Identifying Psychological Distress at Key Stages of the Cancer Illness Trajectory: A Systematic Review of Validated Self-Report Measures

Transcription

1 Vol. 41 No. 3 March 2011 Journal of Pain and Symptom Management 619 Review Article Identifying Psychological Distress at Key Stages of the Cancer Illness Trajectory: A Systematic Review of Validated Self-Report Measures Lucy Ziegler, PhD, Kate Hill, PhD, Liz Neilly, MSc, Michael I. Bennett, ChB, MD, FRCP, FFPMRCA, Irene J. Higginson, BM, BS, BMedSci, PhD, FFPHM, FRCP, Scott A. Murray, MB, CHB, MRCGP, PhD, and Dan Stark, MB, CHB, FRCP, PhD, for and on behalf of the COMPASS Collaborative Psychosocial Oncology and Clinical Practice Research Group (L.Z.) and Section of Oncology and Clinical Research (D.S.), Institute of Molecular Medicine; Health Sciences Library (L.N.), University of Leeds and Psychiatry and Behavioural Sciences (K.H.), Leeds Institute of Health Sciences, University of Leeds, Leeds; International Observatory on End of Life Care (M.I.B.), School of Health and Medicine, Lancaster University, Lancaster; Department of Palliative Care, Policy and Rehabilitation (I.J.H.), King s College London, London; and Primary Palliative Care Research Group (S.A.M.), Division of Community Health Sciences, The University of Edinburgh, Edinburgh, United Kingdom Abstract Context. To enable study of psychological distress along the cancer journey, we need to be able to select and map validated measures through the cancer trajectory. Objectives. To examine the performance of self-report measures for identifying clinically significant levels of psychological distress across the cancer patient trajectory. Methods. Electronic searches of Medline, PsychInfo, CINAHL, EmBase, The Cochrane Library, AMED, BNI, ASSIA, and Web of Science were undertaken. Only studies of self-report measures that used validated diagnostic tools for psychiatric diagnosis as the criterion measure were included. We then further limited our focus to those papers that specified a trajectory stage. Results. Forty-eight different self-report measures were in the 85 papers identified. The Hospital Anxiety and Depression Scale () was the most frequently measure (23 times). Several other measures were between two and four times, but most (37) measures were only once. Twenty-two of the 85 included papers measure performance by trajectory stage. Best performing measures based on validation data available The COMPASS Collaborative is a National Cancer Research Institute-funded Supportive and Palliative Care Research Collaborative. Address correspondence to: Lucy Ziegler, PhD, Psychosocial Oncology and Clinical Practice Research Ó 2011 U.S. Cancer Pain Relief Committee Published by Elsevier Inc. All rights reserved. Group, St. James s Institute of Oncology, Beckett, Street, Leeds LS9 7TF, United Kingdom. L.E.Ziegler@leeds.ac.uk Accepted for publication: June 14, /$ - see front matter doi: /j.jpainsymman

2 620 Ziegler et al. Vol. 41 No. 3 March 2011 could be identified for each trajectory stage: for pretreatment, the for identifying depression; during treatment, the and Mental Health Inventory-5 (MHI-5) together for identifying clinically significant distress; posttreatment, the for identifying depression; and at recurrence and during the palliative phase, the Brief Edinburgh Depression Scale (BEDS) for identifying depression. Conclusion. No single measure had evidence to support use throughout the illness trajectory in a longitudinal study, but the, in combination with the MHI-5, was supported during the cancer treatment phase, and BEDS in the palliative care phase. J Pain Symptom Manage 2011;41:619e636. Ó 2011 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Psychological distress, measurement, self-report, trajectory, cancer Introduction Psychological distress is common in cancer patients, with overall prevalence rates up to 70%, depending on the criteria used to define it. 1e3 Psychological distress can range from normal adaptive emotions through to higher levels of clinically significant symptoms that fulfill the standardized diagnostic criteria for adjustment disorder, anxiety disorders, or major depression. Between 8% and 25% of cancer patients experience a diagnosable depressive syndrome. 4 Despite the frequency, relevance to the provision of holistic treatment, morbidity, 5 and amenability to management, 6,7 psychological distress is often underdiagnosed during cancer care. Attempts to redress this have led to a proliferation of new self-report measures and the application of existing instruments to the measurement of distress, despite often having been developed with a different purpose in mind. To enable the study of psychological distress along the cancer journey, from initial symptoms to diagnosis, treatment, survivorship, and end of life, we need to be able to select and map validated measures through the cancer trajectory. Consideration of longitudinal measurement with the simplest and best-validated instruments is important, as psychological distress has been shown to vary greatly from diagnosis to advanced disease. 1 National Institute for Clinical Excellence (NICE) guidance 8 recommends that assessment be used to plan care and that it should be done at key points in the patient pathway where changes in care management may be indicated: At time of diagnosis; During initial treatment episodes; Following treatment; and At the time of recurrence. The UK National Cancer Survivorship Initiative also requires ongoing assessment of symptoms across the cancer trajectory. The recognition that psychological distress is likely to differ at key points of the cancer trajectory is not only reflected in UK guidance but in the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines for distress management in oncology. 1 With the growing volume of self-report measures available for patient- outcomes, the need for a structured and objective means to evaluate, compare, and select measures has emerged. Previous reviews 4,6,7 have critically appraised the suitability of self-report measures for identifying psychological distress in cancer and end-of-life care, or for the identification of specific aspects of psychological difficulty such as anxiety or depression. For cancer patients who have a diagnosable psychiatric disorder, rigorously developed and validated diagnostic tools for psychiatric diagnosis (Schedules for Clinical Assessment in Neuropsychiatry [SCAN], Psychological Stress Evaluator [PSE], Structured Clinical Interview [] for the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. [DSM-IV] 9 ) are valid within the

