AOAC 2018 Methods

116  B ird et al . : J ournal of AOAC I nternational V ol . 102, N o . 1, 2019

secondary-enrichment broth, followed by subsequent plating and confirmation of isolates using appropriate biochemical, serological, and/or molecular methods. Note: Even a negative sample will not give a zero reading as the system, and 3M MDA 2 – Cronobacter amplification reagents have a “background” relative light unit reading. In the rare event of any unusual light output, the algorithm labels this as “Inspect.” 3M recommends the user to repeat the assay for any Inspect samples. If the result continues to be Inspect, proceed to confirmation test using your preferred method or as specified by local regulations. The collaborative study involved a method comparison evaluation of the 3M MDA 2 – Cronobacter and the ISO 22964:2017 reference method for powdered infant formula (milk based) with probiotics. A total of 11 laboratories throughout the continental United States, Mexico, and Croatia participated in this study. For 3 of the laboratories, two separate analysts participated, and the remaining 10 laboratories each had one participating analyst. Each participant analyzed 36 paired test portions for the 3M MDA 2 – Cronobacter and the ISO 22964:2017 reference method (12 inoculated with a high level of Cronobacter , 12 inoculated with a low level of Cronobacter , and 12 uninoculated controls). In addition to the test portions, all participants set up an APC to determine the total microbial load of the test matrix. The average APC result obtained by the collaborators was 2.5 × 10 1 CFU/g (<1.0 × 10 1 to 7.0 × 10 1 CFU/g). After APC results were received, the matrix was later evaluated by seven collaborators for lactic acid bacterial counts according to the Compendium of Methods for Microbiological Examination of Foods (13). An average count of 1.6 × 10 8 CFU/g was obtained (range 6.4 × 10 7 to 2.9 × 10 7 CFU/g). See Table 1 of the supplementary materials for a summary of results. From the bulk lot of uninoculated test matrix, 10 replicate 10 g test portions were screened using the ISO 22964:2017 reference method for the presence of Cronobacter . All test portions produced negative results for the target analyte. Table 1 in the supplementary information summarizes the collaborative study results for each matrix. As per criteria outlined in Appendix J of the AOAC Validation Guidelines (4), fractional positive results were obtained. Detailed results for each laboratory are presented in Table 2 in the supplementary information. The level of Cronobacter was determined by MPN on the day of initiation of analysis by the coordinating laboratory. The MPN levels obtained, with a 95% confidence interval, was 0.70 CFU/test portion (0.57, 0.87) for the low-inoculum level, and 2.71 CFU/test portion (1.86, 3.97) for the high-inoculum level. MPN results are presented in the second column of Table 2 in the supplementary information. The individual laboratory and sample results are presented in Table 3 in the supplementary information. TheAPC results for each collaborator are presented in Table 2 in the supplementary information. Results of Collaborative Study

(LPOD CC = 0.51) confirming positive. For samples that produced presumptive positive results by the 3M MDA 2 – Cronobacter , 84 out of 168 samples confirmed positive (LPOD C = 0.50; value includes only presumptive positive results that confirmed positive). For the reference method, 86 out of 168 test portions were reported as positive (LPOD R = 0.51). A dLPOD C value of –0.01 with 95% confidence intervals (–0.12, 0.10) was obtained between the candidate and reference method, indicating no statistically significant difference between the two methods. A dLPOD CP value of 0.01 with 95% confidence intervals of –0.10, 0.12 was obtained between presumptive and confirmed results indicating no statistically significant difference between the presumptive and confirmed results. For the high inoculation level, 168 out of 168 test portions (LPOD CP = 1.00) were reported as presumptive positive by the 3M MDA 2 – Cronobacter with 168 out of 168 test portions (LPOD CC = 1.00) confirming positive. For samples that produced presumptive positive results by the 3M MDA 2 – Cronobacter , 168 out of 168 samples confirmed positive (LPOD C = 1.00). For the reference method, 168 out of 168 test portions were reported as positive (LPOD R = 1.00). A dLPOD C value of 0.0 with 95% confidence intervals (–0.02, 0.02) was obtained between the candidate and reference method, indicating no statistically significant difference between the two methods. A dLPOD CP value of 0.0 with 95% confidence intervals (–0.02, 0.02) was obtained between presumptive and confirmed results indicating no statistically significant difference between the presumptive and confirmed results. For the uninoculated controls, 2 out of 168 samples (LPOD CP = 0.01) produced a presumptive positive result by the 3M MDA 2 – Cronobacter with 0 out of 168 test portions (LPOD CC = 0.00) confirming positive. For samples that produced presumptive positive results by the 3M MDA 2 – Cronobacter , 0 out of 168 samples confirmed positive (LPOD C = 0.00). For the reference method, 0 out of 168 test portions were reported as positive (LPOD R = 0.00). A dLPOD C value of 0.00 with 95% confidence intervals (–0.02, 0.02) was obtained between the candidate and reference method, indicating no statistically significant difference between the two methods.AdLPOD CP value of 0.01 with 95% confidence intervals (–0.01, 0.04) was obtained between presumptive and confirmed results, indicating no statistically significant difference between the presumptive and confirmed results. For detailed results of the LPOD statistical analysis, see Table 1 and Figures 1 and 2 in the supplementary information. Upon completion of the collaborative study, feedback was solicited from collaborators on the performance of the candidate method. No negative feedback was provided; several collaborators indicated how the technology would be an improvement to their current workflow. A small number of FP (2.4%) and FN (0.8%) results were observed across several laboratories. Four of the FP and both of the FN results were obtained at the low inoculation level. Although a negative result was observed during the initial screening for Cronobacter by ISO culture method, two of the FP results were observed in the uninoculated test portions. This may be a result of high levels of persistent Discussion

Powdered Infant Formula with Probiotics

For the low inoculation level, 88 out of 168 test portions (LPOD CP = 0.52) were reported as presumptive positive by the 3M MDA 2 – Cronobacter with 86 out of 168 test portions