Review. A substantial body of evidence supports the short- and

Transcription

1 DEPRESSION AND ANXIETY 28 : (2011) Review RELATIVE EFFECTS OF CBT AND PHARMACOTHERAPY IN DEPRESSION VERSUS ANXIETY: IS MEDICATION SOMEWHAT BETTER FOR DEPRESSION, AND CBT SOMEWHAT BETTER FOR ANXIETY? Babak Roshanaei-Moghaddam, M.D., Michael C. Pauly, M.D., David C. Atkins, Ph.D., Scott A. Baldwin, Ph.D., Murray B. Stein, M.D. M.P.H., and Peter Roy-Byrne, M.D. Background: Little is known about whether cognitive behavioral therapy (CBT) or pharmacotherapy is relatively more advantageous for depressive versus anxiety disorders. Methods: We conducted a meta-analysis wherein we searched electronic databases and references to select randomized controlled studies comparing CBT and pharmacotherapy, with or without placebo, in adults with major depressive or anxiety disorders. The primary effect size was calculated from disorder-specific outcome measures as the difference between CBT and pharmacotherapy outcomes (i.e., positive effect size favors CBT; negative effect size favors pharmacotherapy). Results: Twenty-one anxiety (N 5 1,266) and twenty-one depression (N 5 2,027) studies comparing medication to CBT were included. Including all anxiety disorders, the overall effect size was.25 (95% CI: 0.02, 0.55, P 5.07). Effects for panic disorder significantly favored CBT over medications (.50, 95% CI: 0.02, 0.98). Obsessive compulsive disorder showed similar effects-sizes, though not statistically significant (.49, 95% CI: 0.11, 1.09). Medications showed a nonsignificant advantage for social anxiety disorder (.22, 95% CI: 0.50, 0.06). The overall effect size for depression studies was.05 (95% CI: 0.09, 0.19), with no advantage for medications or CBT. Pooling anxiety disorder and depression studies, the omnibus comparison of the relative difference between anxiety and depression in effectiveness for CBT versus pharmacotherapy pointed to a nonsignificant advantage for CBT in anxiety versus depression (B 5.14, 95% CI: 0.14, 0.43). Conclusions: On balance, the evidence presented here indicates that there are at most very modest differences in effects of CBT versus pharmacotherapy in the treatment of anxiety versus depressive disorders. There seems to be larger differences between the anxiety disorders in terms of their relative responsiveness to pharmacotherapy versus CBT. 28: , r 2011 Wiley-Liss, Inc. Key words: meta-analysis; CBT; pharmacotherapy; depression; anxiety VA Puget Sound Health Care System, Seattle, Washington INTRODUCTION A substantial body of evidence supports the short- and long-term efficacy of pharmacotherapy and cognitive behavioral treatments for both mood [1 3] and anxiety [2,4 6] disorders. Although there has been considerable empirical research and debate on the comparative efficacy of the two modalities, on average the extant data show comparable efficacy for both modalities in the short term, [7 10] except for obsessive compulsive disorder The authors report they have no financial relationships within the past 3 years to disclose. Correspondence to: Babak Roshanaei-Moghaddam, VA Puget Sound Health Care System, (S-116 ATC) 1660 S. Columbian Wy., Seattle, Washington. babakr@uw.edu Received for publication 21 January 2011; Revised 9 April 2011; Accepted 18 April 2011 DOI /da Published online 23 May 2011 in Wiley Online Library (wiley onlinelibrary.com). r 2011 Wiley-Liss, Inc.

