Comparative Spatiotemporal Analysis of Global HIV-1 Subtype C Hotspots: Applying Bayesian Hierarchical Modeling, SaTScan, and Getis-Ord Gi* Statistics
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Despite the global importance of HIV-1 subtype C, a global-scale GIS characterization of its geographic clustering and the temporal persistence of hotspots is lacking, and systematic cross-method comparisons are scarce.
Methods
We assembled 2,220 country–year observations from 111 countries (2005–2024), comprising 161,025 subtype C sequences, and generated internally standardized expected counts. We compared hotspot detection using Getis-Ord Gi* statistics (ArcGIS), SaTScan space–time scan statistics, and Bayesian hierarchical models with spatial–temporal smoothing, and quantified temporal persistence and cross-model concordance.
Results
Documented subtype C sequences showed increasing geographic concentration over time, shifting from relatively widespread detection toward progressively localized clustering, with the strongest and intensifying concentration in Southern Africa. SaTScan and Bayesian models identified fewer hotspots but showed greater temporal stability, whereas Gi* detected more localized and short-term spatial fluctuations. High-stability hotspots (sustained multi-year detection) were predominantly in Southern Africa. Zimbabwe was the only country classified as a high-stability hotspot across all three frameworks; Eswatini, Botswana, Malawi, and South Africa showed high stability in at least two models, indicating robust, model-consistent persistence.
Conclusions
Integrating complementary hotspot methods reveals both convergent and method-specific patterns and provides a quantitative basis to prioritize long-term persistence for targeted surveillance, resource allocation, and precision prevention.