3 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 621 cancer population, 6,7 but these are too time consuming to use for detection within routine cancer care. Issues of feasibility of completion of assessments appear especially relevant the more disabled and symptomatic a person is. This is likely to be a particular issue in recurrent disease and palliative care. Therefore, for patients who experience clinically significant levels of psychological distress, there is currently no clear consensus on which measure to use at different points in the cancer trajectory. 6 Most people living with or after cancer who experience significant psychological difficulties do not have a diagnosable psychiatric disorder, but a lesser level of disruptive distress. 4 Lack of consensus regarding prevalence and measure selection may, in part, be because of the absence of a specific definition of psychological distress. Psychological distress associated with cancer has been defined inclusively as a multifactorial, unpleasant experience of an emotional, psychological, social, and spiritual nature, which may affect patients coping response to cancer and the physical manifestations of the illness and treatment. 1 An inclusive definition brings challenges to specific research design and assessment. To date, there are only very limited established conceptual parameters to define psychological distress within oncology and the relationship to the more definable parameters of major depression, anxiety, and adjustment disorder. Further discussion of the challenge of defining psychological distress and the proposal of a conceptual model of distress in oncology is the subject of a parallel paper. Given the lack of consensus about what constitutes general psychological distress in cancer patients, this review focuses on the scope of selfreport tools to measure clinically significant psychological distress across the cancer trajectory. Here, distress incorporates the widely recognized diagnosable psychiatric disordersdadjustment disorder, major depression, minor depression, and anxiety disorderdand also incorporates levels of distress that have been defined as clinically significant by the authors of the studies included in the review. This review has been undertaken as part of a coordinated UK national research group, the Complex Assessment Trials and Implementation of Services Collaborative (COMPASS), which was established to share expertise to increase research capacity and to develop and test complex interventions in a range of settings from diagnosis to survivorship. The review may lead to more systematic and effective selection of measures and a consequent improvement in assessment of patients, together with a better understanding of the natural history of psychological responses to cancer throughout the cancer patient trajectory. A move toward greater consensus on measure selection also could facilitate direct comparison between cohorts and valuable pooling of research data. Methods Study Selection The process for identifying and appraising studies for the review was guided by the Centre for Reviews and Dissemination (CRD) guidance on undertaking systematic reviews, 10 the Cochrane Handbook for Systematic Reviews of Interventions, 11 and from recommendations made by Downs and Black 12 and Moher et al. 13 Electronic searches were conducted using of Medline; PsychInfo; CINAHL; EmBase; The Cochrane Library; AMED; BNI; ASSIA; Web of Science; Dissertations & Theses; and Sociological Abstracts. A core search strategy was developed for Medline, and adapted for each database. Hand searching of relevant journals and conference abstracts also was undertaken. Study Inclusion and Evaluation Only studies exploring the validation of a self-report measure alongside a structured clinical interview for psychiatric disorder were included in the review. Where indicated in the study, time point on the illness trajectory from diagnosis to end of life was noted. Studies conducted in hospitals, hospices, or primary care settings and reporting on any self-report measures delivered by post, , or in person using paper or computer-based questions all qualified for inclusion. Studies published prior to 1960 in a non-english language were excluded. Only papers that measure performance by a definable trajectory stage are included in the subset of papers here. In defining trajectory stages, we were guided by NICE guidance, 8 Feuerstein, 14 and the NCCN Clinical Practice Guidelines for distress management in oncology. 1 This helped

