), citalopram (18 7%), milnacipran (7 1%), mirtazapine (4 4%), fluoxetine (3 4%), venlafaxine (0 9%), duloxetine (0 7%), fluvoxamine (0 4%), paroxetine (0 2%), and reboxetine (0 1%). Conclusions/ Limitations making them less favorable options when prescribing an acute treatment for major depression. Furthermore, in terms of acceptability, reboxetine was the least tolerated agent among the 12 antidepressants and was significantly less effective than all the other 11 drugs. Therefore, reboxetine should not be used as a routine first-line acute treatment for major depression. Limitations: Short duration and follow-up; sponsorship bias; lack of information on randomization and allocation concealment; variations in multipletreatments and direct comparisons; selection bias (placebo-controlled trials mainly designed for regulatory approval tend to include patients with mild form of disease); residual confounding; numbers needed to treat or harm were not calculated, making it difficult to assess the absolute differences in each analysis; statistical significance (defined as p<0.05 in 123 National Adult Depression Clincial Practice Guideline

128 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations this analysis) was not adjusted for multiple comparisons, so many differences could have been due to chance. 124 National Adult Depression Clincial Practice Guideline

129 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations Cipriani et al. (2009b) Escitalopram versus other antidepressive agents for depression (review). Cochrane review. Databases: The Cochrane Depression, Anxiety and Neurosis Group (CCDAN) Controlled Trials Register was searched, together with a supplementary search of MEDLINE, PsycINFO, EMBASE, LILACS, CINAHL, PSYNDEX and hand searches. Trial databases of drug-approving agencies were hand- N: 22 RCTs (14 compared escitalopram vs. SSRI; 8 compared escitalopram vs. (venlafaxine, bupropion and duloxetine) % F: not reported Age: Adults >18 years Duration: 6 24 wks Mean sample size: (range ) Escitalopram was shown to be significantly more effective than citalopram in achieving acute response (OR = 0.67, 95% CI 0.50 to 0.87). Escitalopram was also more effective than citalopram in terms of remission (OR = 0.53, 95% CI 0.30 to 0.93). Significantly fewer patients allocated to escitalopram withdrew from trials compared with patients allocated to duloxetine, for discontinuation due to any cause (OR = 0.62, 95% CI 0.38 to 0.99). Some statistically significant differences favoring escitalopram over other antidepressive agents for the acute phase treatment of MDD were found, in terms of efficacy (citalopram and fluoxetine) and acceptability (duloxetine). There is insufficient evidence to detect a difference 125 National Adult Depression Clincial Practice Guideline

130 Author & Title Search Database / Method searched for published, unpublished and ongoing controlled trials. (July 2008) Search terms: CCDANCTR-Studies: Diagnosis = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Intervention = Escitalopram. CCDANCTR-References: Keyword = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Free-Text = Escitalopram Outcome measures: The primary outcome was response (>50% reduction in HAM-D or MADRS or 1-2 on CGI scores) Study (N) Characteristics Inclusion criteria: RCTs comparing escitalopram against any other antidepressant (including nonconventional agents such as hypericum) for patients with MDD (regardless of the diagnostic criteria used); aged >18 years; studies in which less than 20% of the participants might be suffering from bipolar depression were included Exclusion criteria: Participants diagnosed with comorbid physical or mental conditions; concurrent primary diagnosis of Axis I or II disorders. Results Conclusions/ Limitations between escitalopram and other antidepressants in early response to treatment (after two weeks of treatment). However, the potential for overestimation of treatment effect due to sponsorship bias should also be borne in mind. Limitations: publication bias due to commercial sponsorship; unclear allocations; selective reporting present in many studies; 1) Early response: between 1 and 4 weeks, 2) Acute phase treatment response: between 6 and 12 weeks, 3) Follow-up response: between 4 and 6 months Method: - meta-analysis using random effects model - Responders and remitters to treatment were calculated on the intention-to-treat (ITT) basis - When there were 126 National Adult Depression Clincial Practice Guideline

131 Author & Title Search Database / Method missing data last observation carried forward (LOCF) method was used - heterogeneity was assessed using I 2 and forest plots - Funnel plot was used to assess publication bias - sensitivity analyses were also conducted to account for excluded trials, imputation, wish bias, and sponsorship Study (N) Characteristics Results Conclusions/ Limitations 127 National Adult Depression Clincial Practice Guideline

132 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 128 National Adult Depression Clincial Practice Guideline

133 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 129 National Adult Depression Clincial Practice Guideline

134 Author & Title Nakagawa et al. (2009) Milnacipran versus antidepressive agents for depression (review). Cochrane review. Search Database / Method Databases: Cochrane Collaboration Depression, Anxiety & Neurosis Controlled Trials Registers (CCDANCTR-Studies and CCDAN-References) (searched in December 2006; updated in August 2008). Trial databases of drug-approving agencies were hand-searched for published, unpublished and ongoing controlled trials. Hand searches also conducted. Search terms: CCDANCTR-Studies: Diagnosis = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Intervention = Milnacipran; CCDANCTR-References: Keyword = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Free- Text = Milnacipran. Outcome measures: Response (>50% reduction of baseline HAM-D, MADRS, or who scored much improved on CGI) Method: - meta-analysis using random effects model - Responders and remitters to treatment were calculated on the Study (N) Characteristics N: 16 RCTs (2277 subjects) % F: not reported Age: Adults >18 years Mean Duration: most studies were 6-12 wks Median sample size: 120 (range: ) Inclusion criteria: RCTs comparing milnacipran with any other active antidepressive agents (including nonconventional agents) as monotherapy in the acute phase of major depression; age >18 years. Exclusion criteria: Studies using ICD9 to diagnose depression; participants diagnosed with comorbid physical or mental conditions; concurrent primary diagnosis of Axis I or II disorders. Results No statistically significant differences in efficacy, acceptability and tolerability when comparing milnacipran with other antidepressive agents, with wide CIs. However, compared with TCAs, patients taking milnacipran were associated with fewer dropouts due to adverse events (OR = 0.55; 95% CI 0.35 to 0.85). There was also some weak evidence to suggest that patients taking milnacipran experienced fewer adverse events of sleepiness/ drowsiness, dry mouth or constipation compared with TCAs. Conclusions/ Limitations There is inadequate evidence to conclude whether milnacipran is superior, inferior or the same as other antidepressive agents in terms of efficacy, acceptability and tolerability in the acute phase treatment of major depression. However, there is some evidence in favor of milnacipran over TCAs in terms of dropouts due to adverse events (acceptability) and the rates of experiencing adverse events (tolerability). Limitations: data reporting of most studies was incomplete; selective reporting present in many studies; small number of studies included; may be underpowered to detect significant difference; no sensitivity analysis to examine other modulating effects; high drop-out rates noted which reduce outcome reliability; most studies were funded by industry. 130 National Adult Depression Clincial Practice Guideline

135 Author & Title Search Database / Method intention-to-treat (ITT) basis - When there were missing data last observation carried forward (LOCF) method was used - heterogeneity was assessed using I 2 and forest plots - Funnel plot was used to assess publication bias - sensitivity analyses were also conducted to account for excluded trials, imputation, wish bias, and sponsorship Study (N) Characteristics Results Conclusions/ Limitations 131 National Adult Depression Clincial Practice Guideline

136 Author & Title Omori et al. (2009) Efficacy, tolerability and side effect profile of fluvoxamine for major depression: Meta-analysis. Search Database / Method Databases: Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, CINAHL, pycinfo, PSYINDEX, LILACS and hand searches (thru June 2, 2006). Not limited to English language. Trial databases of drug-approving agencies were handsearched for published, unpublished and ongoing controlled trials. Pharma companies were asked to submit studies that meet inclusion criteria. Search strategy: Diagnosis or Keyword = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective symptoms and Intervention or Free-text = fluvoxamine. Outcome measures: Response at the end of acute phase (between 6 and 12 weeks) (< 50% reduction of HAMDS or MADRS) and tolerability (dropouts for any reason and side effects). Secondary outcome was remission (<7 HAMDS). Method: - RRs were calculated using random-effects model - if significant difference was detected, NNT was Study (N) Characteristics N: 53 RCTs (48 RCTs for efficacy analysis (N=6440); 49 for tolerability analysis (N=8244)) % F: not reported Age: > 18 years Duration: Mean 5.5 wks (range 2 10 wks) Sample size: 38 RCTs had <100 subjects Inclusion criteria: RCTs comparing fluvoxamine with all other active ADs in acute phase treatment of major depression in patients aged 18 or older. The diagnosis must have been made based on established operationalised diagnostic criteria such as DSM-IV. Exclusion criteria: depressive patients with primary diagnosis of other Axis I or Axis II disorders or a serious concomitant medical Illness; depression with psychotic features and those in which more than 20% of the participants suffered from bipolar depression; or fluvoxamine was used as an augmentation strategy. Results Treatment efficacy no significant differences in response and remission rates between fluvoxamine and other ADs as a class (TCAs, heterocyclics, etc.) (refer to table 2) in early or end of acute phase of treatment. Tolerability The total number of dropouts for any reason or for side effects was not significantly different between fluvoxamine and other ADs as a class and between fluvoxamine and individual comparator ADs (refer to table 3). Side-effect profile: there is evidence of differing side-effects profiles, especially when comparing gastrointestinal side effects between fluvoxamine and TCAs. Nausea or vomiting and weight loss or anorexia were experienced significantly more frequently with fluvoxamine than with TCAs and some of other ADs (mianserin, milnacipran and newer ADs). On the contrary, constipation and dry mouth were more common with TCAs than with fluvoxamine. Conclusions/ Limitations No substantial difference exists in the effectiveness between fluvoxamine and any of the ADs including TCAs, such as amitriptyline or clomipramine, in terms of response or remission in any clinical settings. in terms of patients acceptability, there was no difference in dropouts for any reason or for side effects between fluvoxamine and other ADs as a class (TCAs, SSRIs, etc.) or individually. Therefore, the initial selection of an antidepressant medication will and should largely be based on the anticipated side effect profile and patient s preference. Limitations: Probable publication bias; many studies were sponsored by pharmaceutical companies; effects maybe overestimated; several comparisons with small number of subjects were highly heterogeneous; error in reporting due to lack of standardized instruments to measure side effects; 132 National Adult Depression Clincial Practice Guideline

137 Author & Title Search Database / Method calculated - heterogeneity was assessed by I 2 and forest plots - ITT analysis was applied; otherwise, LOCF was used - publication bias assessed by funnel plot - sensitivity analyses for sponsorship, wish bias, and excluded trials were conducted - confidence interval was set at 99% and statitistical significance set at p=0.01 Study (N) Characteristics Results Conclusions/ Limitations and variability in reporting. 133 National Adult Depression Clincial Practice Guideline

138 Author & Title Cipriani et al. (2009c) Sertraline versus antidepressive agents for depression (review). Cochrane review. Search Database / Method Databases: MEDLINE (1966 to 2008), EMBASE (1974 to 2008), the Cochrane Collaboration Depression, Anxiety and Neurosis Controlled Trials Register and the Cochrane Central Register of Controlled Trials up to July Hand searches also conducted. Trial databases of drugapproving agencies were hand-searched for published, unpublished and ongoing controlled trials. Search terms: CCDANCTR-Studies: Diagnosis = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Intervention = Sertraline; CCDANCTR- References: Keyword = Depress* or Dysthymi* or Adjustment Disorder* or Mood Disorder* or Affective Disorder or Affective Symptoms and Free-Text = Sertraline. Outcome measures: Efficacy (the number of patients who responded or remitted), acceptability (the number of patients who failed to complete the study) and tolerability (side-effects). Study (N) Characteristics N: 59 RCTs (55 RCTS and 9303 available for efficacy; 57 RCTs and 9950 for tolerability) % F: not reported Age: Adults >18 years Mean Duration: not reported Sample size: 17 studies included <100 subjects Inclusion criteria: RCTs allocating patients with MDD to sertraline versus any other antidepressive agent; age >18 years. Exclusion criteria: Studies using ICD9 to diagnose depression; participants diagnosed with comorbid physical or mental conditions; concurrent primary diagnosis of Axis I or II disorders. Results A total of 59 studies, mostly of low quality, were included in the review, involving multiple treatment comparisons between sertraline and other antidepressant agents. Evidence favoring sertraline over some other antidepressants for the acute phase treatment of major depression was found, either in terms of efficacy (fluoxetine) or acceptability/tolerability (amitriptyline, imipramine, paroxetine and mirtazapine). However, some differences favoring newer antidepressants in terms of efficacy (mirtazapine) and acceptability (bupropion) were also found. In terms of individual side effects, sertraline was generally associated with a higher rate of participants experiencing diarrhea. Conclusions/ Limitations Results found a trend in favor of sertraline over other antidepressive agents both in terms of efficacy and acceptability, and further suggest that sertraline might be a strong candidate as the initial choice of antidepressant in people with acute MDD. Limitations: Overall low quality of included studies, not all prespecified outcomes were reported in included studies and outcomes of clear relevance to patients and clinicians were not reported in any of the included studies; no analysis of publication bias due to insufficient quantity of studies; only few studies reported remission rates underpowered to detect clinical significance; limited head-to-head comparison studies for individual ADs. Method: - meta-analysis using 134 National Adult Depression Clincial Practice Guideline

139 Author & Title Search Database / Method random effects model - Responders and remitters to treatment were calculated on the intention-to-treat (ITT) basis - When there were missing data last observation carried forward (LOCF) method was used - heterogeneity was assessed using I 2 and forest plots - Funnel plot was used to assess publication bias - sensitivity analyses were also conducted to account for excluded trials, imputation, wish bias, and sponsorship Study (N) Characteristics Results Conclusions/ Limitations 135 National Adult Depression Clincial Practice Guideline

140 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 136 National Adult Depression Clincial Practice Guideline

141 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 137 National Adult Depression Clincial Practice Guideline

142 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 138 National Adult Depression Clincial Practice Guideline

143 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 139 National Adult Depression Clincial Practice Guideline

144 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 140 National Adult Depression Clincial Practice Guideline

145 Author & Title Cipriani et al. (2008) Does randomized evidence support sertraline as firstline antidepressant for adults with acute major depression? A systematic review and metaanalysis. Search Database / Method Databases: MEDLINE, EMBASE, Cochrane Collaboration for Depression, Anxiety and Neurosis Register of Controlled Trials (1966- August 2007). ƒ Search terms: Diagnosis= depress* or dysthymic* or adjustment disorder* or mood disorder* or affective disorder or affective symptoms, and intervention (or free text)= sertraline. English and non-english-language articles. Outcome measures: Treatment response and treatment acceptability. - response was defined as proportion of patients who had reduction of at least 50% from baseline scores on HDRS or MADRS or much improved on CGI. -tx acceptability was proportion of patients who left the study by any cause during first 8 wks. Method: - Responders were calculated according to ITT basis or LOCF Method: - meta-analysis using random effects model - Responders calculated on ITT basis - When there were missing data last observation carried forward (LOCF) method Study (N) Characteristics N: 56 RCTs (8507 patients available for examining efficacy; 8387 available for acceptability) % F: Not reported Duration: 6-24 wks Sample size: 6 studies had <100 subjects Inclusion criteria: RCTs comparing sertraline with all other active antidepressants as monotherapy in acute phase (8-wk tx) tx of depression. Participants were both sex and > 18 years with primary diagnosis of MDD. Exclusion criteria: Quasi-randomized trials; concurrent medical disorder; postpartum depression. Results Treatment response No significant difference in efficacy was detected between sertraline and TCAs [RR: 0.95 (99% CI: )] or other antidepressants, and slightly better than SSRI (refer to Figure 3). Statistically significant difference in efficacy in favor of sertraline over SSRI [RR=0.88 (99% CI: ), p = 0.009; NNT = 17], particularly sertraline over fluoxetine [RR: 0.85 (99% CI: ); p = 0.004; NNT=12] Acceptability No significant difference in acceptability was detected between sertraline and TCAs [RR: 0.83 (99% CI: )], SSRI [RR: 0.9 (99% CI: )], or any other antidepressants (refer to Figure 4). High heterogeneity was detected for comparison of sertraline vs. fluvoxamine (I 2 =53.4 and 57.9%); vs. paroxetine (I 2 =64.2 and 59.3%) for efficacy and acceptability; and vs. other antidepressants for acceptability (I 2 =42.7%) Funnel plots did not suggest evidence of publication bias. In the conclusions section, the authors site cardiovascular physicians belief and clinical care practices in treating depression in patients with cardiovascular disease as observational evidence supporting their conclusion of sertraline as a candidate for initial choice of antidepressant. Conclusions/ Limitations Results found consistent although not statistically significant trend in favor of sertraline over other antidepressants, particularly SSRIs and fluoxetine. Sertraline may be a candidate as the initial choice of antidepressant treatment for people with MDD. Limitations: low generalizability; most included studies were funded by the maker of sertraline; high heterogeneity in some subgroup analysis; some demographic information on patient population were not available; different dosing schedules make it difficult to determine differences (or lack thereof) between equivalent effective doses of ADs; There was substantial overlap of studies in this analysis and the analysis conducted in Cipriani et al. (2009a) thus many of the same limitations (short 141 National Adult Depression Clincial Practice Guideline

146 Author & Title Search Database / Method was used - heterogeneity was assessed using I 2 - Funnel plot was used to assess publication bias - sensitivity analyses were also conducted to account for excluded trials, imputation, wish bias, and sponsorship - NNT was calculated when RRs were statistically significant - confidence interval was set at 99% and statitistical significance set at p=0.01 Study (N) Characteristics Results Conclusions/ Limitations duration of included trials, few trials conducted in primary care settings) apply; Trials including patients with medical disorders were excluded from this analysis, also limiting applicability to the primary care setting. 142 National Adult Depression Clincial Practice Guideline

147 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations Van den Broek et al. (2009) Efficacy of venlafaxine compared with tricyclic antidepressants in depressive disorder: A metaanalysis. Databases: PubMed, Cochrane database of Systematic Reviews, Cochrane Controlled Trial Register and DARE, and manual searches. If necessary, unpublished data were requested from authors. Search strategy: search terms were (( depressive Disorder [TIAB] NOT Medline[SB]) OR depressive Disorder [MeSH Terms] OR depression [MeSH Terms] OR Depression[Text Word]) AND N: 7 RCTs (N=947) % F: not reported Race not reported Age: not reported Duration: range 43 days 8 wks Sample size: range Inclusion criteria: Trials were included if they were double-blind randomized studies comparing venlafaxine with a TCA for the treatment of depression. No overall significance treatment effects were detected for response or withdrawal of TCA vs. venlafaxine (refer to tables 2 and 3). There was heterogeneity between studies in the estimates of response. 143 National Adult Depression Clincial Practice Guideline No significant difference in treatment effect between venlafaxine and TCAs was detected and analysis of tolerability by total treatment dropouts and dropouts because of side effects did not suggest poorer tolerability for venlafaxine or TCAs. However, because of the heterogeneity of the odds ratios, one cannot conclude that they are of equal efficacy

148 Author & Title Search Database / Method ( venlafaxine [Substance Name] OR Venlafaxine[Text Word]) AND ( imipramine [MeSH Terms] OR imipramine[text Word]) AND Randomised Controlled Trial[ptyp] AND humans [MeSH Terms]. Outcome measures: Response (50% reduction in HADRS or MADRS) Method: - analysis was based on ITT method - fixed and random-effect models were used for meta-analysis - heterogeneity was assessed using I 2 Study (N) Characteristics Results Conclusions/ Limitations Limitations: Small sample size; underpowered; short study duration; heterogeneous studies; publication bias assessment was not reported (no funnel plot); majority of studies were sponsored by venlafaxine manufacturer. 144 National Adult Depression Clincial Practice Guideline

149 Author & Title Weinmann et al. (2008) Re-evaluation of the efficacy and tolerability of venlafaxine vs. SSRI: metaanalysis. Search Database / Method Databases: Medline, EMBASE, PsycINFO, PSYNDEX, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, DARE and guideline databases and manual searches (1966-Mar 2007). Search strategy: combination of text and index terms as a modification of the search strategy of the Cochrane Depression and Anxiety group. Outcome measures: Response (50% reduction in HAM-D or MADRS) and remission (HAM-D < 7 or MADRS < 10). Method: - analysis was based on ITT method; otherwise LOCF was used - fixed-effect models was used for meta-analysis - heterogeneity was assessed using I 2 - sensitivity analyses were also preformed on age, type of trial, appropriateness of ITT, treatment setting, dose, and intervention vs. control. - funnel plot was used to evaluate publication bias Study (N) Characteristics N: 17 RCTs (2 had patients > 65 years; N=3523 for response analysis; N=3142 for remission analysis) % F: not reported Race not reported Age: adult >18 years Duration: ranged 6-24 wks (most were 6-8 wks) Sample size: range Inclusion criteria: double-blind RCTs in which venlafaxine was compared to citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine or sertraline with or without a placebo control; depression was diagnosed according to ICD, DSM or Research Diagnostic Criteria (RDC); at least 6 weeks of treatment; used HAM-D or MADRS as outcome parameters Exclusion criteria: conference abstracts unless full-text publication could be obtained; Long-term studies >6 months duration; more than 20% of participants had a primary diagnosis of dysthymia or more than 15% had a primary diagnosis of bipolar disorder; measures based on (CGI) or Patient Global Improvement (PGI) Results Remission rates There was no statistically significant difference between venlafaxine and the SSRI group [random-effect RR: 1.04, (95% CI: ); fixed RR: 1.07 ( ); NNT = 34; I 2 = 18%] Response rates Venlafaxine was not significantly superior to the SSRI group [random-effect RR: 1.05, (95% CI: ); fixed RR: 1.06 ( ); NNT = 27; I 2 =32%] Tolerability The total rate of treatment discontinuation did not differ between venlafaxine and SSRIs (RR: 1.05, 95% CI= , NNH=100). However, there were significantly more dropouts due to adverse effects in the venlafaxine vs. SSRIs group [RR=1.38, 95% CI: , NNH=32] Subgroup analyses did not demonstrate clinically significant effect of venlafaxine compared to SSRIs (refer to table 1) Conclusions/ Limitations The analysis does not support a clinically significant superiority of venlafaxine over SSRIs. However, higher rate of study withdrawal due to AEs was associated with venlafaxine. Limitations: inconsistent remission definitions from included studies; small number of head-to-head comparison studies; underpowered; possible publication bias; and presence of some heterogeneity; no 145 National Adult Depression Clincial Practice Guideline

150 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 146 National Adult Depression Clincial Practice Guideline

151 Author & Title Watanabe et al. (2008) Mirtazapine versus other antidepressants in the acutephase treatment of adults with major depression: systematic review and metaanalysis. Search Database / Method Databases: Cochrane collaboration depression, anxiety and neurosis review group controlled trial registers (CCDANCTR) (- June 2,2006). Trial databases of drug-approving agencies were handsearched for published, unpublished and ongoing controlled trials. Hand searches of references. Pharma companies were asked for studies that meet inclusion criteria. Search strategy: Search terms: depress*, dysthymic*, adjustment disorder*, mood disorder*, affective disorder, and affective symptoms ; and mirtazapine Outcome measures: Primary outcome was response (> 50% reduction in HAMD-D or MADRS scores); secondary outcome was remission (< 8 on HAM- D) Tolerability in terms of dropout rates due to any reason or due to AEs were also considered. Outcomes were stratified by early phase (at 2 wks after tx), conclusion of acute-phase (6-12 wks), and conclusion of continuation-phase (4-6 months) Study (N) Characteristics N: 25 RCTs (N=4842; 9 trials of TCA, 12 of SSRI, 2 of SNRI, and 2 of other) % F: Not reported Age : elderly (>65 years) included in 16 trials; no other info provided Median Duration: 6 wks (range 5-24 wks) Sample size: range Inclusion criteria: RCTs of mirtazapine for acute-phase tx; > 18 years; diagnosed with unipolar MDD based on DSM-IV or other explicit clinical criteria Exclusion criteria: Patients with depression with psychotic feature; 20% of participants with bipolar depression; concurrent Axis 1 or II disorders; presence serious concomitant medical illness. Efficacy Results Early phase (at 2 wks) (refer to Table 2 and Figure 2) Mirtazapine was not superior or inferior to TCAs in terms of response [RR: 0.9 (99% CI: )] or remission [RR: 0.87 (99% CI: )] Mirtazapine was superior to SSRIs in both response [RR: 1.36 (99% CI: ); NNT=11] and remission [RR: 1.68 (99% CI: ); NNT=25] o In addition, mirtazapine was significantly better than paroxetine in response [RR=2.02 (99% CI: ); NNT=8], but not in remission. o While mirtazapine was significantly better than sertraline in remission [RR=1.73 (99% CI: ); NNT=12], but not in response. Mirtazapine was significantly superior to SNRI in terms of response [RR: 1.77 (99% CI: ); NNT=6] but not in remission [RR: 2.21(99% CI: )] o No significant difference was detected in response and remission between mirtazapine and trazodone [response RR: 1.11 (99% CI: ); remission RR: 1.0 (99% CI: )] Sensitivity analysis (included only trials without imputed data) o Mirtazapine was superior to SSRI I response [N=1789; RR: 1.39 (99% CI: ); NNT=11)] and remission [RR: 1.78 (99% CI: ); NNT=17]; significantly better than paroxetine in response [N=726; RR: 2.02 (99% CI: ); NNT=8]; better than sertraline in remission [N=596; RR: 1.73 (99% CI: ); NNT=13] End of acute phase (at 6 wks) o o No statistical significant differences were detected, except for superior remission outcome in comparison of martazapine with paroxetine [RR: 1.34 (99% CI: ); NNT=10) However, significant heterogeneity and publication bias (p=0.01) was observed. End of continuation phase (at 24 wks) o Only 1 study examined mirtazapine with paroxetine at continuation phase and no significant differences were detected (refer to Table 2). Tolerability (Refer to Table 3) o There were no significant differences between patients treated with TCAs (RR: 0.87 (95% CI: )], SSRIs (RR: 1.07 (95% CI: )], SNRI (venlafaxine) [RR: 0.82 (95% CI: )], other AD (trazodone) [RR: 0.93 (95% CI: )] o o Subgroup analysis showed that patients treated with mirtazapine were more likely than those treated with sertraline to withdraw due to any reason [RR: 1.33 (95% CI: ); NNH=14] In terms of AEs, mirtazapine dropouts due to AEs were similar to SSRI [RR: 1.22 (95% CI: )], SNRI [RR: 0.59 (95% CI: )], and trazodine [RR: 0.66 (95% CI: )] o Subgroup analysis found mirtazapine had lower withdrawals due to AEs compared with sertraline [RR: 2.58 Conclusions/ Limitations Although mirtazapine is highly likely to have better efficacy profile than paraoxetine or venlafaxine in terms of early response, in view of similar efficacy of mirtazapine and other ADs, results suggest that clinicians should also focus on other practically or clinically relevant considerations such as differences in the side effect profiles, to tailor treatment to best fit an individual patient s needs. Limitations: None of the trials reported whether allocation concealment was adequately performed; publication bias due to pharmaceutical support for included studies; heterogeneity in study duration and dosing that may lead results to bias toward mirtazapine; and generalizability due to efficacy trials recruiting mainly symptomatic volunteers. 147 National Adult Depression Clincial Practice Guideline

152 Author & Title Search Database / Method Clinical assessments were categorized based on RR = difference of 0.1 as: A (mirtazapine is clinically better than comparator); B (not worse and probably better than comparator); C (uncertain); D (not better and probably worse); and E (worse than comparator) Study (N) Characteristics (95% CI: ); NNH=11] Results Assuming that a difference in RR = of more than 0.1 is clinically important difference in efficacy, mirtazapine was clinically not worse and probably better than paroxetine, sertraline and velafaxine and was uncertain to be clinically significantly better or worse than other ADs in efficacy at 2 week during treatment phase. After acute-treatment phase, mirtazapine was certainly clinically not worse and probably better than fluoxetine, paroxetine and venlafaxine; with uncertain inferiority or superiority relative to other ADs (refer to Table 4). Conclusions/ Limitations Method: - meta-analysis was based on random-effect model - applied ITT analysis; otherwise LOCF were used - Heterogeniety was assessed using I 2 - Funnel plot used to examine publication bias - sensitivity analyses were conduced for exclusion trials, imputation method, and sponsorship - p-value was set at <0.01 and CI 99% for statistical significance 148 National Adult Depression Clincial Practice Guideline

153 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 149 National Adult Depression Clincial Practice Guideline

154 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 150 National Adult Depression Clincial Practice Guideline

155 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 151 National Adult Depression Clincial Practice Guideline

156 Author & Title Arroll et al. (2009) Antidepressants versus placebo for depression in primary care (review). Cochrane review. Search Database / Method Databases: The Cochrane Depression, Anxiety and Neurosis Group (CCDAN) Controlled Trials Register was searched, together with a supplementary search of MEDLINE, PsycINFO, EMBASE, LILACS, CINAHL and PSYNDEX (Sept 2007). Authors of selected studies were asked if they knew of additional published or unpublished studies. Search terms: Not specified. Outcome measures: The primary outcome was depression reduction, and on a continuous measure of depression symptoms Method: - meta-analysis was based on random-effect model when I 2 >50% - applied ITT analysis; otherwise, missing data were approximated - Heterogeniety was assessed using I 2 - Funnel plot used to examine publication bias - sensitivity analyses were conduced for exclusion trials, imputation method, and sponsorship - p-value was set at <0.01 and CI 99% for statistical significance Study (N) Characteristics N: 14 RCTs (10 TCAs, 2 SSRIs, and 2 included both classes vs. placebo) % F: not reported Age: Adults Duration: 4 24 wks Mean sample size: Inclusion criteria: RCTs TCAs or SSRIs versus placebo in adults (>18 years); patients had to be recruited from a primary care setting. For continuous outcomes the HAM-D, MADRS was required. Exclusion criteria: Older Patients (>65 years); participants diagnosed with comorbid physical or mental conditions. Results Both TCAs and SSRIs are significantly more effective than placebo for both discrete and continuous outcomes. Evidence is clear for major depression and levels of depression greater than minor depression. Based on this evidence, tricyclic antidepressants could be considered and the dose kept at or below 100mg per day, and waiting at least 4 weeks for a response may be worth considering. There is no dose information on SSRIs and no comment provided on the appropriate duration of treatment for either TCAs or SSRIs. TCAs RR= 1.24 (95% CI: ) and NNT= 7 to 16 (median NNT 9) (patient expected event rate ranged from 63% to 26% respectively) SSRIs RR = 1.28 (95% CI: ) for SSRIs and NNT= 7 to 8 {median NNT 7) (patient expected event rate ranged from 48% to 42% respectively). An NNT of 7 means that one patient will benefit from treatment and six will not although up to half may get better on placebo. Adverse events There appeared to be more adverse effects with TCAs than with SSRIs, however rates of withdrawal from study medication due to adverse effects were very similar between the two antidepressant classes. Adverse effects not leading to medication cessation seemed to be more common with TCAs than SSRIs. NNH for withdrawal due to side effects ranged from 4 to 30 for TCAs (excluding three studies with no harmful events leading to withdrawal) and 20 to 90 for SSRIs. Conclusions/ Limitations Both TCAs and SSRIs are effective for depression treated in primary care, but more adverse effects can be expected with TCAs. Limitations: Most of the studies were supported by funds from pharmaceutical companies and were of short duration; publication bias were noted in funnel plot; some studies had small sample size; heterogeneity in intervention duration and treatment strategies. 152 National Adult Depression Clincial Practice Guideline