2 Review: Relative Effects of CBT and Pharmacotherapy 561 (OCD), where behavioral treatment seems to be superior. [11] Although some speculate that medication is better than CBT for more severe depressive illness, usually defined as greater symptom severity, such a difference has not been consistently demonstrated. [7,12,13] This issue has been infrequently studied with the anxiety disorders, but some studies have failed to show that severity predicts poorer antidepressant [14] or CBT [15] response. Even fewer long-term studies compare the relative efficacy of CBTand medications in depressive and anxiety disorders. Furthermore, these studies, which tend to show more favorable results for CBT over pharmacotherapy, are difficult to interpret because of their design. Specifically, in these designs, both treatments are discontinued, despite the fact that the two modalities have different carry-over effects (i.e., subjects learn and change their behavior with CBT and this does not necessarily go away as abruptly as do the effects of discontinued medication). In contrast to data showing comparable short-term effects of CBT and pharmacotherapy, there has been little attention to whether one modality is relatively more advantageous for depression versus anxiety. These comparisons are difficult to make because treatment studies of depressive and anxiety disorders often differ with respect to their exclusion criteria for anxiety and depressive disorder comorbidity. Some research suggests that antidepressant medications are less effective for depression with comorbid anxiety. [16] In contrast, the evidence that comorbid depression adversely affects medication response with various anxiety disorders is less consistent. [17] Studies of CBT in various anxiety disorders have generally not found that depressive comorbidity adversely impacts effectiveness. [18 20] We are not aware of any CBT depression treatment studies in adults that have examined the potential moderating effects of comorbid anxiety. Although there are also no data comparing the relative effects of CBT in depression versus anxiety, some reviews and meta-analyses have suggested that there may be more robust evidence for CBT benefits in anxiety versus depression, [21] and that CBT effects in anxiety may be more specific [2] (i.e., effective compared to other active psychotherapeutic treatments) than those effects in depression (at least for adults). This meta-analysis sought to determine the relative effects of CBT versus medication in depression versus anxiety disorders. Although the group of anxiety disorders differ in terms of phenomenology, epidemiology, and possibly pathogenesis, there is broad uniformity in the type of treatments (i.e., antidepressants and CBT) that are effective for these disorders, [22] justifying our pooling of the different disorders for our initial analysis. A secondary aim was to examine whether comorbidity with the other class of disorder (i.e., anxiety or depression) moderated effect sizes. A number of factors were considered in sensitivity analyses as potential moderators. We focused only on short-term studies, as there are fewer longer-term studies, and their designs vary significantly. We reviewed studies that have directly compared the two treatment modalities to each other and include those with and without placebo controls. We extracted effect sizes from studies of major depression, panic, generalized anxiety (GAD), social anxiety, OCD, and posttraumatic stress (PTSD) disorders. METHOD STUDY IDENTIFICATION AND SELECTION To identify relevant studies, we searched computerized databases including PubMed (1900,) Embase (775,) CINAHL (including PsycINFO (318)), and Cochrane Central Register of Controlled Trials (276) from inception to July Our search strategy involved the combination of terms indicative of psychological treatments (psychotherapy, CBT, cognitive therapy, behavioral therapy, exposure), biological treatments [SSRI, SNRI, MAOI, TCA, antidepressant, psychotropic, benzodiazepine, anxiolytic, psychopharmacolog, medication, (including the names of 30 most commonly used antidepressants and anxiolytics)], and psychiatric disorders [depression, anxiety, OCD, panic, PTSD, social phobia, and social anxiety (both MeSH-terms and textwords.)] Additional studies were identified by examining the reference list of relevant meta-analyses. Studies included were: (a) short-term (4 21 weeks), (b) randomized trials, (c) in which the effects of behaviorally oriented psychotherapy (CBT, cognitive therapy, behavior therapy, exposure therapy, or problem solving) were compared to the effects of pharmacotherapy (antidepressant or anxiolytic), (d) with or without a placebo arm, (e) in adults, (f) with major depressive disorder or one of five major anxiety disorders (GAD, social anxiety disorder, panic disorder with or without agoraphobia, OCD, or