Reproducibility ofInterpretation ofGram-Stained Vaginal Smears fortheDiagnosis ofBacterial Vaginosis

Inthediagnostic microbiology laboratory, interpretation ofGram-stained slides ofvaginal swabspecimens isusedtosupport theclinical diagnosis ofbacterial vaginosis. Thereproducibility withwhichtechnologists interpret these Gram-stained slides wasevaluated bypresenting, incodedfashion, 80original slides and80 duplicate slides ofvaginal swabspecimens tothree technologists. Theyeachinterpreted theoriginal slide twice andtheduplicate slide fromthesamespecimen once. Intraobserver andinterobserver agreement wasassessed byuseoftheweighted kappastatistic. Semiquantitation ofLactobacillus andGardnerella morphotypes anda diagnosis ofbacterial vaginosis showed thegreatest intraobserver agreement, withkappa values ranging from 0.772 to1.000. Interobserver agreement wasalsohighforrating Lactobacillus morphotypes andcluecells (kappa values between 0.735 and0.869) butdecreased slightly forGardnerella morphotypes andadiagnosis of bacterial vaginosis (kappa values between 0.656 and0.800). These results indicate that there isgoodagreement fortheinterpretation ofGram-stained slides ofvaginal swabspecimens andthatthis methodalone, without culture, canbeusedreliably tosupport theclinical diagnosis ofbacterial vaginosis. Although thetermbacterial vaginosis hadbeenused previously (17, 18), in1984, Westrom etal.(20) first defined itandproposed thatitbeusedtoreplace thetermnonspecific vaginitis. Clinically, thediagnosis ofbacterial vaginosis isbasedonseveral factors including thepresence ofa characteristic homogeneous, thin, greyvaginal discharge, a vaginal pHof-4.5,apositive amideodortest (release ofa fishy amineodorwhenvaginal fluid ismixedwith10% KOH),andtheidentification ofcluecells (vaginal epithelial