Spatial analyses of infection and disease during chikungunya and Zika epidemics in Managua, Nicaragua.

Chikungunya (CHIKV) and Zika (ZIKV) viruses have caused widespread epidemics across the Americas. After several years of minimal transmission, CHIKV is once again spreading in Nicaragua. We investigated spatial aspects of two chikungunya epidemics (2014, 2015) and one Zika epidemic (2016) in a prospective pediatric cohort in Nicaragua. We used generalized estimating equations, generalized additive and mixed-effects models, Kuldorff's scan statistic, and intracluster correlations to analyze the infection and disease status of ∼3,000 initially uninfected participants during each epidemic. We found that the incidence of infection (# new infections/total population) and disease (# new cases/total population) often had different spatial patterns, demonstrating that high-infection areas may not exhibit much disease. High infection incidence was observed near a large cemetery west of the study site. The large 2015 chikungunya epidemic and 2016 Zika epidemic started near the cemetery, exhibited high infection incidence throughout the study site, and had extensive month-to-month changes in spatiotemporal dynamics. Clusters of excess infections were large and adjacent to the cemetery, while clusters of excess uninfected participants were found in the northern and eastern sections of the study area. Notably, the intracluster correlation of infection was very weak in households (∼0.2), and infection status was correlated across distances <200 meters. Similarities across the epidemics suggest that targeting interventions to the built, non-household environment may reduce epidemic potential in Managua. Overall, spatial analyses of epidemics that only use case data may substantially underestimate the full extent of viral transmission, missing identification of high-infection areas and opportunities for intervention.