Between January 2001 and October 2002, 25 non-hospital based obstetricians in the Vienna area enrolled pregnant women presenting for their routine prenatal visits between 15+0 (15 weeks plus 0 days) and 19+6 weeks (19 weeks plus 6 days) of gestation. Patients' obstetricians determined gestational age on the basis of the date of a woman's last menstrual period and confirmed this by ultrasound before 19 weeks of gestation. The obstetricians included women without subjective complaints (contractions, vaginal bleeding, or symptoms suggestive of vaginal infection) into the study after obtaining individual informed consent. We excluded women erroneously included—those who did not fulfil the inclusion criteria in terms of week of gestation or clinical symptoms of vaginal infection, or those with multiple pregnancies.
In addition to the routine antenatal examinations the women's obstetricians assessed vaginal smears, transferred to a microscopic slide. To diagnose bacterial vaginosis we Gram stained all preparations in a central laboratory and used the scoring system proposed by Nugent et al for our evaluation.19
Evaluations also included screening for the presence of Candida
species and Trichomonas vaginalis
on Gram stain. The study protocol differentiated between bacterial vaginosis (Nugent grade 3), vaginal candidiasis (spores and hyphae), infection with T vaginalis
, or combinations of any of the three.
After the obstetricians had enrolled the women into the study, the case report forms and smear samples went to the central laboratory, where they were randomly assigned to the intervention group or the control group. Randomisation was performed according to a computer generated randomisation list. In the intervention group, obstetricians provided vaginal smear results. All obstetricians and women in the intervention group received their smear results. In the control group, obstetricians and participating women remained blinded to the test results.
Women in the intervention group who were found to have a pathological vaginal flora or another microscopically diagnosed infection received standardised treatment within seven to 10 days of diagnosis. Bacterial vaginosis (Nugent grade 3) was treated for six days with clindamycin 2% vaginal cream. Persistent or recurrent disease was treated with oral clindamycin 300 mg twice daily for seven days. Candidiasis (spores and hyphae) was treated with local clotrimazole 0.1 g for six days. Trichomoniasis was treated with local metronidazole 500 mg for seven days and included treatment of the partner. Women's obstetricians took follow up vaginal smears at the time of the next routine antenatal visit, between 24+0 and 27+6 weeks of gestation. Treatment for persistent or recurrent candidiasis or trichomoniasis was repeated. We used a telephone recall system to check compliance with treatment of the women in the intervention group. In the control group we concealed the test results from the women and their doctors so that they did not influence the women's standard antenatal care programme.
Evaluation of results
The primary outcome variable was the rate of spontaneous preterm delivery (delivery at less than 37 weeks). Secondary outcome variables were spontaneous preterm delivery at less than 37 weeks in combination with birth weights equal to or below 2500 g, 2000 g, and 1500 g, respectively. We defined spontaneous preterm delivery as either vaginal or caesarean delivery due to preterm labour or preterm premature rupture of membranes. We analysed separately all medically indicated (iatrogenic) preterm deliveries due to pre-eclampsia or eclampsia, HELLP (haemolysis, elevated liver, low platelet) syndrome, placenta praevia, abruption of the placenta, or malformation of the fetus. Moreover, we evaluated the rates of miscarriage between 16-22 and 20-24 weeks of pregnancy and intrauterine death and assessed the distribution and prevalence of various forms of vaginal infection and the duration of sick leave and hospitalisation.
The primary outcome measure was spontaneous preterm delivery. We used the χ2 test to test the hypothesis that a smaller rate of spontaneous preterm delivery is expected in the intervention group than the control group. Our significance level was P < 0.05 (two tailed). We based our sample size estimation on an estimated rate of spontaneous preterm delivery of 7% and considered a reduction to (at least) 5% clinically important. On the basis of these assumptions and a power of 80% we considered a sample size of 2000 per group adequate. We tested the secondary outcome measures (spontaneous preterm delivery at less than 37 weeks in combination with birth weights equal to or below 2500 g, 2000 g, and 1500 g, respectively) in the same way. We used SPSS, version 11.0, for all statistical calculations.