Intermittent Lactobacilli-containing Vaginal Probiotic or Metronidazole Use to Prevent Bacterial Vaginosis Recurrence: Safety and Preliminary Efficacy by Microscopy and Sequencing

Bacterial vaginosis (BV) is associated with HIV acquisition and adverse pregnancy outcomes. Recurrence after metronidazole treatment is high. HIV-negative, non-pregnant Rwandan BV patients were randomized to four groups (n=17/group) after seven-day oral metronidazole treatment: behavioral counseling only (control), or counseling plus intermittent use of oral metronidazole, Ecologic Femi+ vaginal capsule (containing multiple Lactobacillus and one Bifidobacterium species), or Gynophilus LP vaginal tablet (L. rhamnosus 35) for two months. Vaginal microbiota assessments at all visits included Gram stain Nugent scoring and 16S rRNA gene qPCR and HiSeq sequencing. All interventions were safe. BV (Nugent 7-10) incidence was 10.18 per person-year at risk in the control group, and lower in the metronidazole (1.41/person-year; p=0.004), Ecologic Femi+ (3.58/person-year; p=0.043), and Gynophilus LP groups (5.36/person-year; p=0.220). In mixed effects models adjusted for hormonal contraception/pregnancy, sexual risk-taking, and age, metronidazole and Ecologic Femi+ users, each compared to controls, had higher Lactobacillus and lower BV-anaerobes concentrations and/or relative abundances, and were less likely to have a dysbiotic vaginal microbiota type by sequencing. Inter-individual variability was high and effects disappeared soon after intervention cessation. Lactobacilli-based vaginal probiotics warrant further evaluation because, in contrast to antibiotics, they are not expected to negatively affect microbiota or cause antimicrobial resistance.


INTRODUCTION 45
Most women have a vaginal microbiota (VMB) that consists predominantly of lactobacilli 1 . The 46 most common type of vaginal dysbiosis is bacterial vaginosis (BV), characterized by a decrease 47 in lactobacilli and increase in other anaerobes, which affects 30-40% of women worldwide 2 . 48 panbacterial 16S rRNA gene copy concentrations of physician-collected samples of enrolled 165 women with valid sequencing results (N=393) were determined by BactQuant qPCR assay as 166 previously described 26 . 167 168 Molecular data processing 169 Reads were demultiplexed, and primer sequences removed using Cutadapt v1.2.1 27 . Error 170 correction, dereplication, denoising, merging, removal of chimeras, and taxonomic assignment 171 were performed in DADA2 v1.4.0 using Silva v128 as the reference database (Supplement 172 1) 28,29 . Further data processing included removal of rare, non-bacterial, and contaminant 173 amplicon sequence variants (ASVs), and identification of vaginal and probiotic sequences that 174 are not included in the Silva database using the NCBI Needleman-Wunsch Global Align 175 Nucleotide Sequences function 15 (Supplement 1). Silva-based taxonomic assignments of non-176 minority ASVs were double-checked using the NCBI Microbial Nucleotide BLAST function, 177 using the non-redundant V3-V4 version of the Vaginal 16S rDNA Reference Database as a 178 tiebreaker 30,31 . We rarefied at 1,111 reads using the GUniFrac 1.0 package in R 32 . This yielded 179 401 unique ASVs in 629 samples, mapping to species (n=255; 63.6%), genus (n=116; 28.9%), or 180 higher taxonomic level (n=30; 7.5%). Concentrations in cells/μl per ASV per sample were 181 estimated by multiplying the ASV-specific copy-normalized relative abundance by the sample-182 specific 16S rRNA gene copies concentration 33,34 . Heatmaps of the 20 most common ASVs by 183 sample are shown in Figure S1A for relative abundances and Figure S1B for concentrations. 184 VMB data reduction was required for molecular efficacy analyses, and was done in three 185 different ways. First, the Simpson diversity index (1-D) was calculated for each sample. Second, 186 each ASV was assigned to one of four 'bacterial groups' based on the published literature 187 (Supplement 2): 1) lactobacilli (all species combined, but with subcategorization into EF+ 188 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https: //doi.org/10.1101/19001156 doi: medRxiv preprint strains, the GynLP strain, and 'natural lactobacilli' in some analyses); 2) BV-anaerobes 189 (anaerobic bacteria that have consistently been associated with BV); 3) pathobionts (bacteria that 190 are considered to have higher intrinsic pathogenicity than BV-anaerobes and have not 191 consistently been associated with BV 35 ); and 4) 'other bacteria' (a