sign in sign up
<<
Home , Cianopramine

BRITISH JOURNAL OF PSYCHIATRY (2007), 190, 287^292. doi: 10.1192/bjp.bp.106.028555 REVIEW ARTICLE AUTHOR’ S P ROOF

Therapeutic effect of follow-up assessments the placebo response but also examined their effect on participants receiving active on and placebo response rates .

in antidepressant efficacy trials METHOD Meta-analysis Sources of data and criteria for review MICHAEL A. POSTERNAK and MARK ZIMMERMAN The collection of studies used here is the same as in our previous meta-analysis which evaluated the time course of Background It remains unclear how Reports in both scientific journals and the improvement on antidepressant medication much various factors contribute to the media have questioned whether the true and placebo (Posternak & Zimmerman, benefits of antidepressant have 2005). These studies were compiled by placebo response. been exaggerated (Goleman, 1995; Fisher reviewing the bibliography of the meta- Aims To estimate the therapeutic & Greenberg, 1997; Horgan, 1998; Kirsch analysis evaluating placebo response rates & Sapirstein, 1999), and a recent review of in antidepressant trials published over the impactof follow-up assessments on the Food and Drug Administration (FDA) past two decades (Walsh et aletal, 2002). To placebo response in antidepressanttrials. database found that that as many as half augment this database, we also systemati- of antidepressant trials yield negative re- cally reviewed each article published from MethodMethod Double-blind, placebo- sults (Khansults(Khan et aletal, 2002). A major hindrance January 1992 through December 2001 in controlled antidepressanttrials that to establishing antidepressant efficacy is the six psychiatric journals (American Journal reported weeklychangesin Hamilton remarkably high rates of improvement of Psychiatry,, Archives of General Psy- Rating Scale for (HRSD) among participants receiving placebo, chiatrychiatry,, British Journal of Psychiatry,, which have been increasing over the past Journal of Clinical Psychiatry,, Journal ofJournalof scores over 6 weeks were selected. two decades (Walsh et aletal, 2002). Factors Clinical Psychopharmacology andand Psycho-Psycho- Included studies (nn¼41) were divided into that have been implicated in the placebo pharmacology Bulletin).). those thatconducted four, five or six response include the instillation of hope, Studies were included if they: (a) were follow-up assessments.Reductionsinassessments.Reductions in response expectancies (Kirsch, 1985), in English; (b) were published from January HRSD scores as a function of the different motivation to please investigators (Orne, 1981 through December 2001; (c) were pri- 1969), the therapeutic impact of marily composed of out-patients with follow-up schedules were compared. assessment contact, rater bias and sponta- major depressive disorder according to Re- ResultsResults An extra follow-up visit at neous improvement (Harrington, 1999). A search Diagnostic Criteria (RDC; Spitzer etet better understanding of how much each alal, 1978); (d) had at least 20 participants in week 3 was associated with a 0.86 further contributes would allow a more accurate the placebo group; (e) randomly assigned reductionreductionin in HRSD score; an extra visit at gauge of the true antidepressant effect and participants to receive a putative antide- week 5 was associated with a 0.67 further could lead to improved trial designs. pressant drug or drugs and placebo; (f) re- reduction.These effects represented In the present study, we sought to ported the total number of participants evaluate the therapeutic impact of frequent assigned to placebo and medication approximately 34^44% ofthe placebo follow-up assessments. In standard anti- group(s); (g) assessed participants under response that occurred over these time depressant trials, participants are usually double-blind conditions; and (h) utilised frames.Two additionalvisits were seen on a weekly basis to assess depression the Hamilton Rating Scale for Depression associated withtwice the reductioninreduction in severity, level of functioning and side- (HRSD; Hamilton, 1960) to assess im- HRSD score than one, suggesting thatthe effects. Such visits typically last 30 min or provement. We excluded studies that did more and are conducted by trained research not report mean baseline HRSD scores, therapeutic impact of assessment visits is assistants over the course of 6 weeks. The did not present weekly or biweekly (every cumulative and proportional. A com- impact of so much contact with a health- other week) changes in HRSD scores, eval- parable therapeutic effect was also found care provider is unknown but could be uated agents with unproven antidepressant inparticipantsreceivingactivemedication. substantial. Furthermore, this amount of properties or evaluated accepted anti- contact is much greater than in routine clin- depressant agents that were used at sub- Conclusions Follow-up assessments in ical practice where two to three 15-min therapeutic doses, or focused on specific antidepressanttreatmenttrialsincur a visits for management of medication are subpopulations of patients such as the the norm (Posternak et aletal, 2002,2002aa). To eval-eval-).To elderly. Forty-seven trials that met these significanttherapeuticsignificant therapeutic effectforeffect for uate the impact of these follow-up assess- inclusion criteria were included in our participants on placebo, and this ments, we conducted a meta-analysis of original meta-analysis. Of these, we represents about 40% ofthe placebo 41 double-blind, placebo-controlled anti- excluded six studies (Claghorn et aletal, 1983;,1983; response. depressant trials published over the past DominguezDominguez et aletal, 1985; Hormazabal et aletal,, two decades. We primarily focused on the 1985; Amsterdam et aletal, 1986; Ferguson etet Declaration of interest None.None. impact that follow-up assessments had on alal, 1994; Khan, 1995) for the present

