Fluoxetine has been widely prescribed for depression in elderly patients. Numerous clinical trials have demonstrated that it is more effective than placebo and is comparable to other antidepressants in this population, with response rates ranging from 35 to 60 percent. However, a high rate of improvement among patients who receive placebo in placebo-controlled trials (e.g. 20 to 40 percent) frequently complicates the determination of true drug effect. Therefore, it would be useful to compare characteristics of placebo-responders with fluoxetine-responders to distinguish between placebo and drug effects. The proposed study will compare responders and nonresponders, with data generated from a multicenter, randomized, double-blind, clinical trial to determine the effectiveness and safety of fluoxetine relative to placebo in elderly patients with major unipolar depression (by DSM III-R criteria). More than 670 patients at least 60 years of age participated in this 6-week trial. Subjects were dropped prior to randomization if they demonstrated a 20 percent or greater improvement on the HDRS during a preliminary, one-week placebo period. Similarities and differences between responders and nonresponders will be evaluated with respect to such clinical and demographic factors as severity of illness, length of illness and age of onset, type of symptoms (i.e., anxiety, sleep disturbances), age, sex, race, and concurrent medical illnesses. In addition, because side effects may indicate a biological sensitivity to a drug, the early side effects will be evaluated as possible predictors of response. A response profile will be constructed from those variables found, by logistic and linear regression models, to be associated with good therapeutic outcome. A response profile will be similarly constructed for placebo-responders. Response profiles will then be compared for both drug and placebo groups. Receiver operator characteristic (ROC) curve analysis will be used to evaluate the ability of response profiles to distinguish between responders and nonresponders. Identification of characteristics of people most likely to improve with fluoxetine or placebo can be used to improve therapeutic guidelines and lead to better-designed clinical trials.
| Ackerman, D L; Greenland, S; Bystritsky, A et al. (2000) Side effects and time course of response in a placebo-controlled trial of fluoxetine for the treatment of geriatric depression. J Clin Psychopharmacol 20:658-65 |
| Ackerman, D L; Greenland, S; Bystritsky, A et al. (1997) Characteristics of fluoxetine versus placebo responders in a randomized trial of geriatric depression. Psychopharmacol Bull 33:707-14 |
