FIXED-EFFECT PANEL COUNTRY-SPECIFIC TRENDS IN HIV PREVALENCE, 1990 -2024 ANALYSIS OF 150 COUNTRIES

Abstract

This study examines changes in HIV prevalence among people aged 15–49 years across 150 countries over 1990–2024. The objective is to estimate country-specific annual changes and identify countries with statistically significant declining trajectories. The World Development Indicators series SH.DYN.AIDS.ZS, sourced from UNAIDS estimates, was reshaped into a country-year panel. A conventional country fixed-effects model with a common time trend was first estimated, followed by a fixed-effects specification allowing country-specific linear trends. Country-specific slopes were estimated with Newey–West heteroskedasticity and autocorrelation-consistent standard errors using three lags; false-discovery-rate adjustment was also examined as a multiple-testing sensitivity check. The common fixed-effects trend was statistically indistinguishable from zero (β = 0.001, cluster-robust SE = 0.008, p = 0.868), while the test for heterogeneous country slopes was highly significant (F(149, 4,944) = 41.86, p < .001). Twenty-four countries out of one hundred and fifty countries recorded statistically significant negative annual trends at the 5% level, and all 24 remained significant after Benjamini–Hochberg adjustment. Zimbabwe had the largest estimated decline (−0.572 percentage points per year), followed by Malawi (−0.225), Kenya (−0.190), Côte d’Ivoire (−0.180), and Uganda (−0.162). The results show that a single pooled trend masks pronounced cross-country heterogeneity. The findings support country-specific surveillance and differentiated HIV-response planning, while cautioning that prevalence is not equivalent to incidence and that a declining prevalence slope cannot by itself establish the causal effect of any intervention.