287287

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use. POSTERNAKPOSTERNAK&ZIMMERMAN & ZIMMERMAN

AUTHOR’SAUTHOR’ S PROOFP ROOF

meta-analysis because they did not conduct extracted from graphs, 456 (95.8%) were comprised 1449 people drawn from 19 outcome assessments at week 6. remeasured by the research assistant within studies and the skip weeks 3 and 5 cohort 0.5 points, suggesting that data extraction comprised 673 participants drawn from 7 Follow-up schedules was performed reliably and without bias. studies. The baseline mean HRSD scores for these three groups were 25.6 For the 41 studies included in the present (s.d.(s.d.¼1.78), 25.9 (s.d.¼1.47) and 24.3 meta-analysis, three types of follow-up Hypotheses (s.d.(s.d.¼2.53) respectively. schedules were used: 15 studies (Cohn & We hypothesised that follow-up assess- For participants randomised to active Wilcox, 1985; Byerley et aletal, 1988; Cohn ments would have a discernible therapeutic medication, the weekly cohort comprised et aletal, 1989; Lineberry et aletal, 1990; Reimherr effect on placebo response rates. Differ- 1507 people from 25 cohorts (some studies et aletal, 1990; Smith et aletal, 1990; Fontaine etet ences in follow-up schedules allowed us to included more than one active medication alal, 1994; Heiligenstein et aletal, 1994; Wilcox compare reductions in HRSD scores in group); the skip week 5 cohort comprised et aletal, 1994; Bremner, 1995; Claghorn & cohorts that met on a weekly basis with 2284 people from 31 cohorts and the skip Lesem, 1995; Fabre et aletal, 1995; Mendels those that by design skipped 1 or 2 weeks. weeks 3 and 5 cohort comprised 820 parti- et aletal, 1995; Claghorn et aletal, 1996;,1996; Our specific hypotheses were: (a) reduc- cipants from 9 cohorts. The baseline HRSD Schatzberg, 2000) conducted weekly tions in HRSD scores from week 4 to week scores for these three groups were 25.6 follow-up assessments over the course of 6 6 will be greater for the weekly cohort com- (s.d.(s.d.¼1.82), 25.9 (s.d.¼1.49) and 25.0 weeks (weekly cohort); 19 studies (Feighner pared with the skip week 5 and skip weeks (s.d.(s.d.¼2.42) respectively. & Boyer, 1989; Versiani et aletal, 1989;,1989; 3 and 5 cohort; (b) reductions in HRSD GelenbergGelenberg et aletal, 1990; Claghorn et aletal,, scores from week 2 to week 4 will be great- Week 5 assessment 1992; Cohn & Wilcox, 1992; Fabre, er for the weekly cohort and the skip week 1992; Kiev, 1992; Rickels et aletal, 1992;,1992; 5 cohort compared with the skip weeks 3 From week 4 to week 6, the mean decrease ShrivastavaShrivastava et aletal, 1992; Smith & Glaudin, and 5 cohort; (c) there will be a propor- in HRSD scores for cohorts receiving 1992; Mendels et aletal, 1993; Cunningham tional and cumulative therapeutic effect of placebo that met at week 5 (the weekly co- et aletal, 1994; Cunningham, 1997; Thase, having multiple extra assessments; to exam- hort) was 1.52 points. For cohorts that did 1997; Khan et aletal, 1998; Rudolph et aletal,, ine this question, we compared reductions not meet at week 5 (the skip week 5 and the 1998; Rudolph & Feiger, 1999; Silverstone in HRSD scores from week 2 to week 6 in skip weeks 3 and 5 cohorts), the mean de- & Ravindran, 1999; Stahl, 2000) con- the skip weeks 3 and 5 cohort, skip week crease in HRSD scores from week 4 to ducted assessments at weeks 1, 2, 3, 4 and 5 cohort, and the weekly cohort; (d) to week 6 was 0.85 points. Thus, participants 6 without an assessment at week 5 (skip confirm that placebo effects do not differ who returned for an extra follow-up visit at week 5 cohort); 7 studies (Feighner et aletal,, between cohorts, we predicted that reduc- week 5 experienced a 0.67 greater reduc- 1983; Merideth & Feighner, 1983; Rickels tions in HRSD scores would be comparable tion in HRSD scores over this 2-week et aletal, 1985; Mendels & Schless, 1986; between cohorts from baseline through period than those who did not have a week RickelsRickels et aletal, 1991; Anonymous, 1994; week 2; because we considered this the 5 visit. This difference represents 44% of LaakmanLaakman et aletal, 1995) conducted assess- most direct method to confirm that there the decrease in HRSD scores over this ments at weeks 1, 2, 4 and 6 without assess- are no random differences in placebo period.period. ments at weeks 3 and 5 (skip weeks 3 and 5 response rates, we deemed it unnecessary cohort). We utilised these differences in to control for potential confounding vari- Week 3 assessment follow-up schedules as a way to focus on ables such as fixed v.v. flexible dose design, From week 2 to week 4, the mean decrease the specific therapeutic effects of follow- year of publication, etc.; (e) if follow-up as- in HRSD scores for cohorts receiving pla- up assessments. sessments are found to convey a therapeutic cebo that met at week 3 (the weekly cohort effect for participants receiving placebo, we and skip week 5 cohort) was 2.56 points. Establishing reduction in HRSD would predict that all of the above findings For cohorts that did not have a scheduled scoresscores would be replicated in cohorts receiving follow-up assessment at week 3 (the skip The method for establishing mean baseline antidepressant medication. weeks 3 and 5 cohort), the mean decrease scores and weekly improvement in HRSD Finally, if follow-up assessments convey a in HRSD scores from week 2 to week 4 scores is the same as in our previous non-specific therapeutic effect, we hypothe- was 1.70 points. Thus, participants who meta-analysis (Posternak & Zimmerman, sised that treatment effect sizes would be returned for an extra follow-up visit at 2005). Baseline HRSD scores and weekly greatergreaterintrialswithfewerfollow-up in trials with fewer follow-up week 3 experienced a 0.86 greater reduc- reductions in HRSD scores were established assessments. However, only a handful of tion in HRSD scores over this 2-week for each study, and all analyses accounted studies published weekly or end-point stand- period than those who did not have a week for differences in sample size between ard deviations. Therefore, we were unable to 3 follow-up visit. This represents 34% of studies. Some studies depicted changes in establish effect sizes or confidence intervals. the decrease in HRSD scores over this HRSD scores graphically. In these in- period.period. stances, weekly changes in HRSD scores RESULTSRESULTS were obtained by measuring each data- Therapeutic impact of multiple point with rounding to the nearest 0.5. A CohortsCohorts extra assessments research assistant who was unaware of the For participants randomised to placebo, the To examine whether there is a cumulative purposes of the study remeasured each weekly cohort comprised 941 people from and proportional therapeutic impact of data-point. Of the 476 data-points 15 separate studies; the skip week 5 cohort multiple extra assessments, we compared

