Background: Ethnic variations in stroke require more European studies, especially as differences are reportedly large.
Methods: We created a retrospective cohort study of 4.65 million people in Scotland linking ethnicity from the census and stroke incidence and mortality from NHS databases. Rate ratios using direct age standardization and risk ratios were calculated, the latter to model the influence of educational qualification in a Poisson regression model.
Results: Age-adjusted rate ratios varied little, compared to the White Scottish group (reference value 100) and the 95% CIs usually included 100, e.g. higher in Pakistani men (120.5, 95% CI 95.2-145.8) and in African men (137.9, 95% CI 91.5-184.4) but not in Pakistani or African women. Stroke rates were low in the Other White British (78.3, 95% CI 75.4-81.2 in men and 84.9, 95% CI 82.0-87.8 in women), Other White (89.8, 95% CI 81.5-98.1 in men and 88.8, 95% CI 80.9-96.7 in women) and Chinese men (70.3, 95% CI 45.7-94.8). Adjusting for highest educational qualification attenuated some and augmented other risk ratios, e.g. in Other White British men, the risk ratio changed from 71.4 to 80.2 (95% CI 74.2-86.6) and in African men from 124.2 to 138.8 (95% CI 107.7-178.8).
Conclusions: Ethnic variations deserve further study, including in White European origin subgroups and the Chinese. Extremely high rates in South Asian and African origin were not corroborated in Scotland. Linkage methods are practical in Europe.