Reproducibility of interpretation of Gram-stained vaginal smears for the diagnosis of bacterial vaginosis.

In the diagnostic microbiology laboratory, interpretation of Gram-stained slides of vaginal swab specimens is used to support the clinical diagnosis of bacterial vaginosis. The reproducibility with which technologists interpret these Gram-stained slides was evaluated by presenting, in coded fashion, 80 original slides and 80 duplicate slides of vaginal swab specimens to three technologists. They each interpreted the original slide twice and the duplicate slide from the same specimen once. Intraobserver and interobserver agreement was assessed by use of the weighted kappa statistic. Semiquantitation of Lactobacillus and Gardnerella morphotypes and a diagnosis of bacterial vaginosis showed the greatest intraobserver agreement, with kappa values ranging from 0.772 to 1.000. Interobserver agreement was also high for rating Lactobacillus morphotypes and clue cells (kappa values between 0.735 and 0.869) but decreased slightly for Gardnerella morphotypes and a diagnosis of bacterial vaginosis (kappa values between 0.656 and 0.800). These results indicate that there is good agreement for the interpretation of Gram-stained slides of vaginal swab specimens and that this method alone, without culture, can be used reliably to support the clinical diagnosis of bacterial vaginosis.