Evaluation of antimicrobial resistance phenotypes for predicting multidrug-resistant Salmonella recovered from retail meats and humans in the United States.

Although multidrug-resistant (MDR) non-Typhi Salmonella (NTS) strains are a concern in food production, determining resistance to multiple antimicrobial agents at slaughter or processing may be impractical. Single antimicrobial resistance results for predicting multidrug resistance are desirable. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value were used to determine each antimicrobial agent's ability to predict MDR phenotypes of human health significance: ACSSuT (resistance to at least ampicillin, chloramphenicol, streptomycin, sulfamethoxazole, tetracycline) in NTS isolates, and MDR-AmpC-SN (resistance to ACSSuT, additional resistance to amoxicillin-clavulanate and to ceftiofur, and decreased susceptibility [MIC >= 2 microg/ml] to ceftriaxone) in NTS serotype Newport. The U.S. National Antimicrobial Resistance Monitoring System determined MICs to 15 or more antimicrobial agents for 9,955 NTS isolates from humans from 1999 to 2004 and 689 NTS isolates from retail meat from 2002 to 2004. A total of 847 (8.5%) human and 26 (3.8%) retail NTS isolates were ACSSuT; 995 (10.0%) human and 16 (2.3%) retail isolates were serotype Newport. Among Salmonella Newport, 204 (20.5%) human and 9 (56.3%) retail isolates were MDR-AmpC-SN. Chloramphenicol resistance provided the highest PPVs for ACSSuT among human (90.5%; 95% confidence interval, 88.4 to 92.3) and retail NTS isolates (96.3%; 95% confidence interval, 81.0 to 99.9). Resistance to ceftiofur and to amoxicillin-clavulanate and decreased susceptibility to ceftriaxone provided the highest PPVs (97.1, 98.1, and 98.6%, respectively) for MDR-AmpC-SN from humans. High PPVs for these agents applied to retail meat MDR-AmpC-SN, but isolate numbers were lower. Variations in MIC results may complicate ceftriaxone's predictive utility. Selecting specific antimicrobial resistance offers practical alternatives for predicting MDR phenotypes. Chloramphenicol resistance works best for ACSSuT-NTS, and resistance to ceftiofur, amoxicillin-clavulanate, or chloramphenicol works best for MDR-AmpC-SN.