Depressive symptoms predict incident stroke independently of memory impairments

Abstract

Background: We evaluated whether depressive symptoms predict the onset of first stroke independently of memory impairment. We conceptualized memory impairment as a marker of preexisting cerebrovascular disease. We hypothesized that if depressive symptoms are causally related to stroke through mechanisms unrelated to cerebrovascular disease, depressive symptoms should predict stroke independently of memory impairment.

Methods: Incidence of first stroke was assessed with self or proxy reports from 19,087 participants in the Health and Retirement Study cohort (1,864 events). Elevated depressive symptoms (3+ on an 8-item Centers for the Epidemiologic Study of Depression scale) and memory impairment (score of ≤6 on a combined immediate and delayed recall of a 10-word list) were used as predictors of incident stroke in Cox survival models with adjustment for sociodemographic and cardiovascular risk factors.

Results: After adjustment for sociodemographic and cardiovascular risk factors, elevated depressive symptoms (hazard ratio = 1.25; 95% confidence interval 1.12-1.39) and memory impairment (hazard ratio = 1.26; 95% confidence interval 1.13-1.41) each predicted stroke incidence in separate models. Hazard ratios were nearly unchanged and remained significant (1.23 for elevated depressive symptoms and 1.25 for memory impairment) when models were simultaneously adjusted for both elevated depressive symptoms and memory impairment. Elevated depressive symptoms also predicted stroke when restricting analyses to individuals with median memory score or better.

Conclusions: Memory impairments and depressive symptoms independently predict stroke incidence. Memory impairment may reflect undiagnosed cerebrovascular disease. These results suggest that depressive symptoms might be directly related to stroke rather than merely indicating preexisting cerebrovascular disease.

Figure 1 Adjusted survival curves by baseline elevated depressive symptoms (A) and memory impairment (B) Survival curves calculated after inverse probability weighting for age at baseline (linear and quadratic), Hispanic ethnicity, race, southern birth, mother's and father's education >8 years, and year of Health and Retirement Study enrollment. Curves for elevated depressive symptoms (left) also weighted by baseline memory impairment; curves for memory impairment (right) also weighted by baseline elevated depressive symptoms.

Figure 2 Hazard ratios for incidence of first stroke associated with each item on the 8-item Centers for Epidemiologic Study of Depression (CES-D) Scale All models were adjusted for age at baseline (linear and quadratic), Hispanic ethnicity, black race, southern birth, mother's and father's education >8 years, year of Health and Retirement Study enrollment, years of education, baseline income, wealth, and marital status. Separate models included only a single CES-D item. Separate plus baseline memory impairment models included that CES-D item plus an indicator for baseline memory impairment. The third set of models “modeled together” included all 8 items from the CES-D in the same model, as well as baseline memory impairment.