International STD Research & Reviews

Background: HIV infection amongst males aged 15-24 is an important health issue that is of global concern and has significant differences across regions. Some of the existing literature is typically based on linear or pooled models that do not adequately account for bounded and heterogeneous information. This research aims to approximate country-specific and global time-varying patterns without incurring the cost of excessively modelling the limited fractional nature of the result and breaking down the heterogeneity between countries. This study quantifies global and regional trends in HIV prevalence among males aged 15-24 and estimates country-level heterogeneity using a multilevel beta regression framework. Also, the observational nature of World Development Indicators (WDI) data and the absence of individual-level behavioural covariates limit causal interpretation.

Methods: The World Bank World Development Indicators longitudinal, unbalanced panel (1990-2024) was analysed using a multilevel beta-mixed-effects model. The prevalence (%) of HIV was converted into a fractional response that was modelled with random intercepts and slope to account for country-level heterogeneity. AIC/BIC was used to select the model, and the results were reported as odds ratios (ORs) accompanied with 95% confidence intervals.

Results: The median HIV prevalence in the world was 0.53% (0.10) with wide geographical variability. In sub-Saharan Africa (1.276% and 0.576%), the highest prevalence and the predicted prevalence were observed. The region-adjusted random-slope beta model (AIC = -64,477.14) was the best-fitting model. HIV prevalence declined over time (OR = 0.952, p = .002). Sub-Saharan Africa had significantly higher odds of prevalence (OR = 3.131; 95% CI: 2.075-4.726), followed by Latin America & Caribbean (OR = 1.566; 95% CI: 1.013–2.421). Large heterogeneity at the country level was noted, with the high-burden countries being Eswatini (6.77%) and Zimbabwe (6.07%).

Conclusion: The prevalence of HIV among males aged 15-24 is decreasing all over the world, but is very uneven. Multilevel beta modelling offers a more precise and policy-relevant estimation and an intervention based on the targeted interventions to reach SDG 3.3. To begin with, governments and international health organisations need to consider focused, data-driven HIV response among young men in high-burden locations and governments and scholars ought to institutionalise sophisticated statistical modelling (e.g. multilevel beta mixed-effects models) in HIV monitoring and surveillance.