A cross-sectional study of cardiovascular disease risk clustering at different socio-geographic levels in India

Anne C Bischops,Jan-Walter De Neve,Sebastian Vollmer,Till Bärnighausen,Pascal Geldsetzer,Ashish Awasthi

doi:10.1038/s41467-020-19647-3

Abstract

Despite its importance for the targeting of interventions, little is known about the degree to which cardiovascular disease (CVD) risk factors cluster within different socio-geographic levels in South Asia. Using two jointly nationally representative household surveys, which sampled 1,082,100 adults across India, we compute the intra-cluster correlation coefficients (ICCs) of five major CVD risk factors (raised blood glucose, raised blood pressure, smoking, overweight, and obesity) at the household, community, district, and state level. Here we show that except for smoking, the level of clustering is generally highest for households, followed by communities, districts, and then states. On average, more economically developed districts have a higher household ICC in rural areas. These findings provide critical information for sample size calculations of cluster-randomized trials and household surveys, and inform the targeting of policies and prevention programming aimed at reducing CVD in India.

Highlights

Despite its importance for the targeting of interventions, little is known about the degree to which cardiovascular disease (CVD) risk factors cluster within different socio-geographic levels in South Asia
Because sociodemographic information in the DLHS-4 and Annual Health Survey (AHS) was collected for all household members from the household head, participants with missing outcome data included eligible household members who were not present at the time of the study team visit
The DLHS-4 and AHS jointly contained data on 515,689 households, 17,841 communities, and 561 districts. 52.5% of participants were female and 42.0% were younger than 36 years (Table 1). 9.1% of participants were obese (BMI ≥ 27.5 kgm−2), 7.7% had a raised blood glucose (BG), and 26.9% had a raised blood pressure (BP)

Summary

Introduction

Despite its importance for the targeting of interventions, little is known about the degree to which cardiovascular disease (CVD) risk factors cluster within different socio-geographic levels in South Asia. Many of the factors that increase one’s probability of developing CVD risk factors— including diet[11], characteristics of the built environment[12], social networks[13], and genetics14—tend to be shared to at least some degree by individuals in the same households and communities It is, unsurprising that studies have shown that diabetes, hypertension, obesity, and smoking tend to co-occur in households and larger community structures, such as neighborhoods[15,16]. We use nationally representative household survey data from India to determine the intracluster correlation coefficients (ICCs) of five major CVD risk factors at each of four different socio-geographic levels (household, community, district, and state). Policymakers need to decide whether to target a screening program for diabetes and hypertension at specific communities or types of households, or if they should instead disregard these socio-geographic units, such as by screening everyone above a certain age threshold

Objectives

Methods

Results

Conclusion