288288

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use. EFFECT OF FOLLOW-UP ASSESSMENT IN ANTIDEPRESSANT TRIALS

AUTHOR’ S P ROOF

reductions in HRSD scores from week 2 to assessment accounted for an additional response, participants who were assessed week 6 in the weekly cohort with reduc- 0.82 points. For the control analysis, we on a weekly basis experienced an overall tions in the skip week 5 and skip weeks 3 again compared reductions in HRSD scores drop in HRSD scores of 9.6 points over and 5 cohorts. The first group had four from baseline to week 2 in the weekly co- the course of 6 weeks. By comparison, scheduled follow-up assessments, the sec- hort with the two cohorts that skipped at participants receiving placebo who were ond group had three and the third group least one follow-up assessment. Reductions assessed only four times experienced only had two. Reductions in HRSD scores were in HRSD scores were 7.78 and 7.61 HRSD a 7.3-point drop in HRSD score. 4.24, 3.33 and 2.49 points respectively. points respectively, again suggesting com- Since follow-up assessments had a dis- Thus, the reduction with one extra parable treatment effects except when there cernible therapeutic effect for participants assessment (skip weeks 3 and 5 cohort v.v. were differences in follow-up schedules. receiving placebo, we expected they would skip week 5 cohort) was 0.84 HRSD points also have a discernible and comparable whereas that with two extra assessments DISCUSSION effect for those receiving active medication. (skip weeks 3 and 5 cohort v.v. weekly co-weeklyco- Indeed, each of our analyses from the hort) was 1.75 HRSD points. This suggests The ubiquitous and robust placebo res- placebo cohorts was replicated for cohorts that the therapeutic impact of follow-up ponse has for years both intrigued and receiving active medication, as each addi- assessments is cumulative and propor- frustrated mood disorder researchers. tional follow-up visit was associated with tional.tional. Although there is general consensus as to a further reduction of 0.97–1.12 in HRSD which factors are responsible for the scores.scores. Control analysis placebo response, it remains unclear how To evaluate whether placebo effects are much each particular component contri- Design of meta-analysis otherwise comparable between the cohorts butes to the overall effect. One exception The ideal method for evaluating the thera- of interest, we compared reductions in to this is the role that spontaneous improve- peutic impact of follow-up assessments on HRSD scores from baseline to week 2 be- ment may play. In a meta-analysis com- the placebo response would be to rando- tween the weekly cohort and the skip week paring treatment effect sizes for people mise participants with depression receiving 5 and skip weeks 3 and 5 cohorts. Because with depression randomised to placebo placebo to different follow-up schedules. all three cohorts received weekly follow-up with those randomised to no treatment, Such a study has not been performed to assessments through week 2, we predicted spontaneous improvement was estimated date and most likely never will. In the that reductions in HRSD scores would be to constitute about one-third of the placebo present meta-analysis, we have in effect similar. The reduction in HRSD scores response (Kirsch & Sapirstein, 1999). randomised cohorts rather than individ- from baseline to week 2 in the weekly co- Other investigators have provided inde- uals. Since the methodology of efficacy hort was 5.35 points. In the two cohorts pendent confirmation of this estimate trials of has remained that subsequently skipped one or two (Posternak & Zimmerman, 2001; largely unchanged over the years (Thase, follow-up assessments, the reduction in PosternakPosternak et aletal, 2006).,2006). 1999), heterogeneity between studies is HRSD scores was 5.41 points. Thus, likely to be minimal: all studies involved placebo effects were comparable between Main results out-patients with moderate-to-severe de- the cohorts when the frequency of follow- In the present study, we isolated one of the pression who received identical treatment up visits was the same. remaining components – the therapeutic (placebo) over the course of 6 weeks using impact of follow-up assessments – to deter- the same outcome measure (the HRSD). Participants receiving active mine the importance of this factor to the re- Where an extra follow-up assessment was medication maining two-thirds of the placebo response. conducted, a clear therapeutic effect was We repeated all the analyses described We found that scheduling an extra follow- associated with that visit as hypothesised. above for participants receiving active up visit at week 3 was associated with an Although it is possible that this could be at- medication. Reduction in HRSD score from additional 0.86-point reduction in HRSD tributable to random differences between week 4 to week 6 for the weekly cohort was scores, whereas scheduling an additional studies, we would argue that this is extre- 2.35 points compared with 1.38 for cohorts week 5 visit was associated with an addi- mely unlikely. The present meta-analysis who did not have a week 5 visit (a differ- tional 0.67 reduction in HRSD scores. included the majority of acute-phase, ence of 0.97 points). Reduction in HRSD These reductions represent approximately placebo-controlled antidepressant trials score from week 2 to week 4 for cohorts 40% of the placebo response that occurred published over the past two decades, and that met at week 3 (the weekly cohort and over their respective time frames. When we our analyses were therefore based on large the skip week 5 cohort) was 3.69 points examined the cumulative effect of schedul- sample sizes. Second, improvement on compared with 2.57 for cohorts that did inging twotwo additional follow-up visits, we placebo was comparable between all three not have a week 3 visit (a difference of found that the therapeutic impact of each cohorts during the first 2 weeks of treat- 1.12 points). Reductions in HRSD scores visit was cumulative and proportional. ment when follow-up assessment from week 2 to week 6 for the weekly co- That is, one extra visit was associated with schedules were identical. As this is the most hort, skip week 5 cohort and skip weeks 3 a 0.84 greater reduction in the HRSD score direct method for evaluating random differ- and 5 cohort were 5.87, 5.05 and 4.29 re- whereas a second extra visit was associated ences in placebo response rates, it would be spectively. One extra assessment visit there- with a 0.91 further reduction in the HRSD superfluous to attempt to control for other fore accounted for a reduction of 0.76 score. As further illustration of the impact potential confounding variables such as HRSD points whereas a second extra of follow-up assessments on the placebo year of publication, episode duration,

289289

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use. POSTERNAKPOSTERNAK&ZIMMERMAN & ZIMMERMAN

AUTHOR’ S P ROOF

comorbidity, etc. Furthermore, all of our Design of trials therapeutic effect for participants receiving findings that supported a clear, therapeutic Considering the relatively modest effect size placebo, and that these assessment visits effect from assessment contact were of FDA-approved antidepressants over account for an estimated 40% of the replicated in cohorts receiving active placebo, that side-effects may unmask placebo response. This does not take into medication. raters in favour of eliciting drug–placebo account the therapeutic effect of the initial We would argue that our results are not differences (Greenberg et aletal, 1992) and that evaluation, which is typically much more undermined by relying solely on published most negative trials never get published, extensive than follow-up assessments and studies. Publication bias is a concern for several investigators have suggested that would be expected to convey a larger ther- many meta-analyses because negative trials the benefits of antidepressant medications apeutic effect. For years, there has been often go unpublished, and attempts to have been exaggerated over the years much speculation as to which ingredients establish effect sizes may consequently (Fisher & Greenberg, 1997; Kirsch & comprise the powerful and seemingly overestimate treatment benefits. The goal Sapirstein, 1999). Although these argu- magical placebo pill, with some investiga- of the present study, however, was to esti- ments are persuasive, we believe an alterna- tors even suggesting that different coloured mate the therapeutic impact of follow-up tive explanation also exists – that the pills may be associated with different assessments. The lack of inclusion of un- methodology used to elicit and establish placebo response rates (Jacobs & Nordan, published studies would only undermine antidepressant efficacy is inefficient. As re- 1979; Buckalew & Coffield, 1982). Our our results if unpublished studies were viewed elsewhere (Posternak et aletal, 2002,2002bb),), findings suggest that, after accounting for found to systematically have less therapeu- the methodology used in antidepressant spontaneous improvement, the placebo tic impact of their assessment visits (for ex- trials evolved largely from traditions estab- response in trials of antidepressants stems ample, if raters in unpublished studies were lished over three decades ago and has never largely from the attention and care received consistently less empathic). Unpublished undergone empirical testing. Our results during the course of the clinical trial. studies, however, by virtue of having failed suggest that the frequent and extensive to separate drug from placebo, would be monitoring that occurs in clinical trials con- REFERENCES expected to have more rather than less fers a significant therapeutic effect for robust placebo response rates, and the ther- participants receiving placebo (and active Amsterdam, J. D., Case,W.G., Csanalosi, E., et aletal apeutic impact of follow-up assessments (19 8 6) A double-blind comparative trial of , medication). High placebo response rates , and placebo in patients with mixed anxiety might, if anything, be more pronounced. reduce treatment effect sizes and increase and depression. Pharmacopsychiatry,, 19,115^119.

