Financial considerations of the sets technique in animal-disease surveillance

Disease-surveillance systems are designed to detect the presence of (or an unusual increase in) the ill health of populations. Once the system correctly identifies an increase in a disease or condition, appropriate preventive action may be taken to alleviate associated losses. However, criteria for determining when an action should be taken are typically subjective. The purpose of this paper was to present and evaluate a technique for detecting clusters or increases in incidence rates: the sets technique. In addition, this technique was modified to permit evaluation of an animal disease-surveillance system, based on the system's detection capability as well as financial impact of the disease and its control.This technique was sensitive to the number of temporally clustered cases and the case-rate increase required to "signal an alarm". A system requiring more vs. fewer clustered cases needed to signal an alarm was always preferred financially for a rare disease (P(Epidemic)=0.001) and (with one exception) for a common disease (P(Epidemic)=0.2). Requiring more temporally clustered cases to signal an alarm, however, resulted in increased costs if the detection level were designed for a 6-fold case-rate increase, or if the true case-rate increase were 2-fold and the system were designed to detect a 2-fold increase of a relatively expensive disease (epidemic cost=$10000 vs. $1000). Case frequencies, costs and other assumptions used in this model may be modified to evaluate specific diseases and their associated surveillance systems.