Revealing the Complex Relationships Between Ethnicity and Health: 29 Outcomes From the Scottish Health and Ethnicity Linkage Study (SHELS) Cohort

Abstract

Objectives: To present a data visualization approach based on epidemiological principles that can be used to illustrate the complexity of health inequalities by ethnicity. Methods: We provide an example of our technique by presenting results from the Scottish Health and Ethnicity Linkage Study (SHELS) on 29 different health outcomes across ten ethnic groups. SHELS is a census-based retrospective cohort study of the general population of Scotland (4.62 million people) linking ethnicity data to health records. We employed different colors to show health advantage, disadvantage or equivalence; different color shades to represent degree of certainty, combining effect size and precision of estimate; and different font sizes for absolute rates, to highlight more common conditions. We ranked health conditions by age-adjusted rate within each ethnic group to identify differences in burden of disease. Results: Data visualization clearly demonstrates that ethnic differences in health vary substantially depending on outcome, sex and ethnic group. Inequalities were present for every outcome but were not uniformly worse for non-White populations. The patterns are also complex within each ethnic group such as the Pakistani population where strong advantage for some outcomes (e.g. RR=0.52 in men for all-cause mortality) coexists with strong disadvantage for others (e.g. RR=1.80 in women for cardiovascular disease). Despite large differences in relative disease rates, ranking conditions within ethnic groups showed that most ethnic groups have similar disease priorities. Conclusions: Our approach demonstrates the complexity of ethnic differences in health. The use of these simple data visualization techniques can facilitate interpretation of this complexity and avoid misleading generalizations that the health of ethnic minorities is worse or better than majority populations. Using absolute rates of disease and ranking conditions within ethnic groups is also useful as large relative differences in rates of disease between ethnic groups may not always translate into different disease priorities.