cyrilt123/Cyril_PHD_Paper2_SA

One Country, Different Clusters? Multimorbidity Patterns Across Rural South Africa, 2012–2022 # One Country, Different Clusters? Multimorbidity Patterns Across Rural South Africa (2012–2022) This repository contains the R code used to clean, harmonise, analyse, and visualise verbal autopsy (VA) data from three rural South African Health and Demographic Surveillance System (HDSS) sites: **Agincourt**, **AHRI**, and **DIMAMO**. The study examines whether **pooled multimorbidity clusters** are consistent with **site-specific clusters** across the three rural South African sites, using descriptive epidemiology, exact condition-combination analyses, and latent class analysis (LCA). --- ## Project overview The main script: - loads and harmonises VA datasets from multiple source files - standardises variable names and classes across sites - derives chronic condition indicators and demographic variables - filters the analytic sample to **Agincourt, AHRI, and DIMAMO** - produces descriptive tables and plots - compares multimorbidity burden across sites - examines exact multimorbidity combinations - supports pooled and site-specific multimorbidity analyses --- ## Study objective The primary aim of this analysis was to assess the extent to which **pooled latent multimorbidity class profiles** differed from **site-specific latent class profiles** among decedents in Agincourt, AHRI, and DIMAMO, rural South Africa, between **2012 and 2022**. Secondary objectives included: - comparing the prevalence of individual chronic conditions across sites - comparing multimorbidity burden across sites - examining exact condition-combination patterns - describing pooled and site-specific latent classes by age group and sex --- ## Repository contents - `SA_sites.R` — main R script for data cleaning, harmonisation, descriptive analysis, and plotting - `README.md` — project description and usage instructions --- ## Data sources This project uses harmonised **verbal autopsy (VA)** data from rural South African HDSS sites. ### Sites included in the final analytic sample - …