Optimization of Epidemiologic Interventions: Evaluation of Spatial and Non-Spatial Methods That Identify Johne’s Disease-Infected Subpopulations Targeted for Intervention

SUMMARY The potential costs and/or benefits associated with two epidemiological methods were compared. Using the same epide-miologic dataset (74 Israeli dairy herds tested for bovine paratuberculosis of which 57 farms were regarded to be infected, and 619 non-tested herds), the efficacy associated with the identification of the target population where control or preventive measures could be applied was evaluated by: 1) A method that applied geographical information systems (GIS), spatial statistics, network analysis (infective spatial links or ISL); and 2) A method that only partially applied spatial techniques. Based on the herd size of tested and non-tested farms, the geographical area of influence of each infected farm was estimated. Using the Euclidean distance between tested farms (distances between 2701 farm pairs), the ISL method calculated two measures of spatial connectivity: the number of links/farm and the ISL index. These measures are analogous to the number of roads connecting a city (links/farm) and the width of a road (index). The more links and/or the greater the average index ("width"), the greater the chances of an infected farm to disseminate an infection (especially to neighboring farms). While not reaching statistical significance, positive indices of Moran's I test for some spatial lags prompted the additional investigation of a subset of 547 farm pairs. This subset included 33 farm pairs (16 individual farms) which displayed > 2 links/farm, and ISL indices >7.5 times greater than average (high ISL farms). Regarding as "cost" the number of infected cows selected to receive an intervention, and as "benefit" the number of susceptible cows within the area of influence of an infected farm, hypothetical interventions implemented on the 16 high ISL farms yielded 39 % greater benefits and occupied a territory 9.5% smaller than decisions based on the 16 farms showing the highest prevalence. The analysis on spatial infec-tive connectivity may lead to earlier, farm-specific and more beneficial, decisions than methods based only on outcomes (later data), such as prevalence.