157 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 153 National Adult Depression Clincial Practice Guideline

158 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 154 National Adult Depression Clincial Practice Guideline

159 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 155 National Adult Depression Clincial Practice Guideline

160 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 156 National Adult Depression Clincial Practice Guideline

161 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 157 National Adult Depression Clincial Practice Guideline

162 Author & Title Hansen et al. (2008) Meta-analysis of major depressive disorder relapse and recurrence with secondgeneration antidepressants. Search Database / Method Databases: MEDLINE, EMBASE, the Cochrane Library, PsycINFO, International Pharmaceutical Abstracts, Center for Drug Evaluation and Research, FDA, and manual searches (January 1980 April 2007). CDER database was also searched for unpublished research submitted to FDA. Search strategy: Not specified. Outcome measures: Loss of response or remission (continuationphase relapse or maintenance-phase recurrence), defined as increased in HAM-D or MADRS scores above predefined cut-off point (set by study). Method: - meta-analysis was based on random-effect model - applied ITT analysis; otherwise LOCF were used - Heterogeniety was assessed using I 2 - Funnel plot used to examine publication bias Study (N) Characteristics N: 27 RCTs (4 head-tohead and 23 placebocontrolled trials) % F: not reported Mean Age : Duration: wks Range sample size: Inclusion criteria: head-to-head trials comparing one antidepressant with another and placebocontrolled trials; adult (>18 years); depressive illness patients that demonstrated response to treatment or remission; reported relapse or recurrence rates, regardless of whether participants were randomly assigned to treatment groups after successful acute-phase or continuation-phase treatment (that is, extension versus randomized substitution trials). For placebo-controlled evidence; only studies that randomly assigned participants after demonstrating either an acute-phase response or lack of relapse during the continuation phase were included. Results Trials shorter than one year: relapse prevention Pooled RR = of relapse was.54 (95% CI=.46.62), and the NNT to prevent one additional relapse over a mean time of 8 months was 5 (CI=4 6). Heterogeneity among these trials was moderate (I2=47%). Trials one year or longer: recurrence prevention Pooled RR = of recurrence was.56 (CI=.48.66) and the NNT to prevent one additional recurrence over a mean time of 16 months was 5 (CI=4 6). Heterogeneity among these trials was moderate (I2=30%). Adverse Events RR = of dropping out for any reason was statistically significantly lower for active treatment than for placebo (relative risk=.75, CI=.69.83). Loss to follow-up because of adverse events was not statistically significantly different between active treatment and placebo (relative risk= 1.42, CI= ). Conclusions/ Limitations This review confirms the benefits of continuation- and maintenance-phase treatment of major depression with second-generation antidepressants. Limitations: limited quality and quantity of head-to-head comparison studies and studies that addressed clinical question; low generalizability; residual publication bias - most studies were sponsored by pharmaceuticals; unclear demographic characteristics of included patients; some heterogeneity among studies; variations in dosage and treatment duration. 158 National Adult Depression Clincial Practice Guideline

163 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 159 National Adult Depression Clincial Practice Guideline

164 Author & Title Nelson et al. (2008) Efficacy of second generation antidepressants in late-life depression: A meta-analysis of the evidence. Search Database / Method Databases: MEDLINE, EMBASE, Cochrane (1966 August 2006). Meeting abstracts from geriatric psychiatric and psychiatric professional society meeting and hand searches were also conduced. Search strategy: Search terms: antidepressants, fluoxetine, sertraline, paroxetine, citalopram, escitalopram, venlafaxine, duloxetine, mirtazapine, bupropion, nefazodone, and trazodone Outcome measures: Response (>50% improvement from baseline on HAMD or MADRS); and remission (as defined by individual studies). Method: - meta-analysis was based on fixed-effect model - Heterogeniety was assessed using I 2 - Funnel plot used to examine publication bias - p-value was set at <0.20 and CI 95% for statistical significance Study (N) Characteristics N:10 RCTs (N=2377 patients) % F: mean ranged 46 76% Age : mean age ranged years Duration: 6-12 wks sample size: Inclusion criteria: Acute phase, parallel group, double-blinded, placebo-controlled trial of 2 nd -gen AD (non- TCAs) marketed in US; patients had nonpsychotic, unipolar MDD not associated with other medical disorder; patients in community dwelling and aged 60+ years; number of patients, outcomes and dropouts obtainable. Trials can include unpublished reports or poster. Results Efficacy Those assigned to drug treatment had significantly greater response [OR: 1.4 (95% CI: )] and remission [OR: 1.27 (95% CI: )] than placebo (refer to figure 2 and 4) Response rates were also higher in the longer trials [10-12 weeks OR: 1.73 (95% CI: )] compared to shorter trials [6-8 weeks OR: 1.22 (95% CI: )] NNT for response =13 and remission = 20 No single class of medication was superior to another as evidenced in the overlapping ORs Tolerability There was an increased in odds for discontinuation on medication [OR: 1.22 (95% CI: ); I2=48.2%], increased discontinuation due to AEs on medication [OR: 1.84 (95% CI: ); I2=61.1%] (refer to figure 3) Funnel plot did not illustrate presence of publication bias. Conclusions/ Limitations Second-generation Ads are more effective than placebo during acute treatment of adults >60 years in terms of response and remission, but the magnitude is small and variable. For every 100 patients treated, 8 would show a response and 5 a remission in excess of placebo and for every 2 patients who responded, one discontinued prematurely because of AEs. Therefore, clinical decision to employ Ads will need to weigh modest benefits with more robust effects on prevention of relapse or recurrence and potential AEs. In addition, duration of treatment appeared to response as well. Limitations: Possible patient selection bias; high heterogeneity; low generalizability; small number of trials included; underpowered; all trials were sponsored by manufacturer of one of the drugs; and possible residual confounding. 160 National Adult Depression Clincial Practice Guideline

165 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 161 National Adult Depression Clincial Practice Guideline

166 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 162 National Adult Depression Clincial Practice Guideline

167 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations Deshauer et al. (2008) Selective serotonin reuptake inhibitors for unipolar depression: a systematic review of classic longterm randomized controlled trials. Databases: MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials (1966- May 2007). Search terms: index terms associated with serotonin uptake inhibitors and the text terms SSRI, fluoxetine, Prozac, sertraline, Zoloft, paroxetine, Paxil, fluvoxamine, Luvox, citalopram and Celexa. N: 6 RCTs (1299 patients) % F: Not reported Duration: 6-8 months Mean sample size: 217 Inclusion criteria: Trials had to involve patients who had a diagnosis of major depression (by DSM- MD), aged >18 years, and who were randomly assigned to receive monotherapy with a SSRI or placebo; Trials reporting a 1- to 2-week placebo run-in period were eligible; there Treatment response Pooled analysis of the 6 trials showed that selective serotonin reuptake inhibitors were superior to placebo in treatment response at 6 8 months [OR: 1.66 (95% CI: ); I 2 = 63.9%] Subgroup analysis showed a statistically significant treatment effect among patients with depression who had no comorbidities [OR: 2.13 (95% CI ); I 2 = 76.8%] but not among those who had comorbidities [OR: 1.32 (95% CI ); I 2 = 30.8%] Remission Pooled difference between SSRIs and placebo was not statistically significant [OR: 1.46 (95% CI ); I 2 = 38%]. However, participants without comorbidities had a significantly higher remission rate if they were taking SSRIs than if they were taking placebo [OR: 2.06 (95% CI ); I 2 = 0%) while whereas the difference for participants with comorbidities was not statistically significant [OR: 0.87 (95% CI: ); I 2 = 0%] Statistically significant improvements in response to treatment but not in remission or acceptability after 6 8 months of SSRI therapy were observed with greater effects in patients without comorbidities. Limitations: Small number of studies identified; underpowered; weak studies due to unclear description of methodological issues, short follow- 163 National Adult Depression Clincial Practice Guideline

168 Author & Title Search Database / Method Outcome measures: Treatment response, remission, and treatment acceptability. - response was defined as proportion of patients who had reduction of at least 50% from baseline scores on HDRS - remission was defined by cut-point (score of <7 on HDRS) -tx acceptability was total number of drop-outs. secondary outcomes: quality of life; admission to hospital; psychotherapy; pharmacotherapy or electroconvulsive therapy; self-harm (including attempted and completed suicide); and back-towork status. Study (N) Characteristics were no comorbidity restrictions. Definition of long-term treatment was treatment over a period of at least 6 months. Trials were limited to those published in English. Results Treatment acceptability No statistically significant difference between SSRIs and placebo [OR: 0.87 (95% CI: ); I 2 =21.3%] Secondary outcomes None of the trials provided information on back-to-work status or the need for specific rescue therapies, including admission to hospital for psychiatric reasons, psychotherapy, pharmacotherapy or electroconvulsive therapy 5 studies reported quality of life at the end of the trial One of the 5 trials reported improvements in all domains of a multi-domain quality-of-life score; one reported a quality-of-life summary statistic favoring SSRIs (p < 0.01); the 3rd reported 10-cm visual analogue scale also favoring SSRIs; and one reported quality-of-life improvements restricted to mood subscales. 2 trials that reported on suicide or self-harm, there was a total of 1 completed suicide among patients receiving placebo and none among patients receiving SSRI Conclusions/ Limitations up, outcomes reported as response to treatment instead of full remission with preset cut-offs, trials could have overestimated active long-term interventions by using last-observation carried forward analysis, and possible publication bias (4 of 6 studies were commercially sponsored). Barbui et al. (2008) Effectiveness of paraoxetine in the treatment of acute major depression in adults: a systematic reexamination of published and unpublished data from randomized Method: - meta-analysis of RR = was based on randomeffect model - Funnel plot was used to assess publication bias - Heterogeniety was assessed using I 2 Databases: Cochrane collaboration depression, anxiety and neurosis review group controlled trial registers, GlaxoSmithKline Clinical Trial Register, MEDLINE, EMBASE ( /2006) were searched for published and unpublished trials. Search terms: Keyword = major N: 40 RCTs (6391 patients; 3704 received paroxetine vs placebo) % F: Not reported Mean duration: not reported Mean sample size: not reported 63% trials were done in Europe and North America. 53 studies included Dropouts No positive effect of paroxetine in terms of the proportion of patients who discontinued treatment for any reason [random effect RR = 0.99, 99% CI )] Efficacy Greater proportion of patients who received paroxetine had a statistical significant improvement of 50% or more compared to placebo [random effect RR = 0.83 (99% CI )]; no statistically significant between-study heterogeneity (I 2 = 9.2%). A statistically significant positive effect of paroxetine in terms of mean difference [standardized mean difference 0.31 (99% CI: 0.40 to 0.22)]; there was no statistically significant between study heterogeneity (I 2 = 31.5%). 164 National Adult Depression Clincial Practice Guideline paroxetine was not superior to placebo in terms of tolerability (participants discontinuing treatment for any reason) and was modestly better than placebo in terms of depression measures (paroxetine exerted a modest antidepressant effect relative to placebo,

169 trials. Author & Title Search Database / Method depression or depression and free-text = paroxetin* ; randomized controlled trial or random allocation or doubleblind method. Outcome measures: Treatment discontinuation was the primary outcome and response was secondary outcome. - response was defined as reduction of <50% from baseline scores on HDRS, MADRS, or who scored much improved on CGI. - tx discontinuation (tolerability) was proportion of patients who left the study early for any reason, because of side effects; with any adverse events; with any serious adverse events; patients who completed or attempted suicide or experienced worsening of suicidal thoughts or emotional liability. Method: - meta-analysis of RR = was based on randomeffect model - Heterogeniety was assessed using I 2 - sensitivity analysis was conducted to evaluate fixed-effect model - p-value was et at 0.01 for statistical significance Study (N) Characteristics people <65 years. Inclusion criteria: only RCTs hat compared any of the following 12 new-generation antidepressants (bupropion, citalopram, duloxetine, escitalopram, fluoxetine, fluvoxamine, milnacipran, mirtazapine, paroxetine, reboxetine, sertraline, and venlafaxine) as monotherapy in acutephase tx of adults with unipolar major depression. Exclusion criteria: placebo groups present and RCTs of women with post-partum depression Results Number needed to treat (NNT) to avoid 1 additional failure = 9 (99% CI: 7 14) Tolerability Significantly more patients assigned to receive paroxetine left the study because of side effects [random effect RR = 1.77 (95% CI: )]; there was no statistically significant between-study heterogeneity (I2 = 2.0%). Number needed to harm (NNH) = 17 (95% CI: 14 25); Significantly more patients assigned to receive paroxetine than of those given placebo reported any adverse event [random effect RR = 1.15 (95% CI: )]; there was no statistically significant between-study heterogeneity (I2 = 17.6%). NNH = 9 (95% CI: 7 11) (Figure 5). No statistically significant difference between paroxetine vs. placebo in terms of patients reporting any serious adverse event (Peto OR = 1.27, 95% CI ); no between-study heterogeneity detected (I2 = 0%). Suicidal tendencies Significantly more patients assigned to paroxetine experienced suicidal tendencies than of those given placebo (US Food and Drug Administration codes 1 to 9) [Peto OR = 2.55 (95% CI: )]; no significant between-study heterogeneity detected (I2 = 0%). NNH=142 (95% CI: ) No statistical significances were detected between trials using 20-mg dose vs. >20-mg dose; among patients with moderate to severe depression vs. mild depression; and published vs. unpublished studies. Funnel plot suggests possible publication bias. Conclusions/ Limitations 11% absolute risk difference). Limitations: Studies were not designed or powered to measure primary outcome; nonadherence; absolute numbers of patients with suicidal tendencies were low and definition was problematic; unclear whether suicidal tendencies are good proxy for suicide attempts. 165 National Adult Depression Clincial Practice Guideline

170 Author & Title Barbui et al. (2009) Selective serotonin reuptake inhibitors and risk of suicide: a systematic review of observational studies. Search Database / Method Databases: MEDLINE, EMBASE, (January 1990 to June 2008). Search strategy: Search terms: antidepressive agents or antidepressive agents second generation AND suicide or suicide attempted Outcome measures: ICD 9-10 definitions of completed or attempted suicide. Suicide attempts had to be sufficiently serious to have led to medical contact. Method: - meta-analysis of RR = was based on rfixed and andom-effect models - publication bias was evaluated by funnel plot - Heterogeniety was assessed using I 2 - sensitivity analyses and metaregression were used to examine study using completed suicide; formal diagnosis of depression, external control group; quality score >7, data for both adolescents and adults; age; and outlier studies. Study (N) Characteristics N: 8 RCTs (>200,000 patients) % F: 8 74% Age : 10+ Duration: not reported Mean sample size: not reported Inclusion criteria: observational cohort and case control studies that reported data on completed or attempted suicide among people exposed to SSRIs and among those who were not exposed to antidepressants; studies that reported relative risk [RR] estimates suitable for re-analysis; only studies that used ICD 9-10 definitions of completed or attempted suicide were retained. Study participants were of both sexes and any age with a diagnosis of major depression. Studies adopting proxy measures to identify patients with depression were included. Results metaregression suggested a promoting effect of SSRI exposure on the risk of suicide among adolescents and a protective effect among adults and elderly individuals Among adolescents, exposure to SSRIs significantly increased the risk of completed or attempted suicide [random effect OR: 1.92 (95% CI ); I 2 =0.0%] Among adults, SSRI exposure significantly decreased the risk of completed or attempted suicide [random-effect OR: 0.57 (95% CI ); I 2 =52.5%] Among elderly people (aged 65 or more years), exposure to SSRIs had a significant protective effect [randomeffect OR: 0.46 (95% CI ) I 2 =0.0%] visual inspection of the funnel plot suggested possible selection bias (lack of small studies that failed to show an excess risk associated with SSRI exposure) Subgroup analysis Among adults, no individual antidepressant was significantly associated with completed or attempted suicide. Among adolescents, exposure to paroxetine [random-effect OR: 1.77 (95% CI ); I2=48.1%] and venlafaxine [random-effect OR: 2.43 (95% CI ); I2=0.0%) was significantly associated with increased risk Conclusions/ Limitations Relation between exposure to SSRIs and the risk of suicide is influenced by age, and prescribing SSRIs to patients with major depression is safe. However, children and adolescents should be followed very closely because of the possibility of increased of risk suicidal thoughts and suicide. Paroxetine and venlafaxine may be better avoided for most adolescents. Limitations: Confounding and biases inherent to observational studies included; confounding by severity of illness, gender, age; presence of publication bias; small sample of studies included; underpowered to detect small differences between drugs lack of longterm head-to-head comparison studies. Patients may have had additional comorbidities that may have influenced selection of treatment and have unaccounted for effects on suicide risk. 166 National Adult Depression Clincial Practice Guideline

171 Author & Title Search Database / Method Study (N) Characteristics Results Conclusions/ Limitations 167 National Adult Depression Clincial Practice Guideline

172 Study Proudfoot, et al. (2004) Clinical efficacy of computerised cognitivebehavioural therapy for anxiety and depression in primary care: randomised controlled trial. Funding: NHS Executive London and Ultrasis UK Ltd. Table 1.2 Inclusion & Exclusion Criteria Inclusion: 274 general practice patients, suffering from depression, mixed anxiety and depression, or anxiety disorder (assessment by GHQ-12 and CIS-R) Exclusion: Concurrent psychological treatment or counseling. Limitations / Age Biases Outcomes measured by self-report. Patients not masked to treatment. No selective attention control. Data analyzers not blinded to treatment. Some data missing. Authors have commercial interests in the specific computerized programs. Unclear how many patients in usual care who did not receive pharmacotherapy received psychotherapy if many of these patients went untreated, it would bias results in favor of the intervention group. Intervention and dose N and Final N Duration Outcome Results NNT Participants randomized into: Rx1: Computerised CBT "Beating the Blues" (N = 146): 15-minute introductory video followed by eight 50- minute therapy sessions. These participants could also receive medication if the general practitioner chose to prescribe it, but not in person psychotherapy. Rx2: Usual Treatment (N = 128): Received whatever therapy general practitioner prescribed. 8 weeks Primary outcome measure was BDI-II score at eight weeks. Secondary outcome measures were Beck Anxiety Inventory, Work and Social Adjustment scale, ASQ CoNeg, ASQ CoPos. Received pharmacotherapy: Rx1: 55.5% Rx2: 55.5% (p not reported) At eight week post-randomization, computerised CBT group had significant improvement over Usual Treatment group in: BDI (p=.0006) Work and Social Adjustment (p=0.002) ASQ, CoNeg (p<0.0001) ASQ, CoPos (p<0.008) Change in BDI scores 3 mos. 5 mos. 8 mos. Rx Rx There was no significant difference found between groups on the BDI (p=0.06). There was no interaction effect found for pharmacotherapy or duration of illness, or severity of illness. BDI scores continued to decline over time (at 3, 5, and 8 month follow-up points), with lower scores in the computerized CBT group. 35% withdrawal rate in the intervention group. p value < < National Adult Depression Clincial Practice Guideline

173 Table 1.3 Study de Jonghe (2004) Psychotherapy alone and combined with pharmacotherapy in the treatment of depression Inclusion & Exclusion Criteria Inclusion: Outpatient adults with DSM-IV criteria for Major Depressive Disorder (with or w/out dysthymia), and baseline HAM-D score of points (mild to moderate depression). Limitations / Age Biases Different instruments produced varying success rates. Intervention and dose N and Final N Duration Outcome Results NNT p value Funding: Supported by an unrestricted grant from Wyeth Nederland. Exclusion: Standard clinical pharmacotherapy research exclusion criteria. Patients and treating physicians not blinded to treatment. Complex protocol may not be generalizable to the real world. More patients in the psychotherapy only arm were previously unresponsive to psychotherapy (26% vs. 14%, biasing study in favor of combined treatment. LOCF data analysis. Multiple measures; Bonferroni adjustment to significance level. Rx1: Psychotherapy alone (N = 107) using Short Psychodynamic Supportive Psychotherapy (SPSP). One refused intervention, therefore N = 106. Rx2: Combined SPSP and pharmacotherapy (N = 101). Antidepressant program was stepped (noradrenaline, SSRI, nortriptyline, and nortriptyline with lithium) to address nonresponse. Sixteen patients refused intervention, therefore N = 85. Primary outcome measure was difference in HAM-D score assessments between baseline and week 24. Secondary measures were scores on CGI-I, CGI-S, and SCL-D between baseline and week 24. Both treatment groups improved over time on several measures (per protocol effect sizes were minimal - moderate, , depending on the measure used, with the lowest effect size on the primary measure). No significant difference between groups on any measure in the Intention-to-treat sample. No significant differences in time to remission between the combined (129 days) and psychotherapy (138 days) groups (p=0.122) No statistically significant differences in dropouts between the two groups. Small but statistically significant differences in HAM-D score at week 12 (1.8 points, p=0.046) in the per protocol group and at weeks 12 and 24 (approximately 2.5 points, p=0.009 and p=0.046) in the observed cases (completers) sample. No difference in HAM-D remission rates in the per protocol sample, significant difference in the observed cases sample at week 12 only (ARR = 6.2%, NNT 16, p=0.043). Variable improvement in different secondary measures at different times in the secondary outcome measures in both the per protocol and observed cases samples. In both the Per Protocol and Observed Cases analyses, there were some small but statistically significant intra-group differences between baseline and week 24, in favor of combined therapy: Per Protocol: CGI-I (p<0.05) Per Protocol SCL-D (p<0.001) < 0.05 < Observed Cases: HAM-D (p<0.046) Observed Cases SCL-D (p<0.001) < < National Adult Depression Clincial Practice Guideline

174 Table 1.4 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Moore, et al. (2005) Prospective, multicentre, randomized, double-blind study of the efficacy of escitalopram versus citalopram in outpatient treatment of Major Depressive Disorder. Funding: Provided by H. Lundbeck A/S through an unrestricted grant. Inclusion: Adult outpatients with DSM-IV criteria for Major Depressive Disorder, baseline MADRS score of at least 30. Exclusion Axis I disorders other than MDD; history of mania, bipolar disorder, schizophrenia or other psychotic disorder, obsessive-compulsive disorder, cognitive disorder including mental retardation or personality disorder; substance abuse or use of antipsychotic, anxiolytic or anticonvulsant medications prior to study No placebo control Statistical significance found may not translate into clinical relevance. (2.1 point difference in MADRS scores of marginal clinical significance). Industry funded. Patients randomly assigned to: Rx1: Escitalopram 10 mg daily during first week; 20 mg daily for remaining 7 weeks, (ITT N = 138, Completer N = 132) Rx2: Citalopram 40 mg daily for 8 weeks, (ITT N = 142, Completer N = 127) 8 weeks Primary outcome measure was change from baseline to previous assessment in MADRS total score. Decrease in MADRS score Rx1: ± 12.9 Rx2: ± point difference in mean endpoint MADRS scores (Rx1 = 13/9, Rx2 = 15.4); calculation not reported by authors Responder rate (50% decrease from initial MADRS score) Rx1: 76.1% Rx2: 61.3% Remission rate Rx1: 56.1% Rx2: 43.6% Tolerability/adverse events Rx1: 14.8% Rx2: 16.4% 7 9 < National Adult Depression Clincial Practice Guideline

175 Study DuRubeis, et al. (2005) Cognitive therapy vs. medications in the treatment of moderate to severe depression Funding: Supported by grants from the National Institutes of Mental Health. GlaxoSmithKline (Brentford, Middlesex, United Kingdom) provided medications and pill placebos for the trial. Table 1.5 Inclusion & Exclusion Criteria Inclusion: Diagnosis of MDD according to DSM-IV criteria, English speaking, score of 20 or higher on modified 17- item HDRS at screen and baseline visits, separated by at least 7 days. Exclusion: History of bipolar disorder, substance abuse/dependence, current or past psychosis, another Axis I DSM-IV disorder; antisocial, borderline or schizotypal disorder; suicide risk requiring hospitalization; medical condition contraindicative to study medications, nonresponse to adequate trial of paroxetine in year preceding. Limitations / Age Biases Placebo control ended at 8 weeks. Unequal number of subjects in treatment groups. Study conducted at two different sites: University of Pennsylvania and Vanderbilt University. Intervention and dose N and Final N Duration Outcome Results NNT Patients randomly assigned to: Rx1: Antidepressant medication, N = 120*, for 16 weeks. Rx2: Cognitive therapy, N = 60, for 16 weeks. Rx3: Pill placebo, N = 60, for 8 weeks** *Rx1 had twice as many patients because responders to Rx1 at week 16 were to be randomized a second time for a companion study of subsequent relapse prevention. **For ethical reasons, pill placebo was discontinued after 8 weeks; patients in this group were offered treatment without cost after 8 weeks. 16 weeks Primary outcome measure was change in HDRS score at 16 weeks. Outcome at 8 weeks Response rate Rx1: 50% Rx2: 43% Rx3: 25% Rx1 vs. Rx3 Rx2 vs. Rx3 Rx1 vs. Rx2 Outcome at 16 weeks Response rate Rx1: 58% Rx2: 58% Remission rate Rx1: 46% Rx2: 40% Site X Treatment Interaction (greater change in Rx1 at Vanderbilt) Authors attribute this difference to variance in patient characteristics and therapists experience levels. p value National Adult Depression Clincial Practice Guideline

176 Study Otsubo, et al. (2005) A comparative study of the efficacy and safety profiles between fluvoxamine and nortriptyline in Japanese patients with Major Depression Funding: Supported in part by grants from the Ministry of Health, Labour, and Welfare in Japan. Table 1.6 Inclusion & Exclusion Criteria Inclusion: DSM-IV criteria for single episode or recurrent Major Depression; either depressed mood or work & interests score of 2 or more in HAM-D. Exclusion: Failure to respond to two or more prior antidepressant trials in current episode; medical contraindications to antidepressant therapy; significant hematologic, endocrine, or cardiovascular disease conditions; acute suicidal tendencies, seizure disorder, psychotic disorder not associated with depression, history of drug or alcohol dependence; experience or receiving fluvoxamine or nortriptaline treatment. Limitations / Age Biases No placebo control. Single blind study design. Dose level of fluvoxamine generally lower than that of previous studies. Small sample size. Intervention and dose N and Final N Duration Outcome Results NNT Patients randomly assigned to one of two treatment groups (singleblind, parallel treatment): Rx1: fluvoxamine, N = 36 Rx2: nortriptyline, N = 38 For both treatment groups: There was a 7 day drugfree washout period. Starting daily dose for first week was 50 mg/day. Thereafter, dose was flexible between mg taken 1-3 times per day. Patients were allowed to take lormetazepam 1-2 mg a day at bedtime for sleep if necessary. 8 weeks Primary outcome measures were changes from baseline in the scores of HAM-D and CGI. Intention-to-Treat Analysis No significant between-group difference for the change in total scores of HAM-D (p = 0.14 NS) Rx1 Rx2 Baseline Endpoint No significant between-group difference for the change in total scores of CGI severity (p = 0.55 NS). Figures not reported. Response rate (HAM-D score reduction of 50% or more) Rx1: 55.6% Rx2: 57.9% Response rate (CGI improvement scale scores of 1 or 2) Rx1: 52.8% Rx2: 44.7% Remission rate (HAM-D score of 7 or less) Rx1: 38.9% Rx2: 26.3% When compared with fluvoxamine, nortriptaline had more patients reporting adverse effects, specifically dysarthria (13.9% vs. 36.8%, p=0.02) and orthostatic dizziness (13.9% vs. 42.1%, p=0.01). p value NS NS NS NS 172 National Adult Depression Clincial Practice Guideline

177 Table 1.7 Study Inclusion & Exclusion Criteria Ag e Limitations / Biases Intervention and dose N and Final N Duratio n Outcome Results NNT p value Wehmeier, et al.(2005) Fluoxetine versus trimipramine in the treatment of depression in geriatric patients Funding: Supported by Lilly Deutschland, Bad Homburg, Germany Inclusion: Geriatric patients with diagnosis of Major Depression per DSM-III- R criteria; score of 16 on HAMD-17. Exclusion: Reduction of HAM-D total score of more than 25% between screening visit and baseline visit; serious suicidal risk; severe organic brain disorder, significant organic illness, history of seizures; history of schizophrenia; recent history of drug or alcohol abuse Study took place in two centers. One center enrolled only inpatients while the other center enrolled both inpatients and outpatients. Small sample size risk of type II error. Low number of male patients (2 total in study vs. 39 females). Findings can only be generalized to elderly patients. Short duration. No placebo control. After a 7 day washout period, patients were randomly assigned to one of two treatment groups: Rx1: Fluoxetine 20 mg/day, N = 20, final N = 15 Rx2: Trimipramine 150 mg/day, N = 21, final N = 18 5 weeks Primary outcome measure was sum score on HAMD-17. Change in HAM-D scores Fluoxetine Trimipramine baseline endpoint response 57.1% 60.0% Tolerability/adverse events Rx1: 54.5% Rx2: 66.7% >0.05 >0.05 > National Adult Depression Clincial Practice Guideline

178 Table 1.8 Study Inclusion & Exclusion Criteria Ag e Limitations / Biases Intervention and dose N and Final N Duratio n Outcome Results NNT p value Sirey, Bruce, and Alexopoulous (2005) The Treatment Initiation Program: An intervention to improve depression outcomes in older adults Funding: Supported by a grant from the National Alliance for Research in Schizophrenia and Affective Disorders and a grant from NIMH to Dr. Sirey. Inclusion: Outpatient geriatric psychiatry patients with diagnosis of DSM-IV Major Depressive Disorder; score of 17 on HAMD-24. Exclusion: Cognitive impairment, existing antidepressant therapy Patients were from Psychiatry only, and not Primary Care. At baseline, the intervention group had a significantly lower HAMD score than the nonintervention group (mean 23.1 vs. mean 26.4, p<0.05), suggesting poor randomization. Lower pretreatment scores in the intervention group might bias remission results in favor of the intervention group, when in fact both groups could have equal changes in HAMD scores. No information given on type of pharmacotherapy (i.e. antidepressant dose, type, and frequency) Participants were randomly assigned to one of two groups: Rx1: Pharmacotherapy as usual, N = 24 Rx2: Pharmacotherapy with Treatment Initiation Program (TIP)*intervention, N = 21 *TIP is an individualized, early intervention to address older adults attitudes about depression and barriers to care. 7 sessions administered over 24 weeks. 24 weeks Primary outcome measure was HAM-D assessment at admission and at 6, 12, and 24 weeks. Intervention group showed greater improvement in depression than nonintervention group. No figures reported. Remission rate Rx1: 42% Rx2: 71% At 12 and 24 weeks after seeking treatment, more intervention patients remained in treatment than nonintervention patients (post hoc analysis; no figures reported) Similar proportions of patients in each group received supportive psychotherapy (numbers not reported) , 0.04 Results only generalizable to elderly population. Small study sample. Lower than usual threshold used to define remission (HAMD <10 instead of <7). 174 National Adult Depression Clincial Practice Guideline