rest group, consisting mostly 192 of skin bacteria and Bifidobacterium species). Within each sample, read counts of ASVs 193 belonging to the same bacterial group were summed. This resulted in four relative abundances 194 (one for each bacterial group) per sample, which sum to 1.0 for each sample. Third, we used 195 hierarchical clustering based on Euclidean distance to pool samples into eight mutually exclusive 196 VMB types (each sample was assigned to only one VMB type): 1) L. iners-dominated (Li,n=247 197  cross-sectional analyses, we used Fisher's exact test for binary and categorical variables, and 208 Kruskal-Wallis and Mann Whitney U tests for continuous variables. For longitudinal analyses, 209 we used incidence rates, incidence rate ratios, and mixed effects models. We conducted intent-to-210 treat (ITT) analyses, and modified ITT analyses limited to women who were BV-negative by 211 both Amsel and Nugent criteria at the time of randomization (n=51 of 68). All mixed models 212 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https: //doi.org/10.1101/19001156 doi: medRxiv preprint included one VMB endpoint at a time as the outcome, participant identification number as the 213 random effect, and randomization group (an indicator variable with the control group as the 214 reference group) as the main fixed effect. Models included samples collected during the 215 intervention period only (including self-sampled time points), and adjusted for covariates that 216 were associated with VMB composition in mixed effects models at p<0.05. 217 218

Ethical statement 219
All participants provided written informed consent. The study was conducted in accordance with 220

Participant disposition 230
Of the 68 randomized women, only four did not complete the trial (Figure 1), resulting in 29.93 231 person-years of data. The median age was 31 (interquartile range (IQR) 27-35) ( Table 1, Table  232 S1). The majority of women (92.6%) had exchanged sex for money or goods, and had had a 233 median of five (IQR 2-18) sex partners, in the past month. All but three women used condoms, 234 but mostly inconsistently. Two-thirds of the women (61.8%) were using hormonal contraception, 235 and four women became pregnant during the trial. Short course metronidazole/tinidazole use for 236 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

VMB compositions at baseline 245
By design, all randomized women at screening had BV (by Nugent and/or Amsel criteria) and/or 246 TV (by wet mount and/or culture): 82.4% had BV alone, 14.7% had BV and TV, and 2.9% had 247 TV alone. Therefore, as expected, most women had dysbiotic VMB types by 16S rRNA gene 248 sequencing at screening: BV_GV (41.8%), LA (17.9%), BV_noGV (11.9%), GV (11.9%), and 249 PB (1.5%). However, 14.9% had a lactobacilli-dominated Li VMB type, of whom 60% were TV-250 negative, and these women would not have needed metronidazole treatment. Also by design, all 251 randomized women were BV-negative (by Amsel criteria) and TV-negative (by wet mount and 252 culture) at the time of randomization. However, almost half of the women (44.8%) were not 253 lactobacilli-dominated by sequencing at the time of randomization, and none of them were L. 254 crispatus-dominated (which is considered the most optimal VMB state): their VMB types were 255 Li (52.2%), Lo (3.0%), LA (26.9%), PB (9.0%), GV (6.0%), and BV_GV (3.0%) ( Table 1). 256 257 At randomization, the mean total bacterial concentration ranged from 5.54-6.34 log10 cells/μl in 258 the four randomization groups, and these ranges were 5.22-5.97 for lactobacilli, 3.36-5.62 for 259 BV-anaerobes, 1.34-2.36 for pathobionts, and 0.57-2.20 for 'other bacteria' (Table S4; relative 260 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the (which was not peer-reviewed) The copyright holder for this preprint . https: //doi.org/10.1101/19001156 doi: medRxiv preprint abundance data in Table S5). The mean richness ranged from 7.5-17.8, and the mean Simpson 261 diversity from 0.15-0.38. Randomization did not completely balance the baseline VMB 262 compositions of the four groups: the concentrations total bacteria and BV-anaerobes were higher 263 in the GynLP group than the control group (Mann Whitney U test p<0.05), and the BV-anaerobes 264 concentration (p<0.01) and Simpson diversity (p<0.05) were lower in the EF+ group than the 265 control group (Table 1). 