Relation between Gram-stain and clinical criteria for diagnosing bacterial vaginosis with special reference to Gram...

The aim of this study was to analyse how the results of Gram-staining vaginal smears correlated with the clinical criteria for determining the existence of bacterial vaginosis (BV) and, in particular, how the category defined as 'intermediate' or Gram grade II did so and its significance. Women attending an antenatal clinic with an abnormal vaginal flora, that is those who had Gram-stains of grades II or III, the latter considered to equate with BV, were given clindamycin or a placebo intravaginally and examined again on up to three occasions. Gram-stain readings of grade III correlated with the clinical criteria for BV on 356 (91.7%) of 388 occasions. Grade II readings covered the spectrum of clinical criteria and correlated with those for BV on 35 (37.2%) of 94 occasions. Grade I, recorded 231 times and seen usually after clindamycin treatment, was associated with BV only once. The sensitivity, specificity, positive predictive value and negative predictive value of the Gram stain for the diagnosis of BV, based on a combination of grades II and III, were 99.7%, 71.6%, 81% and 99.6%, respectively; based on grade III only, the values were 99.7%, 87.7%, 91.6% and 99.6%, respectively. Women reported a malodorous vaginal discharge on 49.2% of the occasions a grade III flora was seen and 13.3% of the times grade II was recorded. It was not associated with grade I and would seem a useful adjunct to the accepted clinical criteria for diagnosing BV. Each of the clinical criteria was found in about equal proportions (87%-91%) for women whose Gram grade was III. For grade II, an increased discharge was noted most often (76.5%) and 'clue' cells least often (24.5%). A positive amine test was the most specific, being associated with <1% of grade I smears. Of women with grade III status, 91% reverted to grade I after treatment with clindamycin for three days. In contrast, of women with grade II status, 53% reverted to grade I, as did 47% of those who were given a placebo. The 'intermediate' (grade II) category is a Gram-stain diagnosis and not one that can be made clinically. It is important to recognize as a distinct entity not only because amalgamation with grade III diminishes the specificity and positive predictive value of the Gram-stain for diagnosing BV, but also because women of grade II status usually fail to respond to clindamycin treatment, whereas those of grade III do not.