4 622 Ziegler et al. Vol. 41 No. 3 March 2011 ensure that our definitions of trajectory stages were in line with the wider theoretical and policy developments, both nationally and internationally, about how best to conceptualize the cancer patient s journey. All identified papers were independently scanned by two of the review authors (K. H., L. Z.). Papers were assessed for inclusion or exclusion from the titles or abstracts on the basis of the criteria specified above. Full copies of papers were obtained if one of the review authors deemed it to be potentially eligible. Any paper about which there was disagreement concerning eligibility was referred to a third review author (D. S.). The retrieved references were downloaded and managed in an Ende Version X database. Extracted data were entered in a Microsoft Office 2003 ACCESS database. Quality Appraisal Two separate quality appraisal processes were undertaken and were guided by the Cochrane Handbook for Systematic Reviews of Interventions 11 and from Downs and Black. 12 The first process informed the decision to include or exclude studies from the review and focused on determining the methodological quality of the validation studies. The second process focused on the validity and quality of the self-report measures. In the first quality appraisal, process data were extracted to help provide an objective indication of the methodological quality of the paper. Extracted data included study design, sample size, sample characteristics (including cancer site, disease stage, trajectory stage, age range, and gender), clinical setting, recruitment process, whether a cognitive screen was undertaken, type of criterion measure used (SCAN, PSE, ), and the interviewer skill level (psychologist, psychiatrist, nurse, or researcher). Rating scales for each criterion were applied using a simple present or absent judgment with a score of 1 or 0 (Yes/No) assigned. Where more precise ratings of quality could be attributed to a criterion, the scoring metric was: 0 ¼ poor or low quality, or criterion absent; 1 ¼ criterion present but moderate or adequate quality; and 2 ¼ criterion present with high or above average quality. The review authors rated the data extracted from studies independently and discrepancies in scores were referred to the third review author for discussion. Only papers with adequate quality scores were included in the review. For papers that demonstrated adequate quality criteria, further data were then extracted relating to the measure in this paper. Extracted data included date and country of development, intended purpose, sample characteristics of validation population, cut-off score for target disorder, sensitivity, specificity, positive predictive value and negative predictive value, reliability scores, item structure, and data relating to feasibility. These data were then subjected to the quality appraisal process detailed above. Data Synthesis Despite their common purpose of validating a self-report measure, many studies were heterogeneous in methods, and consequently, there was variation in the available validation data. This made it difficult to synthesize in a quantitative analysis. 12 Where available, sensitivity and specificity scores are for each measure. A minimum score of 0.7 is taken to indicate good sensitivity and specificity. 6 Measures are presented alongside others used at the same stage of the cancer patient trajectory to facilitate comparison. For palliative care patients, supplementary information on the feasibility of the measures was extracted, as this is particularly relevant to this trajectory stage. Results Identification of Relevant Papers The initial search identified 4160 papers, 177 of which met the inclusion criteria on first screening (Fig. 1). When these papers were subjected to closer examination by two independent reviewers, 84 did not qualify for inclusion. The most frequent reason for exclusion was that the paper did not report self-report measures alongside a. Eleven equivocal papers also were excluded following consultation with a third reviewer and 12 papers identified from hand searching and personal correspondence were added. Eighty-five papers qualified for inclusion, of which 22 findings at a specific point on the illness trajectory. Twenty self-report measures were validated against the, six against Present State

5 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 623 population. This subsection consisted of 22 papers reporting 18 different self-report measures. The performance of these measures is here. Fig. 1. Process of search and retrieval of studies. Examination, and 11 against RCIS. DSM-IV was the most frequently used classification system, used 10 times. DSM-IIIR was used twice. Frequency of Measures Reported In total, 48 different self-report measures were in the 85 papers that fulfilled the initial inclusion criteria for the review. Most measures (37) were only once. The Hospital Anxiety and Depression Scale () was the most frequently measure (23 times). The Hamilton Anxiety Scale was four times. The Rotterdam Symptom Checklist, the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30 (EORTC QLQ-C30), General Health Questionnaire (GHQ-28), and the Beck Depression Inventory (BDI) were each three times. The Cancer Concerns Checklist and the Impact of Events Scale were both twice. For the purposes of this paper, a subsection of papers was identified for which it was possible to assign a trajectory stage to the validation Measures by Trajectory Stage At Diagnosis and Prior to Treatment. Five measures were validated in populations around the time of diagnosis and prior to the commencement of treatment (Table 1). 15e20 The most frequently used was the. This measure was used in five of the validation studies at this trajectory stage. Three were translated versions of the (ese and Italian) and two were English versions used in Australian populations. Although the was developed as a screening instrument for anxiety and depression, in this pretreatment population it was used to screen for major and minor depression, adjustment disorder, and distress. The cut-off scores for the varied between 4 and 8 (subscale scores) and 11 and 19 (whole scale). Optimum sensitivity and specificity scores for identifying major depression were in Kugaya et al. s study 17 as a whole scale score of 19; however, in Love et al. s 19 sample of 303 newly diagnosed breast cancer patients, a cut-off of 19 demonstrated inadequate sensitivity to identify major depressive disorder. A cut-off score of 7 on the depression subscale performed well for identifying major depressive disorder and a score of 11 identified depression. The BDI showed very good sensitivity and acceptable specificity for identification of major and minor depression. The Zung Depression Scale poor sensitivity but good specificity for identifying predromal depression with a cut-off score of 4. A study by Kirsh et al. 16 the social subscale of the Functional Assessment of Cancer TherapydGeneral to be a significant predictor of adjustment disorder. Most of the evidence available to inform the selection of measures at the pretreatment phase of the trajectory relates to the identification of depression. Only one study 16 the usefulness of measures for the identification of anxiety, although anxiety at this stage may be common (12). The Zung Anxiety Scale was used to identify predromal anxiety, and although the sensitivity score was low, the specificity score was only just below the required

