Early and cost-effective identification of high risk/priority control areas in foot-and-mouth disease epidemics.

Geo-referenced data from the 2001 Uruguayan foot-and-mouth disease (FMD) epidemic were explored to assess whether spatial analysis could lead to cost-benefit based policies. Four variables were analysed: (i) location and size of 4022 individual rural land parcels, of which 574 were infected over 60 days, (ii) animal density, (iii) percentage of dairy farms per county, and (iv) road density. Each variable was categorized into two to five classes (e.g. small/medium/large) and the proportion of cases per class reported at days 1-3 of the epidemic was compared with that reported at days 4-6. A higher proportion of cases was found at days 4-6 than at days 1-3 in areas with: small and medium size land parcels, high animal density, > 20% farms specialized in dairy production, and high road density (P < 0.03 for each). Each of these classes showed a greater proportion of cases at days 7-60 than the proportion of the total territory covered by each class's area (early case concentration ratios: 1.14-1.37). Land parcel clusters were indicated by Moran's I-test (P < 0.01). A new region was constructed by intersecting the four spatial classes associated with higher proportions of cases at days 4-6. At days 7-60, this region included 50.4% of all cases and represented 30.6% of the territory under study (final case concentration ratio: 1.65). The final area per case in this region was at least 33% lower and covered at least 45% less territory than any of the four single-variable approaches. Bio-statistical, multivariate spatial analysis of early cases may greatly increase the efficiency of epidemiologic policy.