the risk that an efficacious agent will be Anonymous(19 (1994) 94) A multicenter comparative trial of Limitations deemed ineffective. Although a comparable , and placebo in major therapeutic effect from follow-up visits was depressive disorder. International Clinical Psychopharmacology,, 99,,103^109. 103^109. One limitation of our study is that because found in participants randomised to active few studies published weekly or end-point medication, reducing an equivalent amount Bremner, J. D.(19 (1995) 95) A double-blind comparison of standard deviations of HRSD scores, we of ‘noise’ in both cohorts would have the Org 3770, amitriptyline, and placebo in major depression. Journal of Clinical Psychiatry,, 5656,519^525. were unable to confirm that differences effect of increasing the power to detect dif- between cohorts were statistically signifi- ferences between the active medication and Buckalew, L.L.W. W. & Coffield, K. E.(19 (1982) 82) An investigation of drug expectancy as a function of capsule cant. Although our analyses yielded what control group (Cohen, 1988). appears to be a large and consistent effect color and size and preparation form. Journal of Clinical Knowing the impact that follow-up as- Psychopharmacology,, 22, 245^248. from extra follow-up visits, the lack of sta- sessments have on placebo response rates, tistical confirmation warrants caution in in- Byerley,W.Byerley,W.F., F., Reimherr, F.W.,Wood, D. R., et aletal the design of antidepressant trials could be (19 8 8) , a selective uptake inhibitor, terpreting these findings. We also wondered modified either by reducing the amount of for the treatment of outpatients with major depression. whether the greater therapeutic effect found time devoted to assessing participants in Journal of Clinical Psychopharmacology,, 88,112^115.,112^115. in cohorts that met more frequently might follow-up, reducing the frequency of Claghorn, J. L. & Lesem, M. D.(19 (1995) 9 5) AAdouble-blind double-blind be a consequence of greater retention rates follow-up assessments, or relying more on placebo-controlled study of Org 3770 in depressed outpatients.outpatients. Journal of Affective Disorders,, 3434,165^171. in these cohorts. In most clinical trials, off-site raters or interactive computer rating scores for participants who drop assessment. Of course, consideration of Claghorn, J., Gershon, S., Goldstein, B. J., et aletal out are handled using the last-observation- these changes must be balanced against (19 83) A double-blind evaluation of zimelidine in carried-forward method of analysis. Per- comparison to placebo and amitriptyline in patients ethical concerns of having insufficient mon- with major depressive disorder. Progress ininProgress haps participants who do not present on a itoring over the course of a clinical trial. Neuropsychopharmacology and Biological Psychiatry,, 77,, weekly basis are more likely to drop out This would apply both to participants ran- 367^382.367^382. and therefore not have the opportunity to domised to placebo and to those receiving a Claghorn, J. L., Kiev, A., Rickels, K., et aletal (19 92)9 2) demonstrate improvement. To address this putative antidepressant agent, especially if versus placebo: a double-blind comparison in concern, we evaluated completion rates in there are concerns regarding the potential depressed patients. Journal of Clinical Psychiatry,, 5353,434^, 434^ 438.438. each of the three cohorts and found no for increased suicidal ideation following correlation between frequency of visits the initiation of an antidepressant. Claghorn, J. L., Earl,C.Q.,Walczak, D. D., et al (19 9 6) and completion rates: skip week 3 and 5, maleate in the treatment of depression: a 58.5% (326 of 557); skip week 5, 62.5% single-center, double-blind, placebo-controlled Explaining the placebo response comparison with imipramine in outpatients. Journal of (847 of 1356); weekly, 58.8% (403 of Clinical Psychopharmacology,, 1616,113^120.,113^120. 685). Thus, the therapeutic effect we found Our results suggest that the follow-up Cohen, J.J.Cohen, (19 (1988) 8 8) Statistical Power Analysis for the does not appear to be a function of assessment schedules of standard antide- Behavioral Sciences (2nd edn), pp. 180^181.18 0^181. Lawrence improved adherence. pressant efficacy trials convey a significant Erlbaum.

