Comparison of Gram stain and Pap smear procedures in the diagnosis of bacterial vaginosis.

OBJECTIVE
The purpose of this study was to examine the characteristics of Gram stain versus Pap smear in diagnosis of bacterial vaginosis (BV).


METHODS
One-thousand and sixty women were enrolled in this study. All cases with symptoms of BV were determined by Amsel's criteria, which were accepted as the gold standard for diagnosis of BV. Pap smear and Gram stain evaluations were compared according to Amsel's criteria, without viewing the clinical results of the patients. Gram stain and Pap smear results were determined as negative or positive according to Amsel's criteria. Sensitivity, specifity and positive predictive values were calculated.


RESULTS
After accepting the cases that were diagnosed as BV according to Amsel's criteria as reference cases, the sensitivity of the Gram stain method was calculated as 97% and the sensitivity of the Pap smear method as 93%. Similar specificity rates were obtained with both methods in diagnosis of BV related to the clinical results. There were no statistically significant differences in diagnosis of BV between these two groups.


CONCLUSION
If Amsel's criteria are accepted as the gold standard for diagnosis of BV, Gram stain and Pap smear methods will give similar results in diagnosis.

The most frequently encountered complaint in gynecological out-patient clinics is vaginal discharge. Normal vaginal mucosa does not have any glandular structure. Usually there is a biological balance in the microorganisms living in vaginal mucosa. The most important role in the continuity of this balance -and in preventing the growth of pathogenic microorganisms -is that of Lactobacillus species. Lactobacilli produce an acidic medium in the vagina via hydrogen peroxide (H2O2), which transforms glycogen present in vaginal epithelium to lactic acid 1 . The acidic medium produced by lactobacilli suppresses the growth of other microorganisms. If the balance of vaginal flora is altered against lactobacilli, a clinical picture of itching, pain and vaginal discharge or smell will be observed. In bacterial vaginosis (BV), microorganisms that are short, rod-like and mostly anaerobic, with variable Gram stain properties -such as Gardnerella vaginalis, Bacteriodes, b-streptococci and the Mobilincus-Falcivibrio group -replace the usual vaginal flora formed by lactobacilli 1,2 .
The aim of the presented study is to compare the efficacy of Gram stain and Pap smear with that of Amsel's criteria in the diagnosis of BV -which is continually gaining importance in relation to morbidity and mortality problems -in cervicovaginal specimens obtained in routine controls.

MATERIALS AND METHODS
Routine gynecologic examinations of 1060 patients aged 18-50 years, who were referred to the SSK Ege Maternity Hospital between July and December 1996 with complaints of bad-smelling vaginal discharge, itching and pain, were performed and histories of the patients were taken. Women using oral contraceptives and vaginal spermicides, those who used antibiotics or vaginal suppositories for any reason and those who performed vaginal lavage after coitus were excluded from the study (172 women).
Vaginal inspections were carried out using a dry, sterile speculum without applying antiseptic cleansing liquids and the characteristics of the vaginal discharge were evaluated. The pH of the vaginal discharge was evaluated with litmus paper (Merck KGaA Acilit pH, Darmstad, Germany). An amine test was performed to the appropriately taken vaginal discharge sample with 10% KOH. Samples were taken from the vaginal lateral wall and posterior fornix with brushes, and spread out on three preparations. One of these preparations was examined immediately by light microscope; appropriate fixation, Gram stain and classic Papanicolaou (Pap) stain were applied to the other preparations.
The evaluation of the results obtained from Pap and Gram stains was based on positivity by Amsel's criteria. For a positive diagnosis of BV, the presence of at least three of the following four criteria of Amsel is required: (1) thin, homogeneous gray-white discharge on the vaginal wall, that can be easily sampled; (2) vaginal-discharge pH above 4.5; (3) observation of 'clue' cells in fresh vaginal smear; and (4) amine smell similar to that of stale fish, obtained by dripping 10% KOH on the vaginal discharge 2 .
The aim of the study did not include establishing the prevalence of BV in the population.
In evaluation of Pap stain under´400 magnification, according to the method of Davis and colleagues 3 , cases with the presence of microorganisms other than lactobacilli in a thin film form, the presence of clue cells -which are formed by the covering of the cytoplasmic membranes of squamous cells by non-Lactobacillus microorganisms -and a predominance of non-Lactobacillus bacteria in most areas, were accepted as BV-positive cases. Davis and colleagues reported the condition of absence of lactobacilli as a fourth diagnostic criterion, but the absence of lactobacilli is modified as predominance of non-Lactobacillus bacteria over lactobacilli in the present study.
In evaluation of Gram stain under´400 magnification, according to Thomason and colleagues' 'clue cell' method 4 cases with the presence of non-Lactobacillus bacteria in most areas and the presence of clue cells in at least two of 20 areas in large magnification, were accepted as positive. Cells having only lactobacilli in their cytoplasm were accepted as negative, whereas the covering of cytoplasmic membranes by non-Lactobacillus bacteria was used to identify clue cells.
Cytologic evaluation related to each case was performed, without any data obtained via Amsel's criteria. Specifity, sensitivity and predictive values of the results obtained from Pap and Gram stains were calculated by using the results obtained according to Amsel's criteria as reference values.

