A stochastic evaluation of the validity of an animal-health survey in Uruguay

Abstract

This paper introduces spreadsheet simulation models to evaluate the validity of national or regional disease surveys aimed at identifying infection in populations of farm animals. The process of evaluation includes specification or calculation of cluster-level diagnostic test sensitivity (the proportion of animals with the disease which test positive) and specificity (the proportion of animals without the disease which test negative), which are obtained from two probability distributions of the number of positive tests at individual-level expected from infected and non-infected clusters, respectively. Probability distributions for the number of positive clusters expected in a situation of the herd prevalence are specified and used to define survey properties (the survey being considered a diagnostic system), and receiver operating characteristic curves (consisting of a plot of sensitivity and specificity pairs for different cut-off values) are drawn. The result of a survey implemented to determine the prevalence of infectious bovine rhinotracheitis in dairy cattle in Uruguay, South America was used to illustrate this approach. The models can be adapted to a wide range of survey designs in animal health and production.