266 267 Variables that were associated with at least one VMB composition variable in unadjusted mixed 268 effects models (Table S6) included currently using hormonal contraception or being pregnant 269 (associated with a higher pathobionts concentration), above-average sexual risk taking based on 270 reported numbers of partners and condom use (higher pathobionts concentration), aged 30 years 271 or older (lower BV-anaerobes and pathobionts concentrations), and managing menses with a 272 sanitary pad compared to other methods (higher Nugent score). Reporting current urogenital 273 symptoms was also associated with VMB composition but this likely represents reverse 274 causality. We could not exclude women with ongoing chlamydia and/or gonorrhea infections at 275 randomization because the turn-around time of diagnostic testing was slow, but the VMB 276 compositions after metronidazole treatment of women with and without ongoing chlamydia 277 and/or gonorrhea infection were similar ( Figure S2). As mentioned earlier, short course antibiotic 278 use that was not part of the study interventions was not associated with VMB composition either 279 (Tables S2 and S6; Figure S2). 280 281 Safety 282 Two serious adverse events occurred but these were judged not to be related to study 283 participation: one woman in the oral metronidazole group was hospitalized for typhoid fever and 284 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint one woman in the EF+ group for malaria during pregnancy. Both events occurred after the 285 intervention period and both women recovered completely. Two women reported non-serious 286 social harms that were judged related to study participation. One woman in the control group was 287 beaten by her partner because she engaged in self-sampling; she withdrew from self-sampling 288 but continued participation in the study. One woman in the GynLP group suffered verbal 289 harassment from her partner and her sister for taking part in the study and elected to withdraw. 290 291 During the intervention period, urogenital symptoms were reported by 13.4% of participants 292 (genital itching, burning, and pain during sex but no vaginal discharge) with no differences 293 between groups, and only two speculum exam and no bimanual exam findings were reported by 294 the physician (Table 2). After product cessation, urogenital symptom reporting was similar 295 compared to the intervention period (10.8%), but the number of speculum exam findings 296 increased (32.8%), likely reflecting the high urogenital infection incidence in this cohort. A total 297 of 41 adverse events were spontaneously reported between enrollment and M6. All of them were 298 judged definitely not or unlikely to be related to trial participation, and they were evenly 299 distributed among groups. 300 301 Preliminary efficacy: microscopy endpoints 302 In modified ITT analyses, the BV incidence rate by Nugent score during the intervention period 303 was 10.18 per person-year at risk (PY) in the control group, and lower in the metronidazole 304 (1.41/PY; p=0.004), EF+ (3.58/PY; p=0.043), and GynLP groups (5.36/PY; p=0.220) ( Table 3). 305 Mean Nugent scores during the intervention period were highest in the control group, lowest in 306 the metronidazole group, and in-between in the two vaginal probiotics groups ( Figure 2A). By 307 the end of the intervention period, many women had developed microbiological-BV without 308 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi. org/10.1101/19001156 doi: medRxiv preprint symptoms. In line with standard practice, they were not treated, but they were also no longer 309 included in the 'person-years at risk' denominator because they had already developed the 310 endpoint. BV incidence rates were therefore much lower (1.26/PY overall) after cessation of the 311 intervention, and similar between groups. The results for BV incidence by modified Amsel 312 criteria were similar to those for Nugent scores ( We took the baseline VMB composition imbalances into account by not only showing bacterial 317 group means at each visit but also mean differences with enrollment ( Figure 2 and Table S4 for 318 concentrations, and Figure S3 and Table S5 for relative abundances), and by using mixed effects 319 models including participant identification number as the random effect to determine if any 320 changes were statistically significant (Table 4). Immediately after BV treatment completion, the 321 VMBs of most women gradually worsened (lactobacilli declined and BV-anaerobes expanded) 322 due to the high-risk nature of the cohort. The mean lactobacilli concentration declined to a low of 323 3.86 log10 cells/μl at M2 