Discriminant analysis of antibiotic susceptibility as a means of bacterial identification

Abstract

This study shows that antibiotic susceptibility data can be used effectively in the presumptive identification of bacteria. Using 12 antibiotics and determining the zone sizes for each, 82% of the isolates considered were correctly identified without any other information. If the inability to distinguish between Escherichia coli and Shigella is disregarded, the percentage of correct identification is 92%. The method involves determining a set of discriminant functions and defining each taxon by a unique function. An unknown isolate is identified by evaluating each discriminant function and assigning the isolate to the taxon whose discriminant function has the largest value. A total of 468 isolates were examined. After eliminating the multiply resistant isolates, the remaining 369 isolates were used to determine the discriminant functions for the eight taxa considered.