34 Leon et al. J Clin Psychiatry 2001;62 (suppl 4) © Copyright 2001 Physicians Postgraduate Press, Inc. One personal copy may be printed n antidepressant with an earlier onset of therapeutic effect than other antidepressants could have wide- From the Department of Biostatistics in Psychiatry, Weill Medical College of Cornell University (Dr. Leon); the New York State Psychiatric Institute (Drs. Gorman and Roose), New York, NY; the Department of Psychiatry Brain Institute, University of Florida, Gainesville (Dr. Blier); the Department of Family Medicine, Boston University School of Medicine (Dr. Culpepper); the Department of Outpatient Psychiatry, the Department of Clinical Psychopharmacology (Dr. Rosenbaum), and the Depression and Clinical Research Program (Dr. Nierenberg), Massachusetts General Hospital, Boston; the Department of Psychiatry and Behavioral Sciences, University of Texas Medical Branch, Galveston (Dr. Hirschfeld); the Clinical Neuroscience Research Center, San Diego, Calif. (Dr. Stahl); and the Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas (Dr. Trivedi). Presented at the symposium “Early Onset of Antidepressant Action,” which was held January 12, 2000, in New York, N.Y., and supported by an unrestricted educational grant from Forest Laboratories, Inc. Reprint requests to: Andrew C. Leon, Ph.D., Weill Medical College of Cornell University, Department of Psychiatry, Box 140, 525 E. 68th St., New York, NY 10021. An Ideal Trial to Test Differential Onset of Antidepressant Effect Andrew C. Leon, Ph.D.; Pierre Blier, M.D., Ph.D.; Larry Culpepper, M.D.; Jack M. Gorman, M.D.; Robert M. A. Hirschfeld, M.D.; Andrew A. Nierenberg, M.D.; Steven P. Roose, M.D.; Jerrold F. Rosenbaum, M.D.; Stephen M. Stahl, M.D., Ph.D.; and Madhukar H. Trivedi, M.D. Although various published clinical studies have suggested that some antidepressants may have a more rapid onset of therapeutic effect than others, none of these trials was adequately designed to measure differential time to onset of effect. Thus, existing data do not support claims that one drug reduces the symptoms of depression faster than another. In this article, we propose a study that would be ideal for measuring comparative onset of antidepressant effect. The key features of this ideal trial include (1) a prospective definition of early onset of action, (2) increased frequency of assessment, (3) a data-analytic approach capable of capturing the dynamic nature of symptomatic change, and (4) various strategies to minimize bias and heterogeneity of response. (J Clin Psychiatry 2001;62[suppl 4]:34–36) In devising the ideal trial, we have attempted to address most of the issues and concerns raised in the foregoing ar- ticles and discussion. First, we have prospectively defined early onset of action. Second, we have increased the fre- quency of assessment to detect more subtle changes in symptom severity. Third, we have selected a data-analytic approach capable of capturing the dynamic nature of symptomatic change. Finally, we have adopted various strategies to minimize bias and heterogeneity of response. While it would be impractical, if not impossible, to con- duct a study based on our design, we hope that the exercise will foster fresh thinking and debate on how better to de- fine and measure onset of antidepressant action. A PROSPECTIVE DEFINITION OF EARLY ONSET OF ANTIDEPRESSANT ACTION Our prospective definition of early onset of action com- prises the following elements: (1) it must appear by week 4, (2) the action we are assaying is a significant reduction in depressive symptoms, (3) it must be linked to a clini- cally significant treatment outcome (i.e., early improve- ments should be sustained for the duration of the study), and (4) a differential time to onset of action of a week or more is clinically relevant. Although these criteria are subject to debate, we feel that they accommodate both theoretical and practical con- cerns. For example, few clinicians expect that their pa- tients will respond to an antidepressant after a week or less of treatment, but most expect that the drug will begin to A ranging clinical and economic benefits, from reducing the risk of suicide to shortening hospital stays to returning pa- tients to normal activity sooner. Although data from sev- eral published clinical studies have hinted at time-effect differences among antidepressants, 1–4 none of the trials was adequately conceived or conducted to measure such differences. The aim of the following exercise was to de- sign a study that would be ideal for assessing comparative onset of antidepressant effect.