liciazheng/cross-national-public-health-analysis

HIV prevalence, ART coverage and PMTCT effectiveness in South Africa, Kenya and Botswana, 2007–2016 # Cross-National Public Health Data Analysis — HIV in Sub-Saharan Africa A comparative analysis of HIV prevalence, treatment coverage, and prevention-programme effectiveness in **South Africa, Kenya, and Botswana** from **2007 to 2016**, joined against World Bank economic indicators. The whole pipeline is one script: it joins the raw extracts, runs data-quality checks against them, writes a cleaned dataset, and regenerates every figure below. ```bash pip install -r requirements.txt python analysis.py ``` ## Questions 1. How did HIV prevalence change in these three countries between 2007 and 2016? 2. How far did antiretroviral therapy (ART) coverage expand? 3. How successful were Prevention of Mother-to-Child Transmission (PMTCT) programmes? 4. Does economic development predict HIV outcomes? Sub-Saharan Africa accounts for roughly 70% of global HIV cases. These three countries span a wide range of both epidemic scale and national income, which makes them a useful comparison set: Botswana is small and relatively wealthy with a severe epidemic, Kenya is large and low-income with a milder one, South Africa is large and carries the biggest absolute burden in the world. ## Findings ### Treatment coverage expanded dramatically everywhere ART coverage rose from 10% to 60% in South Africa, 13% to 74% in Kenya, and 29% to 76% in Botswana. It is the clearest signal in the dataset. ### Treatment tracks mortality almost perfectly Within each country, ART coverage and AIDS deaths per 100,000 move together with a correlation of **r = −0.98 to −1.00**. Population-normalised mortality fell 69% in South Africa, 59% in Kenya, and 78% in Botswana. A near-perfect within-country correlation over ten years is not proof of causation — both series trend monotonically, and plenty of other things improved over the same decade. But the direction and consistency across three very different health systems is what the ART literature would predict. ### Income does **not** predict …