in the control group, but less so in the intervention groups (ranging from 324 4.60-5.58 log10 cells/μl at follow-up visits). In unadjusted mixed effects models using data from 325 the intervention period only, metronidazole users had a higher lactobacilli concentration 326 (p=0.043) and relative abundance (p=0.006) than controls, EF+ users had a higher relative 327 abundance (p=0.014) but not concentration than controls, and GynLP users had trends in the 328 same directions that were not statistically significant. The expansion of BV-anaerobes was 329 significantly lower in oral metronidazole users (relative abundance; p=0.023), and in EF+ users 330 (concentration; p=0.041), compared to controls. Mean pathobionts concentrations were low in all 331 groups throughout, ranging from 1.61-3.35 log10 cells/μl at follow-up visits. Mixed effects 332 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint models did not identify any significant associations between randomization groups and 333 pathobionts concentrations or relative abundances, but showed trends (0.05<p<0.1) towards 334 lower pathobionts relative abundances in the two vaginal probiotics groups compared to controls 335 (Table 4). Mean concentrations of 'other bacteria' were also low throughout, but highest in the 336 EF+ group at the D7 and M1 visits (EF+ contains a Bifidobacterium strain). This was significant 337 in unadjusted mixed effects models for relative abundances (p=0.023) but not concentrations. 338

339
The proportions of women in each group at each visit having a particular VMB type 340 corresponded with the concentration and relative abundance data, and additionally showed that -341 in lactobacilli-dominated women -the Li VMB type was far more common than the Lcr and Lo 342 VMB types throughout ( Figure S4). Among women with dysbiosis, the LA and BV_GV VMB 343 types continued to be the most common dysbiosis types during follow-up. In unadjusted mixed 344 effects models using intervention period data only, metronidazole users and EF+ users, each 345 compared to controls, were significantly less likely to have dysbiotic VMB types (BV_GV,346 BV_noGV, and GV combined; p=0.012 and p=0.029, respectively) ( Table 4). 347

348
The associations in unadjusted mixed effects models persisted after adjustment for hormonal 349 contraception use/pregnancy, sexual risk taking, and age, except for the association with 350 Lactobacillus concentration among metronidazole users. Metronidazole users compared to 351 controls had a significantly higher Lactobacillus relative abundance (p=0.014), a significantly 352 lower BV-associated bacteria relative abundance (p=0.049), and were significantly less likely to 353 have BV by Nugent scoring (p=0.031) or by VMB types (BV_GV, BV_noGV, and GV 354 combined; p=0.017) ( Table 4). EF+ users compared to controls had a significantly higher 355 Lactobacillus relative abundance (p=0.025), a significantly lower BV-associated bacteria 356 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Detection of probiotic strains 360
During the intervention period, relevant probiotic strains were detected in 39% of samples from 361 EF+ users and 20% of samples from GynLP users (all swabs combined, including self-sampled 362 swabs). The detection percentages were 58% and 31%, respectively, in sensitivity analyses using 363 non-rarefied data. Some of the EF+ strains cannot be differentiated from naturally occurring 364 strains, and EF+-like strains were therefore detected (at low levels) in all groups at most time 365 points ( Figure 3, Table S4). However, the mean concentrations were highest in the EF+ group 366 during the intervention period (mean concentrations 0.48-1.92 log10 cells/μl per visit for all 367 women combined, and 3.62-4.28 log10 cells/μl per visit for women who did have EF+ strains 368 detected using rarefied data). The GynLP strain was only detected in the GynLP group during the 369 intervention period (mean concentrations 0.25-1.05 log10 cells/μl per visit for all women 370 combined, and 3.72-4.55 log10 cells/μl per visit for women who did have GynLP detected using 371 rarefied data). Inter-and intra-individual differences between participants were high: the highest 372 vaginal probiotic concentration detected in an individual EF+ user was 5.51 log10 cells/μl, and in 373 an individual GynLP user was 6.17 log10 cells/μl. During the intervention period, the mean 374 relative abundances of the probiotic strains were 0.03 in both EF+ and GynLP users, and 0.08 375 and 0.15, respectively, if only samples in which any strains were detected were included. 376