179 Study Table 1.9 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results p value Detke, et al. (2004) Duloxetine in the acute and longterm treatment of Major Depressive Disorder: a placebo- and paroxetinecontrolled trial. Funding: Funding not reported; however all authors employed by Eli Lilly and Co. Inclusion: Adult patients with DSM-IV Major Depressive Disorder; CGI-S 4, HAMD Exclusion: Other Axis I disorders; recent primary diagnosis of anxiety disorder; bipolar disorder, psychosis, or schizoaffective disorder; recent history of substance abuse or dependence; previous of response to two courses of antidepressant therapy, suicidal risk; serious medical illness. 18+ Lack of generalizability: patients had few to no comorbid conditions and, few were taking concomitant medications. Fixed dose design is not customary in clinical practice. No comparison of duloxetine and paroxetine. Secondary analyses likely to be underpowered. Industry funded Duloxetine used at higher doses than currently recommended by the manufacturer; probably not equivalent to paroxetine dose. Participants were randomized into one of the following (doubleblind) groups for 8 weeks: Rx1: Placebo, N = 93 Rx2: Duloxetine 80 mg/day, N = 95 Rx3: duloxetine 120 mg/day, N = 93 Rx4: paroxetine, 20 mg/day, N = 86 Patients who showed a 30% HAMD improvement at 8 weeks (N = 273) entered the 6 month continuation phase: Rx1: Placebo, N = 58 Rx2: Duloxetine 80 mg/day, N = 70 Rx3: duloxetine 120 mg/day, N = 75 Rx4: paroxetine, 20 mg/day, N = 70 8 weeks (acute phase) 6 months (continuation phase) Primary outcome measures were HAMD total score, HAMD subscales, MADRS, HAMA, VAS, CGI-S, and PGI-I. Acute phase HAMD score reduction: Rx1: -8.8 Rx1 vs. Rx2 Rx Rx1 vs. Rx3 Rx3: Rx1 vs. Rx4 Rx Estimated probabilities of response (%) Rx1: 47 Rx1 vs. Rx2 Rx2 70 Rx1 vs. Rx3 Rx3: 77 Rx1 vs. Rx4 Rx4 82 Estimated probabilities of remission (%) Rx1: 30 Rx1 vs. Rx2 Rx2 51 Rx1 vs. Rx3 Rx3: 58 Rx1 vs. Rx4 Rx4 47 Continuation phase Patients in each active treatment group (Rx2, Rx3, Rx4) achieved significant within-group improvement in total HAMD, MADRS, HAMA, CGI-S, and PGI-I. Time to loss of response Rx1 vs. Rx2 Rx1 vs. Rx3 Rx1 vs. Rx4 Frequently reported treatment-emergent adverse events: Rx2: viral infection (5.7%) Rx3: diarrhea (4.0%) Rx4: headache (11.4%) <0.001 < <0.001 <0.001 < National Adult Depression Clincial Practice Guideline

180 Table 1.10 Study Inclusion & Exclusion Criteria Ag e Limitations / Biases Intervention and dose N and Final N Duratio n Outcome Results NNT p value Sechter, et al. (2004) A comparative study of milnacipran and paroxetine in outpatients with Major Depression Funding: Sponsored by Pierre Fabre Medicament, inventors and manufacturers of milnacipran. Inclusion: Adult outpatients with DSM-IV Major Depressive Disorder without psychotic features; MADRS total score 20. Exclusion: Significant suicide risk; lack of response to two adequate antidepressant treatments; history of psychotic disorder; major personality disorder; agoraphobia, social phobia, or obsessive compulsive disorder; current alcohol or drug abuse or dependence No placebo control. Industry funded (milnacipran). Patients were randomized into one of the following groups: Rx1: milnacipran 100 mg/day, N = 149 Rx2: paroxetine 20 mg/day, N = weeks Outcome measures were differences in baseline in: HAMD-17 total score, MADRS total score, CGI score at 7, 14, 28, and 42 days. No significant differences between Rx1 and Rx2 on any of the following measures: HAMD-17 total Rx1 Rx2 Baseline Endpoint MADRS score Rx1 Rx2 Baseline Endpoint Responders (%) Rx1 Rx2 HAMD MADRS CGI Remissions (%) Rx1 Rx2 HAMD MADRS NS NS NS NS Post-treatment discontinuation emergent symptoms: Rx1: 13.0 Rx2: National Adult Depression Clincial Practice Guideline

181 Table 1.11 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Allard, et al. (2004) Efficacy and tolerability of venlafaxine in geriatric outpatients with Major Depression: a double-blind, randomised 6- month comparative trial with citalopram Funding: Wyeth, Lederle Nordiska AB, Sweden Inclusion: Male and female outpatients 65 and older, with DSM-IV Major Depression and in need of antidepressant therapy; MADRS score 20+; 20% change in MADRS score between pre-study and baseline visits. Exclusion: Cognitive impairment, alcohol or drug abuse, psychotic disorder not associated with depression, recent (1 year) psychiatric inpatient treatment; acute suicidality; bipolar disorder, dementia, mental disorders due to a general medical condition, history of seizure disorder, significant cardiovascular or cerebrovascular disorder, uncontrolled hypertension; significant abnormalities assessed in prestudy physical exam No placebo control. Industry funded Participants randomized into one of two groups (double-blind, parallel group design): Rx1: Venlafaxine IR 37.5 mg once daily during first week, ER 75 mg once daily during following two weeks, increase to 150 mg/day during following two weeks if no response, N = 76 Rx2: Citalopram 10 mg once daily during first week, 20 mg once daily during following two weeks, 30 mg during following two weeks if no response, N = 75 6 months Primary outcome measure was change in MADRS score from baseline to 8 weeks, and baseline to 22 weeks Both treatment groups showed comparable improvements in MADRS scores over time. There were no statistically significant differences in MADRS scores between the two groups. MADRS scores Rx1 Rx2 Baseline Week Week MADRS responders Rx1 Rx2 Week % 72.5% Week % 93.2% Spontaneously reported side effects/adverse events: Rx1: 62% Rx2: 43% NS NS 177 National Adult Depression Clincial Practice Guideline

182 Table 1.12 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Montgomery, Huusom, & Bothmer (2004) A randomized study comparing escitalopram with venlafaxine XR in primary care patients with Major Depressive Disorder. Funding: Sponsored by H. Lundbeck A/S. Inclusion: Primary care patients, DSM- IV diagnosis of MDD; 18 score on MADRS. Exclusion: History of mania or bipolar disorder, schizophrenia or any psychotic disorder; currently suffering from obsessive compulsive disorder; eating disorders, mental retardation; development or cognitive disorder; MADRS score 5 on item 10 (suicidal thoughts); alcohol or drug abuse problems past one year; treatment with antipsychotics, antidepressants, psychotropics, serotonin receptor agonists, lithium, carbamazepine, valproate, or alpromide; ECT; treatment with behavior therapy or psychotherapy; pregnant or breast feeding No placebo control. Industry funded Observed cases analysis After one week run-in period, patients were randomized into one of two groups (double-blind design): Rx1: Escitalopram. Initial dose 10 mg/day. At week 2 or 4, dose could be increased to 20 mg/day; N = 146 Rx2: Venlafaxine XR. Initial dose 75 mg/day. At week 2 or 4, dose could be increased to 150 mg/day; N = weeks Primary outcome measures were changes in MADRS and HAMD-17 total scores. MADRS scores (observed cases): Rx1 Rx2 Baseline Week HAMD-17 scores (observed cases) Rx1 Rx2 Baseline Week Rx1 group achieved sustained response significantly faster (4.6 days faster) than Rx2 (p<0.05). Rx1 group achieved sustained remission significantly faster (6.6 days faster) than Rx2 (p<0.001). Rx2 group had significantly more nausea, constipation, and increased sweating (p < 0.05). Rx2 group had significantly more discontinuation symptoms (p < 0.01). No difference in overall discontinuation rates (14.4% vs. 13.3%) NS NS p<0.05 p<0.001 <0.05 < National Adult Depression Clincial Practice Guideline

183 Table 1.13 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention & dose N and Final N Duration Outcome Results NNT p value Rapaport, et al.(2003) Efficacy of controlledrelease paroxetine in the treatment of latelife depression. Funding: GlaxoSmithKline, Research Triangle Park, NC Inclusion: Elderly adults with DSM-IV Major Depressive Disorder; HAMD-17 score 18 at both screening and baseline visits. Exclusion: HAMD total score decreased by 25% or more between screening and baseline; concomitant therapy with psychoactive medication other than chloral hydrate; diagnosis of other axis I disorder; history of brief depressive episodes previousing 8 weeks with spontaneous remission; neurological disorders contributing to secondary depression; dementia; MMSE score 24; serious medical conditions contraindicative to paroxetine; history of seizure disorders; concomitant treatment with warfarin, phenytoin, cimetidine, sumatribtan, type 1C antiarrhythmic agents, or quinidine; history of substance abuse or dependence, ECT within 3 months; unresolved or abnormal ECG findings; suicidal or homicidal tendencies. 60+ Study restricted to elderly outpatients (not those in nursing homes, etc.) Excluded patients with severe or unstable medical illness. Participants randomized into one of three groups (flexible dose, double-blind study) Rx1: Paroxetine CR up to 50 mg/day, N = 104 Rx2: Paroxetine IR up to 40 mg/day Rx3: Placebo 12 weeks Primary outcome measure was change in HAMD total score from baseline to endpoint. HAMD total score change from baseline in Intention-to-Treat, LOCF population: Rx1 Rx2 Rx3 Baseline Endpoint Change Adjusted difference between Rx1 and Rx3 = -2.8 (95% confidence interval = to -0.73, p=0.007). Adjusted difference between Rx2 and Rx3 = -2.8 (95% confidence interval = to -0.99, p=0.03). Response rates (score of 1 or 2 on CGI-I): Rx1: 72% (p<0.002 vs. Rx3) Rx2: 65% (p=0.06 vs. Rx3) Rx3: 52% Remission rates (HAMD 7): Rx1: 43% (p<0.009 vs. Rx3) Rx2: 44% (p=0.01 vs. Rx3) Rx3: 26% Post-hoc analysis revealed that patients with chronic depression (duration > 2 years) responded as well as patients with short-term depression. p= < < Withdrawal due to adverse events: Rx1: 12.5% Rx2: 16.0% Rx3: 8.3% 179 National Adult Depression Clincial Practice Guideline

184 Table 1.14 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Sauer, Uppertz- Helmhold & Dierkes (2003) Efficacy and safety of venlafaxine ER vs. amitriptyline ER in patients with Major Depression of moderate severity Funding: Study granted by Wyeth Pharma GmbH, Germany Inclusion: Adult outpatients with ICD-10 Major Depression of moderate severity; HAMD-21 score at first screening; depressive symptoms present for minimum of 14 days prior to study. Exclusion: Hypersensitivity to or previous unsuccessful treatment with venlafaxine or amitriptyline; bipolar or psychotic disorders, history of convulsive disorders, hypertension, suicidality, HAMD score decrease 4 between screening and baseline; onset of additional psychotherapy; pregnancy or lactation, treatment with sumatriptan or antipsychotic medication, ECT within 30 days prior, treatment with fluoxetine within 30 days prior, treatment with MAO inhibitors within 14 days prior, clinically relevant findings concerning physical exam, ECG or laboratory parameters Noninferiority study. No placebo control. Study conducted in Germany, where antidepressants are prescribed at generally lower doses. Industry funded After a one-week washout period, participants were randomized into Rx1: Venlafaxine ER 75mg, N = 79 Rx2: Amitriptyline ER: 2 mg on day one, 50 mg on day two, 75 mg from day three onwards, N = 77 In cases of insufficient response, doses were increased to 75 mg from day 15 onwards. 6 weeks Primary outcome measure was change in HAMD- 21 score. Significant noninferiority of venlafaxine, as reflected by change in HAMD scores (Intention-to-Treat analysis): Rx1 Rx2 Baseline Endpoint Change Similar results found in ATP (According to Protocol) analysis: Rx1 Rx2 Baseline Endpoint Change Frequency of adverse events: Rx1: 70.9% Rx2: 81.8% p=0.11 Frequency of adverse drug reactions: Rx1: 55.7% Rx2: 71.4 p= National Adult Depression Clincial Practice Guideline

185 Table 1.15 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Kool (2003) Efficacy of combined therapy and pharmacotherapy for depressed patients with or without personality disorders. Funding: Supported by an unrestricted educational grant from Eli Lilly Nederland. Inclusion: Adults with DSM-III-R Major Depression (with or without dysthymia; with or without personality disorder); HAMD Exclusion: Patient considered too ill or too suicidal to participate in a clinical trial; drug abuse or psycho-organic, psychotic, or dissociative disorder; patient not considered reliable enough to participate in a clinical trial Small number of subjects risk of Type II error. Specialty mental health setting --? applicability to primary care settings. Patients randomized into one of two groups: Rx1: Pharmacotherapy with three tiered protocol (steps2 and 3 used if steps 1 and 2 proved ineffective): 1) fluoxetine 20 mg/day 2) amitriptyline mg/day 3) moclobemide mg/day N = 56 Rx2: Combined therapy: pharmacotherapy as described above plus 16 sessions of short psychodynamic supportive psychotherapy (SPSP) starting two weeks after the initiation of medication. N = weeks Outcome measures were CGI-I, CGI-S, HAMD-17, QLDS, and SCL-90. Combined therapy was more effective than pharmacotherapy alone for depressed patients with personality disorders. Combined therapy was not more effective than pharmacotherapy alone for depressed patients without personality disorders. Change in HAMD scores: Baseline Endpoint Rx1 with PD Rx2 with PD Rx1 w/o PD Rx2 w/o PD National Adult Depression Clincial Practice Guideline

186 Table 1.16 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Wade, et al. (2003) A randomized, double-blind, 24- week study comparing the efficacy and tolerability of mirtazapine and paroxetine in depressed patients in primary care. Funding: Study funded by a clinical research grant from Organon Laboratories Ltd., Dambridge Science Park, Cambridge UK. Inclusion: Male and female adults, aged 18 and over; DSM- IV criteria for single or recurrent depressive episode, HAMD-17 score >18. Exclusion: Schizophrenia or bipolar affective disorder, suicidality, abuse of illicit drugs, alcohol dependence; recent treatment with other psychotropic drugs, hypersensitivity to paroxetine or mirtazipine (or use for current depressive episode); fluoxetine within 5 weeks, use of other antidepressants within 2 weeks of study; clinically meaningful renal, hepatic, respiratory, cardiovascular or cerebrovascular disease; pregnancy or lactation; investigator s assessment of unsuitability for trial. 18+ No placebo control. No washout period. High discontinuation rates overall (>50% each group). Industry funded Participants randomized into one of two groups: Rx1: Mirtazipine 30 mg/day (between weeks 5-24 dosage could be increased to 45 mg/day) N = 99 Rx2: Paroxetine 20 mg/day (between weeks 5-24 dosage could be increased to 30 mg/day), N = weeks Primary outcome measure was change in HAMD-17 score from baseline. There was no statistically significant difference in HAMD improvement at 24 weeks. Rx1: Rx2: At weeks 2, and 4 there was a statistically significant difference in HAMD improvement, in favor of mirtazapine: Week 2 Difference -2.0 (95% CI: -3.5 to -0.4; p=0.012) Week 4 Difference -2.0 (95% CI: -3.8 to -0.2; p=0.030) Overall adverse effects: Rx1: 79%, Rx2: 85%. Tolerability/adverse events Rx1 statistically significantly higher incidence of fatigue. Rx2 group reported statistically significantly more sweating (p=0.010), headache (p=0.008) and nausea (p<0.001). HAMD response rates: Rx1: 87% Rx2: 78% HAMD remission rates: Rx1: 61% Rx2: 42% National Adult Depression Clincial Practice Guideline

187 Table 1.17 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Montgomery (2003) The antidepressant efficacy of reboxetine in patients with severe depression (analysis of 4 RCTs) Funding: Data reported supported by grants from Pharmacia Corporation, Peapack, NJ. Inclusion: Adult patients with DSM- III or DSM-IV criteria for Major Depressive Disorder of at least one month s duration and baseline severity score 20 on HAMD-21. Exclusion: Pregnancy or breastfeeding, hypersensitivity to psychotropic medications, recent history of substance abuse; other medical or psychiatric conditions; history of resistance to antidepressant medications Small number of studies analyzed. The four studies were not similar in terms of duration or study population (n). One of the four studies (Versiani et al, 2000) had a much higher baseline mean HAMD score 35.4 vs. 26.4, 27.0, and 27.2). Patients unresponsive to previous antidepressant treatment were excluded. After 4-14 day washout period, patients were randomized into one of two groups: Rx1: Reboxetine initiated at lower doses, but maintained at 8-10 mg/day divided into two daily doses Rx2: placebo days Primary outcome measure was mean change in HAMD score from baseline to previous follow-up evaluation, in each of the four trials. In three of the four trials, the reboxetine group had a significantly greater decrease in HAMD score when compared with placebo. In three of the four trials, antidepressant efficacy occurred significantly faster (at two weeks) in the reboxetine group when compared with the placebo group. In three of the four trials, the responder rate was significantly higher in the reboxetine group (56-74%) compared with placebo (20-52%). <0.001 <0.05 < <0.001 Remission rates not reported All studies included were funded by Pharmacia, the manufacturer of reboxetine; a systematic review was not performed (unpublished studies may not have been included). 183 National Adult Depression Clincial Practice Guideline

188 Table 1.18 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Gasto, et al. (2003) Singleblind comparison of venlafaxine and nortriptyline in elderly Major Depression Funding: None reported. Inclusion: In- and out- patients with DSM-IV Major Depression, with or without endogenous or psychotic features; baseline HAMD-17 21; symptoms present for at least one month. Exclusion: Uncontrolled medical illness; manic or hypo manic episode, history of nonaffective psychosis, substance dependence; ECT therapy within 6 months of recruitment. 65+ Single blind study (psychiatrist only) No placebo control. Elderly patients only. After a two-week washout period, participants were randomized into one of two groups: Rx1: Venlafaxine Starting dose: 75 mg/d Day 4: 150 mg/d Day 8: 225 mg/day Clinician option to increase to 300 mg/day at 2-week evaluation. N = 34 Rx2: Nortriptyline Starting dose: 12.5 mg/d Day 4: 25 mg/d Day 8: 50 mg/day Clinician option to increase to 300 mg/day at 2-week evaluation. Plasma concentrations were assessed after one week and doses were adjusted to achieve a plasma concentration between 80 and 100 mg/ml Maximum dose: 100mg/d N = 34 6 months Primary outcome measures were scores on HAMD-17 and Newcastle Scale. Intention-to-treat analysis (N = 61: Rx1 N = 31, Rx2 N = 30) Remission rates: Rx1: 70.1% Rx2: 70% Efficacy was not dependent on clinical severity of the depressive episode. Treatment groups were similar when subanalyses were performed evaluating the following categories: endogenous, nonendogenous, psychotic, nonpsychotic, severe inhibition, mild inhibition. Both drugs were similarly tolerated. Side effects were frequent (Rx1 = 73.5%, Rx2 =82.3%) but mild to moderate in intensity. On the autonomic side-effects subscale, Rx1 scored lower than Rx2 (2.03 and 2.91 respectively), suggesting a more benign side-effect profile for venlafaxine NS NS National Adult Depression Clincial Practice Guideline

189 Study Table 1.19 Inclusion & Exclusion Criteria Limitations / Biases Intervention and dose N and Final N Outcome Results Hansen, et al. (2005) Efficacy and safety of secondgeneration antidepressants in the treatment of Major Depressive Disorder. (Review) Funding: Dr. Gaynes supported in part by a National Institute of Mental Health K23 Career Development Award, the Robert Wood Johnson Foundation, and the Agency for Healthcare Research and Quality. Efficacy trials, N = 46 Head-to-head comparison of one antidepressant with another; conducted in Primary Care settings; 3 month duration of follow-up; minimal inclusion and exclusion criteria (to represent general population); assessed health outcomes rather than immediate outcomes; adequate sample size to determine a minimally important difference (from a patient s perspective) on a health-related quality of life instrument. Safety and tolerability, N = 21 Head-to-head trials; placebo-controlled; observation studies with large samples (>100) previousing at least one year. Limited quantitative analyses due to inadequate evidence. Many trials sponsored by pharmaceutical companies; possible publication biases. Qualitative analysis of studies Quantitative analysis when more than three head-tohead trials compared the same treatments (primary outcome measure treatment response). If treatment effects differed among studies, potential reasons were assessed. Heterogeneity tested for treatment effects in metaanalyses. Mean incidence and 95% confidence intervals calculated for specific adverse events (significant variance, should be interpreted with caution) Primary outcome measures for head-to-head comparison trials were efficacy, speed of response, quality of life and tolerability/adve rse effects. Primary outcome measure for meta-analyses was responder rate. Head-to-head comparison trials Efficacy Overall, trials reported similar outcomes among the six SSRIs. Differences in treatment effects between some trials that found a significant and those that didn t could be attributed to small sample sizes and publication bias in favor of the sponsoring pharmaceutical company s antidepressant. Trials comparing SSRIs with other second generation antidepressants showed similar efficacy. Speed of response Overall, the trials reported no difference among SSRIs. Evidence of a faster response to citalopram, fluvoxamine, and paroxetine than to fluoxetine is based on results of one trial or is inconsistent with other evidence. Trials comparing SSRIs with other second generation antidepressants showed an average speed of response of 4-6 weeks, with no statistically significant differences among treatments. Evidence in some trials of a faster response rate with venlafaxine is equivocal. Quality of life Overall, trials showed no significant difference among SSRIs in their ability to improve health-related quality of life. One trial did report better sleep quality with fluovoxamine than with fluoxetine. Trials comparing SSRIs with other second generation antidepressants reported no differences in overall quality of life. Tolerability/adverse effects Overall, incidence of adverse effects was similar among antidepressants. Quality of reporting and methods used to report adverse effects differed among studies Meta-analyses Fluoxetine vs. paroxetine Responder rate did not differ significantly between fluoxetine and paroxetine (relative benefit 1.09, CI: 0.97 to 1.21). Fluoxetine with sertraline Pooled results showed a small treatment effect of sertraline compared with fluoxetine (relative benefit, 1.10, CI: 1.01 to 1.22) although no individual trial showed significant differences. Venlafaxine vs. fluoxetine Pooled data from 6 studies showed that venlafaxine yielded more responders than fluoxetine (57.6% vs. 51.1%, ARR = 6.5%, NNT = 15; RR = 1.12, CI: ). Most other individual studies comparisons for venlafxaine vs. other SSRIs did not show a difference, pooled analysis for venlafaxine vs. all SSRIs not performed. 185 National Adult Depression Clincial Practice Guideline

190 Table 1.20 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Arroll, et al. (2005) Efficacy and tolerability of tricyclic antidepressants and SSRIs compared with placebo for treatment of depression in primary care: a meta-analysis Funding: Funding Body Chief Scientist Office, Scotland. Inclusion Randomized, controlled trials using TCAs and/or SSRIs using primary care patients; study population of adults with diagnosis of Major Depression. Exclusion: Studies using predominantly children or the elderly. 18+ (primar ily) Larger number of studies comparing TCAs with placebo than comparing SSRIs with placebo. Studies lacked consistency in patients severity of depression. Three groupings of studies, total N = 15: TCAs vs. placebo, N = 10 SSRIs vs. placebo, N = 3 TCAs and SSRIs vs. placebo, N = 2 N/A Primary outcome measure was efficacy of TCAs and SSRIs in comparison with placebo. Pooled efficacy data showed both TCAs and SSRIs were significantly more effective than placebo: TCAs vs. placebo Relative risk 1.26 (95% CI: ) NNT = 4, NNH (withdrawal)= SSRIs vs. placebo: Relative risk 1.37 (95% CI: ) NNT = 6, NNH (withdrawal)= Both low dose and high dose TCAs were effective National Adult Depression Clincial Practice Guideline

191 Table 1.21 Study, Total n Treatment Groups Size & Intervention Description Study Population Results Comments Detke, M (RCT, double-blind) Follow-up: 9 weeks Initial n: 245 Final n: 225 (based on safety analysis numbers) For efficacy analysis Rx1: Duloxetine, 60 mg/day (n = 121) Rx2: Placebo (n = 115) For safety analysis Rx1: Duloxetine, 60 mg/day (n = 123) Rx2: Placebo (n = 122) Men and women at least 18 years of age Adult patients with DSM-IV MDD baseline score of 15 on the HAM-D-17 (17-item Hamilton Rating Scale for Depression) Baseline score of 4 on the CGI-S (Clinical Global Impressions-Severity) Least square mean change from baseline to previous observation using HAM-D- 17 scale (primary efficacy measure) Rx1: Rx2: p < (in favor of duloxetine) Estimated probability of remission for duloxetine patients = 44% vs. 16% for placebo-treated patients Treatment-emergent adverse events (safety measure) Nausea Rx1: 46.3% of patients Rx2: 9.0% of patients p < (in favor of placebo) Dry mouth Rx1: 27.6% Rx2: 6.6% p < Somnolence Rx1: 21.1% Rx2: 4.9% p < The following adverse effects were also statistically significant: dizziness, diarrhea, insomnia, and constipation Duloxetine in a once daily dose of 60 mg/day was significantly superior to placebo in reducing HAM-D-17 total scores, starting at week 2, but is associated with a significantly increased risk of side effects. 187 National Adult Depression Clincial Practice Guideline

192 Table 1.22 Study, Total n Treatment Groups Size & Intervention Description Study Population Results Comments Goldstein, D (RCT, double-blind) Follow-up: 8 weeks Initial n: 173 Final n: 167 Rx1: duloxetine, titrated in the first three weeks from 40 mg (20 mg b.i.d.) to 120 mg/day (60 mg b.i.d.) (n = 70) Rx2: placebo (n = 70) Rx3: fluoxetine, 20 mg q.d. (n = 33), was used as an internal control Male and female outpatients ages 18 to 65 with DSM-IV Major Depressive Disorder Clinical Global Impressions (CGI)-Severity of Illness score of at least moderate severity ( 4) 17-item Hamilton Rating Scale for Depression (HAM- D-17) total score of at least 15 Least square mean change from baseline to previous visit using HAM-D-17 scale (primary efficacy measure) Rx1: Rx2: Rx3: p = (duloxetine vs. placebo) in favor of duloxetine Estimated probability of remission for duloxetinetreated patients was 56% vs. 32% for placebo-treated patients, p = 0.22 Treatment-emergent adverse events (safety measure) Insomnia Rx1: 20.0% of patients Rx2: 7.1% of patients Rx3: 9.1% of patients p = in favor of placebo (duloxetine vs. placebo) Duloxetine treated patients exhibited a small but statistically significant reduction in body weight relative to placebo (0.59 kg) p = The mean increase in standing diastolic blood pressure for duloxetine-treated patients was 2.80 mm Hg greater than for placebo-treated patients (p = 0.041) Duloxetine is statistically superior to placebo in reducing the symptoms of MDD. Study was not designed to be a comparison of duloxetine and fluoxetine. Fluoxetine was included as an internal control to be sure patients were antidepressant responsive. Clinical significance of changes in BP and weight uncertain. The following adverse effects were not statistically significant: dry mouth, headache, somnolence, dizziness, diarrhea, nausea, and constipation 188 National Adult Depression Clincial Practice Guideline

193 Table 1.23 Study, Total n Treatment Groups Size & Intervention Description Study Population Results Comments Wade, A (RCT, double -blind) Follow-up: 8 weeks, following an initial 1-week single-blind, placebo period Initial n: 380 Final n:320 Rx1: escitalopram (SSRI) 10 mg/day (n = 191) Rx2: placebo (n = 189) Patients were aged between 18 and 65 years, 3:1 ratio of women to men > 97% Caucasian Patients fulfilled DSM-IV criteria for MDD Had a baseline Montgomery- Asberg Depression Rating Scale (MADRS) total score 22 and 40 Adjusted mean change in MADRS total score from baseline to week 8 Rx1: 16.3 Rx2: 13.6 p = (in favor of escitalopram) Tolerability Nausea (% of patients) Rx1: 8.9% Rx2: 3.7% p < 0.05 (in favor of placebo) This difference disappeared within the first two weeks of double-blind treatment Headache (% of patients) Rx1: 12% Rx2: 10.1% Not statistically significant Escitalopram 10 mg/day had a statistically significantly better antidepressant effect than placebo as early as week one, and was safe and well tolerated. The proportion of escitalopram treated patients in complete remission was also statistically significantly higher than that of placebo-treated patients at week eight. (p < 0.01) 189 National Adult Depression Clincial Practice Guideline

194 Study, Total n Table 1.24 Treatment Groups Size & Intervention Description Study Population Results Comments Moncrieff, J (metaanalysis) # studies found: 12 # studies included: 9 Randomized and quasi randomized trials Double-blinded Drugs used were all TCAs. All trials used active placebos containing atropine Participants of either sex of all age groups with primary diagnosis of depressive disorder Inpatients and outpatients Specialty behavioral health settings (Individual studies) Change in mood Daneman, 1961 SMD = 1.1 (0.8 to 1.4) Uhlenhuth, 1963 SMD (unadjusted) = 0.60 (0.02 to 1.2) SMD (adjusted for baseline values) = 0.35 ( to 0.96) Weintraub, 1963 SMD (for hospital director) = 0.14 (-0.34 to 0.62) SMD (for ward doctor) = 0.63 (0.15 to 1.11) Wilson, 1963 SMD = (-1.10 to 0.58) Hollister, 1964 SMD = 0.19 (-0.24 to 0.63) (Combined analysis) Change in mood All nine trials, using the more conservative estimate from Weintraub, 1963 (rating by hospital director) SMD = 0.39 (0.24 to 0.54) Heterogeneity p < Eight trials, excluding Daneman, 1961 (due to high degree of heterogeneity) SMD = 0.17 (0.00 to 0.34) Heterogeneity p = 0.29 Eight trials, excluding Daneman, 1961 and using the higher estimate from Weintraub, 1963 (rating by ward doctor) SMD = 0.23 (0.06 to 0.40) Seven trials, excluding Daneman, 1961 and Murphy, 1984 (participants received cognitive therapy as well as medication) SMD = 0.21 (0.03 to 0.38) Majority of trials found only small differences between antidepressants and active placebos Unblinding effects may inflate the efficacy of antidepressants in trials using inert placebos. Meta-analysis with small number of trials Short duration of trials (minimum three weeks, maximum 12 weeks) Studies old Variety of depression instruments used Limited to TCAs-extrapolation to other antidepressants may not necessarily be true Freidman, 1966 SMD = 0.13 (-0.37 to 0.64) Fixed effect model Hussain, 1970 SMD = 0.79 (0.09 to 1.5) Friedman, 1975 SMD = 0.14 (-0.14 to 0.42) Murphy, 1984 SMD = (-1.0 to 0.28) 190 National Adult Depression Clincial Practice Guideline

