Symptom Clusters as Predictors of Late Response to Antidepressant Treatment

Madhukar H. Trivedi, M.D.; David W. Morris, Ph.D.; Bruce D. Grannemann, M.A.; and Susan Mahadi, M.Ed.

Objective: While there is some indication from studies in the acute phase of antidepressant treatment that there are differences in the timing of improvement in symptoms, relatively little work has explored the patterns of change for specific symptom clusters and the predictability of these changes to signal eventual response during the acute phase of treatment. This article investigates the use of clusters of symptoms on the 17-item Hamilton Rating Scale for Depression (HAM-D-17) to define the pattern of late response versus nonresponse to antidepressant medication.

Method: Using principal component analysis, the HAM-D-17 was divided into 4 symptom clusters (mood, sleep/psychic anxiety, appetite, and somatic anxiety/weight). Data for 996 patients with major depressive disorder (DSM-III-R criteria), who participated in a 12-week acute phase study with nefazodone, were subjected to a post hoc analysis of changes in symptom cluster scores. Patients were divided into 3 groups: early responders (< 4 weeks), late responders (4-12 weeks), and nonresponders (> 12 weeks) as defined by < 50% reduction in HAM-D-17 scores from baseline. The late-responder and nonresponder groups were subjected to the principal component analysis. Data were collected from October 1992 to November 1994.

Results: There were significant differences in the pattern of symptom change on the mood cluster (weeks 3-4) (p < .0001), the sleep/psychic anxiety cluster (weeks 3-4) (p < .003), and the somatic anxiety/weight cluster (weeks 3-4) (p < .01) for the late responders compared to the nonresponders. Using change scores, a discriminant function analysis correctly assigned 127 of the 182 late responders and 85 of the 133 nonresponders, or 70% of the late responders and 64% of the nonresponders, to their final response groups.

Conclusion: Monitoring changes in symptom clusters from the HAM-D-17 during this crucial early stage (first 4 weeks) can be used to distinguish late responders (after week 4) from nonresponders. Successful identification of nonresponders based on symptom cluster change in the first 4 weeks would facilitate a shortening of an ineffective treatment trial and allow for necessary changes in treatment strategy, helping physicians more closely follow treatment guidelines.

(J Clin Psychiatry 2005;66:1064-1070)

Received Aug. 10, 2004; accepted Feb. 7, 2005. From the Department of Psychiatry, The University of Texas Southwestern Medical Center, Dallas.

The study was supported in part by a contract from Bristol-Myers Squibb. This report was funded in part by National Institute of Mental Health (NIMH) grants 1UO1MH61562-01A2 MHT and R01MH064062-01A2 MHT.

Dr. Trivedi has been a consultant for or served on the speakers boards of Bristol-Myers Squibb, Janssen, Eli Lilly, Organon, Pharmacia, Solvay, Wyeth, Pfizer, and Cyberonics and has received grant support from NIMH, Abbott, Akzo (Organon), Bayer, Bristol-Myers Squibb, Eli Lilly, Forest, GlaxoSmithKline, Janssen, Johnson & Johnson, Mead Johnson, Parke-Davis, Pfizer, Pharmacia, Solvay, Wyeth, and the National Alliance for Research on Schizophrenia and Depression. Dr. Morris, Mr. Grannemann, and Ms. Mahadi report no financial affiliation or other relationship relevant to the subject of this article.

The authors appreciate the clerical and administrative assistance of Melissa Haldeman, B.S., and the administrative support of Eric Nestler, M.D., Ph.D., professor and chairman, Department of Psychiatry, University of Texas Southwestern Medical Center, Dallas.

Corresponding author and reprints: Madhukar H. Trivedi, M.D., Mood Disorders Program & Clinic, Department of Psychiatry, University of Texas Southwestern Medical Center, Exchange Park Express, American General Tower, 6363 Forest Park Road, Suite 13.354, Dallas, TX, 75390-9119 (e-mail: madhukar.trivedi@utsouthwestern.edu).