PTSD), (g) provided data using a disorderspecific rather than generic rating scale, and (h) that provided sufficient data to calculate effect sizes. Studies with older patients with Mini Mental State Examination scores of less than 24 were excluded. Manuscripts written in any language other than English were excluded. We also excluded studies aimed at relapse prevention. Comorbid general medical or psychiatric disorders were not used as exclusion criteria. The search process is depicted in Figure 1. META-ANALYSES Our primary research question focused on the relative efficacy of CBT compared to pharmacotherapy. Thus, the primary effect size was the standardized mean difference, defined as the posttreatment difference in outcome between therapy and medication divided by the pooled standard deviation of the posttreatment outcome. We applied Hedges and Olkin s small sample correction. [23] Three studies did not report means and standard deviations so we used alternative formulae to estimate effect sizes based on available data (e.g., estimating effect sizes using t-statistics or odds-ratios for categorical outcomes). [24] The majority of studies reported more than one outcome measure, and relatively few studies reported a prespecified primary outcome variable. Thus, we focused on disorder-specific outcome measures, selecting the most commonly used measures for treatment studies in these disorders (e.g., Y-BOCS for OCD, Hamilton depression rating scale for depression). Plots of effect sizes of individual measures within each study were examined for the possibility of outliers for a particular measure within studies; there were no concerns based on this screening, and all effect sizes based on disorder-specific outcomes within a study were averaged, yielding a single effect size per study. Meta-analyses used a random-effects model, which is appropriate when there is notable heterogeneity of effect sizes and also conceptually reflects that the included studies represent a sample from a larger population of studies. [25] We examined therapy versus

3 562 Roshanaei-Moghaddam et al. Initial screening for potentially relevant studies PubMed: N=1900 Embase: N=775 CINAHL (including PsycINFO:) N=318 Cochrane: N=276 Total: N=3269 Retrieved manuscripts for thorough review N=97 Studies excluded based on title or abstract N=3172 Excluded studies Follow-up or relapse prevention: N=25 Methodology not meeting our inclusion criteria: N=9 No CBT: N=10 (IPT: N=7; Couples: N=2; Social Skills: N=1) Not enough data to calculate effect sizes: N=4 Non-specific measures: N=2 Other medications: N=2 (Buspar and Atenolol) Non-standard treatment: N=1 (Alprazolam for depression) Adolescents: N=1 Mixed depressive disorders: N=1 Total: N=55 Included studies Major depressive disorder: N=21 Anxiety disorders: N=21 Social anxiety disorder: N=7 Panic disorder: N=9 OCD: N=3 GAD: N=1 PTSD: N=1 Total: N=42 Figure 1. Flow diagram of randomized controlled trials included and excluded in meta-analysis. medication differences within each disorder (i.e., testing the overall effect size against zero to answer the question: Is there evidence of superiority within either depression or anxiety?) as well as an overall pooled analysis in which the relative efficacy across disorders was tested (i.e., to answer the question: Is there evidence of superiority of either treatment modality for depression versus anxiety?). In addition, analyses of anxiety disorders considered them both as a single class, as well as individual disorders (i.e., pooling them in a single analyses versus treating them separately). To assess heterogeneity, the I 2 statistic was used, [26] which expresses the percentage of the total variability due to heterogeneity in effect sizes (the remaining variability is due to sampling error). MODERATORS, SENSITIVITY ANALYSES, AND INFLUENCE ANALYSES Moderators were examined using a mixed-effects model for metaregression. [23] (Note that in meta-regression the outcome is typically a difference between two treatments and hence all covariates in a meta-regression are potential moderators of the treatment difference, even though they may not interact with other covariates in the metaregression.) The key moderator was a categorical predictor of disorder (i.e., pooling all effect sizes in a single analysis, asking the question: Is disorder a significant covariate?). In addition, the secondary aim focused on whether comorbidity of the alternative