6 Measure Zung Self- Rating Depression Scale FACT-G Beck Depression Inventory- Short Form Study Participants (Trajectory Stage and Sample Size) 197 breast cancer patients (1/2 sample before start of chemotherapy, 1/ 2 sample at followup postchemotherapy) Pretreatment Bone Marrow Transplant Pretreatment (BMT) 107 newly diagnosed head and neck cancer patients 107 newly diagnosed head and neck cancer patients 303 breast cancer patients, pretreatment, postdiagnosis 227 breast cancer patients, pretreatment, postdiagnosis 227 breast cancer patients, pretreatment, postdiagnosis Table 1 Pretreatment Measures First Author/ Country Purpose in Study Items/Structure Sensitivity Specificity PPV NPV Cut-Off Costantini 15 / Italy Kirsh 16 /USA Kirsh 16 /USA Kugaya 17 / Kugaya 18 / Love 19 / Australia Love 20 / Australia Love 20 / Australia Identify depression To screen for clinically significant levels of depression Cancer specific; assesses four quality-of-life domains To assess psychological distress Screening for adjustment disorder and MDD Identify major and minor depression Accuracy for identifying defined major and minor depression -defined major and minor depression 14-Item Likert scale, two subscales 20-Item Likert scale 27 Items Social subscale of FACT-G significant predictor of adjustment disorder Two seven-item Likert subscales Two seven-item subscales, Likert response structure Two seven-item subscales, Likert response structure 13 Items covering depressive symptoms Two seven-item Likert subscales , Whole scale Standardized Clinical Interview Social subscale of FACT-G significant predictor of adjustment disorder For MDD alone, DS AS T 0.17 MDD 0.94 Depression 0.94 MDD 0.81 Depression 0.16 FACT-G ¼ Functional Assessment of Cancer TherapydGeneral; ¼ Hospital Anxiety and Depression Scale; -T ¼ Hospital Anxiety and Depression Scale Total Score; -DS ¼ Depression Subscale score; -AS ¼ Anxiety Subscale score; MDD ¼ major depressive disorder; ¼ Structured Clinical Interview for DSM-III-R; PPV ¼ Positive Predictive Value; NPV ¼ Negative Predictive Value; BMT ¼ bone marrow transplantation Ziegler et al. Vol. 41 No. 3 March 2011

7 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 625 threshold. In the absence of other evidence to guide the selection of measures for anxiety at this trajectory stage, this measure might still be judged to be potentially useful. During Active Treatment. Seven measures were in populations receiving active treatment (Table 2). 21e24 Only one measure, the EORTC QLQ-C30, was twice. Inconsistencies in available data for studies within this trajectory stage made direct comparisons across measures difficult. Cull et al. 22 is the only study included in the review to attempt to validate a multimeasure approach. In this study, the followed by the Mental Health Inventory-5 (MHI-5), validated in a sample of 172 mixed cancer site patients receiving chemotherapy, performed well to screen for clinically significant levels of distress (defined by the Present State Examination). Pugliese et al. 24 proposed a cut-off score of 8 for depression and 10 for anxiety but do not provide sensitivity and specificity scores at these levels. The 28-item GHQ-28 was by Hughson et al. 23 and demonstrated excellent sensitivity and specificity scores for identifying what the authors characterize as psychosocial morbidity. The validation data were derived from a sample of 75 breast cancer patients on chemotherapy. The study was published in 1988, and given the changes in chemotherapy treatment for breast cancer over the last 20 years and the absence of other validation studies on the GHQ-28 within our review, we were hesitant to recommend it as a preferred measure for the active treatment phase. Post-Treatment. Eight different measures were in the post-treatment trajectory population in the review (Table 3). 15,25e29 Only three measures had comprehensive validation data available. These were the, BDI, and Beck Anxiety Inventory (BAI). All three demonstrated adequate sensitivity for the identification of depression ( and BDI) and anxiety (BAI). However, the BAI and BDI both had specificity scores below the required threshold. Direct comparison was again difficult given the variation in data available. At Time of Recurrence/Palliative Phase. Eleven different measures were in the recurrent/palliative phase of treatment (Table 4). 20,30e35 The ultra-short measures showed low sensitivity and high specificity. Interestingly, when used to identify depression, the with a cut-off of 19 did not perform adequately in a population with less than six months to live. 17 The same measure, however, with the same cut-off and target disorder, showed excellent sensitivity and specificity scores in one study in the pretreatment phase of the trajectory. 15 When used to identify what an author termed any psychiatric disorder rather than depression, the performed well with a cut-off of 20; 31 however, the usefulness of this broad concept within clinical or research applications is questionable. The Brief Edinburgh Depression Scale (BEDS) with a cut-off of 6 also proved acceptable for identifying depression within this trajectory stage and the BDI with a cut-off of 4 demonstrated good sensitivity but low specificity scores, 36 which is consistent with its use in the pretreatment sample. 20 In a recent validation study, 33 three selfreport measuresdthe Distress Thermometer (DT), Brief Symptom Inventory-18 (BSI-18), and GHQ-12dwere delivered by touch screen to 150 palliative care patients and their performance validated against SCAN. The validation data available suggest that the performance of each measure is relatively comparable when optimum cut-off scores are applied (GHQ-12: 6, BSI: 62, and DT: 6). Feasibility Given that palliative care patients may face specific challenges in terms of completing self-report measures, we extracted data that we considered would inform a judgment on the feasibility of measures for use within this population (Table 5). 30e34 This included number of items, time to complete, response structure, and also any data available on attrition from the validation study and recruitment figures. Most measures took less than five minutes to complete and consisted of between 10 and 20 items with a single type of response structure. Most, therefore, could be judged to be feasible in terms of time burden. Where authors that patients were unable to participate, this was generally because they were too drowsy, too ill, or lacked capacity to

