Secondary Logo

Institutional members access full text with Ovid®

Patient Adherence and Medical Treatment Outcomes: A Meta-Analysis

Robin DiMatteo, M. PhD*; Giordani, Patrick J. MA*; Lepper, Heidi S. PhD; Croghan, Thomas W. MD

Original Articles

Background.  Adherence is a factor in the outcome of medical treatment, but the strength and moderators of the adherence-outcome association have not been systematically assessed.

Objectives.  A quantitative review using meta-analysis of three decades of empirical research correlating adherence with objective measures of treatment outcomes.

Method.  Sixty-three studies assessing patient adherence and outcomes of medical treatment were found involving medical regimens recommended by a nonpsychiatrist physician, and measuring patient adherence and health outcomes. Studies were analyzed according to disease (acute/chronic, severity), population (adult/child), type of regimen (preventive/treatment, use of medication), and type and sensitivity of adherence and outcomes measurements.

Results.  Overall, the outcome difference between high and low adherence is 26%. According to a stringent random effects model, adherence is most strongly related to outcomes in studies of nonmedication regimens, where measures of adherence are continuous, and where the disease is chronic (particularly hypertension, hypercholesterolemia, intestinal disease, and sleep apnea). A less stringent fixed effects model shows a trend for higher adherence-outcome correlations in studies of less serious conditions, of pediatric patients, and in those studies using self-reports of adherence, multiple measures of adherence, and less specific measures of outcomes. Intercorrelations among moderator variables in multiple regression show that the best predictor of the adherence-outcome relationship is methodological—the sensitivity/quality of the adherence assessment.

*From the Department of Psychology, University of California, Riverside, California.

†From the Department of Psychology, Drake University, Des Moines, Iowa.

‡From the Health Outcomes Evaluation Group, Eli Lilly and Company, and Indiana University, Indianapolis, Indiana.

Supported by grants from Eli Lilly and Company and the University of California, Riverside, Center for Ideas and Society and Academic Senate Committee on Research.

Address correspondence and reprint requests to: M. Robin DiMatteo, Department of Psychology, University of California, Riverside, CA 92521. E-mail:

Received August 22, 2001; initial review November 5, 2001; accepted February 28, 2002.

© 2002 Lippincott Williams & Wilkins, Inc.