4 Review: Relative Effects of CBT and Pharmacotherapy 563 disorder moderated outcome (i.e., comorbidity of anxiety in depression studies and of depression in anxiety studies). Additional meta-regression analyses were used to examine the sensitivity of findings to various study characteristics and possibly elucidate the source of observed heterogeneity. The focus of these analyses is on potential confounds of observed differences between pharmacotherapy and CBT treatment response in anxiety disorders versus depression. Thus, potential confounds explored in sensitivity analyses included: demographics (average age, percent Caucasian, percent female), medication type (SSRI, TCA, MAOI, benzodiazepine), amount of contact with study therapists or psychiatrists (combining number and length of sessions), placebo-controlled trial (yes/no), single versus multi-site study, whether clinical ratings were double-blind, funding source (pharmaceutical funding or not), allegiance of lead author to CBT versus pharmacotherapy (by degree: Ph.D. assumed allegiance to CBT versus M.D. assumed allegiance to pharmacotherapy), and exclusionary criteria (as indicated by exclusion for various psychiatric and medical comorbidities). Potential confounds were individually added to primary models (i.e., a separate model was run for each confound). Although P-values for the confounds are likely liberal due to multiple comparisons, this was acceptable given the focus of these analyses for potential confounds. Finally, potential reporting bias was examined using funnel plots, and influence was examined via jack-knife analyses (i.e., estimating overall effect iteratively leaving out single studies), studentized residuals, and Cook s D statistic. Statistical significance was based on a two-sided P-value of.05, though effects with P-values between.05 and.10 are also noted given the relatively small sample sizes for some anxiety disorders. All analyses were done in R v [27] and made extensive use of the metafor package for meta-analyses and meta-regression. [28] RESULTS Based on the study selection criteria noted earlier, 21 anxiety studies [8,11,29 47] were included with 625 patients randomized to receive medication and 641 patients randomized to receive therapy. For depression studies, 21 studies were included [7,48 67] with 1,095 patients randomized to medication and 932 patients randomized to therapy. Forest plots of posttreatment CBT minus medication effect sizes are shown in Figures 2 and 3 for anxiety and depression, respectively. As seen in Figure 2, the overall effect size for anxiety disorder studies based on a random-effects metaanalysis was.25 (95% CI: 0.02, 0.55, P 5.07). Thus, there is some evidence that CBT has an advantage over pharmacotherapy across all anxiety disorders included, though this is above the typical value for statistical significance. In addition, there are notable differences by specific anxiety disorder: effects for panic disorder significantly favor CBTover medications (.50, 95% CI: 0.02, 0.98), and OCD shows generally similar effect sizes favoring CBT, though not significantly different from zero (.49, 95% CI: 0.11, 1.09). On the other hand, medications show some advantage (but not statistically significant) for social anxiety disorder (.22, 95% CI: 0.50, 0.06). A post hoc contrast of social anxiety with the other anxiety disorders revealed that it is significantly different from other anxiety disorders in the comparison of CBT and medication, Figure 2. Forest plot of standardized mean differences between medication and therapy at posttreatment for anxiety studies, by disorder and overall (N 5 21 studies). Reference line at zero (solid line) represents an effect size of 0 (i.e., no effect), and the dotted line is at the pooled effect size for all studies. favoring medication (B 5.72, 95% CI: 1.22, 0.22). GAD and PTSD are challenging to evaluate on their own as there are only single studies for each of those disorders. Consistent with the above results, there was notable heterogeneity in effect sizes within anxiety disorders as a whole (I , 95% CI: 69.3, 91.4), as well as within the specific anxiety disorders. Figure 3 reveals that the overall effect size for depression studies was.05 (95% CI: 0.09, 0.19), indicating no advantage for medications or therapy. Like anxiety studies, there was evidence for heterogeneity in effect sizes (I ; 95% CI: 20.2, 82.8). The relative difference in effectiveness for CBT versus pharmacotherapy between anxiety and depression was tested in a pooled meta-regression using an indicator variable for disorder (depression 0, anxiety 1) and a

