Correspondence
Depression — Augmentation or Switch after Initial SSRI Treatment
N Engl J Med 2006; 354:2611-2613June 15, 2006
- Article
To the Editor:
The report on the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial by Rush et al. (March 23 issue)1 advances the evidence base for the treatment of major depression. However, I believe it would be overreaching and incorrect to characterize the results of the study of drug treatment after initial treatment with selective serotonin-reuptake inhibitors (SSRIs) as definitive, owing to the statistical underpowering of the design. This study seems to have the same shortcomings as do other large studies, in that the authors incorrectly estimated the differences in remission among the various therapies that they tested. Rush et al. assumed that there would be a 15 percent difference among treatments and designed the study accordingly.2 Several studies sponsored by industry, independent investigators, and government have detected significant differences of 7 to 10 percent.3-5
Underpowering is a chronic problem in current study design that may lead many practitioners to draw definitive but incorrect conclusions that affect patient care. For example, many reports in the media have misrepresented the STAR*D finding that suggests — but does not conclusively demonstrate — that all treatments are equal. The investigators might have prevented this problem if they had increased the number of participants in the study.
Norman Sussman, M.D.
New York University School of Medicine, New York, NY 10016
sussman01@aol.com Dr. Sussman reports having served as a consultant to GlaxoSmithKline, Shire, and Wyeth and having received honoraria from GlaxoSmithKline, Shire, Wyeth, and AstraZeneca.
5 References1
Rush AJ ,Trivedi MH ,Wisniewski SR , et al. Bupropion-SR, sertraline, or venlafaxine-XR after failure of SSRIs for depression. N Engl J Med 2006;354:1231-1242
Full Text | Web of Science | Medline2
Rush AJ ,Fava M ,Wisniewski SR , et al. Sequenced treatment alternatives to relieve depression (STAR*D): rationale and design. Control Clin Trials 2004;25:119-142
CrossRef | Medline3
Thase ME ,Entsuah AR ,Rudolph RL . Remission rates during treatment with venlafaxine or selective serotonin reuptake inhibitors. Br J Psychiatry 2001;178:234-241
CrossRef | Web of Science | Medline4
Baldomero EB ,Ubago JG ,Cercos CL ,Ruiloba JV ,Calvo CG ,Lopez RP . Venlafaxine extended release versus conventional antidepressants in the remission of depressive disorders after previous antidepressant failure: ARGOS study. Depress Anxiety 2005;22:68-76
CrossRef | Web of Science | Medline5
Cipriani A ,Brambilla P ,Furukawa T , et al. Fluoxetine versus other types of pharmacotherapy for depression. Cochrane Database Syst Rev 2005;4:CD004185-CD004185
Medline
To the Editor:
Trivedi et al. (March 23 issue)1 reported on patients' use of sustained-release bupropion or buspirone, in combination with citalopram, after citalopram alone was ineffective in an initial trial. Approximately 30 percent of patients had a remission, as measured by the 17-item Hamilton Rating Scale for Depression — a figure consistent with the placebo effects reported in other studies of the treatment of depression.2-5 Indeed, the authors noted that the study design did not include a placebo control and, therefore, did “not allow us to exclude spontaneous remission, the nonspecific effects of treatment, or the extended use of citalopram alone as the likely explanation for the present findings.” Nevertheless, they concluded in their very next paragraph that “augmentation of SSRIs with either agent will result in symptom remission.” They also pondered the adoption of such drug combinations as “first-line treatment in an attempt to achieve greater remission rates sooner in more patients than with SSRIs alone.” Similar conclusions describing the clinical usefulness and advantages of the drugs appear in the study abstract (without mentioning limitations). Given the study design, these conclusions seem unsupportable.
5 ReferencesScott C. Williams, Psy.D.
736 Tanager Ln., Geneva, IL 601341
Trivedi MH ,Fava M ,Wisniewski SR , et al. Medication augmentation after the failure of SSRIs for depression. N Engl J Med 2006;354:1243-1252
Full Text | Web of Science | Medline2
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;46:971-982
Web of Science | Medline3
Walsh BT ,Seidman SN ,Sysko R ,Gould M . Placebo response in studies of major depression: variable, substantial, and growing. JAMA 2002;287:1840-1847
CrossRef | Web of Science | Medline4
Moncrieff J . The antidepressant debate. Br J Psychiatry 2002;180:193-194
CrossRef | Web of Science | Medline5
Wampold BE ,Minami T ,Tierney SC ,Baskin TW ,Bhati KS . The placebo is powerful: estimating placebo effects in medicine and psychotherapy from randomized clinical trials. J Clin Psychol 2005;61:835-854
CrossRef | Web of Science | Medline
To the Editor:
In his editorial accompanying the two STAR*D articles, Rubinow1 emphasizes that medications with various mechanisms of action were all roughly equivalent in efficacy and asks what one can possibly infer about the pathophysiology of major depression. In our view, Rubinow puts his finger in an open wound of psychiatry — that is, its insufficient and purely descriptive classification. If we lump together all patients with major depressive syndromes, regardless of whether there is a family history of schizophrenia, unipolar or bipolar depression, a coexisting attention deficit–hyperactivity disorder or borderline personality disorder, or discrete abnormalities on electroencephalography or magnetic resonance imaging, we end up with a mixed bag as a study sample. Such study groups necessarily produce mixed results in etiologic and therapeutic research. We need a more specific and cause-dependent classification system in psychiatry if we want to progress to more specific therapeutic rationales in psychiatric disorders.