195 Table 1.25 Author & Title Last Updated & Search Database Methodology & Study Characteristics Characteristics of Study Participants Heterogeneity Results (95% CI) Wilson, K Most recent Update March 2001 Databases: Cochrane Collaboration Depression Anxiety and Neurosis Review Group (CCDAN) Reference lists of related reviews References of located studies. Contact was made with authors working in the field. Search Terms: Electronic databases were searched using optimally sensitive search terms (exact terms not stated). Number of studies included: 17 trials were included Setting: Not stated. Study Inclusion / Exclusion: The review included all randomized, placebo controlled trials using antidepressant drugs in the treatment of depression in subjects described as elderly, geriatric or senile or in trials where all subjects are over the age of 60. Trials that included subjects under the age of 60 were excluded unless data concerning subjects over the age of 60, or those described as elderly, geriatric or senile, were randomized and analyzed separately. Funding Source: Wirral and West Cheshire Community Trust UK Drug Comparisons: TCAs with placebo SSRIs with placebo MAOIs with placebo Dosages: Not stated Stage of depression: Not stated. Sample: Depressive disorders: n = 1,326 Major Depressive Disorder. Inclusion Criteria: Patients over 60 with depression. Exclusion Criteria: Not stated. Design of Trials: RCT Duration of Trial: Not stated. Method of abstraction: Three reviewers independently assessed the relevance of each trial, blind to decisions made by each other. Each trial was assessed against pre-set criteria and rated on a scoring sheet. In cases of disagreement open discussion was undertaken and decision reached by consensus. Reasons for exclusion and/or inclusion were recorded. Two reviewers are acknowledged expert in the field, the other reviewer wasn t. Reviewers were blind to authorship of trials, journals and institutions from which authors come. Blindness was tested through reviewers 'best guessing' authorship, journal and institution. Other variables associated with the aging process (such as physical ill health) are likely to influence the outcome of trials. The effects of these aspects of clinical heterogeneity were examined and trials combined if appropriate. Where possible, the influence of patient characteristics, for example inpatient vs. outpatient, subtypes of depression, under 75 vs. over 75 was examined. The effect of the trial length was examined. If enough studies were identified, the year of publication and country of publication was examined. A funnel graph to test for publication bias was also used. In addition, a formal test for statistical heterogeneity, the natural approximate chi-square test was conducted. The standardized effect size for the three groups respectively were: TCAs: OR = 0.32 (0.21 to 0.47), SSRIs: OR = 0.51 (0.36 to 0.72), MAOIs: OR = 0.17 (0.07 to 0.39). Adverse Dropout Rates: Not reported. Effects: Not reported. Statistical advice was sought through the Cochrane Depression, Anxiety and Neurosis Review Group and through the Cochrane Statistical Support Group in Liverpool. Efficacy: Seventeen trials contributed data to the analyses comparing the efficacy of antidepressant treatment and placebo. Analyses of efficacy was based on 245 patients treated with TCAs (223 with placebo), 365 patients treated with SSRIs (372 with placebo) and 58 patients treated with MAOIs (63 with placebo). 191 National Adult Depression Clincial Practice Guideline

196 Table 1.26: Effects of Specific Psychological Treatments for Depressive Disorders INTERVENTION EVIDENCE BENEFITS HARMS DISADVANTAGES Cognitive therapy 1 SR (48 RCTs of psychological therapies [2,765 people, mean age 39.3 y] mainly outpatients in secondary care; therefore, probably with mild to moderate depression; people with psychotic or bipolar symptoms were excluded); 20 RCTs compared CT with waiting list or placebo and 17 compared it with drug treatment. 79% of people receiving placebo were more symptomatic than the average person receiving CT (p < ). 65% of people receiving CT were less symptomatic than the average person treated with antidepressant drugs (p < ). No harms reported Requires extensive training. Limited availability. RCTs in primary care suggest limited acceptability to some people. Interpersonal psychotherapy No SR. 1 large RCT (people with mild to moderate depression, mean age 35 y) compared interpersonal psychotherapy vs. either drug treatment, CT or placebo plus clinical management for 16 weeks. Rates of recovery from depression: interpersonal psychotherapy (43%; NNT = 5, 95% CI: 3 to 19), imipramine (42%; NNT = 5, 95% CI: 3 to 22), placebo clinical management (21%). No harms reported Requires extensive training. Limited availability. Problemsolving therapy No SR. 1 large RCT (452 people aged 18 to 65 y with mild to moderate depression or adjustment disorders) compared PS, group treatment, and control. 1 RCT (91 people aged 18 to 64 y with mild to moderate depression) compared problemsolving, placebo, and amitriptyline. PS vs. control significantly increased the proportion of people who were not depressed at 6 mo., but no significant difference at 1 y. PS vs. placebo significantly improved symptoms at 12 weeks, and no significant difference in symptoms with PS vs. amitriptyline. No harms reported Requires extensive training. Limited availability. Nondirective counseling 1 SR (7 RCTs, 772 people aged over 18 y with recent onset psychological problems including depression in UK primary care) compared counseling vs. standard physician care. Counseling vs. standard care significantly improved symptoms in the short term (1 to 6 mo.; WMD -2.03, 95% CI: to -0.24), but no significant difference in the long term (> 6 mo.; WMD -0.03, 95% CI: to +0.32). CT: cognitive therapy, mo.: month, PS: problem-solving, SR: systematic review, y: year Clinical Evidence, November 2003 (70) No harms reported Requires extensive training. Limited availability. 192 National Adult Depression Clincial Practice Guideline

197 Table 1.27 Study, Total n Treatment Groups Size & Intervention Description Study Population Results Comments Casacalenda, N (meta-analysis) Follow-up: 10 to 34 weeks (median = 16 weeks) # of studies found: 6 # of studies included: 6 Intent- to treat analysis Rx1: Medication (n = 261); five studies used tricyclics and one used phenelzine Rx2: Psychotherapy (n = 352); three studies used cognitive behavior therapy, three used interpersonal therapy, one used problem-solving (six studies total, one study used both CBT and IPT) RX3: Placebo (n = 270) All studies were randomized, control trials. Adults All were diagnosed as having nonpsychotic Major Depression Full remission (%) (based on intention-to-treat analysis of the main analysis of the studies) Rx1: 46.4% Rx2: 46.3% Rx3: 24.4% p < (favoring treatment groups) Definition of remission: Definition varied by study. Raskin Depression Scale score 5 Hamilton depression scale score 6 Hamilton depression scale score 7 Intention-to-treat analyses indicated that pharmochotherapy and psychotherapy were significantly more efficacious than control conditions but were not significantly different from each other, when treating mildly to moderately depressed patients Heterogeneity not stated. 193 National Adult Depression Clincial Practice Guideline

198 Table 1.28 Study, Total n Treatment Groups Size & Intervention Description Study Population Results Comments Thase, E (meta-analysis, nonsystematic review) Follow-up : 6 to 8 weeks # of studies found: 8 #of studies included: 8 Rx1: venlafaxine (n = 851), Dose range: venlafaxine IR, 75 to 375 mg/day; venlafaxine XR, 75 to 225 mg/day Rx2: SSRIs (fluoxetine 20 to 80 mg/day, paroxetine 20 to 40 mg/day, fluvoxamine 100 to 200 mg/day) (n = 748) Rx3: Placebo (four studies n = 446) Eight randomized, double-blind studies Patients were at least 18 years old Met the DSM-III-R criteria for Major Depressive Disorder and DSM-IV criteria for MDD for at least one month Minimum score of 20 on HRSD (Hamilton Rating Scale for Depression) or 25 on MADRS (Montgomery-Asberg Depression Rating Scale) Remission rates (Hamilton Rating Scale for Depression, HRSD total score of 7) Rx1: 45% Rx2: 35% Rx3: 25% (NNT = 10) favoring venlafaxine Rx1 vs. Rx3 = 2.2 Rx2 vs. Rx3 = 1.4 Testing for homogeneity of the odds ratio revealed no significant difference (p = 0.28) Drop-outs due to side effects (Rx1 vs. Rx3) p = (significant) (Rx2 vs. Rx3) p = (significant) (Rx1 vs. Rx2) p = 0.185) (not significant) Absolute remission rates were significantly higher (10%) with venlafaxine (SNRI) than with an SSRI Only four of eight trials were placebo controlled All trials reviewed were funded by the manufacturer of venlafaxine. SSRI remission rate lower than seen in other trials, suggesting possible publication bias (nonsystematic review). 194 National Adult Depression Clincial Practice Guideline

199 Table 1.29 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Mulsant, H (RCT, double-blind) Follow-up: 12 weeks Initial n: 116 Final n: 64 Outpatients Rx1: Nortriptyline, 25 mg in the evening, placebo pills in the morning Rx2: Paroxetine, 10 mg in the morning initially and increased to 20 mg after one week, placebo pills in the evening Inpatients Rx1: Nortriptyline, 50 mg in the evening, placebo pills in the morning Rx2: Paroxetine, 20 mg in the morning, placebo pills in the evening Rx1: 54 Rx2: 62 Age 60 and older with MDD 71.6% female 86.2% white 17-item Hamilton Rating Scale for Depression (HAM- D) score of 15 or above Mini-Mental State Exam (MMSE) score of 15 or above Hamilton Rating Scale for Depression for intention-to treat groups Rx1: 9.8 ± 5.2 Rx2: 11.5 ± 7.1 p = 0.16 (not statistically significant) Discontinuation: Patients were twice as likely to discontinue nortriptyline (33%) as paroxetine (16%) due to a significant side effect (p = 0.04), but the overall discontinuation rates did not differ significantly, p = UKU side effects total score for intention-totreat groups Rx1: 12 ± 6 Rx2: 12 ± 7 p = 0.61 (not statistically significant) Efficacy of nortriptyline and paroxetine did not differ significantly in older patients with MDD. Dropouts attributed to a side effect were significantly higher in patients treated with nortriptyline. No difference was noted in overall frequency of side effects or dropouts for all reasons. Overall reasons for dropout included items of questionable clinical relevance for initial decision making when deciding between two medications (i.e., transportation issues). Separate analysis of dropouts due to patient perceived lack of efficacy was not presented. Study results based on both out patients and inpatients outcomes. No separate statistical analysis given for outpatients only. 195 National Adult Depression Clincial Practice Guideline

200 2. Hypericum (St. John s Wort) For Treatment of MDD Problem Formulation 2 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Should Hypericum (St. John s Wort) be used for treatment of adults with MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adults (age 19 and older) with Major Depression MDD Health Intervention: Hypericum (St. John s Wort) vs: Prescription antidepressants (SSRI, SNRI, TCA, NRI, or DA) No treatment Placebo Practitioners: Setting: Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention: KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Sexual problems Drowsiness Headache Nausea Insomnia Agitation/nervousness Dry mouth Seizures Photosensitivity Bleeding Office/UCC/ER visits Hospitalizations Mortality Serotonin syndrome Elevated blood pressure Constipation Diarrhea Abdominal pain Dizziness Blurred vision Weight gain GI bleeding Falls 196 National Adult Depression Clincial Practice Guideline

201 Search Strategy Database: PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Search Date 8/2005 to 9/2007 No. Included / Total Retrieved * 0/129 Cochrane Depression N/A /134 PubMed ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "Major Depressive Disorder"[All Fields]) AND ("hypericum"[mesh Terms] OR(("hypericum"[TIAB] NOT Medline[SB]) OR "hypericum"[mesh Terms] OR St. John's wort[text Word])) Randomized Controlled Trial, All Adult: 19+ years, English, Human 8/2005 to 10/2007 2/6 ((((depression/drug therapy[mesh] OR depression/therapy[mesh]) OR ("depressive disorder/drug therapy"[mesh] OR "depressive disorder/therapy"[mesh])) OR ("dysthymic disorder/drug therapy"[mesh] OR "dysthymic disorder/therapy"[mesh])) AND ("Serotonin Uptake Inhibitors"[MESH] OR "Antidepressive Agents, Tricyclic"[MESH])) Randomized, controlled trial, All Adult: 19+ years English, Human 3/2001 to 9/2007 0/384 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 197 National Adult Depression Clincial Practice Guideline

202 Database: Cochrane Clinical Evidence, Issue 14 PubMed Cochrane Clinical Evidence Search Terms: ((((depression/drug therapy[mesh] OR depression/therapy[mesh]) OR ("depressive disorder/drug therapy"[mesh] OR "depressive disorder/therapy"[mesh])) OR ("dysthymic disorder/drug therapy"[mesh] OR "dysthymic disorder/therapy"[mesh])) AND "hypericum"[mesh Terms]) ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "Major Depressive Disorder"[All Fields]) AND ("hypericum" [MeSH Terms] OR St. John's wort) Depression, Anxiety and Neurosis Group Depressive Disorders ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "Major Depressive Disorder"[All Fields]) AND ("hypericum"[mesh Terms] OR (("hypericum"[tiab] NOT Medline[SB]) OR "hypericum"[mesh Terms] OR St. John's wort[text Word])) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] Depression Depression Article Type and Other Limits: Randomized, controlled trial, All Adult: 19+ years English, Human Meta-analysis, All Adult: 19+ years, English, Human Controlled Clinical Trial OR Comparative Study, All Adult: 19+ years, English, Human Systematic Reviews Systematic Reviews and RCTs Randomized, controlled trials, All adults: 19+ years, English, Human Systematic reviews Systematic reviews & RCTs Search Date 10/2001 to 9/2007 4/03 to 10/2007 4/03 to 10/2007 No. Included / Total Retrieved * 2/ National Adult Depression Clincial Practice Guideline 0/0 0/9 Q4, /131 January /01/03-08/01/05 N/A 1/3 03/05/03 0/295 03/05/03 1/80

203 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * PubMed Search #1 update on AHRQ (((((((((((depression/drug therapy[mesh] OR depression/therapy[mesh]) OR ("depressive disorder/drug therapy"[mesh]or "depressive disorder/therapy"[mesh])) OR ("dysthymic disorder/drug therapy"[mesh] OR "dysthymic disorder/therapy"[mesh])) AND ("Serotonin Uptake Inhibitors"[MESH] OR "Antidepressive Agents, Tricyclic"[MESH])) AND notpubref[sb]) Randomized, controlled trials, All adults: 19+ years, English, Human 01/01/98-03/2001 4/4 ((((((((((depression/drug therapy[mesh] OR depression/therapy[mesh]) OR ("depressive disorder/drug therapy"[mesh] OR "depressive disorder/therapy"[mesh])) OR ("dysthymic disorder/drug therapy"[mesh] OR "dysthymic disorder/therapy"[mesh])) AND "hypericum"[mesh Terms]) AND notpubref[sb]) Randomized, controlled trials, All adults: 19+ years, English, Human 03/ /2001 1/6 ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "Major Depressive Disorder"[All Fields]) AND ("hypericum" [MeSH Terms] OR St. John's wort) Meta-analysis, All Adult: 19+ years English, Human Controlled Trials, All Adult: 19+ years English, Human 01/01/01-04/01/03 01/01/01-04/01/03 0/0 3/3 199 National Adult Depression Clincial Practice Guideline

204 Evidence Tables Table 2.1: Effects of Treatment with Hypericum for Major Depression (RCTs) Name Gastpar et al., 2005 Gastpar et al., 2006 Kasper et al., 2006 Tx (N) Hypericum 612 (123) Sertraline 50 (118) Hypericum 900 (131) Citalopram 20 (127) Mean ±SD Age (yr) % Female Follow-up Rate (%) Follow-up Time Mean ±SD Baseline HAMD Mean ±SD Follow-up HAMD Effect Difference 48.3 ± weeks 22.0± ± ± ± ±6.3 p < noninferiority test Study Quality * Biases 5 N 50.8 ± weeks 21.9± ±6.4 vs C: 0.1 Noninferior 5 N 49.3 ± ± ±6.4 vs P: 2.6 < Placebo (130) 49.4 ± ± ±6.9 vs P: 2.5 < Hypericum 600 (123) Hypericum 1200 (127) 46.3 ± weeks 22.8± ±7.0 vs P: 5.6 < N 46.1 ± ± ±8.3 vs H12 :0.8 NS Placebo (82) 46.6 ± ± ±8.8 vs P: 4.8 <0.001 Comments: Support from pharmaceutical manufacturers Fava et al., 2005 Hypericum 900 (45) Fluoxetine 20 (47) 37.4 ± weeks 19.6± ±6.6 vs P: ± ± ±7.3 vs F: 3.1 <0.05 Placebo (43) 37.8 ± ± ±6.4 vs P: - 1 NS Comments: Support from pharmaceutical manufacturers HAMD = Hamilton Depression Scale; NS, not significant * Study quality measured by Jadad trials scoring system (1 to 5 = low to high). Biases: N: none; 1: sample attrition >15%; 2: sample selection bias; 3: detection bias (e.g., measurement error, power); 4: study procedure biases. 200 National Adult Depression Clincial Practice Guideline

205 Study Table 2.2 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Gelenberg (2004), The effectiveness of St. John's wort in Major Depressive Disorder: A naturalistic Phase 2 follow-up in which nonresponders were provided alternate medication. Men and women with DSM-IV Major Depressive Disorder from tertiary care clinics in academic medical centers. 18+ Naturalistic, follow-up design. Few patients who continued on St. John's wort. Sample limited to outpatients in a tertiary care academic center not necessarily interested in herbal treatment limited generalizability. Small sample. Nonresponders not blinded (open label). Patients who did not respond to hypericum (N = 43) or placebo (N = 55) in Phase 1 treated with antidepressants for 24 weeks (Phase 2). 24 weeks Phase 2 changes in HAM-D scores between nonresponders who received hypericum extract and those who received placebo in Phase 1. Significant time effect for Phase 1 nonresponders over the 24 weeks of Phase 2 treatment, indicating that nonresponders in the initial trial were not treatment resistant (nonresponders in phase 1 did respond in phase 2). HAM-D scores of Phase 1 nonresponders in Phase 2: Baseline Endpoint Hypericum Placebo No significant effect of prior Phase 1 treatment (hypericum or placebo) group. < National Adult Depression Clincial Practice Guideline

206 Study Table 2.3 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Uebelhack (2004). Efficacy and tolerability of Hypericum extract STW 3-VI in patients with moderate depression: a double-blind, randomized, placebo-controlled clinical trial. Inclusion: Outpatients with moderate depression disorder, a total HAM-D score of at baseline. Exclusion: Suicidal patients; those with resistant depression Adequacy of blinding not assessed. Remission rates not assessed. After a 7-day placebo runin period, 140 patients were randomized into one of two groups: Rx1: Intervention (N = 70): Hypericum extract 900mg once daily for six weeks. Rx2: Control (N = 70): Placebo once daily for six weeks. Primary outcome was change in HAM-D score, secondary measure was responder rate (% of patients with >50% decrease from baseline HAM-D score).. 6 weeks Change in total HAM- D score from baseline to end of study. Significant improvement in HAM-D score in hypericum group when compared with placebo group: Baseline Endpoint Hypericum Placebo Hypericum group showed significant improvement in BfS score when compared with placebo group: Baseline Endpoint Hypericum Placebo % of Hypericum group assessed as responders at end of study, compared with 5.7% in placebo group. 23 adverse events reported by patients, none considered serious, and two cases suggested a connection to the study medication. < < National Adult Depression Clincial Practice Guideline

207 Table 2.4 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Szegedi (2005), Acute treatment of moderate to severe depression with hypericum extract WS 5570 (St. John's wort): randomised controlled doubleblind noninferiority trial versus paroxetine. Inclusion: Caucasian male and female adult outpatients with moderate to severe MDD and total HAM-D score 22 Exclusion: Decrease in total depression score of 25% during run-in; diagnoses of schizophrenia, acute anxiety, adjustment disorder, other depressive disorders, organic mental disorder, or substance abuse; suicidality or previous suicide attempt Placebo control not used because of ethical reasons (studying patients with moderate to severe depression) Patients at high risk of suicide excluded. After 3-7 day placebo run in period, participants were randomized into one of two groups: Rx1: Hypericum (N = 122): 900mg hypericum extract three times per day for six weeks. Rx2: Paroxetine (N = 122): 20mg paroxetine once per day for six weeks. If participant showed no response after two weeks, dosage was increased: to 1800mg hypericum or 40mg paroxetine. 6 weeks Change in HAM-D score from baseline to day 42. HAM-D scores decreased for both groups. ITT Analysis: Baseline Endpoint Hypericum Paroxetine Per-Protocol Analysis: Baseline Endpoint Hypericum Paroxetine Authors conclude from these findings that hypericum is not inferior to paroxetine. More patients in hypericum group (50%) than in paroxetine group (35%) showed remission Incidence of adverse effects: Hypericum: Paroxetine: National Adult Depression Clincial Practice Guideline

208 Study Table 2.5 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Bjerkenstedt (2004) Hypericum extract LI 160 and fluoxetine in mild to moderate depression: a randomized, placebo-controlled multi-center study in outpatients. Inclusion: Caucasian males and females with mild to moderate depression (as defined by the DSM-IV) with a total HAM-D score 21. Exclusion: Psychiatric disorders; MAO-I treatment within 14 days prior to entry; history of treatment-resistant MDD, risk of suicide, history of seizure disorder, alcohol or substance abuse; unstable medical illness, impaired renal function, pregnancy, known intolerance to study medications; substantial placebo response; treatment with trial drug prior to study Fixed doses - dosing was not increased for non- or partial response, as it might be in clinical setting. Short duration - only four weeks - due to ERC restriction. Poor estimation of compliance. No statistical test done on fluoxetine vs. hypericum. Dropout rates not reported. After a 3-7 day run-in period, 163 subjects were randomized into one of three groups: Rx1: Hypericum (N = 54): 300mg tid for six weeks Rx2: Fluoxetine (N = 54): 20mg per day for six weeks Rx3: Placebo (N = 55): for six weeks. Per ERC requirement, placebo patients randomly switched to hypericum or fluoxetine after four weeks, until end of study. 6 weeks Change in HAM-D scores from baseline to week 4. Similar improvements among treatment groups on HAM-D. The only statistically significant finding was remission rate (HAM-D total score < 8): Hypericum (24%) vs. placebo (7%) Fluoxetine (28%) vs. placebo (7%) Hypericum better tolerated than fluoxetine Number of adverse events: Hypericum (38) vs. fluoxetine (52) Hypericum (38) vs. placebo (27) Fluoxetine (52) vs. placebo (27) National Adult Depression Clincial Practice Guideline

209 Table 2.6 Study, Total n Hypericum Depression Trial Study Group 2002 (RCT, double-blind) Follow-up: 8 weeks Initial n: 340 Final n: 245 Treatment Groups Size & Drug Rx1: hypericum (900 1,500 mg/dl) (n = 113) Rx2: sertraline ( mg/d) (n = 111) Rx3: placebo (n = 116) One week run-in period Study Population Results Comments At least 18 years old with current diagnosis of Major Depression Minimum total score of 20 on the 17-item Hamilton Depression (HAM-D) scale HAM-D total score (week 8 week 1) Rx1: Rx2: Rx3: (Rx1 vs. Rx3) p = 0.59 (not significant) (Rx2 vs. Rx3) p = 0.18 (not significant) Full response rate (%) Rx1: 23.9 Rx2: 24.8 Rx3: 31.9 (Rx1 vs. Rx3) p = 0.21 (not significant) (Rx2 vs. Rx3) p = 0.26 (not significant) Full response is indicated with a HAM-D score of eight or less and a Clinical Global Impressions Improvement (CGI-I) score of one or two. Adverse Events Diarrhea (Rx1 vs. Rx3) p = 0.81 (not significant) (Rx2 vs. Rx3) p = (significant) Nausea (Rx1 vs. Rx3) p = 0.78 (not significant) (Rx2 vs. Rx3) p = 0.02 (significant) Anorgasmia (Rx1 vs. Rx3) p = 0.04 (significant) (Rx2 vs. Rx3) p = (significant) They found no evidence that hypericum is more effective than placebo in treating moderately severe Major Depression. The principal comparison was between the hypericum and placebo groups. Sertraline served as an active comparator to evaluate the study s sensitivity. 205 National Adult Depression Clincial Practice Guideline

210 Table 2.7 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Kalb, R. 2001, (RCT, double-blind) Follow-up = 6 weeks Initial n: 72 Final n: 64 Rx1: hypericum extract WS 5572 (3 x 300 mg/day) (n = 37) Rx2: placebo (n = 35) Male and female outpatients aged between 18 and 65 years with diagnosis of mild or moderate MDD with single or recurrent episodes according to DSM-IV criteria Total score for the Hamilton Rating Scale for Depression (HAMD, 17 item version) 16 at study entry and during a subsequent baseline investigation (3 to 7 days later). Change (decrease) in HAM-D score (%) Rx1: 54.8% (19.7 ± 3.4 to 8.9 ± 4.3 points, overall change points) Rx2: 29.2% (20.1 ± 2.6 to 14.4 ± 6.8 points, overall change -5.7 points) Difference 5.1 points, p < (in favor of hypericum) % of patients showing at least a 50% reduction in HAM-D scores: Rx1: 62.2% [95% CI: 46.5 to 77.8] Rx2: 42.9% [95% CI: 26.5 to 59.3] p = 0.10 The study concludes that the standardized Hypericum extract WS 5572 has superior efficacy compared with placebo and very good tolerability in the acute treatment of mildly to moderately depressed patients. High placebo response rate. Study may be underpowered to detect differences in 50% reduction in HAM-D scores. % of patients showing at least a 60% reduction in HAM-D scores: Rx1: 51.4% [95% CI: 35.2 to 67.5] Rx2: 17.1% [95% CI: 4.7 to 29.6] p = Table 2.8 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Behnke, K (RCT, double-blind) Follow-up = 6 weeks Initial n: 70 Rx1: hypericum (150 mg twice daily) (n = 35) Rx2: fluoxetine (20 mg twice daily) (n = 35) 18 to 73 years of age with mild to moderate depression Hamilton Rating Scale for Depression of between 16 and 24 Change (decrease) in HAM-D score (%) Rx1: 50% Rx2: 58% The decrease was highly significant (p < 0.001) in both groups but not significantly different (p = 0.23) between groups The study concludes that Hypericum perforatum is therapeutically equivalent to fluoxetine and therefore a rational alternative to synthetic antidepressants. Final n: 61 No placebo control group No placebo control group 206 National Adult Depression Clincial Practice Guideline

211 3. Antidepressants In Patients With MDD Expressing Suicidal Ideation, Intent, or Plan Problem Formulation 3 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention: Which antidepressants should be avoided for treatment of patients with MDD expressing suicidal ideation, intent, or plan? To assist primary care physicians and other health care professionals in treating adults with Major Depression who are expressing suicidal ideation, intent, or plan. All adult (age 19 and older) patients with Major Depression treated with antidepressant medication who are expressing suicidal ideation, intent, or plan Suicide ideation, intent, or plan in primary care patients taking antidepressants Antidepressant treatment (SSRIs, TCAs, DAs, SNRIs, NRIs) KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Hospitalizations Attempted suicide Suicide by overdose Suicide (all cause) 207 National Adult Depression Clincial Practice Guideline

212 Search Strategy Database: PubMed Cochrane Clinical Evidence, Issue 14 PubMed Search Terms: (((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "antidepressive agents/poisoning"[mesh] OR "Major Depressive Disorder"[All Fields]) AND (("suicide"[mesh Terms] OR suicide[text word]) OR "suicide, attempted"[mesh])) Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders (((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "antidepressive agents/poisoning"[mesh] OR "Major Depressive Disorder" [All Fields])AND (("suicide"[mesh Terms] OR suicide[text word]) OR "suicide, attempted"[mesh])) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "3000"[PDAT]) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Article Type and Other Limits: Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Reviews Systematic Reviews and RCTs All Adult: 19+ years, Publication Date from 2003/01/01 to 2005/08/01, English, Humans Search Date 8/2005 to 9/2007 No. Included / Total Retrieved * 0/411 Q4, /131 January /01/03-08/01/05 N/A 1/187 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 208 National Adult Depression Clincial Practice Guideline

213 Database: Cochrane Clinical Evidence PubMed PubMed Suicide Suicide Search Terms: (((((((((((depression/drug therapy [MESH] OR depression/therapy [MESH]) OR ("depressive disorder/ drug therapy"[mesh] OR "depressive disorder/therapy" [MESH])) OR ("dysthymic disorder/drug therapy"[mesh] OR "dysthymic disorder/therapy"[mesh])) AND ("Serotonin Uptake Inhibitors"[MESH] OR "Antidepressive Agents, Tricyclic"[MESH])) AND notpubref[sb]) (((((((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "antidepressive agents/poisoning" [MeSH] OR "Major Depressive Disorder"[All Fields]) AND (("suicide"[mesh Terms] OR suicide[text word]) OR "suicide, attempted"[mesh])) Article Type and Other Limits: Systematic reviews Systematic reviews and RCTs Randomized, controlled trials, All adults 19+ years, English All Adult: 19+ years English, Human No. Included Search / Total Date Retrieved * 06/05/03 0/80 06/05/03 1/ / /01/01-04/01/03 0/222 1/67 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. Buckley (112) did not show up in our search results due to the way PubMed indexing is done. Other Information Source: Evidence report on the Treatment of Depression - Newer Pharmacotherapies. AHRQ (169) 209 National Adult Depression Clincial Practice Guideline

214 Evidence Tables Table 3.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Yerevanian, et al. (2004) Antidepressants and suicidal behavior in unipolar depression Inclusion: In clinical care with senior author from ; DSM-IV criteria for Major Depression and/or dysthymia; minimum 6 month follow-up by senior author; antidepressant monotherapy; charts contained specific information about suicidal behavior. Exclusion: Multiple antidepressant, mood stabilization, or antipsychotic regimens; alcohol and substance abuse. 43 (mean) Treatment adherence not part of methodology. Patient charts (N = 521) were followed and reviewed. N/A Number of the following events during periods of active treatment or discontinuation with SSRIs or TCAs. 1) completed suicide, 2) suicide attempts, or 3)hospitalization for serious suicidal thought or intent Rate of suicidal behavior significantly higher in discontinuation vs. active treatment period (p < ) Rate of suicidal behavior significantly higher in discontinuation vs. active treatment period with TCAs (p < ) Rate of suicidal behavior significantly higher in discontinuation vs. active treatment period with SSRIs (p <.00001) Rate of suicidal behavior similar for active TCA vs. active SSRI treatment (p =0.046) < < < < National Adult Depression Clincial Practice Guideline