5 564 Roshanaei-Moghaddam et al. Figure 3. Forest plot of standardized mean differences between medication and therapy at posttreatment for depression studies (N 5 21 studies). Reference line at zero (solid line) represents an effect size of 0 (i.e., no effect), and the dotted line is at the pooled effect size for all studies. second analysis with each anxiety disorder separately coded (i.e., contrasts between depression and each anxiety disorder). The omnibus comparison pointed to an advantage for CBT in anxiety versus depressive disorders, though this did not reach statistical significance (B 5.14, 95% CI: 0.14, 0.43). In examining contrasts between depression and specific anxiety disorders (for therapy versus medication), there were significantly superior effects for CBT versus medications in panic disorder versus depression (B 5.36, 95% CI: 0.01, 0.71) with a nonsignificant difference for OCD (B 5.43, 95% CI: 0.13, 1.00). The majority of anxiety studies excluded patients with comorbid depression (15/20 excluded with one study not reporting on comorbid disorders), whereas approximately half of depression studies excluded patients with comorbid anxiety disorders (10/21 with six studies not reporting on comorbid disorders). In a meta-regression with comorbid depression (0 excluded, 1 included) as a covariate, there was a negative effect for including depressive disorders that was very close to significance (B 5.57, 95% CI: 1.25, 0.10, P 5.09), suggesting that anxiety disorder studies including comorbid depression favored pharmacotherapy. An important point is that the studies including comorbid depression were not distributed across all anxiety disorders, with 2/5 for social anxiety and 3/5 for panic disorder. Among anxiety disorder studies that did exclude patients with comorbid depression, the overall effect size, in favor of CBT, was.44 (95% CI: 0.05, 0.83). For depression studies, comorbid anxiety did not significantly affect the results (B 5.17, 95% CI: 0.14, 0.48). As noted in the Methods, a variety of study factors were examined as potential confounds in sensitivity analyses. Based on these analyses, the strongest association was for year of study among depression studies (B 5.023, 95% CI: 0.036, 0.010), which can also be seen in Figure 2. Earlier studies were more favorable for CBT, whereas more recent studies have been more favorable for pharmacotherapy. No other potential confounds tested in the sensitivity analyses were significantly related to effect size. Finally, models were scrutinized for potential outliers and influential studies. Among the anxiety studies, the Marks et al. [41] study was identified as having a strong influence on the results via studentized residuals, Cook s D, and jack-knife (i.e., leave one out ) analyses, which is also readily apparent in Figure 1. The authors re-reviewed this study for possible study flaws, or more generally, idiosyncrasies that might clarify its unusually strong effect for CBT over medication, but no clear differences were found. Excluding this study does weaken some of the anxiety findings (e.g., overall anxiety effect size 5.14, 95% CI: 0.09, 0.37), but does not fundamentally alter the basic conclusions of the analyses reported earlier. DISCUSSION These findings do not confirm our initial hypothesis that patients with anxiety disorders would show relatively greater effects from CBT compared to medication, whereas patients with depression would show the opposite trend. Evidence for the former effect (a modest effect size favoring CBT for the aggregate group of anxiety disorder patients compared with depression patients) was not statistically significant (P 5.07), though the effect sizes for one anxiety disorder, panic, was significantly different (i.e., favored CBT) than that seen in patients with depression. The substantial heterogeneity of effect sizes observed in the different anxiety disorders is largely accounted for by the divergent results seen in patients with social anxiety disorder. Patients with this diagnosis showed the strongest trend in favor of benefiting from medication compared with CBT, though this was not statistically significant. Patients with depression showed relatively similar effects for both treatment modalities. The findings of any meta-analysis depend on a number of assumptions and decisions and these constitute many of the limitations of this study. These include: decisions about which population to focus on (we elected to focus only on adults); decisions about which studies to include (we included only studies in which both modalities were compared to each other,

6 Review: Relative Effects of CBT and Pharmacotherapy 565 either with or without a placebo condition, and focused only on studies of short term outcome), and decisions about which outcome measures to include (to accommodate wide variability in the types of assessment measures used we elected to focus first on disorder specific scales to compute our standardized effect sizes, but we also recorded effects sizes for other measures provided and because we did not find any significant differences, we report the pooled or average effect). In addition, there were too few studies of GAD and PTSD to draw conclusions about these disorders. Finally, treatments in each of these studies may not have been totally optimized (e.g., highest doses of medication, most well trained and skilled CBT therapists). We examined potential moderator effects and found relatively few. Effects were similar in studies with and without placebo controls. Demographics, type of medication