RESULTS
The mean age of subjects was 31.7 years (range 18-51 years). Approximately forty-seven percent of all cases (416/888) were pregnant at the time of examination. In the pregnant group, the mean pregnancy number of the 416 cases with a determined pregnancy number was 3.4, and the mean parity of cases with a determined parity was 2.2.
All of the women were suffering from vaginal discharge, which was a yellow-gray discharge in 78% of cases, white in 17% and yellow-green in 4%. Itching was observed in 88% and bad smell was observed in 96%. Post-coital odor was observed in 97%.
Two hundred and forty-two cases were determined as BV-positive by Pap smear, compared with 251 BV-positive cases determined by Gram stain (Tables 1 and 2). Sensitivity and specifity were 93% and 94% respectively with Pap smear and 97%

DISCUSSION
BV is the most frequent cause of vaginitis, and is characterized by increase in growth of anaerobic and aerobic microorganisms due to an unbalanced eco-system in the vagina 2 . Gardner and Dukes 5 were the first to report Haemophilus vaginalis as a cause of nonspecific vaginitis in 1955. It is also reported that there are high sensitivity and specifity values of Gram stain in diagnosis of BV, besides Amsel's criteria 2,4,6-12 .
The clinico-pathologic condition characterized by redness in the vaginal wall, bad-odored discharge and the presence of clue cells in cervicovaginal specimens, which results from the transformation of the acidic vaginal pH to an alkaline pH via metabolic activity of the aforedefined bacteria is known as BV 6 .
It is also reported that BV causes premature rupture of membranes, preterm delivery and endometritis and, according to some authors, BV has possible carcinogenic effects due to abnormal vaginal cytology and mosaic colposcopic pattern in the cervix [13][14][15][16][17] .
Various methods have been recommended for the evaluation of preparations for BV by Gram stain. Spiegel and colleagues 18 counted a total of 20 areas in´1000 magnification in which they reported the diagnosis of BV, in correlation with the number of microorganisms present in each large magnification area. Nugent and colleagues 19 determined a total score over six as BV, by taking Lactobacillus Gardnerella and other microorganisms into account in a total of 20 large magnification areas in´1000 magnification. In the present study, the clue cell method of Thomason and colleagues 4 has been used.
Platz-Christensen and colleagues 9 ascertained the sensitivity of the Pap smear method as 88%, the specifity as 97% and positive predictive value as 97%, following a study comparing Pap smear and Gram stain methods in the diagnosis of BV. These rates were 100%, 97%, and 94% respectively for the Gram stain method 9 . In our study, the observed sensitivity and specificity were 93% and 94% respectively for Pap smear, and 97% and 94% respectively for Gram stain. The positive predictive values were determined as 86% for Pap smear and 88% for Gram stain. The aforementioned rates are similar to rates determined by Platz-Christensen and colleagues 9 . But the specificity rate of our study is smaller than the that determined by Platz-Christensen and colleagues. Also, Platz-Christensen and colleagues reported the sensitivity as 90% and specificity as 97% after the comparison of Pap smear and clinical criteria, in a similar study performed in 1989 20 .
Davis and colleagues 3 reported the sensitivity and specificity rates of Pap smear as 55 and 98% respectively, in a study in which the Gram stain results were accepted as the diagnostic standard. They reached their hypothesis because of lower rates of sensitivity. Davis and colleagues claimed   Platz-Christensen and colleagues obtained similar results in two studies performed 6 years apart 9,20 . In addition, Prey 21 reported the necessity of the presence of coccobacilli-type microorganisms for the diagnosis of BV by Pap smear. Bedrossian and colleagues 22 reported that Pap smear gives results as good as Gram stain in the diagnosis of BV and cervicitis. They also reported that the exclusion of Gram stain from the routine examination method might decrease the economic cost. Lamont and colleagues reported the sensitivity and specificity as 81 and 91% respectively, in Pap smears evaluated by cytotechnicians, but these rates were 87 and 97% respectively in Pap smears evaluated by only one cytopathologist 8 .
Long Gram stain evaluation under´1000 magnification, proposed by Nugent 19 and Spiegel 18 despite Pap smear evaluation at´400 magnification, will cause a significant loss of time especially in countries like Turkey, where cytotechnologists are not employed or do not exist. This time-loss will lead to an increase in the cost per preparation, and possibly a decrease in the time that can be used for the determination of malignancies -especially of cervix carcinomaearlier in Pap smear determinations, if the chance of the Pap smear picking up other infection types is also taken into account. The difference between Gram stain and Pap smear methods, especially in specifity in BV diagnosis, can be neglected. If there is a negligible difference between Gram stain and Pap smear methods it is better to use Pap smears in routine gynecologic cytologic examinations.