Use of the API 20E system to identify non-Enterobacteriaceae from veterinary medical sources

The capability of the API 20E system to identify gram-negative nonfermenters and nonenteric fermenters was evaluated for 272 isolates from veterinary sources. Two different methods were used for interpreting the carbohydrate fermentation reactions on the strip. In method I, weakly fermented (yellow-green) carbohydrates were considered positive for all oxidase-positive organisms, and in method II, yellow-green carbohydrates were considered positive for all organisms requiring incubation for 48 hours. By both methods, the API system correctly identified 62% of the isolates. With method I, 31% of the isolates were misidentified and 6% were not identified. With method II, 21% of the isolates were misidentified and 17% received no identification. Organisms most affected by these 2 methods of interpretation were Pasteurella and Actinobacillus. Identifications reached by the API system were also compared with identifications made by veterinary diagnostic laboratories. The frequency of identifications agreements was not significantly affected by the method of API carbohydrate fermentation reaction interpretation. Generally, 30% of the identifications agreed (diagnostic laboratories vs API) when using only the API Index, whereas 51% agreed when the entire API computer data base identifications were included. The type of identification disagreements between diagnostic laboratories and the API system, however, was significantly affected by the method of API strip interpretation. With method I, 42% of the identifications were different and 6% were not in the API data base. With method II, 33% of the identifications were different and 17% were not in the API data base. Biotype differences between human and veterinary isolates were also compared. Significant differences between the predicted and actual reactions were noted for Pseudomonas aeruginosa and Bordetella bronchiseptica; however, these differences did not affect their correct identification to the API Index. For Pasteurella multocida, most profile numbers were not listed in the API Index because of differences in the actual vs predicted oxidase and nitrate reduction reactions; however, they were correctly identified with the total computer data base.