Bayesian hierarchical spatial ecological model assessing the nexus of vegetation index, insecticide-treated bed-nets and livestock cattle density on lymphatic filariasis transmission

Lymphatic filariasis (LF) remains endemic in Nigeria, exhibiting pronounced spatial heterogeneity. This study employed Bayesian spatio-temporal modelling with a beta-binomial likelihood and a modified Besag-York-Mollie (BYM) random effect on national data from 2013, 2015, 2018, and 2021. It assessed ecological and intervention drivers of LF transmission. Results identified persistent high-risk clusters (RR > 2.25) in northern (Zamfara, Kano, Plateau) and southeastern states (Ebonyi, Enugu), which correlated with high prevalence (10.95-21.84%). Notably, insecticide-treated net (ITN) coverage showed counterintuitive positive associations with LF prevalence in 2013 (RR = 5.31, 95% CI: 0.076-3.26) and 2015 (RR = 7.90, 0.056-4.075), although these effects diminished after 2015. Enhanced vegetation index (VIND) and livestock cattle density (LSCA) demonstrated non-significant, temporally unstable effects, reflecting their context-dependent roles in vector ecology. Spatial dependence remained consistently strong (ρ = 0.75-0.82), with 41-54% of transmission variance attributable to location. Spatial and temporal shifts revealed expanding high-risk local government areas (LGAs) (189 in 2013 to 204 in 2018), mainly driven by conflict-affected northeastern states (Borno, Yobe), with partial declines post-2018 observed in southern regions. The pooled model obscured critical dynamics, inflating spatial heterogeneity (σ = 1.29 vs. annual values of 0.80-0.92). We conclude that LF transmission is influenced by spatially structured ecological and intervention interactions, highlighting the need for adaptive, context-specific strategies. Prioritising longitudinal surveillance and conflict-sensitive interventions in high-risk agro-pastoral zones (North Central, Southeast) will enhance Nigeria's elimination efforts.