Geographically Weighted Regression Modeling of Spatial Clustering and Determinants of Focal Typhoid Fever Incidence

Abstract

Background: Typhoid is known to be heterogenous in time and space, with documented spatiotemporal clustering and hotspots associated with environmental factors. This analysis evaluated spatial clustering of typhoid and modeled incidence rates of typhoid from active surveillance at 4 sites with child cohorts in India. Methods: Among approximately 24 000 children aged 0.5-15 years followed for 2 years, typhoid was confirmed by blood culture in all children with fever >3 days. Local hotspots for incident typhoid cases were assessed using SaTScan spatial cluster detection. Incidence of typhoid was modeled with sociodemographic and water, sanitation, and hygiene-related factors in smaller grids using nonspatial and spatial regression analyses. Results: Hotspot households for typhoid were identified at Vellore and Kolkata. There were 4 significant SaTScan clusters (P