Spatial Co-Clustering of Cardiovascular Diseases and Select Risk Factors among Adults in South Africa.

Abstract

Background: Cardiovascular diseases (CVDs) are part of the leading causes of mortality and morbidity in developing countries, including South Africa, where they are a major public health issue. Understanding the joint spatial clustering of CVDs and associated risk factors to determine areas in need of enhanced integrated interventions would help develop targeted, cost-effective and productive mediations. We estimated joint spatial associations and clustering patterns of 2 CVDs (stroke and heart attack) and 3 risk factors (hypertension, high blood cholesterol (HBC) and smoking) among adults in South Africa. Methods: We used cross-sectional secondary adult (15–64-year olds) health data from the South African Demographic Health Survey 2016. Age and gender standardized disease incidence ratios were analyzed using joint spatial global and local bivariate Moran’s Index statistics. Results: We found significantly positive univariate spatial clustering for stroke (Moran; s Index = 0.128), smoking (0.606) hypertension (0.236) and high blood cholesterol (0.385). Smoking and high blood cholesterol (0.366), smoking and stroke (0.218) and stroke and high blood cholesterol (0.184) were the only bivariate outcomes with significant bivariate clustering. There was a joint stroke-smoking local “hot spots” cluster among four districts in the urban western part of the country (City of Cape Town; Cape Winelands; Overberg and Eden) and a joint “cold spots” cluster in the rural north-western part of the country. Similar joint “hot spots” clustering was found for stroke and high blood cholesterol, which also had “cold spots” cluster in the rural east-central part of the country. Smoking and high blood cholesterol had a “hot spots” cluster among five districts in the urban western part of the country (City of Cape Town; Cape Winelands; Overberg; Eden, and West Coast) and “cold spots” around the rural districts in east-southern parts of the country. Conclusions: Our study showed that districts tended to co-cluster based on the rates of CVDs and risk factors, where higher rates were found in urban places than in rural areas. These findings are suggestive of a more contagious and spatial diffusion process among interdependent districts in urban districts. Urbanization or rurality needs to be considered when intervention initiatives are implemented with more general approaches in rural areas. The finding of “hot spot” co-clusters in urban areas means that integrated intervention programmes aimed at reducing the risk of CVDs and associated risk factors would be cost-effective and more productive.