Annual seroprevalence of Yersinia pestis in coyotes as predictors of interannual variation in reports of human plague cases in Arizona, United States.

Although several health departments collect coyote blood samples for plague surveillance, the association between reported human cases and coyote seroprevalence rates remains anecdotal. Using data from an endemic region of the United States, we sought to quantify this association. From 1974 to 1998, about 2,276 coyote blood samples from four Arizona counties were tested for serological evidence of exposure to Yersinia pestis, the causative agent of plague. Using a titer threshold presumed to be indicative of recent infection (serum titers of ≥1:256), we found a statistically significant relationship between years with >17% sero-positive coyotes and years with two or more human cases reported. Moreover, when the annual coyote seroprevalence rates were dichotomized at 17%, 84% of the years were correctly classified using four biologically relevant meteorological variables in a linear regression. This is the first time a statistically significant temporal association between human plague cases and coyote seroprevalence rates has been shown. However, issues with data resolution and surveillance effort that potentially limit the public health utility of using coyote seroprevalence rates are discussed.