Explaining ethnic inequalities in health in Scotland: a pilot study linking primary care to census data to account for ethnic variations in cardiovascular disease

Abstract

BackgroundEthnic health inequality is striking. Quantifying this inequality is important. The Scottish Health and Ethnicity Linkage Study (SHELS) linked National Health Service hospital discharges and mortality to the 2001 Scottish Census. Our analyses adjusted for age, country of birth, and socioeconomic status. However, the understanding of the role of other risk factors is crucial. The aim of this project was to explore the feasibility and value of linking risk factors from primary care to census and health data in cardiovascular disease. MethodsTen of 17 general practices in Glasgow and Edinburgh provided data relating to cardiovascular disease that were linked to census records using encrypted identifiers (for ethnic group labels, we used Census Scotland terminology). This process created a retrospective cohort study comprising 52 975 people, of whom around 8500 belonged to non-White ethnic groups. This cohort was linked to our pre-existing data on hospital admissions for cardiovascular disease. Risk ratios (RR) for first hospital admission were calculated, by sex, using Poisson regression with robust variance, adjusting for age, and then for smoking or diabetes status, with 95% CIs. The White Scottish population was the reference group (RR=100). The main outcome assessed whether ethnic variations in cardiovascular disease were attenuated on adjustment for smoking and diabetes. Ethics approval (11/MRE00/4) was obtained from Scotland A Research Ethics Committee. FindingsData completeness was similar across ethnic groups. 48 325 (91%) of 52 975 records had a valid smoking status and 2900 (5·5%) people had diabetes. Compared with White Scottish people, Pakistani people had the highest prevalence of diabetes (RR for men 274, 95% CI 238–314; RR for women 364, 305–434). Pakistani men (n=1165) and women (n=1075) were at higher risk of any hospital admission for cardiovascular disease than were White Scottish men (n=9890) and women (n=11 550) (153, 137–170 for men; 193, 158–235 for women). In Pakistani women adjustment for smoking increased the age-adjusted risk (216, 170–274), whereas adjustment for diabetes reduced the risk (161, 134–194). InterpretationThe high rate of hospital admissions for cardiovascular disease among Pakistani people can be explained partly by higher rates of diabetes, but not smoking. Our dataset is a non-representative sample, so results should be treated with caution. However, we have shown potential value (for further research and for health-care planning) in linking primary care to census data to shed light on ethnic variations in cardiovascular disease. FundingChief Scientist Office of the Scottish Government (grant numbers CZH/4/432, CZH/4/648, and CZH4/878).