used, and amount of therapist contact were unrelated to effects. Various other aspects of study design (e.g., blinded assessment, placebo controls) were also unrelated to effect. There was a strong moderating effect for year of study for depression with earlier studies showing much greater effects for CBT than later studies. Our most interesting moderator analysis concerned the effect of comorbidity on effect sizes. Overall, the majority of studies excluded the comorbid class of disorder (i.e., either anxiety or depression), but four anxiety and five depression studies did include the other comorbid class of disorder. This moderator analysis showed that patients in anxiety studies that included comorbid depression were more likely to benefit from pharmacotherapy, whereas patients in anxiety studies that excluded depression were more likely to benefit from CBT. However, the presence of comorbid anxiety in depression studies had no moderating effect on outcome. With increased emphasis on effectiveness and implementation trials, future studies are likely to have a greater incidence of comorbidity and hence, future research should explore whether these findings continue to be observed in new samples. The strong heterogeneity of effects within the group of anxiety disorders is noteworthy. CBT was more effective than medication for panic disorder (with OCD showing a similar, though nonsignificant effect size in this direction), whereas medication was more effective than CBT for social anxiety disorder. Of these three disorders, OCD is distinct from the other two in terms of epidemiology, diagnostic comorbidity, neurobiology, [68] whereas panic and social anxiety disorders have a number of important similarities, These differential treatment response findings are difficult to interpret, and because they focus only on short-term outcome studies, must be taken with caution. On balance, the evidence presented here indicates that there are at best very modest differences in effects of pharmacotherapy versus CBT in the treatment of anxiety versus depressive disorders. Our data indicate, though, that there may be important differences between the anxiety disorders (e.g., panic versus social anxiety disorder) in terms of their relative responsiveness to pharmacotherapy versus CBT. Currently, patient preference and availability of skilled CBT therapists or pharmacotherapists are prime determinants of which treatment modality patients with anxiety and mood disorders receive. Future research efforts are needed to identify individual patient characteristics that might predict preferential response to one or the other of these treatment modalities. REFERENCES 1. Churchill R, Hunot V, Corney R, et al. A systematic review of controlled trials of the effectiveness and cost-effectiveness of brief psychological treatments for depression. Health Technol Assess 2001;5: Chambless DL, Ollendick TH. Empirically supported psychological interventions: controversies and evidence. Annu Rev Psychol 2001;52: Mann JJ. The medical management of depression. N Engl J Med 2005;353: Hofmann SG, Smits JA. Cognitive-behavioral therapy for adult anxiety disorders: a meta-analysis of randomized placebocontrolled trials. J Clin Psychiatry 2008;69: Bandelow B, Seidler-Brandler U, Becker A, Wedekind D, Ruther E. Meta-analysis of randomized controlled comparisons of psychopharmacological and psychological treatments for anxiety disorders. World J Biol Psychiatry 2007;8: Ravindran LN, Stein MB. The pharmacologic treatment of anxiety disorders: a review of progress. J Clin Psychiatry 2010;71: DeRubeis RJ, Hollon SD, Amsterdam JD, et al. Cognitive therapy vs medications in the treatment of moderate to severe depression. Arch Gen Psychiatry 2005;62: Barlow DH, Gorman JM, Shear MK, Woods SW. Cognitivebehavioral therapy, imipramine, or their combination for panic disorder: a randomized controlled trial. J Am Med Assoc 2000;283: Davidson JR, Foa EB, Huppert JD, et al. Fluoxetine, comprehensive cognitive behavioral therapy, and placebo in generalized social phobia. Arch Gen Psychiatry 2004;61: Power KG, Simpson RJ, Swanson V, Wallace LA. Controlled comparison of pharmacological and psychological treatment of generalized anxiety disorder in primary care. Br J Gen Pract 1990;40: Foa EB, Liebowitz MR, Kozak MJ, et al. Randomized, placebocontrolled trial of exposure and ritual prevention, clomipramine, and their combination in the treatment of obsessive-compulsive disorder. Am J Psychiatry 2005;162: Elkin I, Gibbons RD, Shea MT, et al. Initial severity and differential treatment outcome in the National Institute of Mental Health Treatment of Depression Collaborative Research Program. J Consult Clin Psychol 1995;63: DeRubeis RJ, Gelfand LA, Tang TZ, Simons AD. Medications versus cognitive behavior therapy for severely depressed outpatients: mega-analysis of four randomized comparisons. Am J Psychiatry 1999;156: Pollack MH, Rapaport MH, Clary CM, Mardekian J, Wolkow R. Sertraline treatment of panic disorder: response in patients at risk for poor outcome. J Clin Psychiatry 2000;61: Aaronson CJ, Shear MK, Goetz RR, et al. Predictors and time course of response among panic disorder patients treated with cognitive-behavioral therapy. J Clin Psychiatry 2008;69:

7 566 Roshanaei-Moghaddam et al. 16. Fava M, Rush AJ, Alpert JE, et al. Difference in treatment outcome in outpatients with anxious versus nonanxious depression: a STARD report. Am J Psychiatry 2008;165: Overbeek T, Schruers K, Vermetten E, Griez E. Comorbidity of obsessive-compulsive disorder and depression: prevalence, symptom severity, and treatment effect. J Clin Psychiatry 2002;63: Allen LB, White KS, Barlow DH, Shear MK, Gorman JM, Woods SW. Cognitive-behavior therapy (CBT) for panic disorder: relationship of anxiety and depression comorbidity with treatment outcome. J Psychopathol Behav Assess 2010;32: Newman MG, Przeworski A, Fisher AJ, Borkovec TD. Diagnostic comorbidity in adults with generalized anxiety disorder: impact of comorbidity on psychotherapy outcome and impact of psychotherapy on comorbid diagnoses. Behav Ther 2010;41: Storch EA, Lewin AB, Farrell L, et al. Does cognitive-behavioral therapy response among adults with obsessive-compulsive disorder differ as a function of certain comorbidities? J Anxiety Disord 2010;24: Chambless D, Hollon SD. Defining empirically supported therapies. J Clin Consult Clin Psychol 1998;66: Roy-Byrne P, Craske MG, Sullivan G, et al. Delivery of evidencebased treatment for multiple anxiety disorders in primary care: a randomized controlled trial. J Am Med Assoc 2010;303: Hedges LV, Olkin I. Statistical Methods for Meta-analysis. Orlando, FL: Academic Press; Shadish WR, Robinson L, Lu C. ES: A Computer Program and Manual for Effect Size Calculation. Minneapolis, MN: Assessment Systems; Higgins JP, Thompson SG, Spiegelhalter DJ. A re-evaluation of random-effects meta-analysis. J R Stat Soc Ser A Stat Soc 2009;172: Higgins JP, Thompson SG. Quantifying heterogeneity in a metaanalysis. Stat Med 2002;21: R: A language and evironment for statistical computing [Software]. [computer program]. Version. Vienna, Austria: R Foundation for Statistical Computing; Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw 2010;36: Ataoglu A, Ozkan M, Tutkin H, Maras A. Alprazolam and cognitive behavior therapy in the treatment of panic disorder. Turk J Med Sci 2000;30: Bakker A, van Dyck R, Spinhoven P, van Balkom AJ. Paroxetine, clomipramine, and cognitive therapy in the treatment of panic disorder. J Clin Psychiatry 1999;60: Black DW, Wesner R, Gabel J. The abrupt discontinuation of fluvoxamine in patients with panic disorder. J Clin Psychiatry 1993;54: Blanco C, Heimberg RG, Schneier FR, et al. A placebo-controlled trial of phenelzine, cognitive behavioral group therapy, and their combination for social anxiety disorder. Arch Gen Psychiatry 2010;67: Blomhoff S, Haug TT, Hellstrom K, et al. Randomised controlled general practice trial of sertraline, exposure therapy and combined treatment in generalised social phobia. Br J Psychiatry 2001;179: Clark DM, Ehlers A, McManus F, et al. Cognitive therapy versus fluoxetine in generalized social phobia: a randomized placebocontrolled trial. J Consult Clin Psychol 2003;71: Clark DM, Salkovskis PM, Hackmann A, Middleton H, Anastasiades P, Gelder M. A comparison of cognitive therapy, applied relaxation and imipramine in the treatment of panic disorder. Br J Psychiatry 1994;164: Gelernter CS, Uhde TW, Cimbolic P, et al. Cognitive-behavioral and pharmacological treatments of social phobia. A controlled study. Arch Gen Psychiatry 1991;48: Heimberg RG, Liebowitz MR, Hope DA, et al. Cognitive behavioral group therapy vs phenelzine therapy for social phobia: 12-week outcome. Arch Gen Psychiatry 1998;55: Klosko JS, Barlow DH, Tassinari R, Cerny JA. A comparison of alprazolam and behavior therapy in treatment of panic disorder. J Consult Clin Psychol 1990;58: Lindsay WR, Gamsu CV, McLaughlin E, Hood EM, Espie CA. A controlled trial of treatments for generalized anxiety. Br J Clin Psychol 1987;26: Loerch B, Graf-Morgenstern M, Hautzinger M, et al. Randomised placebo-controlled trial of moclobemide, cognitive-behavioural therapy and their combination in panic disorder with agoraphobia. Br J Psychiatry 1999;174: Marks IM, Swinson RP, Basoglu M, et al. Alprazolam and exposure alone and combined in panic disorder with agoraphobia. A controlled study in London and Toronto. Br J Psychiatry 1993;162: Otto MW, Pollack MH, Gould RA, Worthington 3rd JJ, McArdle ET, Rosenbaum JF. A comparison of the efficacy of clonazepam and cognitive-behavioral group therapy for the treatment of social phobia. J Anxiety Disord 2000;14: Rachman S, Cobb J, Grey S, et al. The behavioural treatment of