Enumeration of Enterobacteriaceae in Various Foods with a New Automated Most-Probable-Number Method Compared with Petrifilm and International Organization for Standardization Procedures

Abstract

An automated most-probable-number (MPN) system (TEMPO, bioMérieux, Marcy l'Etoile, France) for enumeration of Enterobacteriaceae (EB) was compared with methods involving violet red bile glucose agar (VRBG) (International Organization for Standardization [ISO] method 21528-2) (ISO-VRBG) and Petrifilm (PF-EB). The MPN partitioning (three different volumes with 16 replicates of each) is done automatically in a disposable card. Bacterial growth is indicated by acid production from sugars, lowering the pH of the medium, and quenching the fluorescence of 4-methylumbelliferone. After incubation, the number of nonfluorescent wells is read in a separate device, and the MPN is calculated automatically. A total of 411 naturally contaminated foods were tested, and 190 were in the detection range for all methods. For these results, the mean (±standard deviation) counts were 2.540 ± 1.026, 2.547 ± 0.995, and 2.456 ± 1.014 log CFU/g for the ISO-VRBG, PF-EB, and automated MPN methods, respectively. Mean differences were −0.084 ± 0.460 log units for the automated MPN results compared with the ISO-VRBG and 0.007 ± 0.450 for the PF-EB results compared with the ISO-VRBG results. The automated MPN method tends to yield lower numbers and the PF-EB method tends to yield higher numbers than does the ISO-VRBG method (difference not significant; Kruskal-Wallis test, P = 0.102). Thus, the average difference was highest between the automated MPN method and the PF-EB method (−0.091 ± 0.512 log units). Differences between the automated MPN and ISO-VRBG results of >1 log unit were detected in 3.4% of all samples. For 3.9% of the samples, one comparison yielded differences of <1 log CFU/g and the other yielded >1 but <2 log CFU/g, which means that the differences are possibly >1 log CFU/g. For the ISO-VRBG method, confirmation of isolates was necessary to avoid a bias due to the presence of oxidase-positive glucose-fermenting colonies. The automated MPN system yielded results comparable to those of the confirmed Enterobacteriaceae ISO-VRBG method but required only 24 h of analysis time.