Iyanuloluwa4/DHS-Analysis-for-Socioeconomic-correlates-of-stroke-risk-factor-profiles-by-sex-in-South-Africa

# DHS-Analysis-for-Socioeconomic-correlates-of-stroke-risk-factor-profiles-by-sex-in-South-Africa # South Africa DHS Stroke Risk Factor Analysis ## Overview This repository contains the complete R workflow used to prepare, analyse, and model nationally representative data from the 2016 South Africa Demographic and Health Survey (DHS). The project investigates sex-specific socioeconomic and behavioural correlates of major modifiable stroke risk factors using complex survey methods. The analyses were conducted as part of a Master of Public Health (International) dissertation at the University of Leeds. --- ## Research Question **What are the socioeconomic correlates of major modifiable stroke risk factors among South African adults, and do these associations differ by sex?** --- ## Objectives 1. Estimate the weighted prevalence of hypertension, diabetes, obesity, and smoking among South African adults. 2. Examine associations between socioeconomic characteristics and each stroke risk factor using survey-weighted logistic regression. 3. Investigate whether socioeconomic characteristics modify the relationship between sex and each stroke risk factor through interaction analyses. --- ## Data Source South Africa Demographic and Health Survey (SADHS) 2016 The analysis uses the following DHS datasets: * Household Recode (HR) * Individual Recode (IR) * Men's Recode (MR) These datasets are **not included** in this repository. Researchers can request access through the DHS Program. --- ## Outcomes The following stroke risk factors were examined: * Hypertension * Diabetes * Obesity * Current smoking Outcome definitions were based on internationally accepted clinical thresholds and DHS biomarker protocols. --- ## Key Features * Complex survey design analysis * Survey weighting * Stratified cluster sampling * Biomarker data processing * Household-to-individual linkage * Interaction modelling * Sex-stratified analyses * Survey-weighted logistic regression …