Use of spatial statistics and monitoring data to identify clustering of bovine tuberculosis in Argentina

The spatial distribution of endemic bovine tuberculosis (TB) in Argentine cattle herds was described using recorded information on the detection of TB-like lesions in cattle slaughtered between March 1995 and 1997 at 126 slaughterhouses with federal inspection. Approximately, 47% (9472396 cattle) of the estimated total number of cattle slaughtered in Argentina during this period was included in the study. Information on the number of cattle per source herd consigned to slaughter, number of cattle with TB-like lesions per herd and the geographical location of counties from which cattle originated was used to investigate spatial clustering of TB. Overall, no evidence of clustering of TB prevalence by county was detected (Moran's autocorrelation statistic I=0.009, P=0.089). However, first- (Cuzick and Edwards' test statistic, T(k)=87, P=0.036) and second-order (T(k)=170, P=0.038) nearest-neighbor case-counties (TB prevalence>median prevalence of all counties, 6.7%) were clustered. Using the spatial scan test based on a Bernoulli model, the most-likely cluster (P=0.001) identified during the study period included 5793 cases of TB (5.2 per 1000 km(2)) in five counties. This cluster coincided with Santa Fe Province, which contains 21% of all dairy cows in Argentina and accounts for 34% of the country's milk production. Several secondary clusters of TB-also located in dairy districts-were identified. Study results demonstrate that bovine TB is clustered in Argentina, and these clusters coincide with dairy cattle production. Identification of clustering can assist efforts to eradicate bovine TB from Argentina. Further spatial investigations need to focus on the reasons why TB is clustered in Argentina. In particular, the relationship between TB clustering and management practices-such as grazing density and production systems-need to be described to assist in the development of disease-control programs. The use of spatial statistics and geographical information systems could meet these needs.