Meta-Analysis: Chronic Disease Self-Management Programs for Older Adults

Although enthusiasm is growing for self-management programs for chronic conditions, there are conflicting data regarding their effectiveness and no agreement on their essential components. PURPOSE: To assess the effectiveness and essential components of self-management programs for hypertension, osteoarthritis, and diabetes mellitus. DATA SOURCES: The authors searched multiple sources dated through September 2004, including the Cochrane Library, MEDLINE, PsycINFO, and Nursing and Allied Health databases, and bibliographies of 87 previous reviews. STUDY SELECTION: Randomized trials that compared outcomes of self-management interventions with a control or with usual care for diabetes mellitus, osteoarthritis, or hypertension; outcomes included hemoglobin A1c level, fasting blood glucose level, weight, blood pressure, pain, or function. DATA EXTRACTION: Two reviewers independently identified trials and extracted data regarding whether the intervention used tailored adjustments to meet individual patient needs, a group setting, feedback, and psychological services, and whether the intervention was provided by the patient's usual physician. DATA SYNTHESIS: Of 780 studies screened, 53 studies contributed data to the random-effects meta-analysis (26 diabetes studies, 14 osteoarthritis studies, and 13 hypertension studies). Self-management interventions led to a statistically and clinically significant pooled effect size of -0.36 (95% CI, -0.52 to -0.21) for hemoglobin A1c, equivalent to a reduction in hemoglobin A1c level of about 0.81%. Self-management interventions decreased systolic blood pressure by 5 mm Hg (effect size, -0.39 [CI, -0.51 to -0.28]) and decreased diastolic blood pressure by 4.3 mm Hg (effect size, -0.51 [CI, -0.73 to -0.30]). Pooled effects of self-management interventions were statistically significant but clinically trivial for pain and function outcomes for osteoarthritis. No consistent results supported any of the 5 characteristics examined as essential for program success.

Studies had variable quality, and possible publication bias was evident.