8 Measure Study Participants (Trajectory Stage and Sample Size) Center for 115 Breast cancer Epidemiologic patients Studies undergoing Depression radiotherapy Scale / MHI-5 together EORTC QLQ-C30 Concerns Checklist GHQ Mixed cancer site patients undergoing chemotherapy 172 Mixed cancer site patients undergoing chemotherapy 172 Mixed cancer site patients undergoing chemotherapy 75 Breast cancer patients undergoing chemotherapy First Author/ Country Aukst- Designed to Margetic 21 / measure Croatia depressive symptoms in community Cull 22 /UK Cull 22 /UK Cull 22 /UK Hughson 23 / UK Table 2 During Active Treatment Purpose in Study Items/Structure Sensitivity Specificity PPV NPV Cut-Off Screening for clinically significant levels of distress Screening for clinically significant levels of distress Screening for clinically significant levels of distress Identify psychosocial morbidity 20-Item self-report scale dtwo 7-item subscales MHI-5 5 questions; patients report how much in the last 4 weeks they have been feeling like this on 1e6 scale (none of the time to all the time). 4-item scale; responses given on the 4-point scale 14-Item checklist referring to 13 cancerrelated concerns 28 Items in four subscales Score of $16 indicated clinically relevant depression MHI-5 ¼ 11, use. Then if $9 ¼ case Performance inferior to and MHI-5 Performance inferior to MHI-5 and GHQ-28 Cut-off Performance inferior to and MHI-5 Performance inferior to MHI-5 and GHQ-28 Cut-off Whole scale 10. Anxiety and insomnia subscale 6; severe depression subscale 2; somatic symptoms subscale 6; and social dysfunction subscale 7. Standardized Clinical Interview PSE PSE PSE PSE 626 Ziegler et al. Vol. 41 No. 3 March 2011

9 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 627 Eight for depression, 10 for anxiety Two subscales, 14 items As previously Pugliese 24 / Italy 98 Colorectal patients undergoing chemotherapy 30-Item self-report Five functioning scales Pugliese 24 / Italy 98 Colorectal patients undergoing chemotherapy EORTC QLQ- C30 ¼ Hospital Anxiety and Depression Scale; MHI-5 ¼ Mental Health Inventory-5 items; PSE ¼ Present State Examination; GHQ-28 ¼ General Health Questionnaire-28 items; EORTC QLQ- C30 ¼ European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-30 items; ¼ Structured Clinical Interview for DSM-III-R; PPV ¼ Positive Predictive Value; NPV ¼ Negative Predictive Value. participate in the element of the study, rather than complete the self-report measure. The feasibility of measures delivered by touch screen appears to be at least comparable to paper completion questionnaires. Discussion When validation studies are undertaken, justification for measure selection is provided by the study author very infrequently. This suggests that choice of measure for validation is often arbitrary, based on familiarity, personal preference, or perceived relevance. Arguably, selection of measures for validation should be based on an absence of inconsistencies in existing validation data. Some measures identified in this review have been subjected to closely replicated validation studies as many as 23 times within an oncology population (), suggesting that absence of existing data is not determining selection. Although it has been possible to identify measures that have adequate validation data available (detailed below) to support their selection at specified stages of the cancer patient trajectory, these validation data should be considered in the context of the above. Within the pretreatment population, the performed well for identifying adjustment disorder, and major and minor depression, although suggested cut-off scores for the same target disorder differed. No attempts were made in any validation studies included here to explore the validity of the to identify anxiety disorder. The feasibility data extracted suggest that the is a highly feasible measure that is generally acceptable to patients and takes less than five minutes to complete. In the active treatment phase, the MHI-5 and the were validated against the Present State Examination and performed well in combination. Although both measures are highly feasible and acceptable to patients independently, using the combined approach may introduce some feasibility issues relating to time burden for both patients and staff. During the post-treatment phase, the (cut-off score 10), validated against with a target disorder of major depression, was the only measure that had adequate validation data available to support its selection. The cutoff score within this population was significantly

10 Measure GHQ-12 BAI BDI-II Kellner s Symptom Questionnaire Mental Adjustment to Cancer Profile of Mood States Study Participants (Trajectory Stage and Sample Size) 131 Mixed cancer site patients postchemotherapy 131 Mixed cancer site patients postchemotherapy 197 Breast cancer patients; 1/2 sample before start of chemotherapy, 1/2 sample at follow-up postchemotherapy 207 Breast cancer patients 8 weeks postsurgery, mastectomy/ lumpectomy 207 Breast cancer patients 8 weeks postsurgery, mastectomy/ lumpectomy One day to six months postmastectomy/ lumpectomy 74 Early stage breast cancer patients three or more years postsurgery Table 3 Post-Treatment First Author/ Country Purpose in Study Items/Structure Sensitivity Specificity PPV NPV Cut-Off Atesci 25 / Atesci 25 / Costantini 15 / Italy Dausch 26 / USA Dausch 26 / USA Grandi 27 / Italy Matsuoka 28 / 74 Early stage breast Matsuoka 28 / cancer patients three or more years postsurgery Identify psychiatric disorder Identify psychiatric disorder Identify depression (for purposes of this study) To assess significant distress To assess significant distress Depression, anxiety, anger-hostility, and somatic symptoms To examine the possibility of using cancer-related intrusive thoughts (CITs) as an indicator of psychological distress and adjustment after breast surgery To examine the possibility of using CITs as an indicator of psychological distress and adjustment after breast surgery Standardized Clinical Interview Likert (0123) 10 Likert scale, 14 items Likert scale, 14 items, two subscales 13 Items covering depressive symptoms 92 Items, most items yes/no response 40-Item scale developed in UK to determine adjustment to cancer 65-Item selfrating scale 10 Anxiety/7 depression Whole scale 0.83 Any depressive disorder 0.75 Any anxiety disorder 0.51 Any depressive disorder 0.60 Any anxiety disorder 0.30 w w Ziegler et al. Vol. 41 No. 3 March 2011

