Relative effectiveness of Irish factories in the surveillance of slaughtered cattle for visible lesions of tuberculosis, 2005-2007

From these animals, 8,178 suspect lesions were submitted for laboratory confirmation. Lesions from 5,456 (66.7%) animals tested as positive, and 269 (3.2%) were inconclusive for bovine TB. Logistic regression was used to determine adjusted submission and confirmation risks for each factory while controlling for confounding factors. Factory rankings based on adjusted and crude risks were similar. The average crude submission risk for all the factories was 25 lesions per 10,000 animals slaughtered, ranging from 0 to 52. The crude confirmation risk varied between 30.3% and 91.3%.Substantial variation in the effectiveness of lesion submission and subsequent confirmation as bovine TB was found among the 37 factories. Compared to previous years (2003-2004), there was an increased bovine TB lesion submission and confirmation risk. Continued monitoring of the effectiveness of slaughter surveillance in Ireland is recommended; emphasis should be placed on efforts to improve bovine TB surveillance in factories with lower rankings.A programme to eradicate tuberculosis (TB) from cattle was begun by the Irish government in the 1950's [1]. The detection of gross (visible) tuberculous lesions at slaughter has proved to be an essential component of the overall bovine TB surveillance system for the cattle population [1]. With respect to effectiveness of factory surveillance for bovine TB, several studies have been conducted, some based on univariable analysis [2-5] and two using multivariable analysis [6,7]. The multivariable approach is preferable because it helps ensure that measures of surveillance effectiveness are adjusted for factors that can affect the TB status (e.g. age and source) of the animals slaughtered at different factories. This study, using data from 2005-2007, is an update of the evaluation of the effectiveness of factory surveillance in Ireland, first undertaken by Martin et al. [6] using data from 2001-2002 and subsequently by Frankena et al. [7] using data from