215 Study Martinez, et al. (2005) Antidepressant treatment and the risk of fatal and nonfatal self-harm in first episode depression: nested case control study. Table 3.2 Inclusion & Exclusion Criteria Age Limitations / Biases Patients (N = 146,095) from Primary Care in the UK with a new diagnosis of depression who were prescribed antidepressants for the first time between 1995 and Outcome under study might in itself be the reason one drug was prescribed over another. No consistent monitoring of adherence to medication. Large number of hypothesis tests could have led to change findings of associations. No comparison between those prescribed SSRIs and those with equivalent morbidity not receiving treatment (i.e. no control). Intervention and dose N and Final N Duration Outcome Results NNT p value Clinical records (N = 146,095) from the General Practice Research Database were acquired and reviewed. N/A Risk of nonfatal selfharm and suicide between users of SSRIs and users of TCAs cases of self-harm 69 suicides Adjusted odds ratio of nonfatal self-harm in people prescribed SSRIs vs. TCAs was 0.99 (95% CI: ) Adjusted odds ratio of suicide in people prescribed SSRIs vs. TCAs was 0.57 (95% CI: ) No evidence risk of nonfatal selfharm varied among different SSRIs. No evidence risk of nonfatal selfharm varied among different TCAs. NS NS No age-related differences in nonfatal self harm between patients prescribed SSRIs and TCAs. 211 National Adult Depression Clincial Practice Guideline

216 Table 3.3 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Fergusson, et al. (2005) Association between suicide attempts and selective serotonin reuptake inhibitors: systematic review of randomised controlled trials. All RCTs (N = 702) comparing SSRI with placebo or active non-ssri control. Treatment condition not limited to Major Depression. Underreporting of suicide attempts in RCTs. High rates of losses to follow-up. Small trial sizes and durations. Some trials restricted eligibility to patients who previously tolerated SSRIs well. Three separate meta-analyses were conducted to evaluate the association between suicide attempts and the sue of SSRIs: 1) SSRIs compared with placebo 2) SSRIs compared with TCAs 3) SSRIs compared with other active forms of treatment. Number of fatal and nonfatal suicide attempts. Significant increase in odds of suicide attempts for patients receiving SSRIs compared with placebo (odds ratio 2.28, CI: 1.14 to 4.55). No difference in odds of suicide attempts in patients receiving SSRIs compared with TCAs (odds ratio 0.88, CI: ). Increase in odds of suicide attempts when comparing SSRIs with therapeutic interventions other than TCAs (odds ratio 1.94, 1.06 to 3.57). 684 (NNT to harm) 239 (NNT to harm) 0.02 Some trials did not have a sufficient washout period. Did not compare TCAs with placebo, or other active treatment with placebo. 212 National Adult Depression Clincial Practice Guideline

217 Study Gibbons, et al. (2005) The relationship between antidepressant medication use and rate of suicide. Table 3.4 Inclusion & Exclusion Criteria Age Limitations / Biases Suicides in all US counties between ; adjusted for sex, race, age and income. 5+ Medication estimates based on outpatient use. Uncontrolled variability in suicide rate data (definition of suicide, qualifications of medical examiner, extent of case investigation, relationship between prescription rates and actual taking of medication, quality of data) Possible selection bias (different classes of antidepressants may have been selectively prescribed to different patient groups). No data on cause of suicide. Intervention and dose N and Final N Duration Outcome Results NNT Analysis of relationship between antidepressant pharmacy prescription volumes and overall suicide rate. N/A Primary outcome measure was number of suicides for given population size. Overall relationship between all prescribed antidepressant drugs and suicide rate was not statistically significant (p = 0.14). For individual classes of antidepressants, there was a significant positive association between TCAs and suicide rate (p < 0.001). Combination of SSRI and non-ssri or non-tca prescriptions had a significant negative association with suicide rate (p < 0.001). p value 0.14 < < National Adult Depression Clincial Practice Guideline

218 4. Second-Line Treatement of MDD Problem Formulation 4 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention: What strategies should be used in adults with MDD whose symptoms do not resolve after the first-line treatment? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) with Major Depression who are unresponsive to first-line treatment MDD Change antidepressant medication Increase existing antidepressant dose Switch to psychotherapy Add psychotherapy Add another antidepressant to existing antidepressant Add an augmenting agent to the existing antidepressant (lithium, beta-blocker, buspirone, Cytomel, carbamazepine, valproic acid, methylphenidate, ethyl-eicosapentaenoate (E-EPA), folate, inositol, atypical antipsychotic, risperidone, aripiprazole rtms Vagus nerve stimulation KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Sexual problems Drowsiness Headache Nausea Insomnia Agitation/nervousness Dry mouth Seizures Elevated blood pressure Office/UCC/ER visits Hospitalizations Mortality Constipation Diarrhea Abdominal pain Dizziness Blurred vision Weight gain GI bleeding Falls 214 National Adult Depression Clincial Practice Guideline

219 Search Strategy Database: PubMed Cochrane Systematic Reviews PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR metaanalysis"[all Fields])) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Search Date 8/2005 to 9/2007 Depression N/A 2005 to 2007 ((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "Major Depressive Disorder"[All Fields]) AND ((((("Antidepressive Agents/adverse effects"[mesh] OR "Antidepressive Agents/therapeutic use"[mesh]) OR apiprazole [text word] OR atypical antipsychotic [text word] OR "carbamazepine"[mesh terms] OR "valproic acid"[mesh terms] OR "buspirone"[mesh terms] OR "triiodothyronine"[mesh terms] OR Cytomel[text word] OR "adrenergic betaantagonists "[MESH terms] OR beta blocker[text word] OR "methylphenidate"[mesh terms] OR "psychotherapy"[mesh terms]) OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms])) OR "interpersonal therapy"[all Fields]) OR ("Problem-solving"[MESH Terms] AND "therapy"[mesh Subheading]))) Randomized controlled trials, only items with abstracts, Humans, English, All Adult: 19+ years 8/2005 to 9/2007 No. Included / Total Retrieved * 0/129 0/134 5/271 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 215 National Adult Depression Clincial Practice Guideline

220 Database: National Institute for Clinical Excellence (NICE) Depression Search Terms: Article Type and Other Limits: Systematic Review, Metaanalyses and RCTs Cochrane Depression, Anxiety and Neurosis Group Systematic Reviews Clinical Evidence, Issue 14 PubMed Cochrane Chapter: Depressive Disorders (((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/ therapy" [MESH]) OR "Major Depressive Disorder"[All Fields]) AND ((((("Antidepressive Agents/adverse effects"[mesh] OR "Antidepressive Agents/therapeutic use"[mesh]) OR "carbamazepine"[mesh terms] OR "valproic acid"[mesh terms] OR "buspirone"[mesh terms] OR "triiodothyronine "[mesh terms] OR Cytomel[text word] OR "adrenergic betaantagonists "[mesh terms] OR beta blocker[text word] OR "methylphenidate"[mesh terms] OR "psychotherapy"[mesh]) OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms])) OR "interpersonal therapy"[all Fields]) OR ("Problem-solving"[MeSH Terms] AND "therapy"[mesh Subheading]))) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Depression Systematic Reviews and RCTs Randomized, controlled trials Systematic reviews Search Date Decembe r 6, 2004 No. Included / Total Retrieved * N/A Q4, /131 January /01/05-08/01/05 N/A 6/239 03/05/03 0/ National Adult Depression Clincial Practice Guideline

221 Database: Clinical Evidence PubMed PubMed Depression Search Terms: (((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "Major Depressive Disorder"[All Fields]) AND ((((("Antidepressive Agents/adverse effects"[mesh] OR "Antidepressive Agents/therapeutic use"[mesh]) OR "carbamazepine" [mesh terms] OR "valproic acid" [mesh terms] OR "buspirone" [mesh terms] OR "triiodothyronine " [MESH terms] OR Cytomel [text word] OR "adrenergic betaantagonists " [MESH terms] OR beta blocker [text word] OR "methylphenidate" [MESH terms]or "psychotherapy"[mesh]) OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms])) OR "interpersonal therapy"[all Fields]) OR ("Problemsolving"[MESH Terms] AND "therapy"[mesh Subheading]))) /therapeutic use"[mesh]) OR "carbamazepine" [mesh terms] OR "valproic acid" [mesh terms] OR "buspirone" [mesh terms] OR "triiodothyronine " [mesh terms] OR Cytomel [text word] OR "adrenergic beta-antagonists " [mesh terms] OR beta blocker [text word] OR "methylphenidate" [mesh terms]or "psychotherapy"[mesh]) OR (cognitive[all Fields] AND "behavior therapy"[mesh Terms])) OR "interpersonal therapy"[all Fields]) OR ("Problem-solving"[MeSH Terms] AND "therapy"[mesh Subheading]))) Article Type and Other Limits: Systematic reviews and RCTs Meta-analysis, All Adult: 19+ years English, Human Controlled Trials, All Adult: 19+ years English, Human Meta-analysis, All Adult: 19+ years English, Human Controlled Trials, All Adult: 19+ years English, Human Search Date No. Included / Total Retrieved * 03/05/03 1/80 01/01/80-12/31/00 01/01/80-12/31/00 01/01/01-04/01/03 01/01/01-04/01/03 0/12 0/576 0/5 3/129 The rationale statement from the November 2005 KP Inter-regional New Technologies evidencebased review on vagus nerve stimulation was incorporated into these guidelines. Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. The Peet (136) study did not show up in the search results due to the way PubMed indexing is done. The GDT decided to include this study, since it met the inclusion criteria. The Perez (138) study did not show up in the search results because pindolol was indexed as serotonin receptor antagonist rather than as a beta-blocker or an adrenergic betaantagonist. 217 National Adult Depression Clincial Practice Guideline

222 Evidence Tables Table 4.1: RCTs of Second-Line Treatments for Depression Name Thase, 2007 Mean ±SD Age (yr) N AUG 40.0±12.8 SW 42.2±13.7 N % Female Follow -up Rate NR Follow -up Time 14 weeks Augmentation with Cognitive Therapy % Remissions N Augmentation with Medication % Remissions N Switch to Cognitive Therapy % Remissions N Switch to New Medication % Remissions N p Study Quality NS 1 3, 4 Cognitive therapy not independently validated. No placebo control group included. Mean Follow Follow Bupropion-SR Sertraline Venflaxine-XR ±SD Age (yr) % -up -up % % % Study Name N Female Rate Time Remissions N Remissions N Remissions N p Quality Biases* Rush, 41.8± NR weeks NS 1 4 Adverse effects were similar among treatment groups. No placebo control group included. Bupropion-SR Buspirone Name Mean ±SD Age (yr) N % Female Follow -up Rate Follow -up Time % Remissions N % Remissions N p Study Quality Biases* Trivedi, 41.1 ± NR weeks NS 1 4 Adverse effect burden was similar among treatment groups; however, fewer participants taking citalopram plus sustained-release bupropion stopped treatment because of intolerance than did those taking citalopram plus buspirone (12.5 percent vs percent; 2 = 6.86; p<0.001). No placebo control group included. Mean Follow Follow Risperidone Placebo Name ±SD Age (yr) N % Female -up Rate -up Time % Remissions N % Remissions N p Study Quality Biases* Mahmoud, Risperidone % Placebo weeks Adverse effects were similar among treatment groups. Funded by manufacturer of risperidone. AUG, augmentation; NR, not reported; NS, not significant; SW, switch Study quality measured by Jadad scoring system (1 to 5 = low to high *Biases: N: none; 1: sample attrition >15%; 2: sample selection bias; 3: detection bias (e.g., measurement error, ITT analysis, power); 4: study procedure biases (e.g., blinding, randomization) Biases* 218 National Adult Depression Clincial Practice Guideline

223 Table 4.2: Meta-analyses of Augmentation with Atypical Antipsychotic Medications Studies Total N Papakostos GI 2007 (meta-analysis) # studies found: 137 # studies included: 10 Total N = 1500 Heterogeneity: heterogeneity addressed by use of randomeffects model Study Population Group Size & Drug Results Comments All patients diagnosed with treatmentresistant depression All trials compared adjunctive treatment with atypical antipsychotics against placebo treatment. 4 trials from published medical literature 6 reports from scientific meetings 5 reports of olanzapine 3 reports of quetiapine 2 reports of risperidone Remission: RR = 1.75 (95% CI, ; p<. 0001) for active treatment versus placebo Pooled remission: Active treatment: 47.4% Placebo: 22.3% Pooled response: Active treatment: 57.2% Placebo: 35.4% Discontinuation due to adverse effects:rr = 3.38 (95% CI, ; p< ) for active treatment versus placebo No difference noted in overall rate of discontinuation: RR = 1.18 (95% CI, ; p =0.929) Higher discontinuation rate due to side effects with active treatment: RR = 3.38 (95% CI, ; p<0.001) Authors conclusions: These results support the utility of augmenting therapy for treatment-resistant depression with atypical antipsychotic. Biases, etc.: Studies involved only olanzapine, quetiapine, and risperidone. Table 4.3: RCTs of Augmentation with Atypical Antipsychotic Medications Follow Placebo Apiprazole Name Mean ± SD Age (yr) N % Female Followup Rate -up Time % Remissions N % Remissions N p Study Quality Biases* Berman, 2007 Placebo 44.2± Apiprazole 46.5± weeks < None 81.9% of active-treatment patients reported adverse effects. Study quality measured by Jadad scoring system (1 to 5 = low to high *Biases: N: none; 1: sample attrition >15%; 2: sample selection bias; 3: detection bias (e.g., measurement error, ITT analysis, power); 4: study procedure biases (e.g., blinding, randomization) 219 National Adult Depression Clincial Practice Guideline

224 Table 4.4: RCTs of Third-Line Treatments for Depression Mean ± SD Age (yr) Followup Follow- Mirtazapine Nortriptyline Study Name N % Female Rate up Time % Remissions N % Remissions N p Quality Biases* Fava, ± NR 14 weeks NS 1 3,4 Adverse effects were similar among treatment groups. No placebo control group included. Mean ± SD Age (yr) Followup Follow- Lithium Triiodothyronine Study Name N % Female Rate up Time % Remissions N % Remissions N p Quality Biases* Nierenberg, ± NR 9.6 weeks NS 1 3, 4 Fewer side effects were seen with triiodothyronine. No placebo control group included. NR, not reported; NS, not significant Study quality measured by Jadad scoring system *Biases: N: none; 1: sample attrition >15%; 2: sample selection bias; 3: detection bias (e.g., measurement error, ITT analysis, power); 4: study procedure biases (e.g., blinding, randomization) Table 4.5 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Schatzberg (2005) Medication (Nefazodone) or psychotherapy (CBASP) is effective when the other is not Inclusion: Adult outpatients in a current Major Depressive Disorder episode and scoring 20+ on the HAM-D Exclusion: Organic mental syndromes, psychotic disorders, panic disorder, PTSD, substance abuse, eating disorders, suicidality, medical contraindications to antidepressants, unstable general medical disorders, nonresponse to nefazodone, recent treatment with benzodiazepines, fluoxetine, MAO-I, ECT, neuroleptics; psychotherapy outside trial No placebo control; no nonresponder control group continuing in previous (initial) therapy. Treatments only switched for nonresponders able to complete acute trial. Excluded patients with greater levels of comorbid conditions. Nonresponders who entered crossover trial may not have been comparable since they were not initially randomized. 681 patients randomized into acute phase of nefazodone (N = 226), CBASP (N = 228) or combination therapy (N = 227). After 12 weeks, nonresponders of nefazodone (N = 73) were switched to CBASP and nonresponders of CBASP (N = 83) were switched to nefazodone. 24 weeks HAM-D score of nonresponders to nefazodone or CBASP 12 weeks after they switched to the other treatment. Both groups improved from baseline of crossover to endpoint. In the intention-to-treat analysis, patients who switched from nefazodone to CBASP showed significantly greater improvement than those who had switched from CBASP to nefazodone. In the completer analysis, there was no significant between-group difference National Adult Depression Clincial Practice Guideline

225 Table 4.6 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Parker, Brotchie, & Parker (2005) Is combination olanzapine and antidepressant medication associated with a more rapid response trajectory than antidepressant alone? Inclusion: Male and female outpatients with a first or new episode of DSM-IV nonpsychotic Major Depression. Exclusion: Antidepressant in preceding two weeks, ECT in previous one month. Mean age 49 Small number of subjects. Patients (N = 20) randomized into one of two groups: Rx1: antidepressant plus olanzapine (N = 10) Rx2: antidepressant plus placebo (N = 10) [After two weeks, Rx2 nonresponders (N = 4) were given late olanzapine augmentation.] 2 weeks HAM-D scores at 2 weeks. Nonsignificant HAM-D improvement: Endpoint Baseline (Day 14) Rx Rx Four nonresponders of antidepressant plus placebo who were given olanzapine at 2 weeks, showed distinctive improvement from baseline at day 21. Endpoint Baseline (Day 21) Rx Table 4.7 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Perry, et al. (2004) Pindolol augmentation in depressed patients resistant to selective serotonin reuptake inhibitors: a double-blind, RCT Inclusion: Male and female outpatients with Major Depression (per DSM IV criteria) and minimum HAM-D baseline score of 25. Exclusion: Neuroleptics, psychosis, substance abuse, pregnancy, contraindications to beta blockers, reactive airway disease, diabetes, significant medical illness Type II error. Inconsistent dosing in acute phase. Confounding patient characteristics. Patients initially received at least 20 mg fluoxetine, 20mg paroxetine, or 50mg sertraline per day for a minimum of 6 weeks prior to study. Forty-two patients who failed to respond sufficiently to acute trial of SSRIs were continued on SSRI treatment and randomly assigned to one of the following groups for 3 weeks: Rx1 pindolol 2.5mg tid (N = 21) Rx2 placebo tid (N = 17) 3 weeks Primary outcome measure was change in HAM-D score from baseline to the end of week 3. No significant differences in antidepressant response between the two groups: Partial response rates for pindolol (19%) and control (24%) were comparable at three weeks. Pindolol and control groups demonstrated mean decreases in HAM-D scores of 6.5 and 9.7, respectively, at three weeks. 221 National Adult Depression Clincial Practice Guideline

226 Study Table 4.8 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Kauffmann (2004) Slow right prefrontal transcranial magnetic stimulation as a treatment for medicationresistant depression: a double-blind, placebocontrolled study Inclusion: Eleven females, one male (mean age 52) with Major Depression (per DSM-IV criteria) who failed to respond to at least two standard antidepressants given at adequate doses for at least 8 weeks. Exclusion: Pre-existing neurological and/or cardiac diseases. 18+ Small number of subjects. Short duration (10 days). Limited evidence. Effectiveness of blinding not assessed. Control group stimulation similar to treatment group stimulation. Unadjusted differences in baseline HAM-D scores: Active group 21.9, control group Patients randomly assigned to one of two groups: Rx1: Right transcranial magnetic stimulation (rtms), N = 7 Rx2: "sham" rtms, N = 5 10 days Primary outcome measure was HAM-D score after 10 days. Mean HAM-D score for Rx1 decreased significantly from to (p<.02) Mean HAM-D score for Rx2 group decreased from to (not significant). (p>.09) On follow-up, Rx1 relapsed after 2-3 months, whereas Rx2 relapsed after 2 weeks. < 0.02 > 0.09 Group x time interaction not assessed. 222 National Adult Depression Clincial Practice Guideline

227 Study Table 4.9 Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Rumi (2005) Transcranial magnetic stimulation accelerates the antidepressant effect of amitriptyline in severe depression: a double-blind, placebo-controlled study. Inclusion: Outpatient adults with severe nonpsychotic Major Depressive Disorder (DSM-IV) and minimum baseline score of at least 22 on HAM-D Exclusion: Neurological conditions, personality disorders, suicide risk, severe uncontrolled organic disease, substance abuse, abnormal lab tests, pacemaker, history of seizure, major head trauma, risk factors for TMS procedure. Mean age 39 Efficacy of blinding not assessed. Short-term study. Longer-term difference in response and remission rates not assessed. Seven days prior to treatment, all patients (N = 46) were started on amitriptyline which were maintained during the 4- week trial. Patients (N = 46) were randomized into one of two groups: Rx1: rtms (N = 22) - 5 sessions a week for four weeks. Rx2: sham rtms (N = 24) - 5 sessions a week for four weeks. Clonazepam was allowed for sedative purposes; its use was monitored throughout the trial. 4 weeks Primary outcome measures were HAM-D and MADRS scores, CGI changes, and VAS at four weeks. Overall response ratio (HAM-D decrease 50%) was significantly higher in rtms group than in sham group (95% and 46% respectively). Remission (HAM-D 7) was significantly higher in rtms group than in sham group (54% and 12% respectively). Similar findings were observed for MADRS and CGI scales, and subjective assessments through VAS. Neck pain and burning, and burning in scalp was significantly predominant in rtms group. The need for clonazepam was significantly smaller in rtms group than in sham group (p<0.001). < < < National Adult Depression Clincial Practice Guideline

228 Table 4.10 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Hirschfeld, M (RCT, open-label trial) Follow-up: 12 weeks Initial n: 681 Final n: Not stated Rx1: nefazodone at 200 mg/day (in 2 divided doses) during the first week and increased to 300 mg/day during week two, dose adjustments were made thereafter in weekly increments of 100 mg/day up to 600 mg/day. Rx2: Cognitive Behavioral Analysis System of Psychotherapy (CBASP) drawing on techniques from behavioral, cognitive, and interpersonal forms of psychotherapy Rx3: Combined nefazodone/cbasp Patients met criteria for either chronic Major Depressive Disorder, current Major Depressive Disorder superimposed on antecedent dysthymic disorder, or recurrent MDD with incomplete interepisode recovery (by definition, these patients would have failed first-line treatment). Men and women between the ages of 18 and 75 (Mean of 43 ± 11) 65% female, 91% white, 43% married/cohabiting Score of 20 on the 24-item Hamilton Rating Scale for Depression (HAM-D) HAM-D score analysis (numbers obtained from initial Keller study) p = 0.68 (Rx1 vs. Rx2) (no difference) p (Rx1 vs. Rx3) (in favor of combined treatment) p (Rx2 vs. Rx3) (in favor of combined treatment) SF-36 general health without HAM-D p = 0.22 (Rx2 vs. Rx1) (no difference) p = (Rx3 vs. Rx1) (in favor of combined treatment) p = 0.02 (Rx3 vs. Rx2) (in favor of combined treatment) SF-36 social functioning with HAM-D as covariate: p = 0.94 (Rx2 vs. Rx1) (no difference) p = 0.02 (Rx3 vs. Rx1) (in favor of combined treatment) p = 0.02 (Rx3 vs. Rx2) (in favor of combined treatment) There was greater improvement in psychosocial functioning and general health when combined treatment was used. Combined treatment was statistically superior to psychotherapy or nefazodone alone in reducing the HAM-D score in patients with chronic Major Depressive Disorder. Patients were not blind to treatment received No placebo control group 224 National Adult Depression Clincial Practice Guideline

229 Study, Total n Table 4.11 Treatment Groups Size & Drug Study Population Results Comments Fava, M (RCT, double-blind) Follow-up: 4 weeks Initial n: 101 Final n: 88 Rx1: High-dose fluoxetine (40 60 mg/day) (n = 33) Rx2: fluoxetine (20 mg/day) plus desipramine (25 50 mg/day) (n = 34) Rx3: fluoxetine (20 mg/day) plus lithium ( mg/day) (n = 34) Outpatients with Major Depressive Disorder (52 men and 49 women, between 18 and 65 years old) who were either partial responders or nonresponders to eight weeks of treatment with fluoxetine 20 mg/day. Initial HAM-D-17 score of 16 Mean change in HAM-D-17 score (from visit 1-endpoint)for all patients with partial response or nonresponse to fluoxetine 20 mg/day Rx1: 5.1± 5.3 Rx2: 3.5 ± 5.6 Rx3: 3.6 ± 6.2 p = 0.4 (not significant) Response rate % (at least 50% reduction in HAM-D score) Rx1: 50% Rx2: 33.3% Rx3: 33.3% p = 0.5 (not significant) Most common side effects: Rx1: gastrointestinal distress (54.5%), headache (42.4%), dizziness (30.3%), dry mouth (27.3%), sedation or fatigue (18.2%) Rx2: dry mouth (55.9%), gastrointestinal distress (47.1%), dizziness (35.3%), insomnia (32.4%), agitation (29.4%), sedation or fatigue (26.5%) Rx3: gastrointestinal distress (50.0%), dry mouth (38.2%), insomnia (35.3%), sedation or fatigue (32.4%), headache (26.5%) For patients who fail to respond to a trial of a standard dose of fluoxetine (20 mg/day), there are no significant differences in efficacy among patients whose fluoxetine dose is raised to 40 to 60 mg/day, patients on fluoxetine plus low-dose desipramine (25 50 mg/day), and patients on fluoxetine plus low-dose lithium ( mg/day). Patients who partially respond to eight weeks of fluoxetine 20 mg/day may respond more completely to higher doses of fluoxetine (with a response rate of 50%), although the difference in response rates across the three treatment groups was not statistically significant. No placebo control group does not allow comparison with continuing initial dose of antidepressant for a longer period of time, and does not exclude placebo response to augmentation. May have been underpowered. There were no significant differences in dropout rates across the three groups. 225 National Adult Depression Clincial Practice Guideline

230 Study, Total n Table 4.12 Treatment Groups Size & Drug Study Population Results Comments Peet, M (RCT, double-blind) Follow-up: 12 weeks Initial n: 70 Final n: 60 Intention-to treat Rx1: ethyl-eicosapentaenoate 1 g/d plus three placebo capsule (n = 17) Rx2: ethyl-eicosapentaenoate 2 g/d plus two placebo capsules (n = 18) Rx3: ethyl-eicosapentaenoate 3 g/d plus one placebo capsule (n = 17) Rx4: Liquid paraffin placebo, four capsules (n = 18) Patients of either sex, aged 18 to 70 HAM-D-17 score of 15 or more despite ongoing treatment with a standard antidepressant at an adequate dose Change in HDRS (17-item Hamilton Depression Rating Scale) score for the intent -to- treat (ITT) population (comparing active treatment with placebo) Rx1: 9.9, p = 0.02 (statistically significant) Rx2: 5.8, p = 0.88 (no difference) Rx3: 5.4, p = 0.44 (no difference) Rx4: 5.1 Change in Montgomery-Asberg Depression Rating Scale (MADRS) score for ITT population (comparing active treatment with placebo) Rx1: 11.2, p = (statistically significant) Rx2: 3.0, p = 0.41 (no difference) Rx3: 8.5, p = 0.15 (no difference) Rx4: 5.4 Change in Beck Depression Inventory (BDI) score for the intention-to- treat (ITT) population (comparing active treatment with placebo) Treatment with ethyl-eicosapentaenoate at a dosage of 1 g/d, but not higher doses is effective in treating depression in patients who remain depressed despite adequate standard therapy (suggests a possible therapeutic window effect). Small n Rx1: 12.5, p = (statistically significant) Rx2: 5.7, p = 0.99 (no difference) Rx3: 9.3, p = 0.24 (no difference) Rx4: National Adult Depression Clincial Practice Guideline

231 Table 4.13 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Nemets, B (RCT, double-blind) Follow-up: 4 weeks Initial n = 20 Final n = 19 Rx1: Ethyl- eicosapentaenoate (E-EPA), 2 g/day Rx2: Placebo 17 women and 3 men (18 75 years old) with a current diagnosis of Major Depressive Disorder 24-item Hamilton Depression Rating Scale of 18 or higher Patients under standard antidepressant treatment for at least three months except for one. Change in mean Hamilton Depression Rating Scale Score from baseline to week four Rx1: 12.4 Rx2: 1.6 p = not stated No clinically relevant side effects were reported. Highly significant benefits of the addition of the omega-3 fatty acids compared with placebo were found by week 3 of treatment, Small n Table 4.14 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Perez, VS (RCT, double-blind) Follow-up: 10 days Initial n: 80 Final n: 78 Rx1: 5-HT reuptake inhibitors plus pindolol (2.5 mg tid) (n = 40) Rx2: 5-HT reuptake inhibitors plus placebo (tid) (n = 40) Adults 18 to 65 years Existence of a Major Depressive Disorder, single or recurrent (DSM-IV) with a current episode resistant to pharmacological treatment 17-item Hamilton depression-rating scale score of 16 All patients except 2 were outpatients. Change in HAM-D score from day 0 to day 10 Rx1: 2.2 ± 4.4 Rx2: 3.7 ± 5.9 Difference not significant, p value not given The study does not support the hypothesis that the addition of pindolol results in a rapid augmentation of the effects of SSRIs in depressed patients resistant to initial treatment with antidepressants. 227 National Adult Depression Clincial Practice Guideline

232 5. Length of Treatment With Antidepressants In Patients With MDD Problem Formulation 5 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: How long should adults with MDD continue taking antidepressant medication? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients with Major Depression who receive treatment with antidepressant medication with: One lifetime episode of MDD Two or more episodes of MDD MDD Continue antidepressant (SSRIs, TCAs, DAs, SNRIs, or NRIs) treatment for appropriate length of time Discontinue antidepressant treatment KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Health Outcomes (associated with the intervention): Change in symptoms Quality of life Missed school/work days Office/UCC/ER visits Hospitalizations Mortality 228 National Adult Depression Clincial Practice Guideline

233 Search Strategy Database: PubMed PubMed National Institute for Clinical Excellence (NICE) Cochrane Clinical Evidence, Issue 14 Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "depressive disorder/drug therapy"[mesh] OR "depressive disorder/prevention and control"[mesh] OR "Antidepressive Agents/ administration and dosage"[mesh] OR "depression recurrent"[all Fields] OR "depression chronic"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ("drug administration schedule"[mesh] OR "Recurrence/prevention and control"[mesh] OR "Treatment duration"[all Fields] OR "treatment discontinuation"[all Fields] OR "treatment continuation"[all Fields]) Depression Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Review, Metaanalyses and RCTs Systematic Reviews Systematic Reviews and RCTs Search Date 8/2005 to 9/2007 8/2005 to 9/2007 December 6, 2004 No. Included / Total Retrieved * 0/129 0/72 N/A Q4, /131 January 2006 N/A * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 229 National Adult Depression Clincial Practice Guideline

234 Database: PubMed Cochrane Clinical Evidence Search Terms: ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "depressive disorder/drug therapy"[mesh] OR "depressive disorder/prevention and control"[mesh] OR "Antidepressive Agents/administration and dosage"[mesh] OR "depression recurrent"[all Fields] OR "depression chronic"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ("drug administration schedule"[mesh] OR "Recurrence/prevention and control"[mesh] OR "Treatment duration"[all Fields] OR "treatment discontinuation"[all Fields] OR "treatment continuation"[all Fields]) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "3000"[PDAT]) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Depression Depression Article Type and Other Limits: Randomized, controlled trials Systematic reviews Systematic reviews & RCTs Search Date 01/01/03-08/01/05 No. Included / Total Retrieved * 3/81 03/05/03 0/295 09/22/03 1/ National Adult Depression Clincial Practice Guideline