290

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use. EFFECT OF FOLLOW-UP ASSESSMENT IN ANTIDEPRESSANT TRIALS

AUTHOR’ S P ROOF

Cohn, J. B. & Wilcox, C.(19 (1985) 8 5) Acomparisonof fluoxetine, imipramine, and placebo in patients with MICHAEL A. POSTERNAK, MD,MARKMD, MARK ZIMMERMAN, MD,DepartmentMD, Department of Psychiatry and Human Behaviour, major depressive disorder. Journal of Clinical Psychiatry,, Brown University School of Medicine, Rhode Island Hospital, Providence, RI 02905,USA 46,,26^31. 26^31.

Cohn, J. B. & Wilcox, C. S.(19 (1992) 92) Paroxetine in major Correspondence:Drondence: Dr Michael A.Posternak,Depression Clinical and Research Program,Department of depression: a double-blind trial with imipramine and Psychiatry,Harvard Medical School, Massachusetts General Hospital, 50 Staniford Street, Suite 401, placebo.placebo. Journal of Clinical Psychiatry,, 53 (suppl. 2), 52^56. Boston,Boston,MA MA 02114,USA.02114,USA.Email: Email: mposternak@@partners.org

Cohn, J. B.,B.,Collins,G., Collins, G., Ashbrook, E., et aletal (19 8 9) AA comparison of fluoxetine imipramine and placebo in (First received10received 10 July 2006, accepted 31October31 October 2006) patients with bipolar depressive disorder. International Clinical Psychopharmacology,, 44, 313^322.,313^322.