11 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory Itemsd7 in intrusion subscale and 8 in avoidant To examine the possibility of using CITs as an indicator of psychological distress and adjustment after breast surgery Matsuoka 28 / 74 Early stage breast cancer patients three or more years postsurgery Impact of Events Scale 9 > On depression subscale 11 > On anxiety subscale Likert scale, 14 items For screening of depression and anxiety in nonpsychiatric samples Singer 29 / Germany 189 Laryngectomy patients who have undergone treatment 1971e present ¼ Hospital Anxiety and Depression Scale; BAI ¼ Beck Anxiety Inventory; BDI ¼ Beck Depression Inventory; GHQ-12 ¼ General Health Questionnaire-12 items; ¼ Structured Clinical Interview for DSM-III-R; PPV ¼ Positive Predictive Value; NPV ¼ Negative Predictive Value. lower than in the pretreatment population, which may have been necessary to account for the compounding physical side effects of treatment. Within the recurrent disease/palliative phase of illness, the performance of the was inferior to its performance in the pretreatment phase, with sensitivity scores typically below the required threshold. The reason for this may be related to physical morbidity and the need for patients to respond to the questions by relating their present feelings and well-being to normal function. When patients are receiving palliative care only, the concept of normal function can be particularly distorted. The palliative phase studies largely recruited patients as consecutive admissions to hospice care, either as inpatients or day patients. The samples in this review are, therefore, more representative of palliative phase patients who are engaged with specialist hospice services rather than palliative patients who may be experiencing fewer symptoms. Within the recurrent disease/palliative phase of illness, the short form of the BEDS (six-item version) with a cut-off score of 6 performed well for identifying depression. Measures delivered by touch screen demonstrated good feasibility at this trajectory stage and showed promise for identifying distress, although further validation data is required from future studies to support their selection. Although no previous reviews have the performance of measures for psychological distress by cancer trajectory stage, it is useful to explore the scope and findings of some closely related reviews. Two previous reviews have comprehensively explored the performance of self-report measures for psychological distress and the findings in the context of a cancer population 7 and a palliative care population. 6 In the most recent review, 7 the criterion measure was not, as it is here, required to be a for psychiatric disorder and this enabled a significant number of studies exploring the performance of ultra-short measures to be explored. A single-item question Are you anxious? was studied as a screening tool for psychological distress in palliative care, but showed insufficient specificity to rule out nonanxious patients. The Distress Thermometer

12 Measure Two single items: Are you depressed? and Have you lost interest? Distress Thermometer (via touch screen) Brief Symptom Inventory-18 (via touch screen) Table 4 During Palliative Care Study Participants (Trajectory Stage and Sample Size) Author Purpose in Study Items/Structure Sensitivity Specificity PPV NPV Cut-Off 209 Consecutive patients recruited at time of registration with palliative care unit 209 Consecutive patients recruited at time of registration with palliative care unit 77 Mixed cancer site new hospice admissions 100 Palliative care patients with a prognosis of six months or less 150 Consecutive patients with advanced disease referred for inpatient or outpatient care 150 Consecutive patients with advanced disease referred for inpatient or outpatient care Akechi 30 / Akechi 30 / Le Fevre 31 / UK Lloyd- Williams 32 / UK Thekkumpurath 33 / UK Thekkumpurath 33 / UK Widely validated measure for screening anxiety and depression Brief measure for identifying depression Explore usefulness as a screening instrument for the detection of psychiatric disorders in a palliative care inpatient setting Explore suitability to identify depression Explore utility to identify distress in palliative care Explore utility to identify distress in palliative care Likert scale, 14 items Two single items MDD and AD -T DS 0.78 MDD HDS-T DS Standardized Clinical Interview w As previous Revised Clinical Interview Schedule As previous w 19 Whole score A single-item visual analogue scale 18 Items (cancerspecific scale) divided into three dimensions: depression, anxiety, and somatoform. Likert scale Cut-off $ Cut-off $ Cut-off $ Cut-off $ Cut-off $ Cut-off $ $4 $5 $6 $61 $62 $63 PSE SCAN SCAN 630 Ziegler et al. Vol. 41 No. 3 March 2011

