The goal of this study was to develop a profile, using multivariate statistical techniques, which would predict the probability of a patient's response to placebo. The interest was in baseline factors (demographic, physical and psychiatric history, and psychiatric rating scale scores) that might distinguish placebo responders from non-responders in clinical trials in depression. The patient population studied consisted of 1048 patients who were treated only with placebo during randomized, double-blind clinical trials investigating the efficacy of two new antidepressant compounds. All subjects were outpatients, suffering from moderate to severe major depressive episodes. A positive placebo responder was defined as a patient who, at his or her final evaluation during the six-week study, showed a decrease of at least 50% in the baseline total score of the Hamilton Psychiatric Rating Scale for Depression. All other patients were classified as non-responders. Thirty-two percent of the patients (344) were classified as responders and 68% (704) were non-responders. Baseline differences between responders and non-responders were analyzed by univariate and multivariate methods, including logistic regression. Patients were randomly divided into two groups. The first multivariate logistic regression model was formulated using the data from one group and validated with data from the second group. The final model was formulated using the entire group; due to missing values, this model was based on the data from 1026 patients. This final model included: age, family adjustment, precipitating stress, previous psychiatric treatment, and race. Those patients who responded better to placebo were younger, had an adequate family adjustment, had not been previously treated, and were suffering from definite or probable precipitating stress. Those who had poor family adjustment, had been previously treated, and had no precipitating stress had a lower probability of responding. In general, those patients with a more deep-seated, severe depression were less likely to respond to placebo. This model is indicative only of response trends in this population and should not be used to predict an individual's possible response to placebo until it has been assessed in additional populations.
| Identifer | oai:union.ndltd.org:UMASS/oai:scholarworks.umass.edu:dissertations-7761 |
| Date | 01 January 1990 |
| Creators | Woodin, Karen Estelle |
| Publisher | ScholarWorks@UMass Amherst |
| Source Sets | University of Massachusetts, Amherst |
| Language | English |
| Detected Language | English |
| Type | text |
| Source | Doctoral Dissertations Available from Proquest |
Page generated in 0.0637 seconds