Comparison of culture, ELISA and PCR techniques for salmonella detection in faecal samples for cattle, pig and poultry
National Veterinary Institute, Department of Bacteriology, SE-751 89 Uppsala, Sweden
BMC Veterinary Research 2007, 3:21 doi:10.1186/1746-6148-3-21Published: 22 September 2007
Performances of different salmonella detection methods were evaluated by applying them to of artificially contaminated faecal specimens from cattle, pigs and poultry. The NMKL71 method, being the standard reference method for detection of salmonella in the official Swedish control program, was compared with the proposed ISO method using MSRV-selective enrichment for culturing, and also with three commercial ELISA- based systems, Bioline Selecta, Bioline Optima and Vidas, a commercial PCR-based method, BAX® system, and three different strategies using PCR detection using a non-commercial PCR system.
Altogether, 391 samples were tested, and the overall results clearly indicate that, when faeces from all animal species and all serotypes were included, the MSRV performed best, with a calculated accuracy of 99% and a calculated sensitivity of 98%. The second most sensitive and specific method was the BAX® system, using the modified enrichment protocol as recommended by the manufacturer for faecal samples. However, this protocol includes one additional day of work, as compared with the standard procedure for food sample analysis by the same method. The different strategies for salmonella detection using non-commercial PCR showed a sensitivity and specificity in the same range as the BAX® method; furthermore, results were obtained more quickly. The various commercial ELISA methods and the NMKL method showed the poorest performance of the methods included in the study, and were closely dependent on the origin of the faeces used and on which salmonella strain was to be detected.
The study showed that the sensitivity of the different methods depended to a great extent on the origin of the faecal matrices and the salmonella strains used to "spike" the samples.