235 Database: PubMed Search Terms: ((((("depression"[mesh Terms] OR depression[text Word]) AND ("drug administration schedule"[mesh Terms] OR DRUG ADMINISTRATION SCHEDULE[Text Word])) ("depression/drug therapy"[mesh] OR "depression/therapy"[mesh] OR "depressive disorder/drug therapy"[mesh] OR "depressive disorder/prevention and control"[mesh] OR "Antidepressive Agents/administration and dosage"[mesh] OR "depression recurrent"[all Fields] OR "depression refractory"[all Fields] OR "depression chronic" [All Fields] OR "depression first episode"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ("drug administration schedule"[mesh] OR "treatment protocols"[mesh] OR "Recurrence/prevention and control"[mesh] OR "Treatment duration"[all Fields] OR "treatment discontinuation"[all Fields] OR "treatment continuation"[all Fields]) Article Type and Other Limits: Randomized, controlled trials, English, Randomized Controlled Trial, All Adult: 19+ years, English, Human Meta-Analysis, All Adult: 19+ years, English, Human Review, All Adult: 19+ years, English, Human Search Date / /01/ /01/ /01/ /01/ /01/ /01/2003 No. Included / Total Retrieved * 0/344 5/308 0/4 0/44 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. The Keller (135) study did not show up in our search results due to the way PubMed indexing is done. Other Information Sources: Evidence report on the Treatment of Depression-Newer Pharmacotherapies. AHRQ (2) The British Medical Journal (151) 231 National Adult Depression Clincial Practice Guideline

236 Evidence Tables Table 5.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Keller, et al. (2005) Relapse prevention with gepirone ER in outpatients with Major Depression Inclusion: Adults with primary diagnosis of recurrent Major Depression (per DSM-IV); screening and baseline HAM-D score 20+. Exclusion: HAM-D score that decreased by > 20% between screening and baseline, history of treatmentrefractory depression, pregnancy, substance abuse, pre-trial treatment with MAO-I (within 3 weeks), fluoxetine (within 5 weeks), other psychotropic agents (within 2 weeks) Limited external validity and generalizability due to exclusion of : a) patients with severe comorbidities b) patients with prior treatment resistance or who responded insufficiently to gepirone open label treatment. In acute phase, patients (N = 420) were given openlabel gepirone ER for 8 or 12 weeks, depending on time to remission. Patients who achieved remission (N = 250) were randomized into a doubleblind continuation phase: a) gepirone ER (N = 126) b) placebo (N = 124) 44 weeks Primary outcome measure was number of patients with relapse, defined as HAM-D score 16+ at endpoint (44 weeks). In continuation phase, there was a statistically significant lower percentage of overall relapses in gepirone ER (23%) group than in placebo (34.7%). (p=0.024). Differences were statistically significant from week 24 to week 44. (p<0.05) Gepirone well tolerated - 3.2% in continuation phase dropped out due to adverse events < National Adult Depression Clincial Practice Guideline

237 Table 5.2 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Montogomery, et al. (2004), Venlafaxine versus placebo in the preventive treatment of recurrent Major Depression Inclusion: Adult outpatients with Major Depression (per DSM III-R criteria) who had a history of recurrent Major Depression ( 1 previous episode in the last 5 years with a symptom free period > 6 months between episodes) and symptoms of depression for > 1 month before study entry. Exclusion: Recent MI; history of drug or alcohol dependence, hepatic or renal disease, seizure disorder, psychotic disorder, bipolar disorder, hypersensitivity to venlafaxine; pregnancy; concomitant psychiatric disorders 18+ Disproportionate contribution of number of participants (53 from Europe, remaining from USA). Some investigators might have inadvertently been delivering psychotherapy, in some way. Short taper period for those who switched to placebo. Phase 1 Open label treatment (N = 495): mg/day venlafaxine for 6 months Phase 2 Maintenance (N = 225): Those who responded to treatment in Phase 1 were randomized into one of two groups - a) venlafaxine continuation mg/day (N = 109) or b) placebo (N = 116) - for 12 months. 12 months Primary outcome was number of patients experiencing recurrence of Major Depression in the 12 months after a successful Phase 2. 22% cumulative probability of recurrence in venlafaxinetreated patients after 12 months compared with 55% for placebo group. 21% venlafaxine-treated patients discontinued treatment because of lack of efficacy, compared with 48% placebo-treated patients. < < National Adult Depression Clincial Practice Guideline

238 Table 5.3 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Rapaport, Bose & Zheng (2004), Escitalopram continuation treatment prevents relapse of depressive episodes Inclusion: Adults diagnosed with Major Depressive Disorder who had completed open label phase. Exclusion: Principal Axis I disorder (other than MDD), history of schizophrenia or any other psychotic disorder; suicidality, concomitant psychotropic medication, pregnancy Two different phases, each with separate randomization. Exclusion of patients with comorbid diagnoses (may not be representative of patients in routine practice) Impact of lead-in treatment on subject outcomes. Phase 1 Open label treatment (N = 502): mg/day escitalopram, 20-40mg/day citalopram, or placebo for 8 weeks. Phase 2 Double-blind (N = 274): Those who responded to treatment in Phase 1 were randomized into one of two groups - a) escitalopram (N = 181) or b) placebo (N = 93) for 36 weeks. 36 weeks The primary efficacy parameter was time to relapse from the start of the double-blind treatment. Time to relapse was significantly longer and cumulative rate of relapse was significantly lower in escitalopram group (26% escitalopram vs. 40% placebo) through 36 weeks. Continuing treatment with escitalopram significantly decreased percentage of patients meeting DSM-IV criteria for Major Depressive Disorder: Escitalopram 23% vs. Control 35% Table 5.4 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Weihs, K.L (RCT, double-blind) Follow-up: 44 weeks (11 months) Initial N: 816 Initial N (continuation phase) : 417 Final N: 103 All patients were treated for an initial 8- weeks (acute phase), open- label phase in which they received bupropion SR (150 mg to 300 mg/day). Patients who responded to the treatment and continued to meet the selection criteria, entered a 44-week randomized, double-blind, placebo controlled continuation phase. Rx1: bupropion SR, 300 mg/day (n = 207) Rx2: Placebo (n = 210) Men and women of at least 18 years old with moderate to severe, recurrent Major Depression based on Diagnostic and Statistical Manual-IV (DSM-IV) criteria Minimum score of 18 on the 21- item Hamilton Depression Scale (HAMD) Patient s current episode must have been preceded by at least one other episode within the last 60 months. Time to depression relapse (time from randomization to intervention) Rx1: 44 weeks Rx2: 24 weeks p = (log-rank test) favoring continuation treatment Survival estimates indicated 52% of placebo treated patients and 37% of bupropion patients would have become depressed by the end of the study (p = favoring bupropion). By the end of one year treatment, the odds of placebo-treated patients requiring treatment intervention for a relapse of depression were 1.83 times greater that those of bupropion SRtreated patients. Treatment with bupropion SR for up to 44 weeks decreases the risk of depression relapse. 234 National Adult Depression Clincial Practice Guideline

239 Table 5.5 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Reimherr, F.W (RCT, double-blind) Follow-up: 50 weeks after acute phase Initial N: 839 Initial N : (continuation phase) = 395 Final N: Not given Study phases: Baseline: A one week (5 to 9 day) medication-free baseline phase Open label acute phase: 12 to 14 weeks of fluoxetine, 20 mg/day Continuation phase (Randomized double-blind allocation) Rx1: 14 weeks of continuation therapy with fluoxetine followed by 38 weeks placebo(n = 97) Rx2: 38 weeks of continuation therapy with fluoxetine followed by 12 weeks placebo (n = 100) RX3: 50 weeks of continuation therapy with fluoxetine (n = 102) RX4: 50 weeks of placebo (no continuation phase) (n = 96) Relapse rates were calculated during 12 week periods after each double-blind transfer from fluoxetine to placebo (weeks 12, 26, and 50). Male and female outpatients 18 to 65 years of age who met the DSM- III-R criteria for Major Depression with a duration of at least one month All patients had a modified 17-item Hamilton Depression Rating Scale score of at least 16. Relapse rate% (Kaplan-Meier estimates) After 12 weeks of continuation treatment: Rx1: 26.4% Rx 4: 48.6% p < (favoring continuation treatment) After 26 weeks of continuation treatment: Rx2: 9.0% Rx4: 23.2% p < (favoring continuation treatment) After 50 weeks of continuation treatment: Rx3: 10.7% Rx4: 16.2% p = 0.54 (no statistical difference) The study shows significantly lower relapse rates for fluoxetine-treated patients compared with placebo-treated patients for at least 26 weeks after 12 weeks of successful acute treatment (38 weeks of treatment in total). There was no statistical difference (p = 0.54) in the relapse rate between the fluoxetine group (10.7%) and placebo (16.2%) after 50 weeks of continuation treatment (possibly due to lack of statistical power due to patient attrition, as only 53 patients were included in this analysis). 235 National Adult Depression Clincial Practice Guideline

240 Table 5.6 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Hochstrasser, B (RCT, double- blind) Follow-up: 48 to 77 weeks Initial N: 264 (maintenance phase) Final N: Not given Intention-to-treat analysis Rx1 citalopram, dose range 20 to 60 mg (n = 132) Rx2 placebo (n = 132) Patients were treated with citalopram (20 to 60 mg) for 6-9 weeks, acute phase (n = 427), if responding, continued for 16 weeks (continuation phase) (n = 327) before randomized to doubleblind maintenance treatment with citalopram or placebo for 48 to 77 weeks. Patients 18 to 65 years of age (inand out-patients) with recurrent unipolar Major Depression (DSM- IV), a Montgomery Asberg Depression Rating Scale score of 22 and two or more previous depressive episodes, one within the past five years. Intent- to-treat population consisted of 29% males and 71% females Time from randomization to recurrence of a new depressive episode (using Kaplan-Meier estimates) : Time difference was significant, favoring citalopram group p < Rx1: 0.22 recurrences/person year at risk Rx2: 0.76 recurrences/person year at risk Time to recurrence was longer in patients taking citalopram than in patients taking placebo (p < 0.001) 236 National Adult Depression Clincial Practice Guideline

241 Table 5.7 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Gilaberte, I (RCT, double-blind) Follow-up: 48 weeks of maintenance treatment Initial N: 253 Initial N (maintenance phase): 140 Final N: 78 Intention-to-treat analysis All patients completed 32 weeks of open-label treatment phase consisting of both an acute (8 weeks) and a continuation period (6 months). Maintenance phase Rx1: fluoxetine, 20 mg/day (n = 70) Rx2: placebo (n = 70) Male and female outpatients, 18 to 65 years of age, who met DSM-III- R criteria for unipolar Major Depression and had at least one previous major depressive episode in the last 5 years. Score of at least 18 on the 17-item Hamilton Rating Scale for depression (HAM-D-17) and at least 4 on the Clinical Global Impression (CGI) severity scale. Recurrence Rate (%) Rx1: 20% Rx2: 40% p = Mean HAM-D-17 score at the end of maintenance phase: Rx1: 6.5 ± 8.6 Rx2: 9.9 ± 9.4 p = The symptom-free period was significantly longer for patients treated with Fluoxetine versus placebo (295 days vs. 192 days; Kaplan-Meier estimates, log-rank test, p = 0.002) Long-term maintenance treatment with fluoxetine appeared to be effective in the prevention of recurrent depression. Mean number of previous episodes for fluoxetine group were 2.3 ± 1.2, and for the placebo group 2.6 ± National Adult Depression Clincial Practice Guideline

242 Table 5.8 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Kocsis, J.H (RCT, double-blind) Follow-up: 16 weeks for continuation phase 24 months for maintenance phase Initial N: 129 Initial N (maintenance phase): 53 Final N: Not given in both cases Intention-to-treat analysis All patients treated with desipramine hydrochloride during the acute phase (10 weeks) and continuation phase (16 weeks), dose range 50 to 325 mg/d). Maintenance phase Rx1 desipramine (n = 28) Rx2 Tapered by 25% per week in the first month to placebo (n = 25) Out patients who met DSM-III diagnostic criteria for pure dysthymia (40%), chronic Major Depression (10%) and dysthymia with current Major Depression (50%) 88% white, 6% black, 4% Hispanic, 4% Asians Mean age 36 to 37, 51 to 64% female. Only 41% of patients initially enrolled entered maintenance phase treatment Relapse rates (%), maintenance phase Rx1: 15% (11% in abstract) Rx2: 52% ARR = 37%, NNT = 3 p = 0.01 favoring maintenance treatment (p listed as in abstract) Long-term maintenance treatment with desipramine appeared to be effective in the prevention or postponement of relapse of depression in patients who responded to desipramine during the acute and continuation phases. Most placebo relapses occurred during the first 6 months of maintenance therapy. 40% of patients studied did not have MDD, although results did not differ by subgroup. The three subtypes of chronic depression responded similarly to desipramine treatment in the acute, continuation, and maintenance phases. Small sample size for maintenance phase Table 5.9 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Keller, M.B. 1998(RCT, double-blind) Follow-up: 76 weeks of maintenance phase Initial N: 426 Initial N (maintenance phase): 161 Final N: 59 Intention-to-treat analysis Patients completed 12 weeks of treatment (acute phase) and 4 months of continuation phase treatment with sertraline 50 to 200 mg/d Maintenance phase Rx1 sertraline hydrochloride, daily dose of 50 to 200 mg (n = 77) Rx2 placebo (n = 84) Outpatients meeting a structured clinical interview diagnosis of Chronic Major Depression (of at least 2 years in duration) or Dysthymic Disorder with a concurrent diagnosis of Major Depression (double depression) based on DSM-III-R. Minimum baseline severity of 18 on the 24-item Hamilton Depression Scale (HAM-D) Rx2 n (%) Rx1 n (%) Log-Rank P Experienced 19 (23) 5 (6) recurrence by strict protocol criteria Experienced 42 (50) 20 (26) depression reemergence by Consensus assessment Showed 50 (60) 26 (34) first symptoms of reemergence of depression by consensus assessment. Strict protocol criteria: requiring 3 weeks of treatment and a confirmatory second evaluation one week later by an independent senior investigator Long-term maintenance treatment with sertraline protects against recurrence or reemergence of depression and considerably extends the time in remission for these high-risk patients. 238 National Adult Depression Clincial Practice Guideline

243 6. Follow-Up For Patients In The Acute Phase (First Three Months) of Treatment For MDD Problem Formulation 6 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): What is the recommended follow-up for patients in the first three months of treatment for MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients in acute phase treatment of Major Depression MDD Follow-up* in the first 12 weeks of treatment ( acute phase ) for patients diagnosed with MDD: Follow-up within four weeks and again at eight to twelve weeks Three times in the first 12 weeks (HEDIS measure) Other follow-up intervals No follow-up KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Adherence Office/UCC/ER visits Hospitalizations Mortality * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. 239 National Adult Depression Clincial Practice Guideline

244 Search Strategy Database: PubMed PubMed Cochrane Clinical Evidence, Issue 14 PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) (((((((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "depression"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders (((((((("depression/drug effects" [MESH] OR "depression/drug therapy"[mesh]) OR "depression/ therapy"[mesh]) OR "depression" [All Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) AND English[Lang]) AND "adult"[mesh Terms]) AND "hominidae"[mesh Terms]) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Reviews Systematic Reviews and RCTs Randomized, controlled trials Search Date 8/2005 to 9/2007 8/2005 to 9/2007 No. Included / Total Retrieved * 0/129 0/87 Q4, /131 January /01/03-08/10/05 N/A 1/20 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 240 National Adult Depression Clincial Practice Guideline

245 Database: Cochrane Clinical Evidence PubMed Depression Depression Search Terms: (((("depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) AND ("office visits"[mesh Terms] OR OFFICE VISIT[Text Word])) (((((((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy"[mesh]) OR "depression" [All Fields] OR"Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) Article Type and Other Limits: Systematic reviews Systematic reviews and RCTs All publication types, All adults 19+ years, English, Human All publication types, All adults 19+ years, English, Human No. Included Search / Total Date Retrieved * 03/05/03 0/295 03/05/03 0/ /25/ /01/01-04/01/03 0/58 0/246 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. 241 National Adult Depression Clincial Practice Guideline

246 Evidence Table Table 6.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Swindle, et al. (2003), Integrating clinical nurse specialists into the treatment of primary care patients with depression. Inclusion: diagnosis of Major Depression, dysthymia or partially remitted Major Depression; 2 General Medicine Clinics visits past one year, telephone access, informed consent Exclusion: Incompetent for interview, nursing home residence, active suicidality, seen in VAMC (Veterans Affairs Medical Center) mental health program, substance abuse, history of bipolar disorder, terminal illness. Mean age 56 CNS's often disagreed with the PRIME-MD diagnosis and did not believe treatment was appropriate or necessary for many of the subjects (affecting the fidelity of the intervention). Subjects of study were veterans (96% men) with medically complex issues. Unit of randomization was the firm and not the participant. No true control group (all clinicians received educational program). Post hoc rather than a priori outcome assessment. Lacked power to detect differences between intervention and control for patients with Major Depressive Disorder. Control firm (N = 134): Diagnosis of depression documented in chart. Intervention firm (N = 134): At initial visit, Clinical Nurse Specialist (CNS) would create an initial treatment plan based on medical record and responses to PRIME- MD. Components of the treatment plan could include medication, cognitive behavioral therapy, and or referral to Mental Health. Monitoring through telephone and/or in person contacts was to occur at two weeks, one month, and two months after initial visit. 12 months Primary outcomes assessed at 3 and 12 months were depressive symptoms and patient satisfaction. No significant effects of the CNS intervention on overall depressive symptoms at 3 months. No significant effects of the CNS intervention on overall depressive symptoms at 12 months. Post hoc, within group analyses of BDI scores revealed that intervention group patients with only Major Depression improved significantly at 3 and 12 months. Differences between intervention and control groups in patient satisfaction were small and not significant Data on usual care patient follow-up not provided. 242 National Adult Depression Clincial Practice Guideline

247 7. Follow-Up For Patients In The Continuation Phase (Months Four to 12) of Treatment For MDD Problem Formulation 7 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): What is the recommended follow-up for patients in remission during months four to 12 of treatment for MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients in continuation phase treatment of Major Depression MDD Follow-up* for patients during months four to 12 of treatment ( Continuation Phase ) for Major Depression: Follow-up at five to six months Additional follow-up visits Other follow-up intervals No follow-up KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Adherence Office/UCC/ER visits Hospitalizations Mortality * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. 243 National Adult Depression Clincial Practice Guideline

248 Search Strategy Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * PubMed ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) Only items with abstracts, Humans, English, All Adult: 19+ years 8/2005 to 9/2007 0/129 PubMed (((("depression/drug effects"[mesh] OR "depression/drug therapy" [MESH]) OR "depression/therapy" [MESH]) OR "depression"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years 8/2005 to 10/2007 0/88 PubMed (((("depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) AND ("office visits"[mesh Terms] OR OFFICE VISIT[Text Word])) Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years 8/2005 to 10/2007 0/31 PubMed "follow up" AND ("telephone" OR "office visit" or remission ) AND ("depressive disorder" [MeSH Terms]) Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years 8/2005 to 10/2007 0/29 Cochrane Depression, Anxiety and Neurosis Group Systematic Reviews Q4, /131 Clinical Evidence, Issue 14 Chapter: Depressive Disorders Systematic Reviews and RCTs January 2006 N/A * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 244 National Adult Depression Clincial Practice Guideline

249 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * PubMed (((((((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/ therapy"[mesh]) OR "depression"[all Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) AND English[Lang]) AND "adult"[mesh Terms]) AND "hominidae"[mesh Terms]) AND ("2001/01/01"[PDAT] : "2003/04/01"[PDAT])) AND "2003/07/ "[MHDA] : "2005/08/ "[MHDA] Randomized, controlled trials 01/01/03-08/10/05 0/20 Cochrane Depression Systematic reviews 03/05/03 0/295 Clinical Evidence Depression Systematic reviews and RCTs 03/05/03 1/80 PubMed (((("depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) AND ("office visits"[mesh Terms] OR OFFICE VISIT[Text Word])) All publication types, All adults 19+ years, English, Human /25/01 0/5 (((((((("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depression/therapy" [MESH])OR "depression" [All Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) All publication types, All adults 19+ years, English, Human 01/01/01-04/01/03 0/246 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. 245 National Adult Depression Clincial Practice Guideline

250 Evidence Table Table 7.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Swindle, et al. (2003), Integrating clinical nurse specialists into the treatment of primary care patients with depression. Inclusion: diagnosis of Major Depression, dysthymia or partially remitted Major Depression; 2 General Medicine Clinics visits past one year, telephone access, informed consent Exclusion: Incompetent for interview, nursing home residence, active suicidality, seen in VAMC (Veterans Affairs Medical Center) mental health program, substance abuse, history of bipolar disorder, terminal illness. Mean age 56 CNS's often disagreed with the PRIME-MD diagnosis and did not believe treatment was appropriate or necessary for many of the subjects (affecting the fidelity of the intervention). Subjects of study were veterans (96% men) with medically complex issues. Unit of randomization was the firm and not the participant. No true control group (all clinicians received educational program). Post hoc rather than a priori outcome assessment. Lacked power to detect differences between intervention and control for patients with Major Depressive Disorder. Control firm (N = 134): Diagnosis of depression documented in chart. Intervention firm (N = 134): At initial visit, Clinical Nurse Specialist (CNS) would create an initial treatment plan based on medical record and responses to PRIME- MD. Components of the treatment plan could include medication, cognitive behavioral therapy, and or referral to Mental Health. Monitoring through telephone and/or in person contacts was to occur at two weeks, one month, and two months after initial visit. 12 months Primary outcomes assessed at 3 and 12 months were depressive symptoms and patient satisfaction. No significant effects of the CNS intervention on overall depressive symptoms at 3 months. No significant effects of the CNS intervention on overall depressive symptoms at 12 months. Post hoc, within group analyses of BDI scores revealed that intervention group patients with only Major Depression improved significantly at 3 and 12 months. Differences between intervention and control groups in patient satisfaction were small and not significant Data on usual care patient follow-up not provided. 246 National Adult Depression Clincial Practice Guideline

251 8. Follow-Up For Patients In Maintenance Phase (Beyond 12 Months) Treatment Of MDD Problem Formulation 8 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): What is the recommended follow-up for patients in remission who need ongoing treatment for MDD beyond 12 months? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients in maintenance phase treatment of Major Depression MDD Frequency of follow-up * for patients with MDD who need ongoing treatment beyond 12 months ( maintenance phase ): Annual follow-up Other follow-up intervals No follow-up KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Adherence Office/UCC/ER visits Hospitalizations Mortality * Follow-up contact may include in-person visits, phone calls or between patient and clinician, or phone calls/ between patient and a care manager. The use of between patients and providers is relatively new, and has not been a widely utilized means of communication to date. However, it is being increasingly advocated as part of a patient-centered, more efficient ( less visit dependent ) model of care. At least one member of the GDT uses this modality regularly and deems it effective for follow-up contacts. 247 National Adult Depression Clincial Practice Guideline

252 Search Strategy Database: PubMed PubMed PubMed PubMed Cochrane Clinical Evidence, Issue 14 Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) (((("depression/drug effects" [MESH] OR "depression/drug therapy"[mesh]) OR "depression/ therapy"[mesh]) OR "depression" [All Fields] OR "Major Depressive Disorder"[All Fields]) AND ((("office visits"[mesh] OR "office visits"[text word]) OR "telephone"[text word]) OR "visits"[text word])) (((("depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) AND ("office visits"[mesh Terms] OR OFFICE VISIT[Text Word])) "follow up" AND ("telephone" OR "office visit" or remission ) AND ("depressive disorder" [MeSH Terms]) Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Reviews Systematic Reviews and RCTs Search Date 8/2005 to 9/2007 8/2005 to 10/2007 8/2005 to 10/2007 8/2005 to 10/2007 No. Included / Total Retrieved* 0/129 0/88 0/31 0/29 Q4, /131 January 2006 N/A * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 248 National Adult Depression Clincial Practice Guideline

253 Evidence Table Table 8.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Swindle, et al. (2003), Integrating clinical nurse specialists into the treatment of primary care patients with depression. Inclusion: diagnosis of Major Depression, dysthymia or partially remitted Major Depression; 2 General Medicine Clinics visits past one year, telephone access, informed consent Exclusion: Incompetent for interview, nursing home residence, active suicidality, seen in VAMC (Veterans Affairs Medical Center) mental health program, substance abuse, history of bipolar disorder, terminal illness. Mean age 56 CNS's often disagreed with the PRIME-MD diagnosis and did not believe treatment was appropriate or necessary for many of the subjects (affecting the fidelity of the intervention). Subjects of study were veterans (96% men) with medically complex issues. Unit of randomization was the firm and not the participant. No true control group (all clinicians received educational program). Post hoc rather than a priori outcome assessment. Lacked power to detect differences between intervention and control for patients with Major Depressive Disorder. Control firm (N = 134): Diagnosis of depression documented in chart. Intervention firm (N = 134): At initial visit, Clinical Nurse Specialist (CNS) would create an initial treatment plan based on medical record and responses to PRIME- MD. Components of the treatment plan could include medication, cognitive behavioral therapy, and or referral to Mental Health. Monitoring through telephone and/or in person contacts was to occur at two weeks, one month, and two months after initial visit. 12 months Primary outcomes assessed at 3 and 12 months were depressive symptoms and patient satisfaction. No significant effects of the CNS intervention on overall depressive symptoms at 3 months. No significant effects of the CNS intervention on overall depressive symptoms at 12 months. Post hoc, within group analyses of BDI scores revealed that intervention group patients with only Major Depression improved significantly at 3 and 12 months. Differences between intervention and control groups in patient satisfaction were small and not significant Data on usual care patient follow-up not provided. 249 National Adult Depression Clincial Practice Guideline

254 9. Discontinuation of Antidepressants In Patients With MDD Problem Formulation 9 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention How should antidepressants be discontinued in patients with MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and older) patients with Major Depression MDD Gradual tapering of antidepressants Stopping antidepressants without tapering KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Medication withdrawal symptoms Office/UCC/ER visits Hospitalizations Mortality 250 National Adult Depression Clincial Practice Guideline

255 Search Strategy Database: PubMed PubMed PubMed Cochrane Clinical Evidence, Issue 14 PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) (("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depressive disorder/drug therapy"[mesh]) OR "Major Depressive Disorder"[All Fields] AND ("discontinuation" OR "withdrawal" OR "tapering") ("antidepressive agents"[mesh]) AND ("discontinuation" OR "withdrawal" OR "tapering") Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders (("depression/drug effects"[mesh] OR "depression/drug therapy"[mesh]) OR "depressive disorder/ drug therapy"[mesh]) OR "Major Depressive Disorder"[All Fields] AND (("antidepressive agents/adverse effects"[mesh] OR "serotonin uptake inhibitors/adverse effects"[mesh]) OR "substance withdrawal syndrome"[mesh]) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "3000"[PDAT]) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Randomized Controlled Trial, All Adult: 19+ years, English, Human Randomized, controlled trial, All Adult: 19+ years English, Human Systematic Reviews Systematic Reviews and RCTs Randomized, controlled trials Search Date 8/2005 to 9/ /2005 to 10/ /2005 to 10/2007 No. Included / Total Retrieved * 0/129 0/37 0/40 Q4, /131 January /01/03-08/01/05 N/A 1/77 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 251 National Adult Depression Clincial Practice Guideline

256 Database: Cochrane Clinical Evidence PubMed Depression Depression Search Terms: ((((((("depression"[mesh Terms] OR depression[text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) AND ("drug administration schedule"[mesh Terms] OR DRUG ADMINISTRATION SCHEDULE [Text Word])) AND Randomized Controlled Trial[ptyp]) AND English[Lang]) AND ("1980"[PDat] : "3000"[PDat])) (("depression/drug effects"[mesh] OR "depression/drug therapy" [MeSH]) OR "depressive disorder/ drug therapy"[mesh]) OR "Major Depressive Disorder"[All Fields]) AND (("antidepressive agents/adverse effects"[mesh] OR "serotonin uptake inhibitors/adverse effects"[mesh]) OR "substance withdrawal syndrome"[mesh])) Article Type and Other Limits: Systematic reviews Systematic reviews and RCTs Clinical trials, All adults 19+ years, English All publication types, All adults 19+ years, English, Human No. Included Search / Total Date Retrieved * 03/05/03 0/295 09/22/03 0/ /01/01 01/01/01-04/01/03 2/347 1/286 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. 252 National Adult Depression Clincial Practice Guideline

257 Evidence Tables Table 9.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Sunder, et al. (2004), Postpartum Depression Recurrence Versus Discontinuation Syndrome: Observations from a Randomized Controlled Trial Twenty two pregnant women (mean age 32) who had recovered from at least one prior episode of Postpartum Major Depressive Disorder (PMDD) Mean age 32 Postpartum women only. Only 11 out of 22 patients initially treated for postpartum MDD entered the tapering phase: Sertraline (N = 14) or placebo (N = 8) administered for 17 weeks post birth: 25mg/day for first four days, 50mg/day through first four weeks, 75mg/day through week 17. Intervention tapered weeks 18-20: 50 mg/day in week 18, 25 mg/day week 19, 25 mg/day week 20. Discontinued after week 20. One year follow-up phase began after RCT; assessments for depression every two months. 20 weeks Depression symptoms assessed every two months in follow-up phase. No significant difference between sertraline and placebo treated groups on sum of ASE-derived symptoms in taper phase National Adult Depression Clincial Practice Guideline

258 Table 9.2 Judge, R (RCT, double-blind) Follow-up: Not given Initial N: 150 Final N: 141 Study, Total n Treatment Groups Size & Drug Study Population Results Comments At visit one all patients received blinded drug equivalent to their current daily maintenance dose (fluoxetine mg or paroxetine mg) Rx1: fluoxetine (n = 75) Rx2: paroxetine (n = 75) These patients were then randomized to: active therapy at visit 2, placebo at visit 3, original active therapy at visit 4 placebo at visit 2, active therapy at visit 3 and the rest of the study Periods of treatment interruption lasted between 3-5 days. Male or female outpatient aged 18 Diagnosis of unipolar depressive disorders for which the current effective maintenance therapy with fluoxetine or paroxetine was prescribed Montgomery-Asberg Depression Rating Scale (MADRS) score of 12 or less Current continuous maintenance treatment for depression (fluoxetine 20 to 60 mg/day or paroxetine 20 to 50mg/day) lasting more than 4 months and less than 24 months All white, 77% female Mean change following treatment interruption: Mean change ± SD for Discontinuation Emergent Signs and Symptoms (DESS), Rx1: -0.2 ± 2.5 Rx2: 2.2 ± 5.5 p = Mean change ± SD for Montgomery-Asberg Depression Rating Scale (MADRS) Rx1: 0.2 ± 2.9 Rx2: 1.8 ± 5.0 p = Mean change ± SD for Clinical Global Impressions-Severity (CGI-severity scores) Rx1: 0.0 ± 0.5 Rx2: 0.3 ± 0.8 p = Comparison between treatment groups following treatment interruption indicates paroxetine-treated patients had significantly greater increases in MADRS and CGI- severity than did fluoxetinetreated patients. The deterioration from baseline within the paroxetine treatment group was also significant for both MADRS (p = 0.010) and CGI- Severity scores (p = 0.001). Eighteen interruption-emergent symptoms were reported by significantly more paroxetine-treated patients than fluoxetine-treated patients Did not examine for treatment withdrawal side effects beyond 5 days of discontinuation, when side effects of the longer half life medication fluoxetine might be expected to emerge. No evaluation of side effects between low and higher doses of fluoxetine. Industry funded (Eli Lilly, maker of fluoxetine). 254 National Adult Depression Clincial Practice Guideline