Cunningham,Cunningham, L. A.(19 (1997) 9 7) Once-daily extended release (XR) and venlafaxine immediate release (IR) in outpatients with major depression. AnnalsAnnals Horgan, J.J.Horgan, (19 (1998) 9 8) Science triumphant? Not so fast. and placebo in patients with primary affective disorders. of Clinical Psychiatry,, 99,157^164. New York Times,19, 19 January, p. A1917. Acta Psychiatrica Scandinavica,, 68 (suppl. 308), 70^79. Cunningham,Cunningham, L. A., Borison, R. L., Carman, J. S., Hormazabal, L., Omer, L. M. O. & Ismail, S.(19 (1985) 8 5) Orne, M. T.(19 (1969) 69) Demand characteristics and the et aletal (19 94) A comparison of venlafaxine, , Cianopramine and amitriptyline in the treatment of concept of quasi-controls. In Artifact in Behavioral and placebo in major depression. Journal of Clinical depressed patients ^ a placebo-controlled study. ResearchResearch (eds R. Rosenthal & R. L. Rosnow), pp. 143^ Psychopharmacology,, 1414, 99^106.,99^106. Psychopharmacology,, 86, 205^208.,205^208. 179. Academic Press. Dominguez, R. A., Goldstien, B. J., Jacobson, A. F., Jacobs,K.W.&Nordan,F.M.(1979)Jacobs, K.W. & Nordan, F. M. (19 79) Classification of Posternak, M. A. & Zimmerman, M. (2001)(2001) Short- et aletal (19 8 5) A double-blind placebo-controlled study of placebo drugs: effect of color. Perceptual and Motor Skills,, term spontaneous improvement rates in depressed fluvoxamine and imipramine in depression. Journal of 49, 367^372.,367^372. outpatients.outpatients. Journal of Nervous Mental Disorders,, 6262,, Clinical Psychiatry,, 4646, 84^87. 799^804.799^804. Khan, A.,Upton,V., Rudolph, R. L., et aletal (19 9 8) The Fabre,L.F.(1992)Fabre,L.F.(19 92) A 6-week, double-blind trial of use of venlafaxine in the treatment of major depression Posternak, M. A. & Zimmerman, M. (2005)(2005) Is thereIsthere paroxetine, imipramime. and placebo in depressed and major depression associated with anxiety: a dose^ a delay in the antidepressant effect? A meta-analsyis. outpatients.outpatients. Journal of Clinical Psychiatry,, 5353 (suppl. 2), response study. Journal of Clinical Psychopharmacology,, 1818,, Journal of Clinical Psychiatry,, 66,148^158.,148^158. 40^43.40^43. 19^25.19^25. Fabre, L. F., Abuzzahab, F. S., Amin, M., et aletal (19 95) Posternak, M. A., Zimmerman, M. & Solomon, D. A. safety and efficacy in major depression: a Khan, A., Khan, S. & Brown,W. A.(2002) AreAre (2002(2002aa)) Integrating outcomes research into clinical double-blind fixed-dose comparison with placebo. placebo controls necessary to test new antidepressants practice: a pilot study. Psychiatric Services,, 5353, 335^336. Biological Psychiatry,, 3838,592^602., 592^602. and ? International Journal of Neuropsychopharmacology,, 55, 193^197.,193^197. Posternak, M. A., Zimmerman, M., Keitner,G. I., Feighner, J. P. & Boyer,W.Boyer,W.F. F.(1989) Paroxetine in the et aletal (2002(2002bb)) A reevaluation of the exclusion criteria treatment of depression: a comparison with imipramine Khan, M. C.(19 (1995) 95) A randomised, double-blind, used in antidepressant efficacy trials. American Journal of and placebo. Acta Psychiatrica Scandinavica,, 80 (suppl. placebo-controlled, 5-weeks’study of Org 3770 PsychiatryPsychiatry,, 159159,191^200.,191^200. 350),125^129. () in major depression. Human Psychopharmacology,, 10 (suppl. 2), S119^S124.S119 ^ S12 4. Posternak, M. A., Solomon, D. A., Leon, A. C., et aletal Feighner, J. P., Aden, G. C., Fabre, L. F., et aletal (19 83) (2006)(2006) The naturalistic course of untreated major Comparison of , imipramine, and placebo in Kiev, A.A.Kiev, (19 (1992) 92) A double-blind, placebo-controlled depressive disorder. Journal of Nervous and Mental the treatment of depression. JAMAJAMA,, 249249, 3057^3064. study of paroxetine in depressed outpatients. JournalJournal DisordersDisorders,, 194, 324^329.,324^329. Clinical of Psychiatry,, 53 (suppl. 2), 27^29. Ferguson, J. M., Mendels, J. & Manowitz, N. R. Reimherr, F.W., Chouinard, G., Cohn, C. K., et al (19 94)94)(19 Dothiepin versus in major depression: Kirsch, I.I.Kirsch, (19 (1985) 8 5) Response expectancy of experience (1990)(19 9 0) Antidepressant efficacy of sertraline: a double- results of a multicenter, placebo-controlled trial. JournalJournal and behavior. American Psychologist,, 40,1189^1202. blind, placebo- and amitriptyline-controlled, multicenter of Clinical Psychiatry,, 5555, 258^263. comparison study in outpatients with major depression. Kirsch, I. & Sapirstein, G.