A "hazard model" using risk-weighted surveillance for first detection of chronic wasting disease

Surveillance for emerging diseases can be enhanced through incorporation of risks and hazards to identify areas on the landscape with higher likelihoods of disease introduction and spread while increasing confidence that samples are collected from locations and animals with the highest probability of disease detection. A primary example of this situation is wildlife surveillance programs for chronic wasting disease (CWD) in free-ranging white-tailed deer (Odocoileus virginianus) in jurisdictions where it is not yet known to exist. But knowledge gaps in areas that lack sufficient disease testing and the nonexistence of data depicting disease introduction risks have impeded the ability to detect disease at the earliest intrusion into wild herds. We developed a novel method to conduct wildlife disease surveillance by considering how disease introduction likelihood may increase in the presence of risk factors, such as certain human activities and dense deer populations. In the absence of empirical risk data, we solicited perceptions from subject matter experts to develop a risk assessment (survey) characterizing the likelihood of disease introduction from anthropogenic activities. We overlaid these summarized perceptions with independent harvest data on the demographic attributes of wild cervid herds. We further incorporated previously published surveillance weights representing the differential disease information gained by testing each age/sex segment of deer. We applied the resulting surveillance design ('Hazard Model') in New York during the 2013-2014 hunting season and in Tennessee during the 2018-2019 hunting season. In both states, the Hazard Model suggested that counties with large deer populations, high-risk cervid businesses, or those in close proximity to infections in neighboring states were at greatest risk for introduction of CWD and therefore should be sampled with the greatest intensity. After a brief outbreak of CWD in New York in 2005, wildlife officials in New York did not re-discover CWD in their state, while officials in Tennessee discovered their first case of CWD within four months. The Hazard Model was developed with logistics and constraints as primary considerations, so implementation is sufficiently flexible to accommodate specific operational needs of the wildlife agency.