Yellow fever virus is a mosquito-borne pathogen that persists through a zoonotic transmission cycle in non-human primates in South America and Africa. Humans in these regions were once subject to large, urban outbreaks of yellow fever, but now enjoy considerable protection from vaccination. Because small clusters of cases in humans continue to occur on a regular basis, there is interest in mapping spatial patterns of zoonotic spillover to better target ongoing efforts to vaccinate against this sometimes-fatal disease.
Results/Conclusions
Although the majority of recent cases in humans have been attributed to zoonotic spillover, we show that existing methods for mapping zoonotic spillover of yellow fever virus are biased by failing to consider the possibility of urban transmission. To correct for this bias, we introduce a method that treats the total number of yellow fever virus infections in humans as the sum of two random variables, one for infections attributable to zoonotic transmission and another for infections attributable to urban transmission. This method acknowledges that many infections in humans may be unobserved, including some involved in undetected stuttering chains of urban transmission. Combining this approach with a branching process model of urban transmission byAedes aegypti mosquitoes, we quantify the potential magnitude of this bias in conventional approaches to mapping zoonotic spillover of yellow fever virus. Given an appropriate model of location-specific transmission, this approach has the potential to be extended to other pathogens with the potential for outbreaks following cross-species transmission.