Abstract
We compared the accuracy of the Gram-stained smear, the acridine orange-stained smear, and the AutoMicrobic system (AMS; Vitek Systems, Inc., Hazelwood, Mo.) in screening for bacteriuria, as detected by conventional cultures. For 1,024 clinical specimens, results with the acridine orange-stained smear and the Gram-stained smear were very similar. When read for the presence of one or more microorganisms or leukocytes per 20 oil immersion fields, both smears were highly sensitive (92.1 and 93.3%, respectively) and moderately specific (70.0 and 61.7%, respectively). Sensitivity was greater for specimens yielding greater than or equal to 10(5) CFU/ml (96.1 and 98.9%, respectively) than for those with 10(3) to 10(4) CFU/ml (81.4 and 78.0%, respectively). Preliminary classification based upon the tinctorial and morphological characteristics of the Gram-stained smear was compatible with culture results in nearly all cases. The accuracy of the Gram-stained smears was not influenced by special cleaning of the microscopic slides, or the level of expertise of the microscopist. For 715 specimens, the sensitivity of the AMS in detecting bacteriuria (91.5%) was very similar to that of the stained smears (92.1 and 95.7%, respectively), but the specificity was significantly higher (83.2% versus 42.6 and 70.0%). Detection of microorganisms by the AMS took an average of 6.3 +/- 3.0 h. These data suggest that the Gram-stained smear is easily interpreted, very sensitive, acceptably specific, and still the optimal rapid method for screening for bacteriuria in most clinical microbiology laboratories.