13 GHQ-12 (via touch screen) Brief Edinburgh Depression Scale Beck Depression Inventory- Short Form Mental Adjustment to Cancer EORTC QLQ-C30 EORTC QLQ- BR Consecutive patients with advanced disease referred for inpatient or outpatient care 100 Palliative care patients with a prognosis of six months or less 227 Metastatic breast cancer patients 72 Patients at first recurrence of breast cancer 72 Patients at first recurrence of breast cancer 72 Patients at first recurrence of breast cancer Thekkumpurath 33 / UK Lloyd- Williams 34 / UK Love 20 / Australia Okamura 35 / Okamura 35 / Okamura 35 / Explore utility to identify distress in palliative care Short measure of depression Accuracy for identifying defined major and minor depression was examined To assess psychological response to having cancer 30-Item self-report questionnaire covering functional and symptom-related aspects of QOL for cancer patients The breast cancer module, 23 questions assessing disease symptoms, adverse treatment events, body image, and future perspective Used to screen for psychological morbidity and capture the construct of distress. Each item is scored on a scale of increasing severity (0e4) Originally a 10-item short scale adapted to a six-item scale for use in a palliative population 13 Items covering depressive symptoms 40-Item self-rating scale consisting of five subscales (fighting spirit, anxious preoccupation, fatalism, helplessness/ hopelessness, and avoidance) Multidimensional questionnaire consisting of nine scales 23 Questions, self-rating Cut-off $ Cut-off $ Cut-off $ $4 $5 $6 SCAN PSE Depression 0.84 MDD here here here DSM-II-R þ IV ¼ Hospital Anxiety and Depression Scale; GHQ-12 ¼ General Health Questionnaire-12 items; EORTC QLQ-C30 ¼ European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-30 items; ¼ Structured Clinical Interview for DSM-III-R; PPV ¼ Positive Predictive Value; NPV ¼ Negative Predictive Value; PSE ¼ Present State Examination; SCAN ¼ Schedules for Clinical Assessment in Neuropsychiatry; MDD ¼ Major Depressive Disorder; DSM-IV ¼ Diagnostic and Statistical Manual of Mental Disorders, 4th ed; QOL ¼ quality of life. Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 631

14 Table 5 Feasibility Data Available on Self-Report Measures Validated in Palliative/Recurrent Disease Populations Measure Time to Complete Response Structure/Number of Items Mode of Completion Attrition Rates and Recruitment Five minutes 14 Items in total; two subscales of seven items each. Two single items Distress Thermometer Approx. one to two minutes Approx. one minute BSI-18 Approx. 10 minutes Paper-based questionnaire delivered in hospice setting. There was no requirement to complete other self-report measures alongside the in Le Fevre et al. s study, 31 although in Akechi et al. s 30 and Lloyd-Williams et al. s 32 study, the was completed alongside the DT and BDI, respectively. In Le Fevre s 31 validation study of the, only 79 were recruited out of 282 patients approached (refused n ¼ 69, too unwell n ¼ 134). In Lloyd-Williams et al. s validation study of the, of 100 patients approached participated. In Akechi et al. s 30 validation study of the, out of a possible 704 participants, only 209 participated and only 85 completed the study. Two questions, yes/no response. Paper based and completed alongside. In Akechi et al. s 30 validation study of the and two single-item questions, out of a possible 704 participants, only 209 participated; 85 completed the study. An 11-point visual analogue scale (0 ¼ no distress 10 ¼ extreme distress). 18-Item multidimensional questionnaire specifically for use in a cancer population. There are three dimensions: depression, anxiety, and somatoform. Total score represents a global severity index. GHQ-12 Five minutes A short version of the General Health Questionnaire designed for use in the general population to screen for psychological morbidity and capture the construct of distress. BEDS Approx. three minutes A 10-item brief measure for depression, originally developed as a measure for postnatal depression but here adapted in terms of cut-off score to identify depression within a palliative care population. Touch screen computer, completed at patient s bedside. Other measures completed were BSI-18 and GHQ-12. Touch screen computer, completed at patient s bedside. Other measures completed were DT and GHQ-12. Touch screen computer, completed at patient s bedside. Other measures completed were DT and BSI-18. Self-report. In Thekkumpurath et al. s study, consecutive referrals to inpatient or outpatient palliative care services were approached, 174 consented and 150 completed the study. Total attrition n ¼ 24 (too ill: 14; too drowsy: 4; lacks capacity: 3; physical disability: 2; unknown: 1). As above. As above. Two hundred twenty-six patients participated in the study, 47% of those eligible (485) to take part. Reasons given by nonparticipants (some gave multiple responses) were too busy (61; 27%); coping satisfactorily (41; 18%); health and the demands of treatment (46; 20%); practical issues (35; 16%); not being a group person (20; 9%); and wanting to get on with life (22; 10%). The remaining 55 (24%) gave assorted other reasons. Forty-one patients declined to participate and 20 patients were excluded because of cognitive impairment. At the end of the study period, 190 (77.2%) patients had died. Prevalence of depression identified by the PSE was 30%. 632 Ziegler et al. Vol. 41 No. 3 March 2011

