Sonnenwirth, Alex C.

An automated and computerized system (Automicrobic System [AMS]) for the detection of frequently encountered bacteria in clinical urine specimens was tested in a collaborative study among six laboratories. The sensitivity, specificity, reliability, and reproducibility of the AMS were determined, and the system was compared with conventional detection and identification systems. In this study, pure cultures and mixtures of pure cultures were used to simulate clinical urine specimens. With pure cultures, the sensitivity of the AMS in identifying the nine groups of organisms most commonly found in urine averaged 92.8%. The specificity averaged 99.4%, and the reliability of a positive result averaged 92.1%. The latter value was strongly influenced by a relatively high occurrence of false positive Escherichia coli results. The AMS was capable of detecting growth of most organisms, including those which it was not designed to identify. However, it identified some of these incorrectly as common urinary tract flora. Reproducibility of results, both within laboratories and among different laboratories, was high. Fast-growing organisms, such as E. coli and Klebsiella/Enterobacter species, were detected often at cell populations well below the AMS enumeration threshold of 70,000/ml. In mixed culture studies, high levels of sensitivity and specificity were maintained but when Serratia species were present in mixtures with other organisms, there was often a false positive report of E. coli. The overall performance of the AMS was considered satisfactory under the test conditions used.

Susceptibility tests using 7 antimicrobial agents (carbenicillin, chloramphenicol, clindamycin, penicillin, cephalothin, metronidazole, and tetracycline) were run against 35 anaerobes including Bacteroides fragilis (17), other gram-negative bacilli (7), clostridia (5), peptococci (4), and eubacteria (2). Results in triplicate obtained by the microbroth dilution method and the aerobic modification of the broth disk method were compared with those obtained with an agar dilution method using Wilkins-Chalgren agar. Media used in the microbroth dilution method included Wilkins-Chalgren broth, brain heart infusion broth, brucella broth, tryptic soy broth, thioglycolate broth, and Schaedler's broth. A result differing by more than one dilution from the Wilkins-Chalgren agar result was considered a discrepancy, and when there was a change in susceptibility status this was termed a significant discrepancy. The microbroth dilution method using Wilkins-Chalgren broth and thioglycolate broth produced the fewest total discrepancies (22 and 24, respectively), and Wilkins-Chalgren broth, thioglycolate, and Schaedler's broth had the fewest significant discrepancies (6, 5, and 5, respectively). With the broth disk method, there were 15 significant discrepancies, although half of these were with tetracycline, which was the antimicrobial agent associated with the highest number of significant discrepancies (33), considering all of the test methods and media.