259 10. Treatment Preferences For MDD In Different Ethnic Groups Problem Formulation 10 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): Side Effects Associated With the Intervention: Do adult patients from different ethnic groups have a preference among treatment options for MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. Adults (age 19 and older) patients from diverse ethnic backgrounds who have Major Depression MDD Antidepressant treatment (SSRIs, TCAs, DAs, SNRIs, NRIs) Psychotherapy (Interpersonal Therapy, Cognitive Behavioral Therapy, Problem-Solving Therapy) Combination of antidepressants & psychotherapy Cultural healers Traditional/herbal remedies Clergy No treatment KP primary care physicians, physician assistants, nurse practitioners, pharmacists, and health educators Outpatient office visit, emergency department, urgent care clinics Change in symptoms Quality of life Missed school/work days Sexual problems Drowsiness Headache Nausea Insomnia Agitation/ nervousness Dry mouth Seizures Noncompliance Office/UCC/ER visits Hospitalizations Mortality Patient satisfaction Patient dissatisfaction Elevated blood pressure Constipation Diarrhea Abdominal pain Dizziness Blurred vision Weight gain GI bleeding Falls 255 National Adult Depression Clincial Practice Guideline

260 Search Strategy Database: PubMed PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) ((("depression/drug therapy"[mesh Terms] OR "depressive disorder/ drug therapy"[mesh Terms])OR "Major Depressive Disorder"[All Fields]) AND ((((((("cultural diversity"[mesh Terms] OR "cultural diversity"[text Word]) OR "culture"[mesh Terms] OR "ethnic groups"[mesh Terms]) OR "ethnic groups"[text Word]) OR "Asians"[text word]) OR "Blacks"[Text Word]) OR "Hispanics"[Text word]) OR "social class"[mesh Terms] OR "health knowledge, attitudes, practice"[mesh Terms])) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Randomized Controlled Trials, Humans, English, All Adult: 19+ years Cochrane Depression, Anxiety and Neurosis Group Systematic Reviews Clinical Chapter: Depressive Disorders Systematic Evidence, Reviews and Issue 14 RCTs PubMed ((("depression/drug therapy"[mesh Terms] OR "depressive disorder/ drug therapy"[mesh Terms])OR "Major Depressive Disorder"[All Fields]) AND ((((((("cultural diversity"[mesh Terms] OR "cultural diversity"[text Word]) OR "culture"[mesh Terms] OR "ethnic groups"[mesh Terms]) OR "ethnic groups"[text Word]) OR "Asians"[text word]) OR "Blacks"[Text Word]) OR "Hispanics"[Text word]) OR "social class"[mesh Terms] OR "health knowledge, attitudes, practice"[mesh Terms])) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "3000"[PDAT]) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Search Date 8/2005 to 9/2007 8/2005 to 9/2007 No. Included / Total Retrieved * 0/129 0/129 Q4, /131 January 2006 All 01/01/03-08/01/05 N/A 1/74 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 256 National Adult Depression Clincial Practice Guideline

261 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * Cochrane Depression Systematic reviews 03/05/03 0/295 Clinical Evidence Depression Systematic reviews and RCTs 03/05/03 0/80 PubMed (("depression"[mesh Terms] OR DEPRESSION[Text Word]) AND ((("therapy"[subheading]or"therapeutics" [MeSH Terms]) OR TREATMENT[Text Word]) AND PREFERENCE[All Fields])) Clinical trials, All adults 19+ years, English, Human /2001 2/170 PsychInfo Major Depression/Preferences/Blacks/Asians/Hisp anics/american Indians Systematic reviews and RCTs /1/03 0/43 PubMed [All Fields]) AND ((((((("cultural diversity" [MeSH Terms] OR "cultural diversity"[text Word]) OR "culture" [MeSH Terms]OR "ethnic groups"[mesh Terms]) OR "ethnic groups"[text Word]) OR "Asians"[text word]) OR "Blacks"[Text Word]) OR "Hispanics"[Text word])or "social class" [MeSH Terms] OR "knowledge, attitudes, practice"[mesh Terms])) All Adult: 19+ years English, Human 01/01/01-04/01/03 0/9 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. The Cooper (168) study did not show up in our search results due to the way PubMed indexing is done. Other Information Source: Evidence report on the Treatment of Depression-Newer Pharmacotherapies. AHRQ. (2) 257 National Adult Depression Clincial Practice Guideline

262 Evidence Table Table 10.1 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT p value Dwight-Johnson, et al. (2004), Using conjoint analysis to assess depression treatment preferences among lowincome Latinos Inclusion: English and Spanish speaking patients with depression or dysthymia age 18 and older seen as outpatients in general medicine and women's clinics Exclusion: Suicidality, history of bipolar or psychotic disorder, evidence of gross cognitive impairment. 18+ Small sample. Low response rate. Most participants were women (95%), and most were Latino from Mexico and El Salvador (93%). Most completed the survey in Spanish (86%) No comparison group. N/A N/A Conjoint analysis survey administered to 42 participants to assess: 1) Treatment choice (counseling, antidepressants, or combination) 2) Treatment format (individual or group) 3) Treatment location (primary care or mental health clinic) 4) Barrier reduction strategies including system navigation (help making appointments vs. no help), reducing transportation barriers (bus pass, telephone appointments, or no assistance); and educational interventions (individual education session, group education session, pamphlets, or videos) Study patients expressed: Preference for individual over group treatment. (p<0.001) Preference for combination therapy over counseling (p=0.009) Preference for combination therapy over medication alone. No comparison for preference of psychotherapy alone vs. medication alone. No significant preference for treatment setting (p=0.833) Helpful barrier reduction strategies included: written educational information < < < telephone appointment = bus pass < help making appointments < National Adult Depression Clincial Practice Guideline

263 11. Patient Self-Management Strategies For Improving Symptoms of MDD Problem Formulation 11 Clinical Question: Population: Health Intervention: Most Important Health Outcomes: What patient self-management strategies are effective in improving the symptoms of MDD? All adult (age 19 and over) patients in maintenance phase treatment of Major Depression Exercise/physical activity Community resources Internet resources Support groups/befriending Light therapy Life review therapy Change in symptoms Quality of life Missed school/work days Adherence to treatment plan Bibliotherapy Music therapy Computer- or internet-based self-study materials No patient self-directed intervention Office/UCC/ER visits Hospitalizations Mortality 259 National Adult Depression Clincial Practice Guideline

264 Search Strategy Database: Cochrane Systematic Reviews Clinical Evidence PubMed PubMed Cochrane Search Terms: Depression Depression ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR meta-analysis"[all Fields])) ("depression" [MeSH] OR "depressive disorder" [MeSH] OR "Major Depressive Disorder" [all fields]) AND ("self care" [MeSH] OR "self efficacy" [MeSH] OR "patient participation" [MeSH] OR "choice behavior" [MeSH] OR "bibliotherapy" [MeSH] OR "befriending" [all fields] OR "behavioral activation" [all fields] OR "exercise" [Mesh] OR "community resources" [all fields] OR "self-help groups" [MeSH] OR "Internet"[MeSH]) Depression, Anxiety and Neurosis Group Article Type and Other Limits: N/A Systematic Reviews and RCTs; Adults Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Reviews Search Date 2005 to /2007 to 9/14/2009 1/2006-9/2009 8/2005 to 9/2007 9/2007 to 9/14/2009 8/2005 to 9/2007 9/2007 to 9/2009 No. Included / Total Retrieved * 0/134 0/70 0/2 0/129 2/135 1/75 1/84 Q4, /131 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 260 National Adult Depression Clincial Practice Guideline

265 Database: Clinical Evidence PubMed Cochrane Clinical Evidence PubMed Search Terms: Chapter: Depressive Disorders ("depression"[mesh] OR "depressive disorder"[mesh] OR "Major Depressive Disorder"[all fields]) AND ("self care"[mesh] OR "self efficacy"[mesh] OR "patient participation"[mesh] OR "choice behavior"[mesh] OR "bibliotherapy"[mesh] OR "befriending"[all fields] OR "exercise"[mesh] OR "community resources"[all fields] OR "self-help groups"[mesh] OR "Internet"[MeSH]) AND Randomized Controlled Trial[ptyp] AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Depression Depression ("depression" [MeSH] OR "depressive disorder" [MeSH] OR "Major Depressive Disorder" [all fields]) AND ("self care" [MeSH] OR "self efficacy" [MeSH] OR "patient participation" [MeSH] OR "choice behavior" [MeSH] OR "bibliotherapy" [MeSH] OR "befriending" [all fields] OR "behavioral activation" [all fields] OR "exercise" [Mesh] OR "community resources" [all fields] OR "self-help groups" [MeSH] OR "Internet"[MeSH]) Article Type and Other Limits: Systematic Reviews and RCTs All Adult: 19+ years, English, Randomized Controlled Trial, Humans Systematic reviews Systematic reviews and RCTs All publication types, All adults 19+ years, English, Human Search Date January /01/03-08/01/05 No. Included / Total Retrieved * N/A 4/62 03/05/03 0/295 03/05/03 1/80 01/01/ /01/2003 9/ National Adult Depression Clincial Practice Guideline

266 Evidence Tables Table 11.1: Included Studies of Patient-Self-Management Strategies for Improving Depression Symptoms Study Salkovskis et al (2006) A randomized controlled trial of the use of self-help materials in addition to standard general practice treatment of depression compared to standard treatment alone. Craft et al (2007) Intervention study of exercise for depressive symptoms in women. N initial: 32 N Final: not reported Inclusion & Exclusion Criteria Inclusion: depressive disorder; recently prescribed antidepressant medication; aged years. Exclusion: difficulty reading English; severe (lifethreatening) medical illness, history of psychosis or bipolar disorder, current alcohol or substance dependency, and taking antidepressants for > 4 weeks. 32 low-income, minority women volunteers from Boston s center for Excellence in Women s Health. Inclusion: English speaking; aged between 18-55; DSM- IV criteria for MDD, dysthymia, or depressive disorder not otherwise specified as diagnosed and scoring >9 on BDI-II; sedentary Age Mean age: 40.2 ± ±13.3 % Female: 83.0% for self-help vs. 78.3% for treatment as usual Mean age: Female: 100% Race: 68.8% black, 18.7% white, and 12.5% Latino. Limitations / Biases Small sample size Majority were women Possible selection bias No control group Short duration and follow-up Small sample size Low power No blinding Some women were also under other treatment for depression. Possible selection bias. Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value SELF-HELP MATERIALS Self-help materials (SH, N = 50) Treatment as usual (TAU, N = 46) EXERCISE Participants stratified by current depression treatment (no txt, medication, therapy, medication and therapy) and randomized into one of two groups: Clinic-based exercise (N = 16): provided feedback and information on exercise; 4-week clinicbased training for 2 times/wk + once/wk at 26 weeks Beck Depression Inventory (BDI) score 3 months Outcome measures were changes in BDI-II (Beck Depression Inventory), BMI, % fat, and cardiovascular fitness. Compliance: 76% Follow-up rate: 85% Baseline values: SH: 27.1±10.5 TAU: 27.5±9.8 Follow-up values: SH: 14.3±12.5 TAU: 12.6±9.6 Absolute effect: % of all participants experienced a > 50% reduction in depressive symptoms. Overall effect size for exercise on depressive symptoms were large in both groups between preand post-intervention (d= vs for homeand clinic-based, respectively). At 3-months follow-up, there were no significant 262 National Adult Depression Clincial Practice Guideline NA NA P >0.5 NS

267 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value lifestyle; and average or below average fitness level by exercise testing. Exclusion: CHD; diabetes; mental health diagnosis; and contraindication to exercise. Low generalizability. home; followed by 8- wks transitioned to home-based phase. Home-based exercise (N = 16): feedback and info on exercise and one instructional session. difference between the two groups on BDI-II depression score, selfreported physical activity, BMI, % body fat, and cardiovascular fitness. At 3-months, clinic-based group walked further each day than home-based according to pedometer readings. <0.05 A home-based program can increased physical activity and reduce depressive symptoms similar to clinic-based program. Dunn (2005), Exercise treatment for depression: efficacy and dose response Inclusion: Sedentary men and women with mild (HAM-D 12-16) to moderate (HAM-D 17-25) MDD. (75% women, 75% white); aged Exclusion: Not receiving any other treatment for depression; 160% over ideal weight, 21 drinks per week, suicidal risk, suicide attempt last two years, hospitalization or psychiatric disorder last five years, substance abuse, Mean age: Participants not blinded to treatment assignment. Participants required to exercise under supervision; might have compromised external validity. Sample size small compared with pharmacologic treatment studies. Participants randomized into one of five groups: LD/3: Low dose exercise (7 kcal/kg/week) three times per week. (N = 16) LD/5: Low dose exercise (7 kcal/kg/week) five times per week. (N = 18) PHD/3: Public health dose (17 kcal/kg/week) three times per week. (N = 17) 12 weeks HAM-D score after 12 weeks. Adjusted mean HAM-D scores at 12 weeks were reduced 47% from baseline for PHD, a significant difference compared with 30% for LD 29% for control The LD condition was not significantly different from the control condition. The 3 day/week condition was not significantly different from the 5 day/week condition. NNT* for reduction of HAM-D at 12 wks compared to participant s baseline: PHD vs. LD: 5.9 PHD vs. Cntrl: 5.6 LD vs. Cntrl: National Adult Depression Clincial Practice Guideline

268 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value PHD/5: Public health dose (17 kcal/kg/week) five times per week. (N = 16) inability to exercise, planned or current pregnancy. Placebo control: 3 days/week of stretching flexibility exercise for minutes per session. (N = 13) Rx1: Exercise group (n = 43) Mather, A.S (RCT) Effects of exercise on depressive symptoms in older adults with poorly responsive depressive disorder: randomized controlled trial. Initial N: 86 Final N: 86 Intention-to-treat analysis Singh, N.A (RCT) The efficacy of exercise as a longterm antidepressant in elderly subjects: a randomized, Out-patients 53 years and older with a diagnosis of depression Score of at least 10 on the Geriatric Depression Scale (GDS) Have been taking a therapeutic dose of antidepressant therapy for at least 6 weeks No ongoing structured psychotherapy No regular exercise more than twice weekly 16% male, 84% female Age 60 and older with a BDI score of >12 Diagnosis of either unipolar major or minor depression or dysthymia according to DSM-IV diagnostic criteria Excluded if suicidal, currently seeing a psychiatrist or had been Preponderance of women in the exercise group Actual HRSD pre and post scores were not given Short term Definition of threshold for response (30% reduction) is lower than usual 50% seen in most studies Small sample size Population elderly Final difference between BDI scores between groups may not be clinically significant. Rx2: Nonexercise social control (health education talks) (n = 43) All exercise session lasted for 45 minutes and comprised predominantly weightbearing exercise; 2 sessions per week All patients continued to take antidepressant therapy throughout the trial Phase I (weeks 1-10) Rx1: Supervised exercise in laboratory Rx2: Control group (health education lectures) Phase II (weeks 11-20) Follow-up: 10 weeks Follow-up: 20 weeks, with followup at 26 months Change in 17- item Hamilton Rating Scale for Depression (HRSD) from baseline at 10 weeks (%) BDI score Change in 17-item Hamilton Rating Scale for Depression (HRSD) from baseline at 10 weeks (%) Rx1: 55% Rx2: 33% Response was defined as a 30% reduction in HRSD score from baseline This RCT found that 10 weeks of twice-weekly exercise was associated with a modest reduction in depression symptoms in a group of older people with depression. BDI score at 20 weeks Rx1: Decreased from 21± 2.0 to 9.2 ± 2.8 Rx2 : Decreased from ± 1.8 to 11.0 ± 2.36 BDI score after 26 months Rx1: Decreased from NNT* for >30% reduction in HRSD score from baseline: Rx1 vs. Rx2: 4.5 NNT* for participants classified as nondepressed (BDI<9) at 20 wks: Exercise vs. Control: 2.7 p = 0.05 (favoring exercise group) p= (favoring exercise) p= National Adult Depression Clincial Practice Guideline

269 Study Inclusion & Exclusion Criteria Age on antidepressant drugs within the last three months Subjects participating in aerobic exercise more than twice a week in the month prior to enrollment were also excluded Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value controlled trial. Initial N: 32 Final N: 29 37% male, 63% female Mean age: No long-term difference between groups if exercise not continued Response occurred over time even in control group Rx1: Unsupervised exercise at home, laboratory or health club setting (N = 15), 18±2 weight lifting exercise sessions Rx2: Control group (N = 14) Phase III (Months 6-26) Rx1: No study requirements Rx2: No study requirements 21± 2.0 to 13 ± 2.2 Rx2: Decreased from 18.4 ± 1.7 to 14.4 ± 2.2 The relative improvements in depression scores in the exercisers were 1.5 to 2.5 times greater than those in the controls at the end of Phase II. The exercise group showed significantly reduced depression compared with control group at both 20 weeks and 26-month follow-up. BDI score (total) Rx1: decreased from 16.6 to 11.1 Rx2: decreased from 16.0 to 11.8 Rx3 : decreased from 15.2 to 14.7 (favoring exercise group) McNeil, J.K (RCT) The effect of exercise on depressive symptoms in the moderately depressed elderly. Initial N: 30 Final N: 30 Elderly outpatients with a mean age of 72.5 years BDI score of Elderly patients Short-term results Small n Rx1: Exercise, walking between minutes, three times a week Rx2: Social Contact, two home visits (20-40 minutes) each week by an undergraduate psychology student Rx3: Wait-list (control group) Follow-up: 6 weeks Total and psychosocial subscale of the Beck Depression Inventory. BDI score (somatic symptoms) Rx1: decreased from 7.7 to 5.3 Rx2: decreased from 6.9 to 6.0 Rx3 : decreased from 6.8 to 6.4 NA p < 0.05 (favoring exercise or social contact) p < 0.5 (favoring exercise) Exercise and social contact both resulted in significant reductions in both total and the psychosocial subscale of the Beck Depression Inventory. The exercise 265 National Adult Depression Clincial Practice Guideline

270 Study North, T.C (Metaanalysis) Effect of exercise on depression. Inclusion & Exclusion Criteria Age Depressed patients 11 to 55 years with a mean of 31.8 ± studies were all males, 16 studies were all females. # studies included in effect of exercise on depression = 80 # studies included in effect of age = 63 # of studies included for the effect of gender = 36 Limitations / Biases There are a lot of differences in the study designs of the studies used in this metaanalysis. That includes differences in the mode of exercise, length of the exercise programs and types of control groups). Meta-analysis excludes older patients Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value Rx1: exercise (aerobic endurance and muscular strength) varying in number of sessions Rx2: comparison groups (control, leisure activity, and psychotherapy) Follow-up: Varied from 4 weeks to 24 weeks with majorities of studies being between 5-12 weeks Exercise scores condition, however, resulted in decreased somatic symptoms of the BDI. The overall Effect Size of the studies included was 0.53 ± That indicates that patients in exercise groups (Rx1) decreased their scores an average of 0.53 of a standard deviation unit more than subjects in comparison groups (Rx2). A significant, negative correlational relationship was found between mean age and ES. The negative relationship suggests that the older the subjects, the greater the decrease in depression with exercise. The ESs for males and females were not significantly different, suggesting that exercise had an equally beneficial effect in improving symptoms of depression in both genders. The overall results indicated that exercise has a beneficial effect on improving symptoms of depression. NA p < (p < 0.05) (p > 0.05) 266 National Adult Depression Clincial Practice Guideline

271 Study Meyer et al. (2009) Follow-up: 6 months N initial: 396 N at 9 weeks: 216 N at 18 weeks: 146 N at 6 months: 99 Inclusion & Exclusion Criteria Participants were recruited via internet depression forum in Germany. Inclusion/ exclusion: provided consent, above 18 years, and completed at least half of baseline depression questionnaire. Female: 76% Race: Not reported Many of the participants were incapacitated in terms of symptom severity and social dysfunction. Over half of the sample was currently unemployed, more than half reported currently being in treatment (medication and/or psychotherapy), and 85% stated they had been feeling depressed for several months (29%) or even several years (56%). Age Mean age: Limitations / Biases No placebo-control group Short duration and follow-up Participants were not clinically diagnosed with MDD High attrition rate (at least 55% at 9 weeks) Some participants were also under other treatment for depression. Possible selection bias participants were more experienced/comfortabl e with internet/computer; heterogeneous sample of users. Low generalizability. Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value COMPUTER / INTERNET Internet program as add-on to usual care (N = 320): Consisted of 10 content modules representing different psychotherapeutic approaches. Wait-list control plus usual care (N = 76): No internet access until 9 weeks after the treatment group has completed. Duration: 6 months Outcome measures were changes in the adapted German BDI (Beck Depression Inventory), and work and social adjustment scale (WSA). Change in depression scale: Using last-observationcarried forward ITT analysis, those in the internet program demonstrated a reduction of 3.11 BDI points (d=0.29) vs. no change observed in participants assigned to wait-list controls (BDI change: -0.04). between-groups effect size at 9 weeks, using this ITT sample, was Cohen s d =.30. In completers, effect size was reported at Cohen s d =.64 (CI 95% = ). Change in work and social scale: Similarly, ITT analysis found small decrease in internet group (d=0.17), versus no significant change. Between-group effect size at T1 was Cohen s d =.36. Follow-up: Improvements in depression symptoms were maintained at 6- months follow-up for completers (d=0.74). NNT* clinical significant improvement/recovered (change of at least 8.46 points on BDI, with posttest score of <14.29): Internet vs. wait-list controls: 4.3, p<0.001 NS (0.96) p< % vs. 1.9% of those in the internet group and 267 National Adult Depression Clincial Practice Guideline

272 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value the wait-list control reached the criteria of clinical significant improvement/recovered (change of at least 8.46 points on BDI, with posttest score of <14.29). Users who engaged more often and intensively with the program were more likely to complete the follow-up assessment and to benefit from the program. Control and intervention groups had statistically equivalent baseline depression (CES-D) and SF-12 scores. Clarke, et al. (2005) Overcoming Depression on the Internet (ODIN) (2): A Randomized Trial of a Self-Help Depression Skills Program With Reminders Inclusion: Kaiser Permanente Northwest HMO adults with either 1) depression received depression medication or psychotherapy, and had a chart diagnosis, or 2) no previous depression diagnosis, but age and gender matched to the first group Exclusion: None stated. 18+ implied, not stated Subjects lost to followup were older, more depressed, and less likely to be in the control group. Low enrollment (255 patients out of over 12,000 initially eligible) and completion (57% in intervention group vs. 80% in usual care) rates. Only one self-reported measure of depression. Brief follow-up period 16 weeks. Rx1 Treatment as usual without access to the internet site.(n = 100) Rx2 Access to depression internet site ( ) with postcard reminders (N = 75) Rx3 Access to depression internet site with brief telephone reminders delivered by nonclinician study staff (N = 80) All participants were permitted to obtain any physical or mental health services during the study. 16 weeks Primary outcome measure was score on Center for Epidemiological Studies (CES- D) self-report questionnaire at 5, 10, and 16 weeks after enrollment In the intention-to-treat analysis, intervention participants (Rx2 and Rx3) reported greater reductions in CES-D depression scores compared with the control group (Rx1) Baseline 16 weeks Rx Rx Rx Intervention had a greater effect on those who were more severely depressed at baseline. NNT* for patients who moved over time from the disordered to the non disordered CES- D ranges: Rx3 vs. Rx1 = Baseline 16 weeks 268 National Adult Depression Clincial Practice Guideline

273 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value Rx Rx Rx No significant difference between treatment groups with regard to outcome measures. In Intent to treat analysis, CES score improvement (in points) was: Blue Pages 3.9 MoodGYM 4.2 Control 1.0 Christensen, Griffiths, & Jorm (2004) Delivering interventions for depression by using the internet: a randomized controlled trial Men and women from the community of Canberra, Australia. Participants had scored 22 or higher on the Kessler scale and were not being clinically treated by a psychologist or psychiatrist Short follow-up time (six weeks) does not offer information about the sustainability of internet interventions. High drop-out rate for MoodGYM. No information on whether or not patients were concurrently taking antidepressants. 525 participants were randomized into one of three groups: Internet "Blue Pages" (N = 165): provided depression literacy, offering evidencebased information on depression and its treatment. Internet "Mood GYM" (N = 182): offered cognitive behavioral therapy for the prevention of depression. Control "Attention placebo" (N = 178): weekly contact with a lay interviewer to discuss lifestyle factors such as exercise, education, and health habits. 6 weeks Primary outcome was score on Center for Epidemiological Studies (CES) scale. In Completer analysis, CES score improvement (in points) was: Blue Pages 4.9 Mood Gym 5.8 Control 1.2 Therefore, in both analyses: MoodGYM was significantly more effective than control. (p<0.05) Blue Pages was significantly more effective than control. (p<0.05) MoodGYM was slightly (but not significantly) more effective than Blue Pages. NA < 0.05 < 0.05 > National Adult Depression Clincial Practice Guideline

274 Study Clarke, G (RCT) Overcoming depression on the Internet (ODIN): a randomized controlled trial of an Internet depression skills intervention program. Initial N: 299 Final N: 177 Intention-to treat analysis Inclusion & Exclusion Criteria Age Combination of depressed and nondepressed cases Rx1: Mean age of 43.3, with an SD of 12.2, 73.6% females Rx2: Mean age of 44.4 with an SD of 12.4, 77.4% females Limitations / Biases High attrition rate (41%) Heterogeneous sample. Over three-quarter of sample was female. Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value Rx1: Access to Depression Internet site (n = 144), depression cases (n = 107) Rx2: Usual Care only (n = 155), depression cases (n = 116) Much of the internet site content was adapted from group CBT psychotherapy manuals Follow-up: 32 weeks Change in CES-D score Change in CES-D score (baseline to week 32 Depressed cases only: mean(sd) Rx1: 30.7(12.9) to 22.2(12.8) Rx2: 31.3(11.5) to 25.2(14.2) Stratified analyses by high vs. low baseline CES-D scores, gender, and age greater or less than 45 all showed no statistically significant between group differences. NA p = 0.12 The study found no differences between the control and experimental conditions on self-reported depression (CES-D) over the study period, indicating a lack of treatment effect. LIGHT THERAPY Martiny, et al. (2005) Adjunctive bright light in nonseasonal Major Depression: results from clinician-rated depression scales Inclusion: Adults, 18 and older, fulfilling diagnostic criteria for Major Depression according to the DSM-IV; HAMD score 13. Exclusion: Seasonal Depression (SAD); fulfillment of seasonal pattern specifier; psychotic disorder; organic brain 18+ No follow-up to determine lasting effects. Unequal duration of treatments (30 vs. 60 minutes). Percentage of patients in ongoing treatment with antidepressants at inclusion or who had stopped treatment within the last month before All patients received a fixed dose of 50mg sertraline daily. Increments or dose reductions, oxazepam, and mianserin were allowed as necessary. Patients were randomized into one of two groups: Rx1: Red dim light (50 lux, 30 minutes daily), 5 weeks Primary outcome measure was change in HAMD-17 score from baseline to endpoint. Rx2 showed a significantly greater decrease in HAM- D scores at five weeks. Baseline 5 weeks Rx Rx Response rates at 5 weeks: Rx1: 38.9% Rx2: 70.8% Remission rates at 5 weeks: NNT* for response (50% or greater reduction of baseline scores on the HAM-D6) at 5 wks: Bright vs. Dim light: 3.1 < National Adult Depression Clincial Practice Guideline

275 Study Serrano & Latorre (2004), Life review therapy using autobiographical retrieval practice for older adults with depressive symptomatology Inclusion & Exclusion Criteria disorder; mental retardation, alcohol or drug abuse; pregnancy; history of light-induced migraine or epilepsy; retinal blindness or severe cataract; glaucoma; retinal diseases; ongoing treatment with antipsychotics; suicidal ideation; severe agitation; score of <13 on HAMD. Volunteer clients (men and women of Social Services in Almansa, Spain. All participants receiving one hour of social services per day, five days per week. Participants had to have clinically significant symptoms of depression (CED-D >16), no evidence of dementia, and could not be receiving pharmacological treatment for depression. Patients were 70% women. Age Limitations / Biases inclusion was noticeably higher in bright light group (46% vs. 30%). This difference was determined not to be statistically significant, but no analysis of difference (p value) reported No placebo control where participants received same amount of attention, but without intervention. Prompting questions during intervention targeted primarily positive memories. No follow-up to indicate maintenance of changes. Small sample size. Not all participants had MDD (although many had significant depressive symptoms). Limited to older population receiving social services, in Spain. Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value N = 54 Rx2: Bright white light (10,000 lux, 60 minutes daily), N = 48 LIFE REVIEW THERAPY Participants were randomized into two groups: Experimental Group (N = 20): given life review therapy while continuing with social services. Life review therapy consisted of autobiographical retrieval practice focusing on a particular period (childhood, adolescence, adulthood, summary) each week for four consecutive weeks. Control group (N = 23): social services as usual. 8 weeks Primary outcome measures were scores on CES- D, LSIA, and BHS. Secondary outcome was change in specificity of memories from pretest to posttest. Rx1: 18.5% Rx2: 41.7% Significant Time x Group effect indicated experimental group improvements in CES-D, LSIA, and BHS. Mean CES-D change for treatment group vs. 0 for control group. Experimental group showed significant increase in recall of specific memories. This was found to be a significant predictor of posttest hopelessness This was also found to be a significant predictor of posttest life satisfaction. There was a trend in the same direction for posttest NNT* for improvement in CES-D to below major depression among those with CES- D diagnosed depression: Experimental vs. control: NNT = 5 < < National Adult Depression Clincial Practice Guideline

