A Bayesian spatio-temporal framework to identify outbreaks and examine environmental and social risk factors for infectious diseases monitored by routine surveillance

Abstract

Spatio-temporal disease patterns can provide clues to etiological pathways, but can be complex to model. Using a flexible Bayesian hierarchical framework, we identify previously undetected space-time clusters and environmental and socio-demographic risk factors for reported giardiasis and cryptosporidiosis at the New Zealand small area level. For giardiasis, there was no seasonal pattern in outbreak probability and an inverse association with density of dairy cattle (β^1 = −0.09, Incidence Risk Ratio (IRR) 0.90 (95% CI 0.84, 0.97) per 1 log increase in cattle/km2). In dairy farming areas, cryptosporidiosis outbreaks were observed in spring. Reported cryptosporidiosis was positively associated with dairy cattle density: β^1 = 0.12, IRR 1.13 (95% CI 1.05, 1.21) per 1 log increase in cattle/km2 and inversely associated with weekly average temperature: β^1 = −0.07, IRR 0.92 (95% CI 0.87, 0.98) per 4 °C increase. This framework can be generalized to determine the potential drivers of sporadic cases and latent outbreaks of infectious diseases of public health importance.