Monte Carlo simulation of testing requirements for chronic wasting disease (CWD) surveillance via hunter-harvest in Illinois, USA

Abstract

Managing chronic wasting disease (CWD) – a fatal, transmissible prion disease of cervids – has many challenges, such as determining the number of cervids to test to accurately estimate true prevalence with apparent prevalence. Options range from basic equations to complex techniques like agent-based modelling, but the former does not account for parameter uncertainty, and more detailed models may be less accessible for managers. Our study used Monte Carlo simulation to create a reference table showing the number of hunter-harvested deer to test to approximate herd true prevalence with a ± 2% margin of error and 95% confidence. A three-step process was simulated in which deer from the total population were harvested (testable subpopulation), and then a subset of those harvested deer were tested for CWD (tested group) with diagnostic testing error, with group assignment determined by random sampling. We evaluated true prevalences ranging from 0.5–20% and testable subpopulation sizes up to 5,000 deer from a total population of 20,000. An example from the reference table shows that detecting a 1% true prevalence in a testable subpopulation of 300 deer requires testing ≥280 deer, and that the only way to accurately detect a true prevalence of 0.5% is to test 600 out of 5,000 deer. Evaluating IDNR’s CWD surveillance efforts found that 16.14% of counties had adequate sample sizes in a given year and could detect true prevalences as low as 1%, showing that these guidelines are achievable.