Depression after stroke: a prospective epidemiological study.

OBJECTIVES: To elucidate the relationship between stroke and depressive symptoms and to determine whether disability or cerebrovascular risk factors mediate that relationship. DESIGN: A prospective longitudinal epidemiological survey. SETTING: The mid-Monongahela Valley, a rural, nonfarm, low-socioeconomic-status community. PARTICIPANTS: Random sample of 1,134 subjects aged 65 and older. MEASUREMENTS: The dependent variable was clinically significant depressive symptoms, as defined by five or more symptoms on the modified Center for Epidemiological Studies Depression scale. The independent variables were demographics (age, sex, education), stroke, number of impaired instrumental activities of daily living (IADLs), diabetes mellitus, hypertension, atherosclerotic heart disease, and smoking. Logistic regression analyses were conducted for cross-sectional and longitudinal models examining whether stroke was associated with or predicted depressive symptoms, with other associated factors included as covariates. RESULTS: Clinically significant depressive symptoms were cross-sectionally associated with stroke (odds ratio (OR)=3.5, 95% confidence interval (CI)=1.4-8.3), diabetes mellitus (OR=2.8, 95% CI=1.7-4.6; P