VMB transitions 378
The stacked graph and alluvial diagrams ( Figure S4) show that transitions from one VMB type to 379 another were common. As expected, most transitions during the intervention period were from 380 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

DISCUSSION 392
This trial showed that all three interventions were safe. Our preliminary efficacy results confirm 393 that intermittent use of metronidazole reduces BV recurrence 10-12 , and suggest that intermittent 394 use of lactobacilli-containing vaginal probiotics may also reduce BV recurrence. We also found 395 that vaginal probiotic use is acceptable and feasible in African settings (to be reported 396 elsewhere). These findings are important because many women and clinicians would prefer safe 397 and efficacious vaginal probiotics to metronidazole as they are not expected to negatively affect 398 other body niche microbiotas or cause antimicrobial resistance, and may have fewer side effects. 399 400 Our trial was funded as a pilot study and therefore had a modest sample size. Despite this, many 401 of the preliminary efficacy associations for intermittent metronidazole and EF+ use reached 402 statistical significance. While this was not the case for intermittent GynLP use, all of the trends 403 in this group were in the same directions. We believe that the GynLP group suffered a few 404 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint disadvantages compared to the other randomization groups, which may explain the lack of 405 statistical significance. Randomization imbalances commonly occur in small trials 36 , and in our 406 trial, this led to women in the GynLP group being more dysbiotic at baseline than controls. We 407 ameliorated this disadvantage in our analysis strategy, but we may not have been able to 408 eradicate it. In addition, GynLP users were less adherent on average than metronidazole and EF+ 409 users. Differences in adherence using triangulated data did not reach statistical significance, but 410 probiotic strain detection rates (58% for EF+ samples and 31% for GynLP samples using non-411 rarefied data) support this claim. Biose has since simplified the dosing regimen of next 412 generation Gynophilus products to two fixed days a week to boost adherence. Finally, GynLP 413 dosing (once every four days) was similar to metronidazole dosing (twice weekly) but less 414 frequent than EF+ dosing (once per day for the first five days followed by thrice weekly for the 415 remainder of the intervention period). 416 417 Probiotic detection rates were 58% for EF+ samples and 31% for GynLP samples, and inter-and 418 intra-individual variabilities were high (with probiotic concentrations ranging from zero to 6.17 419 log10 cells/μl). This detection variability is consistent with most other vaginal probiotic studies 420 that used sampling at non-daily intervals and molecular assessment methods 21,37-39 . A major 421 drawback of all of those studies, including ours, is that product use was not directly observed but 422 self-reported, and precise information about the time period between last product insertion and 423 sample collection was lacking. The average total bacterial load of a healthy vagina is currently 424 not known 40 . The average vaginal surface area was estimated to be 87.5 cm 41 . One Dacron swab 425 head in this study absorbed about 10 6 /μL bacteria. If we assume that one swab head absorbs on 426 average 200 μL 42 , and that this covers about one cm 2 of a total of about 100 cm 2 vaginal surface, 427 the total bacterial load in the vagina would be in the order of 2x10 10 bacteria. The vaginal 428 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint probiotic strain(s) in our trial were both applied at about 1.5x10 9 CFU per dose, which would be 429 about 7.5% of the total vaginal load after application if all probiotic bacteria were to remain in 430 the vagina. We detected mean relative abundances of 7.7% for EF+ strains and 15.1% for GynLP 431 when only samples with any relevant probiotic strains detected during product use were included 432 (thereby eliminating any potential non-adherence). A recently published study of Gynophilus 433 Slow Release tablet (which is almost identical to GynLP) in which women self-sampled every 434 day showed that mean vaginal concentrations of Lcr35 by qPCR were between 10 4 and 10 6 435 CFU/μl in women who used the tablet once every four or five days 43 . Using our