1 ReferencesLudger Tebartz van Elst, M.D.
Dieter Ebert, M.D.
Bernd Hesslinger, M.D.
University Clinic Freiburg, 79104 Freiburg, Germany
tebartzvanelst@uniklinik-freiburg.de 1
Rubinow DR . Treatment strategies after SSRI failure -- good news and bad news. N Engl J Med 2006;354:1305-1307
Full Text | Web of Science | Medline
Author/Editor Response
Sussman raises the important question of whether the results would have revealed significant differences had the sample been larger, despite the fact that this is the largest second-step clinical trial ever conducted. We respectfully disagree with the notion that our findings are the result of an underpowered study. Given the three-group design, the study had 80 percent power to detect an effect size of 0.12 in remission rates, which is considered to be a small effect.1 The translation of the effect size into a detectable difference in a three-group trial is not as straightforward as in a two-group trial. If one were conducting a two-group trial with a similar sample size, there would be 80 percent power to detect a difference of approximately 10 percent in remission rates. Although one can argue that more modest differences are clinically important, we believe that a difference of 10 percent or less in remission rates is probably not clinically meaningful.
Williams asks whether the 30 percent rate of remission in our study may be due to the placebo effect, as in efficacy trials. The STAR*D remission rate is probably not due to placebo for several reasons. First, patients in step 2 of STAR*D had already undergone a full trial with an SSRI.2 Second, patients reported numerous psychiatric and general medical coexisting conditions and long durations of illness; most were chronically depressed. Third, remission rates are low in real-world clinical trials of outpatients with major depressive disorder. Both the Texas Medication Algorithm Project,3,4 involving psychiatric patients, and the Improving Mood Promoting Access to Collaborative Care Treatment (known as IMPACT) trial,5 involving primary care patients, showed remission rates of 8 percent with usual care over a one-year observation period. These factors all suggest that spontaneous remission is likely to be low in a 14-week trial, with patients probably entering remission because of the medication itself. Finally, given the remission rate of 30 percent with the treatment in step 2, the question of whether augmentation in step 1 has more benefits than waiting until step 2 remains to be addressed.
5 ReferencesA. John Rush, M.D.
Madhukar H. Trivedi, M.D.
University of Texas Southwestern Medical Center at Dallas, Dallas, TX 75390-9086
john.rush@utsouthwestern. edu Stephen R. Wisniewski, M.D.
University of Pittsburgh, Pittsburgh, PA 15261for the STAR*D trial investigators
1
Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, N.J.: Lawrence Erlbaum, 1988.
2
Trivedi MH ,Rush AJ ,Wisniewski SR , et al. Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. Am J Psychiatry 2006;163:28-40
CrossRef | Web of Science | Medline3
Rush AJ ,Trivedi M ,Carmody TJ , et al. One-year clinical outcomes of depressed public sector outpatients: a benchmark for subsequent studies. Biol Psychiatry 2004;56:46-53
CrossRef | Web of Science | Medline4
Trivedi MH ,Rush AJ ,Crismon ML , et al. Clinical results for patients with major depressive disorder in the Texas Medication Algorithm Project. Arch Gen Psychiatry 2004;61:669-680
CrossRef | Web of Science | Medline5
Unutzer J ,Katon W ,Callahan CM , et al. Collaborative care management of late-life depression in the primary care setting: a randomized controlled trial. JAMA 2002;288:2836-2845
CrossRef | Web of Science | Medline
Author/Editor Response
One would hope that there are few in our field who would disagree with the plea of Tebartz van Elst et al. for a classification system that would permit greater sample homogeneity and more precise prediction of the course of illness and treatment response. Indeed, much of the recent interest in intermediate phenotyping reflects the hope that by parsing psychiatric disorders into component processes, we may be better able to identify the roles played by the many genes that make small but important contributions to the variance of these disorders. Nonetheless, genetic studies to date demonstrate that genes can contribute to susceptibility without being “causative.” Thus, it is premature to insist that the classification system be “cause-dependent,” and it is unnecessary for an effective nosology to be based on etiology. We can diagnose depression without knowing its cause, just as we can treat depression with antidepressants without knowing why they work. Such is the beauty of syndromal diagnoses. Clearly, the goal is to identify the historical, clinical, and molecular factors that will improve the quality and efficacy of our therapeutic efforts.
David R. Rubinow, M.D.
University of North Carolina, Chapel Hill, Chapel Hill, NC 27599- Citing Articles (3)
Citing Articles
1
2010. References. , 165-193.
CrossRef2
L. Tebartz van Elst, P. Ludaescher, T. Thiel, M. Büchert, B. Hesslinger, M. Bohus, N. Rüsch, J. Hennig, D. Ebert, K. Lieb. (2007) Evidence of disturbed amygdalar energy metabolism in patients with borderline personality disorder. Neuroscience Letters 417:1, 36-41
CrossRef3
Phil Skolnick, Anthony S. Basile. (2006) Triple reuptake inhibitors as antidepressants. Drug Discovery Today: Therapeutic Strategies 3:4, 489-494
CrossRef
