Table 2. Key findings of included studies.

Objective 1: Studies assessing BV cure
Reference	Study design	Intervention	Comparator	No. women randomised	Outcome measure	Duration of follow-up	BV cure results in intervention vs comparator
Andersch, 1986 [34]	RCTa	Lactal gel 5ml PV/night x 7 nights	Oral MTZ 500 mg bid x 7 days	Lactal = 32	≤2 of 3 Amsel criteriab	1 week after start of treatment	31/31 (100%, 95%CI 89–100) vs
				MTZ = 22			17/17 (100%, 95%CI 90–100)
Boeke, 1993 [35]	RCTa	Oral placebo bid x 7 days and lactic acid vaginal suppository/night x 7 days	Two comparator groups:	Lactic acid = 41c	≤2 of 4 Amsel criteria	2 weeks after start of treatment	18/37 (49%, 95%CI 32–66) vs
				MTZ = 44			35/42 (83%, 95%CI 69–93)
			1) Oral MTZ 500 mg bid x 7 days and placebo vaginal suppository/night x 7 days	Placebo = 40			16/34 (47%, 95%CI 30–65)
						4 weeks after start of treatment	11/33 (33%, 95%CI 18–52) vs
							27/38 (71%, 95%CI 54–85)
			2) Oral placebo bid x 7 days and placebo vaginal suppository/night x 7 days				12/35 (34%, 95%CI 19–52)
						3 months after start of treatment	12/32 (38%, 95%CI 21–56) vs
							29/37 (78%, 95%CI 62–90)
							11/32 (34%, 95%CI 19–53)
Simoes, 2006 [32]	Double-blind RCT	Acidform gel 5g PV/day x 5 days	10% MTZ gel PV/day x 5 days	Acidform = 13	≤2 of 4 Amsel criteria	12–17 days after start of treatment	3/13 (23%, 95%CI 5–54) vs
				MTZ = 17			15/17 (88%, 95%CI 64–99)
						33–40 days after start of treatment	1/13 (8%, 95%CI 0–36) vs
							9/17 (53%, 95%CI 28–77)
Fredstorp, 2015 [36]	Open-label RCT	Two intervention groups:	Untreated control group	Once/week = 37	≤2 of 4 Amsel criteria	1 week after start of treatment	24/34 (71%, 95%CI 53–85)
				Twice/week = 35			28/35 (80%, 95%CI 63–92) vs
		1) OMLA pessary applied once/week for 1 week		Control = 33			3/30 (10%, 95%CI 2–27)
		2) OMLA pessary applied twice/week for 1 weekd
Objective 2: Studies assessing vaginal microbiota composition
Reference	Study design	Intervention	Comparator	No. women randomised	Outcome measure	Reported results
Keller, 2012 [33]	Single-blind RCT	Acidform gel 5g PV bid x 14 days	HEC placebo gel PV bid x 14 days	Acidform = 18	qPCR assays:	In 35e women without BV, no significant changes were observed in the prevalence or concentration of L. crispatus, L. jensenii, Megasphaera (type 1 & type 2) or BVAB2 following 14 days of gel use in either the Acidform or placebo group (compared to baseline values).
				Placebo = 18	L. crispatus
					L. jensenii
					G. vaginalis
					Megasphaera (type 1 & type 2)
						There was a non-significant trend towards a decrease in G. vaginalis concentration in the Acidform group following 14 days of gel use compared to baseline (median of 1.36x106 to 3.66x104 DNA copies/swab, p = 0.083), but not in the placebo group (median of 9.8x105 to 4.4x106 DNA copies/swab, p-value not reported).
					BVAB2
Gottschick, 2017 [29] f	Double-blind RCT	Oral MTZ 2g single dose. After 7–28 days, WO3191 pessary applied twice-weekly x 3 weeks	Oral MTZ 2g single dose. After 7–28 days, Vagisan® pessary applied twice-weekly x 3 weeks	WO3191 = 18	16S rRNA gene sequencing of V1-V2 regions	In 36g women initially treated for BV with oral metronidazole, no significant changes in vaginal microbiota composition were reported during or following use of either WO3191 or Vagisan®.
				Vagisan® = 26
						The cumulative relative abundance of Lactobacillus spp. (L. crispatus, L. iners and L. gasseri) was 73% in the WO3191 group prior to starting pessary use, 77% after 3 weeks of pessary use, and had decreased to 59% 12–14 weeks after last pessary use.
						The cumulative relative abundance of Lactobacillus spp. was 75% in the Vagisan® group prior to starting pessary use, 69% after 3 weeks of pessary use, and 73% 12–14 weeks after last pessary use. There was a non-significant increase in the relative abundance of L. crispatus in Vagisan® group from 18% prior to starting pessary use to 33% 12–14 weeks after last pessary use.
						There was no difference in microbiota diversity (as measured by Shannon diversity index) between women randomised to WO3191 and women randomised to Vagisan®.
van der Veer, 2019 [37]	Single arm prospective cohort	Participants were followed for 3 menstrual cycles. Etos® douche was applied 3/per week for duration of cycle 2 starting on day 1 of menses.	NA	29	16S rRNA gene sequencing of V3-V4 regions	In 25h women without BV there was a non-statistically significant increased odds of having a diverse anaerobic vaginal microbiota relative to an L. crispatus microbiota during (odds ratio: 1.4; 95% CI 0.9–2.1) and after douching with Etos® (odds ratio: 1.7; 95%CI 0.9–3.1), compared to before douching, following adjustment for menses.
						Douching with Etos® had no effect on microbiota diversity as measured by Shannon diversity index.

No., number; BV, bacterial vaginosis; RCT, randomised controlled trial; PV, intravaginal; MTZ, metronidazole; bid, twice a day; OMLA, oligometric lactic acid; CI, confidence interval; qPCR, quantitative PCR; NA, not applicable.

^a Details of blinding not provided.

^b The three criteria evaluated were: positive amine test, clue cells, pH≥5.0.

^c 168 women randomised, but post-randomisation, 43 women were found to be ineligible and excluded, thus randomisation numbers presented reflect the 125 eligible women included in analyses.

^d OMLA pessary is designed to release lactic acid over a 72hr period.

^e One woman allocated to Acidform did not receive the intervention.

^f Both the intervention (WO3191) and the comparator (Vagisan®) contain lactic acid.

^g 36 women were included in microbiota analyses, n = 13 receiving WO3191 and n = 23 receiving Vagisan®.

^h Twenty-nine women were recruited, four were excluded and 25 women completed the study.