estimated 436 concentration data, we detected a similar concentration range in samples with any GynLP 437 detected. This consistency is reassuring, but the question remains whether probiotic 438 concentrations of this order of magnitude optimally prevent BV recurrence in the long-term. 439 Furthermore, all studies referenced in this paragraph, including ours, have shown that probiotic 440 strains do not persist in the vagina after dosing has ceased. The second question then is whether 441 the colonization capacity of probiotic bacteria should be improved. Our data suggest that 442 probiotic lactobacilli may boost 'natural' lactobacilli indirectly, which may be sufficient. Indirect 443 effects may include increased localized lactic acid production, modulation of cervicovaginal 444 mucosal immune responses, and/or inhibition of biofilm formation, by probiotic bacteria 3,44,45 . 445 446 Additional limitations of our study include the high urogenital infection risk of this cohort 447 (which makes prevention more challenging), and our inability to fully control for potential 448 confounders. However, the mixed effects models were controlled for some of the best known 449 VMB determinants (hormonal contraception, pregnancy, sexual risk taking, and age) 1,46-48 . We 450 were not able to exclude women with gonorrhea and/or chlamydia infection at the time of 451 randomization due to the slow laboratory turn-around time, but the VMB compositions of 452 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint women with and without infection were similar, and we therefore think that this did not 453 negatively affect our results. microscopy. The addition of sequencing methods showed that many more women than 459 previously thought are not lactobacilli-dominated after standard antibiotic BV treatment, that 460 host responses to antibiotic and probiotic treatment are highly variable, and that it is possible to 461 differentiate between probiotic strains and 'natural' lactobacilli. Furthermore, others have shown 462 that quantifying relative abundance data in the same manner as we have done in this study 463 correlates well with species-specific quantitative PCRs of non-minority species 33,34 . This then 464 allows for microbiota data reduction into quantitative variables that can be analyzed in mixed 465 effects models that adjust for repeated measures and confounding. We recommend that future 466 trials incorporate these or other rigorous methods, optimize dosing and timing of product 467 insertion versus sample collection, and enroll women with various urogenital risk profiles. 468 Ideally, these trials would also evaluate the effects of interventions on vaginal biofilm formation, 469 and -eventually -the impact on pregnancy complications, HIV epidemics, and other adverse 470 outcomes. 471 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint

Figure 1. Flow diagram of numbers of women seen and study procedures at each visit
Abbreviations: BV, bacterial vaginosis; D7, day 7 visit; M1/2/6, month 1/2/6 visit; RU, Rinda Ubuzima; STI, sexually transmitted infection; TV, Trichomonas vaginalis; UTI, urinary tract infection. *Totals to 110 reasons among 102 women because there could be more than one reason per woman. †No speculum exam performed; molecular testing of self-sampled vaginal swabs. ‡Reasons: outside of metronidazole treatment window (n=5), enrollment target already met (n=4), has a mental disorder (n=1), did not complete screening procedures and was subsequently lost to follow=up (n=1), withdrew consent during the Screening visit because she thought the reimbursement was too low (n=1). §Three women in each randomization group were selected for self-sampling every other day during the first month of follow-up.
¶Reasons: moved away from Kigali (n=2), lost interest because symptoms resolved (n=1), and was verbally harassed by partner and sister about study participation (n=1).
. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint .
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint , and M2 visits). Selfsampled samples were also taken during product use, but were not Gram stained nor tested by 16S rRNA gene qPCR (see Methods). ‡Including all valid samples during product use (D7, M1, and M2 visits, and self-sampled samples). §Model adjusted for hormonal/pregnancy status, sexual risk taking, and age (see Methods).
. CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/19001156 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi. org/10.1101/19001156 doi: medRxiv preprint