Validity of self-reported versus hospital-coded diagnosis of stroke: a cross-sectional and longitudinal study.

Background: Population-based studies, as well as clinicians, often rely on self-report and hospital records to obtain a history of stroke. This study aimed to compare the validity of the diagnosis of stroke by self-report and by hospital coding according to their cross-sectional association with prevalent vascular risk factors, and longitudinal association with recurrent stroke and major cardiovascular outcomes in a large cohort of older Australian men.

Methods: Between 1996 and 1999, 11,745 older men were surveyed for a self-reported history of stroke as part of the Health in Men Study (HIMS). Previous hospitalization for stroke was obtained with consent from linked medical records via the Western Australian Data Linkage System (WADLS). Subjects were followed by WADLS until December 31, 2010, for hospitalization for stroke, cardiovascular events, and all-cause mortality. The primary outcome was hospitalisation for stroke during follow-up. Secondary outcomes included incident vascular events and composite vascular endpoints.

Results: At baseline, a history of stroke was reported by 903 men (7.7%), previous hospitalisation for stroke was recorded in 717 (6.1%), both self-report and hospitalisation in 467 (4.0%), and no history of stroke in 10,696 men (91.1%). Prevalent cardiovascular disease and peripheral arterial disease were more common among men with previous hospitalisation for stroke than a history of self-reported stroke (p < 0.001). In longitudinal analyses, incident aortic aneurysm was also more common among men with baseline history of hospitalization for stroke (adjusted hazard ratio (HR) 1.71, 95% CI 1.12-2.60) than among men with self-reported stroke (HR 0.88, 95% CI 0.56-1.36) compared to men with no history of stroke. With regard to the primary outcome, the rate of hospitalisation for stroke during follow-up was significantly higher among men with self-reported stroke (HR 2.44, 95% CI 2.03-2.94), hospital-coded stroke (adjusted HR 3.02, 2.42-3.78) and both self-reported and hospital-coded stroke (adjusted HR 3.33, 2.82-3.92) compared to participants with no previous stroke. Time to recurrent stroke was similar among different methods of initial stroke diagnosis (p = 0.067).

Conclusions: Self-reported stroke and hospital-coded stroke have a similar prognostic value for predicting the risk of recurrent stroke. This supports the use of these ways of assessing a history of stroke for the clinical purposes of secondary prevention and for further epidemiological studies.