(1999) Listening to Prozac Journal of Clinical Psychiatry,, 5151 (suppl.12),18^27.(suppl. 12), 18^27. Fisher, S. & Greenberg, R. P.(eds) (1997) From but hearing placebo: a meta-analysis of antidepressant Placebo to Panacea.JohnWileyandSons. medications. In How Expectancies Shape Experience (ed.(ed. Rickels, K., Feighner, J. P. & Smith,W.T.(19 (1985) 8 5) I.Kirsch), pp. 303^320. American Psychological Alprazolam, amitriptyline, doxepin, and placebo in the Fontaine, R., Ontiveros, A., Elie, R., et aletal (1994)(19 94) AA Association.Association. treatment of depression. Archives of General Psychiatry,, double-blind comparison of , imipramine, 42,134^141. and placebo in major depression. Journal of Clinical Laakman, G., Faltermaier-Temizel, M., Bossert- PsychiatryPsychiatry,, 5555,234^241., 234^241. Zaudig, S., et aletal (19 95)95)(19 Treatment of depressive Rickels,Rickels,K.,London,J.,Rox,I., K., London, J., Rox, I., et al (19 91) outpatients with , alprazolam, amitriptyline , , imipramine, and placebo in Gelenberg, A. J.,Wojcik, J. D., Falk,W.Falk,W.E., E., et aletal (19 9 0) and placebo. Psychopharmacology,, 120,109^115. major depressive disorder: a controlled study. Clovoxamine in the treatment of depressed outpatients: Pharmacopsychiatry,, 2424,127^131.,127^131. a double-blind, parallel-group comparison against Lineberry,C.Lineberry, C. G., Johnston, A., Raymond, R. N., et aletal amitriptyline and placebo. Comprehensive Psychiatry,, 31,, (19 9 0) A fixed-dose (300 mg) efficacy study of Rickels, K., Amsterdam, J., Clary,C., et aletal (19 992) 2) The 307^314.307^314. and placebo in depressed outpatients. JournalJournal efficacy and safety of paroxetine compared with placebo Goleman, D.(19 (1995) 95) Psychologists dispute value of of Clinical Psychiatry,, 5151,194^199.,194^199. in outpatients with major depression. Journal of Clinical PsychiatryPsychiatry,, 5353 (suppl. 2), 30^32. antidepressants. New York Times, 29 November, p.C1910. Mendels, J. & Schless, A. P.(19 (1986) 8 6) Comparative Greenberg, R. P., Bornstein, R. F., Greenberg, M. D., efficacy of alprazolam, imipramine, and placebo Rudolph, R. L. & Feiger, A. D.(1999) AAdouble-blind, double-blind, et aletal (19 992) 2) A meta-analysis of antidepressant outcome administered once a day in treating depressed patients. randomized, placebo-controlled trial of once-daily under ‘blinder’ conditions. Journal of Consulting Clinical Journal of Clinical Psychiatry,, 47, 357^361.,357^361. venlafaxine extended release (XR) and fluoxetine for PsychologyPsychology,, 60, 664^669.,664^669. the treatment of depression. Journal of Affective Mendels, J., Johnston, R., Mattes, J., et aletal (19 93)93)(19 DisordersDisorders,, 5656,171^181.,171^181. Hamilton, M.(19 (1960) 6 0) A rating scale for depression. Efficacy and safety of b.i.d. doses of venlafaxine in a Journal of Neurology Neurosurgery and Psychiatry,, 2323,, dose^responsestudy.dose ^ response study. Psychopharmacology Bulletin,, 29,, Rudolph, R. L., Fabre, L. F., Feighner, J. P., et al (19 9 8)8)(19 56^62.56^62. 169^174. A randomized, placebo-controlled, dose-response trial of venlafaxine hydrochloride in the treatment of major Harrington, A.(19 (1999) 9 9) The Placebo Effect.Harvard Mendels, J., Reimherr, F., Marcus, R. N., et aletal (19 95) depression. Journal of Clinical Psychiatry,, 59,116^122. University Press. A double-blind, placebo-controlled trial of two dose ranges of nefazodone in the treatment of depressed Schatzberg, A. F. (2000)(2000) Clinical efficacy of Heiligenstein, J. H., Tollefson, G. D. & Faries, D. E. outpatients.outpatients. Journal of Clinical Psychiatry,, 5656 (suppl. 6), in major depression. Journal of Clinical Psychiatry,, 6161 (19 94)94)(19 Response patterns of depressed outpatients with 30^36.30^36. (suppl.10),(suppl. 10), 31^38. and without melancholia: a double-blind, placebo- controlled trial of fluoxetine versus placebo. Journal ofofJournal Merideth, C. H. & Feighner, J. P.(19 (1983) 83) Adouble-Adouble- Shrivastava, R. K., Shrivastava, S. P., Overweg, N., Affective Disorders,, 3030,163^173. blind, controlled evaluation of zimelidine, imipramine et aletal (19 92) A double-blind comparison of paroxetine,

