Abstract
The diagnostic accuracy of a PCR used to identify horses shedding Salmonella spp. in their feces during hospitalization was estimated, relative to bacterial culture of serially collected fecal samples, using longitudinal data. Five or more fecal samples were collected from each of 116 horses admitted as inpatients, for reasons other than gastrointestinal disease, between July 26, 2001 and October 25, 2002. All 873 fecal samples collected were tested with a PCR based on oligonucleotide primers defining a highly conserved segment of the histidine transport operon gene of Salmonella typhimurium, and each sample was cultured for Salmonella spp. One or more samples from 87 (75%) horses were PCR positive, and Salmonella was cultured from 1 or more samples from 11 (9.5%) horses. All culture-positive horses had at least 1 PCR-positive result, whereas only 29 (28%) culture-negative horses were PCR negative on all fecal samples tested. The PCR was most specific, relative to bacterial culture of serially collected fecal samples, when used to test samples from Quarterhorse or breeds other than Thoroughbred or Standardbred, or from clinical (vs. healthy, accompanying horses) cases. Overall, the PCR had the greatest agreement (70%), compared with bacterial culture of serially collected fecal samples, using a cutoff of 2 or more positive PCR test results to define a Salmonella-positive horse. The reasons why some fecal samples, from which Salmonella organisms cannot be isolated, are PCR positive need to be determined before the PCR can be incorporated into Salmonella surveillance programs for hospitalized equine populations.