Dengue viruses cause serious disease in populations throughout the tropical and subtropical regions of the world. While control efforts have at times been successful, they have also been limited by inefficiency and cost. Better understanding of the dynamics of transmission may lead to the design of more effective interventions. Here we address the potential role of climate in driving changes in dengue transmission. While there is strong biological support for potential effects, in vivo associations have not been convincingly demonstrated. First, we employ wavelet analysis to robustly show that there is little association between multi-year climate variation and dengue incidence in Puerto Rico, Thailand, and Mexico. We then directly address the issue of whether any correlation exists between temperature, precipitation, and dengue incidence in Puerto Rico. To do this, we first develop weather models to estimate fine-scale temperature and precipitation throughout the island from the limited observations that have been made. We model municipio-level dengue incidence over time using lagged estimated weather variables while controlling for seasonality via a spline smoothing function. Municipio-level effect estimates are then pooled in second-level models and adjusted with municipio-level effect modifiers. We find that temperature and precipitation are positively associated with dengue incidence and that the local strength of these effects is determined by local climate characteristics. Given the absence of strong multi-annual correlations, we hypothesize that climate variation drives the seasonal timing of dengue incidence in Puerto Rico, but does not determine the overall intensity of transmission.