291

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use. POSTERNAKPOSTERNAK&ZIMMERMAN & ZIMMERMAN

AUTHOR’ S P ROOF

imipramine, and placebo in major depression. Journal ofofJournal Spitzer, R. L., Endicott, J. & Robins, E. (1978) Versiani, M., Oggero,U., Alterwain, P., et aletal (19 8 9)9)(19 Clinical Psychiatry,, 53 (suppl. 2), 48^51. Research diagnostic criteria: rationale and reliability. A double-blind comparative trial of moclobemide v. Archives of General Psychiatry,, 3535, 773^782.,773^782. imipramine and placebo in major depressive episodes. Silverstone, P.H.P. H. & Ravindran, A.(19 (1999) 9 9) Once-dailyOnce-daily British Journal of Psychiatry,, 155155 (suppl. 6), 72^77. venlafaxine extended release (XR) compared with Stahl, S. M.(2000) Placebo-controlled comparison of fluoxetine in outpatients with depression and anxiety. the selective serotonin reuptake inhibitors Walsh, B. T., Seidman, S. N., Sysko, R., et aletal (2002)(2002) Journal of Clinical Psychiatry,, 60, 22^28.,22^28. and sertraline. Biological Psychiatry,, 48, 894^901.,894^901. Placebo response in studies of major depression. Thase, M. E.(19 (1997) 9 7) Efficacy and tolerability of once- Variable,substantial,andgrowing.JAMAJAMA,, 287287,1840^ Smith,W.T. & Glaudin,V.(19 (1992) 92) AAplacebo-controlled placebo-controlled daily venlafaxine extended release (XR) in outpatients 181847. 47. trial of paroxetine in the treatment of major depression. with major depression. Journal of Clinical Psychiatry,, 58,, Journal of Clinical Psychiatry,, 5353 (suppl. 2), 36^39. 393^398.393^398. Wilcox, C. S., Cohn, J. B., Katz, B. B., et aletal (19 94) Smith,W.T., Glaudin,V., Panagides, J., et aletal (19 9 0) Thase, M. E. (1999)(1999) How should efficacy be evaluated A double-blind, placebo-controlled study comparing Mirtazapine vs. amitriptyline vs. placebo in the in randomized clinical trials of treatments for and amitriptyline in moderately depressed treatment of major depressive disorder. depression?depression? Journal of Clinical Psychiatry,, 60 (suppl. 4), outpatients.outpatients. International Clinical Psychopharmacology,, 99,, Psychopharmacology Bulletin,, 20,191^196. 23^31. 271^279.

292

Downloaded from https://www.cambridge.org/core. 30 Sep 2021 at 07:03:00, subject to the Cambridge Core terms of use.