Trajectories of Treatment Response in Late-Life Depression

Abstract

The authors examined the effect of psychosocial and clinical variables on treatment response trajectory in elderly patients with major depressive disorder. Three studies provided data on treatment response in 360 elderly depressed subjects who participated in protocols using either nortriptyline or paroxetine as monotherapy or, in 2 studies, combined with interpersonal psychotherapy. Treatment response was assessed with the Hamilton Rating Scale for Depression-17 Item (HRSD-17) score over 12 weeks of acute treatment in each study. The mixture-modeling method of trajectory analysis was used to identify different subpopulations of response, and to determine whether baseline HRSD-17 score, depressive illness course (single or recurrent), current episode duration, Interpersonal Self Evaluation List-Self-esteem factor, age at study entry, and medical burden were risk factor covariates associated with response trajectory. As a contrast, logistic regression was used to assess the association between the same covariates and the probability of response (defined as HRSD-17 < or =10 and 50% reduction from baseline). In each study, there were 2 response trajectories with similar course, but with different speed. We classified the trajectories as "rapid response" and "slower response." Baseline HRSD-17 score was a significant predictor of response trajectory, with higher initial score related to slower response trajectory. Higher self-esteem was associated with more rapid response trajectory. In the logistic regression analysis, in two of the studies, higher baseline HRSD-17 score was a significant risk factor for nonresponse. In the study without psychotherapy, higher self-esteem was associated with responding to treatment. Thus, trajectory analysis can identify different trajectories of responders and determine psychosocial and clinical variables associated with response trajectory in the acute treatment of geriatric depression. Further study focusing on risk factors associated with slower response may help optimize treatment in elderly patients who do not respond quickly to first-line therapies.