276 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value CES-D. BIBLIOTHERAPY Smith, N.M. 1997(Observational) Initial N: 50 Final N: Not given Mean age of 41.6 ±10.0 Small sample size No control group Sensitivity analysis not included to account for patients who completed the initial study but did not participate in the follow-up (some of whom may have relapsed) Observational study of bibliotherapy only; no comparative group. Follow-up: 3-year HRSD and BDI scores Follow-up analysis revealed a significant decrease in HRSD and BDI scores from pretreatment to follow-up No changes from post treatment to follow-up Treatment gains were maintained over the 3- year follow-up period and participants were less depressed than when they began the program. NA p <0.05 each p>0.05 each Study suggests that there maybe long-range benefits to participants in structured, minimalcontact bibliotherapy programs for depression. Jamison, C (RCT) The outcome of cognitive bibliotherapy with depressed adults. Initial N: 80 Final N: 65 Mean age of 40 years Score of 10 or greater on the Hamilton Rating Scale for Depression 21-item version (HRSD) Score of 10 or greater on the 21- item Beck Depression Inventory (BDI) Met the DSM-III-R criteria for a mild or moderate Major Depression Rx1: 30% females, 10% males Rx2: 37% females, 3% males Participants and researchers not-blinded. Wait-list control showed marked symptom improvement in f/u phase Treatment phase (4 weeks) Rx1: Cognitive bibliotherapy (n = 33) Rx2: Waiting list (control group) (n = 39) Follow-up phase (3 months) Rx1: Cognitive bibliotherapy (n = 31) Rx2: Waiting list (control group) Follow-up: 4 weeks of treatment and 3 months of follow-up HRSD and BDI scores Treatment phase HRSD score Rx1: decreased from 20.2 to 9.9 Rx2: decreased from 19.6 to 19.0 BDI score Rx1: decreased from 21.9 to 9.2 Rx2: decreased from 20.9 to 19.5 Follow-up phase HRSD score Rx1: decreased from 10.0 NNT* of clinical improvement at 3- months follow-up (BDI cutoff < 11; HRSD cutoff < 12): BDI: 2.17 HRSD: 1.7 p < 0.05; favoring bibliotherapy p < 0.05; favoring bibliotherapy p < 0.05 (favoring treatment) when starting from baseline 272 National Adult Depression Clincial Practice Guideline

277 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value (n = 34) to 9.2 Rx2: decreased from 18.7 to 10.0 BDI score Rx1: decreased from 9.6 to 7.7 Rx2: decreased from 18.9 to 11.1 p < 0.05 (favoring treatment) when starting from baseline. Study suggests that cognitive bibliotherapy for depression is an effective treatment for depression. Scogin, F Two-year follow-up of bibliotherapy for depression in older adults. Initial N: 30 Final N: Not given 60 years and older, mean of 67.2 Mean age: years Race: 3 Black, 27 White Gender: 30 female, 0 male No male participants No control group Small sample size Sensitivity analysis not included to account for patients who completed the initial study but did not participate in the follow-up (some of whom may have relapsed) Rx1: Behavioral and cognitive bibliotherapy (n = 22) those who had read (n = 16) those who had not (n = 14) Follow-up: 2-year HRSD scores HRSD score change from post -treatment to 2-year follow-up Rx1: 8.1 to 7.4, no significant effect for time Structured bibliotherapy s effect for mild to moderate depression in older adults is maintained 2 years after treatment Unknown how often participants referred to materials over time NA p < 1.00 Scogin, F (RCT, doubleblind) Comparative efficacy of cognitive and behavioral bibliotherapy for mildly and moderately depressed older adults. 60 years and older Score of 10 or higher on the Hamilton Rating Scale for Depression (HRSD) Score of 8 or higher on the Mental Status Questionnaire Not being on psychotropic medication or, if so, then stabilized Only included older adults Predominantly female Small sample size Short follow-up Rx1: Behavioral bibliotherapy (n = 23), reading a copy of Control Your Depression (Lewinsohn, et al. 1986) book Rx2: Cognitive bibliotherapy (n = 22), reading a copy of Follow-up: Initial study 4 weeks of treatment and a 6- month follow-up HRSD Score changes HRSD Score changes from time 1 to 2 to 3 Rx1: 17.8 to 9.7 to 9.1 Rx2: 16.3 to 7.5 to 8.9 Rx3: 16.4 to 15.9 to 7.2 No significant differences between Rx1 and Rx2. Treatment gains were NNT* for clinical improvement from bibliotherapy (scores outside range of dysfunctional population): 2.13 (Rx2 vs. Rx3) p < 0.05, favoring cognitive bibliotherapy 273 National Adult Depression Clincial Practice Guideline

278 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value Initial N: 67 Final N: 44 Scogin, 1990 (RCT, double-blind) on the medication Not being in psychotherapy 57 females and 10 males Believing Good (Burns, 1980) book Rx3 : Delayed treatment (control group) (n = 22) The Rx1 and Rx2 were assessed at: Pre-treatment =Time 1 Immediately following treatment = Time 2 6-month follow-up = Time 3 maintained 6 months after treatment for Rx1 and Rx2. There was no control group for follow-up after the delayed treatment group received treatment. This study supports the efficacy of self-paced bibliotherapy for mild to moderate depression in older adults. The Rx3 participants were assessed at time 1 and at 1 month following time 2, at which point their own treatment began, and a third time immediately following treatment. MUSIC THERAPY Hanser, S.B 1994 (RCT) Effects of a music therapy strategy on depressed older adults. Initial N: 30 Final N: 28 Older adults (61 to 86 years old) 77% female Diagnosis of major or minor depressive disorder, based on a structured interview using the Schedule of Affective Disorders and Schizophrenia Small sample size Predominantly female Relative short follow-up Rx1: Home based music therapy (n = 10) Rx2: Self-administered music therapy (n = 10) Rx3: Wail-list control group (n = 10) Follow-up: 8 weeks Change in Geriatric Depression Scale sore Change in Geriatric Depression Scale sore (pretest to posttest) Rx1: to 7.70 Rx2: to Rx3: to Follow-up results at nine months revealed no significant differences on the GDS score from post test results in all three groups. NA p < 0.05 (favoring the two music therapy groups) This study supports the use of music therapy strategies with older adults experiencing symptoms of depression. 274 National Adult Depression Clincial Practice Guideline

279 Study Inclusion & Exclusion Criteria Age Limitations / Biases Intervention and dose N and Final N Duration Outcome Results NNT P/ χ 2 value EXERCISE & LIGHT THERAPY Leppamaki, et al. (2004), Drop-out and mood improvement: a randomized controlled trial with light exposure and physical exercise. 120 Volunteers from Occupational Health Centers: men and women free from prior mental disorders and with HAM-D Participants were not a random sample of the population. Participants did not have a clinical diagnosis of Major Depression; they had varying degrees of depressive symptoms. Unclear on how many participants may have had Seasonal Affective Disorder. No placebo control group. Participants were randomized into one of three groups: Aerobics training in bright light (N = 40) Aerobics in normal lighting of gym (N = 38) Relaxation and stretching in bright light (N = 42) 8 weeks Primary outcome measures were changes in HDRS, ATYP, and SIGH- SAD-SR Based on HDRS, 42 responders (50% improvement); 83% of these responders received light therapy, which was significant. Based on ATYP, 51 responders; 73% of these responders received light therapy. Based on SIGH-SAD-SR, 45 responders; 82% of these responders received light therapy, which was significant The NNT for light was 3.8. χ 2 = 0.02 χ 2 = 0.02 No information on whether patients were using other treatments. INTERNET AND TELEPHONE Patten, S.B (RCT) Prevention of depressive symptoms through the use of distance technologies. Initial N: 786 Final N: 764 Volunteers were screened for diagnosis of current Major Depression by using the Composite International Diagnostic Interview (95% CIDI) and eligible participants were assigned to treatment or control groups. No other information were provided. Gender: 90.1% female, and 9.9% male Mean age 45.2 ± 11.9 Included predominantly older women. Possible selection bias participants may be more comfortable with the internet. Rx1: Active group, using web and telephone-based program (n = 420) Rx2: Control group (n = 366) The preventive intervention was an interactive computer program accessible through the Web or by interactive voice telephone. Follow-up: 3 months Changes in depression rating scale (CES-D) % of patients exceeding Center for Epidemiological Studies Depression rating scale (CES-D) score of more than 15 Rx1: 37.4% (baseline 35.7%) Rx2: 34.1% (baseline 33.3%) Difference statistically not significant. The study did not find any significant differences between the groups. NNT* for reduction in CES-D score of >15: Rx1 vs. Rx2: 111 NS (p not stated) 275 National Adult Depression Clincial Practice Guideline

280 12. Behavioral Health Education Classes For Adults With MDD (Cognitive Behavioral Skills or Problem-Solving Classes) Problem Formulation 12 Clinical Question: Intended Use of the Guideline: Population: Health Problem: Health Intervention: Practitioners: Setting: Health Outcomes (associated with the intervention): Are behavioral health education classes (cognitive behavioral skills or problem-solving classes) effective in improving the symptoms of MDD? If so, are they recommended for adults with MDD? To assist primary care physicians and other health care professionals in treating adults with Major Depression. All adult (age 19 and over) patients with Major Depression MDD Multi-Session behavioral health education classes: Cognitive behavioral skills for managing depression Problem-solving No behavioral health education classes KP primary care physicians, physician assistants, nurse practitioners, pharmacists, health educators, and social workers Outpatient office visit, emergency department, urgent care clinics, health education, social work Change in symptoms Quality of life Missed school/work days Adherence to treatment plan Office/UCC/ER visits Hospitalizations Mortality Relapse prevention Patient satisfaction Note: Behavioral Health Education Classes (BHE) differ in content, structure, and scheduling from protocolized psychotherapy delivered to patients with specifically defined diagnoses in a specialty behavioral health setting. BHE delivers a specific educational curriculum (as opposed to a structured, tailored psychotherapeutic intervention) focusing on coping and self-care skills, targets patients with mild to moderate depressive symptoms (including patients without Major Depression), and excludes patients with more severe Major Depression. 276 National Adult Depression Clincial Practice Guideline

281 Search Strategy Database: PubMed Pubmed Cochrane Clinical Evidence, Issue 14 PubMed Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR metaanalysis"[all Fields])) ("depression"[mesh] OR "depressive disorder"[mesh] OR "Depressive disorder/therapy"[mesh] OR "Major Depressive Disorder"[all fields]) AND ("health education"[mesh] OR "patient education"[mesh] OR "behavior therapy/methods"[mesh] OR "patient satisfaction"[mesh] OR "cognition"[mesh] AND "psychoeducation"[all fields] OR "behavioral classes"[all fields]) Depression, Anxiety and Neurosis Group Chapter: Depressive Disorders ("depression"[mesh] OR "depressive disorder"[mesh] OR "Depressive disorder/therapy"[mesh] OR "Major Depressive Disorder"[all fields]) AND ("health education"[mesh] OR "patient education"[mesh] OR "behavior therapy/methods"[mesh] OR "patient satisfaction"[mesh] OR "cognition"[mesh] AND "psychoeducation"[all fields] OR "behavioral classes"[all fields]) AND English[Lang] AND "adult"[mesh Terms] AND "humans"[mesh Terms] AND ("2003/01/01"[PDAT] : "2005/08/01"[PDAT]) Article Type and Other Limits: Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, Randomized Controlled Trial, English, All Adult: 19+ years Systematic Reviews Systematic Reviews and RCTs All publication types, All adults 19+ years, English, Human Search Date 8/2005 to 9/2007 8/2005 to 9/2007 No. Included / Total Retrieved * 0/129 0/0 Q4, /131 January /01/03-08/01/05 N/A 0/4 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. Systematic reviews and meta-analyses with search dates outside the range of current searches were excluded. 277 National Adult Depression Clincial Practice Guideline

282 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * Cochrane Depression Systematic reviews 03/05/03 0/295 Clinical Evidence Depression Systematic reviews and RCTs 03/05/03 0/80 PubMed ("depression" [MeSH] OR "depressive disorder" [MeSH] OR "Depressive disorder/therapy" [MeSH] OR "Major Depressive Disorder" [all fields]) AND ("health education" [MeSH] OR "patient education" [MeSH] OR "behavior therapy/methods" [MeSH] OR "patient satisfaction" [MeSH] OR "cognition" [MeSH], "psychoeducation" [all fields] OR "psychosocial classes" [all fields] OR "behavioral classes" [all fields]) All publication types, All adults 19+ years, English, Human 01/01/ /01/2003 2/704 Some studies found and used in this guideline did not appear in PubMed search results due to the way studies are indexed in PubMed. 278 National Adult Depression Clincial Practice Guideline

283 Brown, R.A (RCT) Evidence Tables Table 12.1 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Follow-up: One and 6 months Initial N: 63 Final N: Not Stated (4.6% drop-out rate) Rx1: Class (n = 25) Rx2: Individual tutoring (n = 13) Rx3: Phone contact (N = 14) Rx4: Wait-list control (N = 11) Coping with Depression course format was standardized across the three treatment conditions. It included a text, Control Your Depression from which reading assignments were made. Course consisted of 12 sessions over an 8 week period plus a visit or phone contact (Rx3 group) at one and 6 month post treatment Patients receiving concurrent treatment for depression were included Adults years old 70% female All met Research Diagnostic Criteria (RDC) for a current episode of unipolar depression Mean Beck Depression Inventory (BDI) score of 22.2 (consistent with mild to moderate Major Depression) Unclear if patients with severe Major Depression or chronic Major Depression excluded. Patients with concurrent substance abuse excluded. Participants in the three treatment groups improved more during the 8-week period than did the wait-list control group, p<0.05 There were no statistically significant differences among the three treatment groups, when compared at 8weeks (treatment duration) and one and 6 months follow-up sessions Among the participants in the three treatment groups, only 25% still met the criteria for depression at the 6-month follow-up session Those with residual depression tended to have depression onset at an earlier age than those who responded From an economic point of view, the class condition was by far the most cost effective Proportions of individuals receiving concurrent treatment for depression in each group not specified Difference in response between those receiving concurrent treatment and those not receiving other treatment for depression not specified 34 of 122 individuals chose not to participate in the study, differences between enrolled and not enrolled group not specified 279 National Adult Depression Clincial Practice Guideline

284 Dowrick, C (RCT) Table 12.2 Study, Total n Follow-up: 6 and 12 months Initial N: 452 Final N: 212 Intention-to-treat-analysis Treatment Groups Size & Drug Study Population Results Comments Rx1: Six individual sessions of Problem-Solving treatment delivered in a setting convenient for the patient (usually the patient s home) by behavioral health, allied health, or nursing personnel (N = 128) Rx2: Eight group sessions (2.5 hours each) of a course on prevention of depression (n = 108) Rx3: Control group (n = 189) Adults ages 18 to 65 Diagnosis of depressive disorder (71%) according to DSM-IV, dysthymia (16%), adjustment disorder (4%) or other depressive disorders (9%) Suicidal patients and patients with concurrent substance abuse excluded. Unclear if patients with severe Major Depression who were not suicidal were excluded or included. Mean difference in Beck Depression Inventory (BDI) score Rx1 at 6 months =-2.63 (-4.95 to 0.32), p=0.026 (significant) At 12 months =-1.00 (-3.31 to 1.31), p=0.398 (not significant) Rx2 at 6 months = 1.50 (-4.16 to 1.17), p=0.272 (not significant) At 12 months = 1.11 (-1.30 to 3.52), p=0.901 (not significant) There were significant improvements in SF36 scores (mental role, social function, and mental health) at 6 months for patients in the depression prevention course (p=0.042, 0.048, and respectively), when compared with baseline scores There were significant improvements in SF36 scores (mental role, social function, and mental health) at 6 months for patients in the Problem-Solving therapy group compared with controls at 6 months (p=0.030, 0.012, and respectively). There was no significant differences in SF36 scores at 12 months At 6 months there was a 17% difference in numbers of depressed patients between participants assigned to Problem-Solving and controls (NNT = 6). At 12 months there was no difference between Rx1 and Rx3 At six months there was a 14% difference in numbers of depressed patients between participants assigned to prevention of depression and controls (NNT = 7). At 12 months there was no difference between Rx2 and Rx3. Unweighted (complete case analysis) odds ratio for RX1 at 6 months : 0.51 (0.27 to 0.97) RX1 at 12 months : 0.92 (0.48 to 1.77) Rx2 at 6 months : 0.50 (0.21 to 1.15) Rx2 at 12 months: 1.02 (0.46 to 2.23) More patients completed Problem-Solving therapy ((63%) than the course (44%) p = Study concludes both Problem-Solving treatment and the course on prevention of depression reduces the severity and duration of depressive disorders and improves subjective mental and social functioning. Mixed population (patients with different types of depression, not all Major Depressive Disorder) with the results not stratified by condition The number of different individuals delivering the Problem-Solving therapy was not specified, the problem-solving therapy protocol was not specified, and the inter-provider variability of adherence to the therapy protocol was not analyzed. Therefore, this may not be directly comparable to the Problem- Solving therapy delivered by trained mental health personnel in traditional behavioral health settings. 280 National Adult Depression Clincial Practice Guideline

285 Brown, R.A (RCT) Table 12.3 Study, Total n Treatment Groups Size & Drug Study Population Results Comments Follow-up: One and 6 months Initial N: 63 Final N: Not Stated (4.6% drop-out rate) Rx1: Class (n = 25) Rx2: Individual tutoring (n = 13) Rx3: Phone contact (N = 14) Rx4: Wait-list control (N = 11) Coping with Depression course format was standardized across the three treatment conditions. It included a text, Control Your Depression from which reading assignments were made. Course consisted of 12 sessions over an 8 week period plus a visit or phone contact (Rx3 group) at one and 6 month post treatment Patients receiving concurrent treatment for depression were included Adults years old 70% female All met Research Diagnostic Criteria (RDC) for a current episode of unipolar depression Mean Beck Depression Inventory (BDI) score of 22.2 (consistent with mild to moderate Major Depression) Unclear if patients with severe Major Depression or chronic Major Depression excluded. Patients with concurrent substance abuse excluded. Participants in the three treatment groups improved more during the 8-week period than did the wait-list control group, p<0.05 There were no statistically significant differences among the three treatment groups, when compared at 8weeks (treatment duration) and one and 6 months follow-up sessions Among the participants in the three treatment groups, only 25% still met the criteria for depression at the 6-month follow-up session Those with residual depression tended to have depression onset at an earlier age than those who responded From an economic point of view, the class condition was by far the most cost effective Proportions of individuals receiving concurrent treatment for depression in each group not specified Difference in response between those receiving concurrent treatment and those not receiving other treatment for depression not specified 34 of 122 individuals chose not to participate in the study, differences between enrolled and not enrolled group not specified 281 National Adult Depression Clincial Practice Guideline

286 Dowrick, C (RCT) Table 12.4 Study, Total n Follow-up: 6 and 12 months Initial N: 452 Final N: 212 Intention-to-treat-analysis Treatment Groups Size & Drug Study Population Results Comments Rx1: Six individual sessions of Problem-Solving treatment delivered in a setting convenient for the patient (usually the patient s home) by behavioral health, allied health, or nursing personnel (N = 128) Rx2: Eight group sessions (2.5 hours each) of a course on prevention of depression (n = 108) Rx3: Control group (n = 189) Adults ages 18 to 65 Diagnosis of depressive disorder (71%) according to DSM-IV, dysthymia (16%), adjustment disorder (4%) or other depressive disorders (9%) Suicidal patients and patients with concurrent substance abuse excluded. Unclear if patients with severe Major Depression who were not suicidal were excluded or included. Mean difference in Beck Depression Inventory (BDI) score Rx1 at 6 months =-2.63 (-4.95 to 0.32), p=0.026 (significant) At 12 months =-1.00 (-3.31 to 1.31), p=0.398 (not significant) Rx2 at 6 months = 1.50 (-4.16 to 1.17), p=0.272 (not significant) At 12 months = 1.11 (-1.30 to 3.52), p=0.901 (not significant) There were significant improvements in SF36 scores (mental role, social function, and mental health) at 6 months for patients in the depression prevention course (p=0.042, 0.048, and respectively), when compared with baseline scores There were significant improvements in SF36 scores (mental role, social function, and mental health) at 6 months for patients in the Problem-Solving therapy group compared with controls at 6 months (p=0.030, 0.012, and respectively). There was no significant differences in SF36 scores at 12 months At 6 months there was a 17% difference in numbers of depressed patients between participants assigned to Problem-Solving and controls (NNT = 6). At 12 months there was no difference between Rx1 and Rx3 At six months there was a 14% difference in numbers of depressed patients between participants assigned to prevention of depression and controls (NNT = 7). At 12 months there was no difference between Rx2 and Rx3. Unweighted (complete case analysis) odds ratio for RX1 at 6 months : 0.51 (0.27 to 0.97) RX1 at 12 months : 0.92 (0.48 to 1.77) Rx2 at 6 months : 0.50 (0.21 to 1.15) Rx2 at 12 months: 1.02 (0.46 to 2.23) More patients completed Problem-Solving therapy ((63%) than the course (44%) p = Study concludes both Problem-Solving treatment and the course on prevention of depression reduces the severity and duration of depressive disorders and improves subjective mental and social functioning. Mixed population (patients with different types of depression, not all Major Depressive Disorder) with the results not stratified by condition The number of different individuals delivering the Problem-Solving therapy was not specified, the problem-solving therapy protocol was not specified, and the inter-provider variability of adherence to the therapy protocol was not analyzed. Therefore, this may not be directly comparable to the Problem- Solving therapy delivered by trained mental health personnel in traditional behavioral health settings. 282 National Adult Depression Clincial Practice Guideline

287 13. Antidepressants To Avoid During Pregnancy or Breastfeeding Problem Formulation 13 Clinical Question(s) Which antidepressants should be avoided in women who are pregnant or breastfeeding? Population Patients diagnosed with MDD who may be pregnant or breastfeeding Health Intervention(s) Antidepressant treatment (SSRIs, TCAs, DAs, SNRIs, NRIs) No treatment Most Important Health Outcomes Hospitalizations Mortality Spontaneous abortion Congenital malformations Perinatal complications (mother and child) Long-term behavioral sequelae 283 National Adult Depression Clincial Practice Guideline

288 Search Strategy Database: Cochrane Systematic Reviews PubMed PubMed Depression Search Terms: ("Depression"[MeSH] OR "Major Depressive Disorder"[All Fields] AND ( systematic"[all Fields] OR systematic review"[all Fields] OR metaanalysis"[all Fields])) "depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) OR "Antidepressive Agents"[Mesh] AND neonate OR neonatal OR pregnancy OR lactation OR breastfeeding (((("depression"[mesh Terms] OR DEPRESSION[Text Word]) OR "depressive disorder"[mesh Terms]) OR depressive disorder[text Word]) OR "Antidepressive Agents"[Mesh]) AND (((((( neonate OR neonatal ) OR pregnancy ) OR lactation ) OR breastfeeding ) OR "prenatal") OR "antenatal") Article Type and Other Limits: N/A Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, English, All Adult: 19+ years Only items with abstracts, Humans, English, All Adult: 19+ years Search Date 2005 to 9/2007 9/2007 to 10/01/2009 8/2005 to 9/2007 9/2007 to 10/01/ /2007 9/2007 to 10/01/2009 Hand search NA 9/2007 to 10/01/2009 No. Included / Total Retrieved* 0/134 0/58 0/129 5/137 8/262 5/483 1 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. 284 National Adult Depression Clincial Practice Guideline

289 Hemels MEH 2005 (meta-analysis) Evidence Tables Table 13.1: Meta-Analyses of Antidepressants in Pregnancy Studies Total N # studies found: 156 # studies included: 6 Total N = 3567 Heterogeneity: overall heterogeneity was nonsignificant (used fixed-effects model?) Lattimore KA 2005 (meta-analysis) # studies found: 194 # studies included: 9 Total N = 326 Heterogeneity: a nonlinear mixed effect model was used in statistical analysis due to the absence of homogeneity Moses-Kolko EL 2005 (meta-analysis) # studies found: not reported # studies included: 9 Total N = 1,111 What model used for MA? Study Population Groups Size & Drug Results Comments Observational cohort studies including pregnant women on antidepressant therapy 1 study of trazodone or nefazodone 1 study of venflaxine 1 study of sertraline/paroxetine/fluvoxamine 2 studies of fluoxetine 1 study of TCA Prospective cohort studies of second and third trimester exposure to SSRIs. 1 study of paroxetine only 2 studies of fluoxetine only Cohort studies of patients who had been exposed to SSRIs in the final trimester of pregnancy that included assessment of neonatal outcomes compared with control groups who had early or no exposure 1 study of paroxetine only 3 studies of fluoxetine only Rate of spontaneous abortion in women exposed to all antidepressants was 12.4%, compared with 8.7% in women not exposed to antidepressants (RR = 1.45) Outcome Prematurity Low birth wt SNU/NICU Poor adaptation NNT Outcome Neonatal behavioral syndrome SNU/NICU Respiratory difficulty Seizure Nonsignificant differences seen among drug classes. OR (95% CI) 1.85 (0.79, (1.01, 13.08) 3.30 (1.45, 7.54) 4.08 (1.20, 19.93) OR (95% CI) 3.0 ( ) 2.6 ( ) 2.3 ( ) 4.1 ( ) p Authors conclusions: Maternal exposure to antidepressants is associated with an increased risk of spontaneous abortion. Biases, etc. Time of exposure to antidepressants varied, studies may represent overlapping populations, no verification made of spontaneous abortion versus elective termination of pregnancy. Authors conclusions: SSRI exposure appears to increase the incidence of prematurity, low birth weight, special-care nursery admissions, and the diagnosis of poor neonatal adaptation. Biases, etc. Some studies included smokers and women using other psychotropic medications. Authors conclusions: There is an increased risk of neonatal behavior syndrome and other CNS symptoms following gestational exposure to SSRIs. Biases, etc. No information regarding heterogeneity of data, 285 National Adult Depression Clincial Practice Guideline

290 Name design Pedersen et al (2009) Cohort study Table 13.2: Cohort and Case-Control Studies of Antidepressants in Pregnancy Sample characteristics Not reported. Information from all live-born children in Denmark between 1/ /31/2005. Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI Exposure: SSRI (Fluoxetine, Citalopram, Paroxetine, Sertraline) during 1 st trimester of pregnancy Outcomes: Birth defects including major and minor malformations, septal heart defects, noncardiac malformations p- value Study Quality Fair 3,4 Biases* Sub-analyses included women with one or more redemptions for antidepressants: 3010 women redeemed one or more SSRI, 265 for tricyclic antidepressants, and 150 for venlafaxine." 286 National Adult Depression Clincial Practice Guideline

291 Name design Andrade et al (2009) Case-control study Sample characteristics Females >15 years who were admitted between for delivery. Exposure and nonexposure women were matched by age, health plan, year of admission. Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI Exposure: Antidepressants used (SSRIs, tricylics, miscellaneous) during 3 rd trimester. Outcomes: Infant persistent pulmonary hypertension (PPHN) RR = for antidepressants used: 0.67 RR = for SSRIs: p- value Study Quality Poor 3,4 Biases* Wichman et al (2009) Retrospective cohort study Women presented at Mayo Clinic from 1993 July 15, Exposure: SSRI use (fluoxetine, paroxetine, sertraline, citalopram, escitalopram, venlafaxine) during pregnancy. Poor 3,4 Outcomes: Congenital heart disease (CHD) and PPHN. 287 National Adult Depression Clincial Practice Guideline

292 Name design Toh et al (2009) Case-control study Sample characteristics Women who gave birth to nonmalformed an non-hypertensive babies and who participated in the Slone Epidemiology Center Birth Defects Study from Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI Exposure: SSRI use during pregnancy Outcomes: Gestational hypertension and preeclampsia p- value Study Quality Poor-Fair 3,4 Biases* Jordan et al (2008) Retrospective cohort study Pregnant women with psychiatric illness in prenatal clinic from Sept Aug Women not treated with SSRI considered controls Exposure: SSRI use during pregnancy Outcome: Neonatal behavioral syndrome (NBS) irritability, jitteriness, hypoor hypertonia, hyperreflexia, oxygen requirement, apnea, flaring, grunting, retractions, vomiting, poor RR = 1.67** NA NA Poor 3,4 288 National Adult Depression Clincial Practice Guideline

293 Name design Sample characteristics Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI p- value Study Quality Biases* feeding or hypoglycemia. Ramos et al (2008) Case-control study Pregnant women aged in Quebec; have at least 1 diagnosis of psychiatric disorder before pregnancy. Exposure: Antidepressants (SSRIs, - citalopram, fluoxetine, fluvoxamine, paroxetine and sertraline; tricyclic amitriptyline, clomipramine, deipramine, doxepin, imipramine, nortriptyline, trimipramine; and new antidepressants bupropion, mirtazapine, moclobemide, nefazodone, trazodone, venlafaxine) during 1 st trimester of pregnancy. Fair 4 Outcome: Major congenital malformation. Diav-Citrin et al (2008) Prospective cohort study Pregnant women who contacted the Israeli, Italy and Germany Teratology Information Service (TIS). Exposure: Paroxetine or fluoxetine or nonteratogenic exposure (e.g., antibiotics, oral contraceptives, Table 2. Pregnancy outcomes. Poor 2,3,4 289 National Adult Depression Clincial Practice Guideline

294 Name design Sample characteristics Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI p- value Study Quality Biases* etc.). Outcome: Major congenital abnormalities (i.e. structural abnormalities with medical, surgical or cosmetic consequences; VSD, neurodevelopmen tal or functional problems) Table 3. Logistic regression analysis for miscarriage rate and cardiovascular anomalies. 290 National Adult Depression Clincial Practice Guideline

295 Name design Sample characteristics Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI p- value Study Quality Biases* Salkeld et al (2008) Nested casecontrol study Women aged years in Ontario who received government-funded drug coverage 2 years prior to delivery. Cases were matched with controls in ratio 1:10 Exposure: Antidepressant (SSRIs fluoxetine, fluvoxamine, sertraline, paroxetine, and citalopram; and non-ssris cyclic antidepressants, bupropion and venlafaxine) use in 3 rd trimester. Poor 2,3,4 Outcomes: Postpartum hemorrhage Kallen and Olausson (2008) Prospective cohort study Swedish women in Register who gave birth between to infants diagnosed with PPHN. Exposure: SSRI during pregnancy. Outcome: PPHN Poor Fair 3,4 291 National Adult Depression Clincial Practice Guideline

296 Name design Sample characteristics Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI p- value Study Quality Biases* Oberlander et al (2008) Retrospective cohort study Records of live births between 4/1/97 3/31/2002 in British Colombia from women diagnosed with depression. Data derived from hospital, pharmacy and physician billing registries. Exposure: 1 st trimester SSRI (Citalopram, Fluoxetine, Fluvoxamine, Paroxetine, Sertraline, And Venlafaxine) and/or benzodiazepines (BZ). Poly drug exposure was defined as SSRI and BZ were dispensed for same day during 1 st trimester. Fair 3 Outcome: Major congenital anomalies and subset of cardiovascular defect (VSD and ASD). 292 National Adult Depression Clincial Practice Guideline

297 Name design Sample characteristics Exposure/ EXPOSED NONEXPOSED Outcome of interest Cases N Cases N OR/ RR 95% CI p- value Study Quality Biases* 293 National Adult Depression Clincial Practice Guideline