Estimation of the risk of Salmonella shedding by finishing pigs using a logistic model obtained from a survey

Date
2007-01-01
Authors
Fablet, C.
Robinault, C.
Jolly, J.
Dorenlor, V.
Eono, F.
Eveno, E.
Labbé, A.
Bougeard, S.
Fravalo, Philippe
Madec, F.
Major Professor
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Department
Abstract

An analytic epidemiological survey was carried out in 105 French farms to identify factors associated with Salmonella shedding by finishing pigs. This study gave out a list of 7 risk factors using a logistic model. The aim of the present survey was to validate this model on a second sample of batches of pigs in order to estimate their Salmonella status. The validation study was carried out from April 2003 to August 2005 on 64 finishing pig batches distinct from those used originally to generate the logistic model. In each farm, Salmonella shedding of a batch of pigs at the end of the finishing phase was assessed using swabs as described in the analytical study. Questionnaires were filled in with the farmer to collect data related to management routines. Blood samples from10 growing and 10 finishing pigs were taken to assess sanitary risk factors: status vs Lawsonia intracellularis and Porcine Respiratory Coronavirus. Salmonella contamination status of a finishing room before loading, a further identified risk factor, was tested by environmental swabbing procedure. The estimated risk with the standard error, of Salmonella shedding was calculated using the logistic model and compared to the bacteriological Salmonella status of each batch. Several thresholds are proposed and sensitivity, specificity, positive and negative predictive values related to each cut-off value were calculated. A cut-off value of 0.34 maximised both sensitivity (76.9%) and specificity (68.6%) of the model. Whatever the threshold, the accuracy of the Salmonella non-shedding predicted status is better than the Salmonella shedding predicted status. In a bacteriological sampling programme, this model could be a useful tool to identify batches with low risk of Salmonella shedding and to focus attention on those getting a high probability for being positive.

Comments
Description
Keywords
Citation
Source