15 Vol. 41 No. 3 March 2011 Identifying Psychological Distress During the Cancer Illness Trajectory 633 Self-report delivered alongside. As above. The BDI-SF has 13 items covering depressive symptoms. Patients rate (0e3) the severity of each symptom over a seven-day period and the results are summed. A score above 4 is considered indicative. BDI-SF Approx. 5e10 minutes BDI-SF ¼ Beck Depression Inventory-Short Form; ¼ Hospital Anxiety and Depression Scale; GHQ-12 ¼ General Health Questionnaire-12 items; BSI-18 ¼ Brief Symptom Inventory-18; BDI ¼ Beck Depression Inventory; BEDS ¼ Brief Edinburgh Depression Scale. was in 15 validation studies, 11 of which used the as a criterion measure. Overall congruence was as weak to moderate, although some studies good performance, including good reliability and sensitivity to change. This illustrates that when data on measure performance are derived from validation studies that include all cancer sites and stages, pooling the data and reporting on the measure s performance across the population as a whole is complex. It is not always easy to interpret the findings to determine a measure s potential application for a specific cancer site or stage of disease. The second review 6 on measures for psychological distress validated against a psychiatric interview within a palliative care population. Our findings support their recommendation that the BEDS showed promise as a screening tool for depression within the palliative phase. The authors recommended that the, if used within this population to identify psychological distress, should be calculated as a whole scale score. We acknowledge limitations in this work. Determining the suitability of measures for use at specific points of the cancer patient trajectory was challenging because of the diversity in available data. Some authors defined their study populations by trajectory stage but did not use a common approach to splitting the trajectory. In the absence of a universal definition of cancer patient trajectory, the definition of trajectory used in this review was informed by Feuerstein. 14 In studies where a trajectory stage could not be assigned, the data will be by cancer site and stage in a subsequent paper. The average sample size of included studies is 161 (range 31e503). The relatively small sample sizes may relate to all included studies conducting a Standardized Clinical Interview alongside data collection by self-report measure. The time burden of conducting a Standardized Clinical Interview makes larger sample sizes generally less feasible. The absence of a framework for the appraisal of self-report measures (such that exists for the appraisal of research studies) hinders the objective assessment of measures and made it difficult to determine criteria on which to compare measures. Given that instruments of this type are in widespread use, it is problematic that there is not an established

16 634 Ziegler et al. Vol. 41 No. 3 March 2011 format for evaluating the soundness of their construction and measurement. Sensitivity and specificity scores provided a good indication of the measure s utility as a screening instrument; however, the value of such data in terms of judging the measure s suitability for use in a longitudinal study is less clear. The was to be a valid measure at two of the four trajectory stages. However, because there is a charge associated with its use, financial constraints might make it a less feasible choice for some. The different cancer sites are not well represented within the populations used to validate the self-report measures. Most studies are in breast cancer patients, one is in head and neck cancer patients, and the remaining are in populations of mixed cancer sites. Ultra-short measures, such as single- or twoitem questions and visual analogue scales such as the Distress Thermometer, 6 are increasingly popular. These are not well represented here as they are most frequently validated against other self-report measures (such as ), rather than against the Standardized Clinical Interview. Furthermore, in studies where this was not the case, it was usually not possible to assign a trajectory stage to the study sample. Conclusions A number of factors presented challenges when attempting to draw accurate conclusions about the most appropriate measures to use at each trajectory stage. The validation data presented were variable, and measures were validated against different target disorders (depression, distress, adjustment disorder) and in different populations in terms of cancer site. This made accurate comparisons across measures difficult. Nonetheless, best performing measures for specified target disorders could be identified for each trajectory stage. For pretreatment, the with a cut-off of 7 (depression subscale) or 15 (whole scale) performed well for identifying major depressive disorder. During active treatment, the and MHI-5 together performed best; a cut-off >11 on MHI-5 followed by the >9 (¼case) indicated clinically significant distress. The GHQ-28 (cut-off 10) also performed well for identifying psychosocial morbidity, but reservations about whether the data should be viewed as contemporaneous have been stated earlier. At the post-treatment phase, the with a cut-off between 9 and 11 on the depression subscale was useful for identifying depression. At the recurrence/palliative stage, the six-item version of the Edinburgh Depression Scale with a cut-off score of 6 performed well for identifying depression. No single measure had evidence to support use throughout the trajectory in a longitudinal study, but the during pretreatment followed by in combination with MHI-5 was supported during the cancer treatment phase and BEDS in the palliative care phase. There were clear knowledge gaps identified in the validated assessment of anxiety, adjustment and undifferentiated distress pretreatment, depression during treatment, and anxiety and distress both after treatment and at recurrence. These areas should be prioritized for future specific study. We propose a longitudinal study of psychological distress in cancer using the alongside the MHI and BEDS at the end of life. Disclosures and Acknowledgments This review was funded by the COMPASS Collaborative, a National Cancer Research Institute Supportive and Palliative Care Research Collaborative. None of the authors have any personal, professional, or financial conflicts of interest to declare. References 1. Holland JC. NCCN practice guidelines for the management of psychosocial distress. Oncology 1999;13:113e Portenoy RK, Thaler HT, Kornblith AB, et al. The Memorial Symptom Assessment Scale: an instrument for the evaluation of symptom prevalence, characteristics and distress. Eur J Cancer 1994;30A: 1326e Breitbart W, Bruera E, Chochinov H, Lynch M. Neuropsychiatric syndromes and psychological symptoms in patients with advanced cancer. J Pain Symptom Manage 1995;10:131e Hotopf M, Chidgey J, Addington-Hall J, Ly KL. Depression in advanced disease: a systematic review. Part 1: prevalence and case finding. Palliat Med 2002;16:81e Gilbody SM, House AO, Sheldon TA. Routinely administered questionnaires for depression and anxiety: systematic review. BMJ 2001;322:406e409.