Stroke is a leading cause of mortality among Canadian adults. There is currently very little information on the utility of stroke prediction algorithms for the Canadian population. The purpose of this study was to determine if the Framingham stroke risk prediction model (D’Agostino et al.
1994;25:40-43) was associated with stroke mortality in a population-based sample of Canadian adults.
The sample included 3,163 adults 55-74 y of age who participated in the Canadian Heart Health Surveys (1986-95) and were free from stroke at baseline. The analysis includes all deaths that occurred between baseline data collection and December, 2004. A modified Framingham Stroke Score was assigned to each participant at baseline based on their age, sex, treated or untreated systolic blood pressure levels, diabetes, smoking, and history of cardiovascular disease. Data on history of atrial fibrillation and left ventricular hypertrophy were not available. Hazard ratios for all-cause and stroke mortality were estimated using Cox proportional hazards regression with exam year as a covariate, and multivariable models also included body mass index, alcohol consumption and education level as covariates.
There were 915 deaths (70 from stroke) over 12.2 y (0.5 to 16 y) of follow-up. The hazard ratios (95% C.I.) for all-cause mortality associated with the Framingham Stroke Score were 1.12 (1.10 - 1.15) in men and 1.18 (1.15 - 1.22) in women. The hazard ratios for stroke mortality were 1.15 (1.06 - 1.25) in men and 1.22 (1.11 - 1.36) in women. When comparing men and women with scores greater than 8 and 9, respectively (median) to those with lower scores, the hazard ratios were 2.34 (1.20 - 4.58) in men and 4.81 (1.85 - 12.50) in women. The additional inclusion of body mass index, alcohol consumption and education to the models did not appreciably change the results.
The Framingham Stroke Score is a significant predictor of all-cause and stroke mortality in Canadian adults, and the risk is independent of body mass index, alcohol consumption and education level.