FoodNet conducts active laboratory-based surveillance for 9 pathogens transmitted commonly through food, including Shiga toxin-producing E. coli (STEC). Adoption of CIDTs has allowed for rapid identification of Shiga toxin or Shiga toxin genes, but incorporating multiple test results with differing sensitivity and specificity complicates treatment decisions and public health surveillance. Between 2007 and 2017, FoodNet reported increases in the use of CIDTs and decreases in rates of confirmation by culture.

We examined STEC cases reported to FoodNet during 2012−2017 with a positive immunoassay (IA) or polymerase chain reaction (PCR) test performed at a clinical laboratory, followed by positive or negative test at a state public health laboratory. Three test type combinations were assessed (IA/IA, PCR/PCR, and IA/PCR) by state, symptoms, test discordance, and culture (cx) result.

During 2012−2017, 8,298 (76% of all STEC reported) specimens were tested by IA or PCR at both a clinical and a public health laboratory, 58% by IA/PCR, 27% by IA/IA, and 25% by PCR/PCR; some specimens had more than one test at each laboratory. Among these, 8,132 (98%) were also tested by cx. Among the IA/PCR test results, 20% were discordant and 75% of these were cx-negative. Even more of IA/IA (27%) and PCR/PCR (24%) results were discordant, and 75% of these were cx-negative. A median of 24% of test results were discordant (range by state, 13%–44%). Persons with discordant test results were less likely to have diarrhea (91% vs. 97%) and bloody diarrhea (33% vs. 57%). During 2012–2017, discordant results increased for IA/PCR (14% to 22%), IA/IA (17% to 34%), and PCR/PCR (6% to 25%). Most (85%) specimens with discordant results were cx-negative and 8% did not have a cx.

Almost a quarter of results were discordant, with marked variation by state, and most of these infections could not be confirmed by culture at the public health laboratory. Discordant results can pose problems for patient management. Including or excluding patients with discordant results also affects our ability to measure trends. Sensitivity and specificity of test types, test targets, and specimen transport must be considered when interpreting test results.

All authors: No reported disclosures.