Microscopy of stained urine smears to determine the need for quantitative culture

Consecutive specimens (2,564) of urine were cultured quantitatively, and Gram-stained smears were prepared from centrifugates as well as from the uncentrifuged specimens. About half of the specimens harbored organisms, and the quantity seen in smears of the centrifugates correlated reasonably well with the numbers of viable organisms cultured from the specimens. For example, smears of 900 centrifugates had one or more organisms per oil immersion field; 712 of their respective specimens proved to have colony counts of 10(5) or greater, and 120 had 10(4) to less than 10(5). For 188 specimens, however, the smears falsely predicted greater than or equal to 10(5) colony-forming units (many of these fell into the group that had 10(4) to less than 10(5)) but failed to predict clinically significant concentration of 10(5) or greater in only 31. Predictive values are presented also for lesser quantities of organisms seen in such smears. With rare exception, smears of centrifugates were superior to those of whole specimens as predictors of the concentration of viable organisms. Evaluation of smears proved not to be biased by the operator; values did not deviate significantly depending upon whether one or several microbiologists performed the evaluations.