obsessional-compulsive disorders, with and without clomipramine. Behav Res Ther 1979;17: Davidson J, Yaryura-Tobias J, DuPont R, et al. Fluvoxaminecontrolled release formulation for the treatment of generalized social anxiety disorder. J Clin Psychopharmacol 2004;24: Frommberger U, Stieglitz R-D, Nyberg E, et al. Comparison between paroxetine and behavior therapy in patients with PTSD: a pilot study. Int J Psychiatry Clin Pract 2004;8: Hendriks GJ, Keijsers GP, Kampman M, et al. A randomized controlled study of paroxetine and cognitive-behavioural therapy for late-life panic disorder. Acta Psychiatr Scand 2010;122: Sousa MB, Isolan LR, Oliveira RR, Manfro GG, Cordioli AV. A randomized clinical trial of cognitive-behavioral group therapy and sertraline in the treatment of obsessive-compulsive disorder. J Clin Psychiatry 2006;67: Blackburn IM, Bishop S, Glen AI, Whalley LJ, Christie JE. The efficacy of cognitive therapy in depression: a treatment trial using cognitive therapy and pharmacotherapy, each alone and in combination. Br J Psychiatry 1981;139: Blackburn IM, Moore RG. Controlled acute and follow-up trial of cognitive therapy and pharmacotherapy in out-patients with recurrent depression. Br J Psychiatry 1997;171: Blom MB, Jonker K, Dusseldorp E, et al. Combination treatment for acute depression is superior only when psychotherapy is added to medication. Psychother Psychosom 2007;76: David D, Szentagotai A, Lupu V, Cosman D. Rational emotive behavior therapy, cognitive therapy, and medication in the treatment of major depressive disorder: a randomized clinical trial, posttreatment outcomes, and six-month follow-up. J Clin Psychol 2008;64: Dimidjian S, Hollon SD, Dobson KS, et al. Randomized trial of behavioral activation, cognitive therapy, and antidepressant medication in the acute treatment of adults with major depression. J Consult Clin Psychol 2006;74: Elkin I, Shea MT, Watkins JT, et al. National Institute of Mental Health Treatment of Depression Collaborative Research Program. General effectiveness of treatments. Arch Gen Psychiatry 1989;46: ; discussion Hollon SD, DeRubeis RJ, Evans MD, et al. Cognitive therapy and pharmacotherapy for depression. Singly and in combination. Arch Gen Psychiatry 1992;49:

8 Review: Relative Effects of CBT and Pharmacotherapy Jarrett RB, Schaffer M, McIntire D, Witt-Browder A, Kraft D, Risser RC. Treatment of atypical depression with cognitive therapy or phenelzine: a double-blind, placebo-controlled trial. Arch Gen Psychiatry 1999;56: Keller MB, McCullough JP, Klein DN, et al. A comparison of nefazodone, the cognitive behavioral-analysis system of psychotherapy, and their combination for the treatment of chronic depression. N Engl J Med 2000;342: McLean PD, Hakstian AR. Clinical depression: comparative efficacy of outpatient treatments. J Consult Clin Psychol 1979;47: Miranda J, Chung JY, Green BL, et al. Treating depression in predominantly low-income young minority women: a randomized controlled trial. J Am Med Assoc 2003;290: Mohr DC, Boudewyn AC, Goodkin DE, Bostrom A, Epstein L. Comparative outcomes for individual cognitive-behavior therapy, supportive-expressive group psychotherapy, and sertraline for the treatment of depression in multiple sclerosis. J Consult Clin Psychol 2001;69: Murphy GE, Carney RM, Knesevich MA, Wetzel RD, Whitworth P. Cognitive behavior therapy, relaxation training, and tricyclic antidepressant medication in the treatment of depression. Psychol Rep 1995;77: Murphy GE, Simons AD, Wetzel RD, Lustman PJ. Cognitive therapy and pharmacotherapy. Singly and together in the treatment of depression. Arch Gen Psychiatry 1984;41: Mynors-Wallis LM, Gath DH, Day A, Baker F. Randomised controlled trial of problem solving treatment, antidepressant medication, and combined treatment for major depression in primary care. Br Med J 2000;320: Scott AI, Freeman CP. Edinburgh primary care depression study: treatment outcome, patient satisfaction, and cost after 16 weeks. Br Med J 1992;304: Shamsaei F, Rahimi A, Zarabian M, Sedehi M. Efficacy of pharmacotherapy and cognitive therapy, alone and in combination, in major depressive disorder. Hong Kong Journal of Psychiatry 2008;18: Thompson LW, Coon DW, Gallagher-Thompson D, Sommer BR, Koin D. Comparison of desipramine and cognitive/ behavioral therapy in the treatment of elderly outpatients with mild-to-moderate depression. Am J Geriatr Psychiatry 2001;9: Tollefson G, Hause E, Garvey M, Evans MD, Tuason V. 24 hour urinary dehydroepiandosterone sulfate in unipolar depression treated with cognitive and/or pharmacotherapy. Ann Clin Psychitry 1990;2: Rush AJ, Beck A, M K, Hollon SD. Comparative efficacy of cognitive therapy and pharmacotherapy in thetreatment of depressed outpatients. Cogn Ther Res 1977;1: Stein DJ, Fineberg NA, Bienvenu OJ, et al. Should OCD be classified as an anxiety disorder in DSM-V? Depress Anxiety 2010;27: