Kiss 2004.

Methods	Randomised trial with computer‐generated randomisation sequence. All pregnant women presenting for antenatal care were screened, with smear samples sent to the central laboratory where they were randomly assigned to the intervention or control group. Women in the intervention arm with a positive screening result received treatment. Women in the control group were blinded to screening results and received routine antenatal care. Description of withdrawals: yes. Intention‐to‐treat analysis: not used.
Participants	4429 pregnant women (mean age 28.9, SD 5.6) presenting for routine prenatal visits between 15 and 19 weeks' gestation (mean 17, SD 1.6). Intervention group: n = 2058 ; control group: n = 2097. Inclusion criteria: gestational age 15‐19 weeks without subjective complaints (e.g. contractions and vaginal bleeding). Exclusion criteria: clinical symptoms of vaginal infection, multiple pregnancies. Location: Vienna, Austria.
Interventions	Intervention group: vaginal smears (Gram stain and evaluated by the scoring criteria proposed by Nugent 1991) screening for bacterial vaginosis, Trichomonas vaginalis and Candida species and received standard antibiotic treatment if positive screening test, i.e. 2% for 6 days local clindamycin for bacterial vaginosis, 300 mg twice daily for seven days oral clindamycin for recurrent bacterial vaginosis, 0.1 g for 6 days local clotrimazole for candidiasis, and 500 mg for 7 days local metronidazole for trichomoniasis (including treatment of the partner). Control group: were smeared, but the results of testing were not made available to the women's care providers and did not have any effect on the standard clinical antenatal care program routine antenatal examination.
Outcomes	Primary outcome: spontaneous preterm delivery at less than 37 weeks' gestation. Secondary outcomes: low birthweight: preterm birth with birthweight below 2500 g; very low birthweight: preterm birth with birthweight below 1500 g; rates of miscarriage between 16‐22 and 20‐24 weeks; intrauterine death; prevalence of various forms of vaginal infections; duration of sick leave and hospitalisation.
Notes	4429 randomised, 274 excluded from analysis, 140 lost to follow up, 68 did not fulfill all inclusion criteria, 66 multiple pregnancies. We have contacted the author and are waiting for a reply for our request for additional data (secondary outcomes e.g. neonatal necrotizing enterocolitis, neonatal sepsis, neonatal death, duration of neonatal admission to NICU/hospital). We will incorporate these additional data in an update to the review, should they be forthcoming.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	A pre‐established computer generated randomisation list was used to allocate patients to the treatment groups.
Allocation concealment (selection bias)	Unclear risk	No information on allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	All obstetricians and women in the intervention group received their smear results and different treatment regimens.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Not described, but assessors would not have influenced the objective outcome of birthweight.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	4429 pregnant women: 2058 in the treatment group, 2097 in the control group. There were 274 patients excluded from the study with group allocation not stated (140 lost to follow up, 68 did not fulfill all inclusion criteria, 66 multiple pregnancies). Intention to treat analysis was not described.
Selective reporting (reporting bias)	Unclear risk	No information available because protocol is not accessible; we have contacted authors for additional outcome data.
Other bias	Low risk	The study seems to be free of other types of bias.

NICU: neonatal intensive care unit
 SD: standard deviation