In addition, consistent associations between depression
and stroke were observed in a variety of supplemental
models. In models that excluded early stroke
events, depression continued to be prognostic of
stroke. In models that excluded individuals with heart
disease or diabetes at baseline, the depression-stroke
connection was similarly unaffected. Wassertheil-
Smoller et al. (17) and Penninx et al. (42) showed that
only change in depression over time, rather than baseline
depression, was predictive of future cardiovascular
events. This finding implies that vascular disease
could be a potential cause of both subsequent depression
and stroke. Because baseline depression was predictive
of stroke in the analyses reported here, and
because the impact of excluding early stroke cases (up
to 10 years after the NHANES I examination) or subjects
with baseline heart disease or diabetes on these
analyses was minimal, we conclude that baseline vascular
disease leading first to depression and subsequently
to stroke is not a likely explanation for the
observed associations. Baseline age was a strong predictor
of stroke incidence, with incidence rates rising
sharply with increasing age. However, analyses stratified
by age showed that depression was consistently
associated with stroke incidence in both th