Estimating drivers of chikungunya virus transmission and spread in its invasion of the Americas
Chikungunya is an emerging mosquito-borne virus that has caused explosive outbreaks in Africa and Asia for decades and began an ongoing invasion of the Americas in December 2013. In just over a year, chikungunya spread to 45 countries where local transmission was documented, infecting nearly 1.3 million people in total. Our goal was to make use of weekly, country-level case reports provided openly online by the Pan American Health Organization to infer relationships between local transmission and putative drivers thereof, including land surface temperature, enhanced vegetation index, and land use classification. To link fine-scale spatial information about these drivers (i.e., 5 km2or below) to case data aggregated at a national level, we developed a novel framework that weights data about putative spatial drivers according to projections from a niche model for the dominant mosquito vector and spatial data on human population densities under arbitrary assumptions about human movement. To account for movement of the virus across national borders, we fit a connectivity model that made use of airline passenger flows, border traffic, and geographic distance. The relationships between these drivers and transmission were fit to case data under the time series SIR (TSIR) model that has been applied most famously to measles as well as to other infectious diseases. We furthermore adapted the TSIR model to more realistically account for the relatively lengthy serial interval necessitated by the presence of a vector.
We found significant relationships between transmission and linear and quadratic terms for the environmental variables we considered and a linear term for log incidence during the previous pathogen generation. The latter effect is strongly nonlinear at the country level, however, due to an estimated mixing parameter of 0.74. Relationships between transmission and the environmental variables that we estimated were biologically plausible and in agreement with expectations. Our analysis suggests that local transmission can be successfully correlated with putative drivers, even at the coarse scale of countries, when weighted appropriately. Our analysis likewise highlights the importance of drivers of international spread such as airline passenger flow, and it appears that this information, combined with case data, is somewhat predictive of future outbreaks at the country level. Overall, this analysis suggests that successfully forecasting the future trajectory of the chikungunya epidemic in the Americas may be possible. We will conclude with up-to-date forecasts of chikungunya transmission and a discussion of the implications thereof.