Bacterial Vaginosis Incidence Following a Single Hyperosmolal Vaginal Lubricant Exposure: A Comparison of Two Observational Cohorts.

Of the 121 and 135 participants enrolled in the GALE and UMB-HMP studies, 63 and 76 were excluded from all analyses, respectively. Further details on sample sizes and reasons for participant exclusion for each analysis can be found in Figure S1 and Figure 1 , respectively. In each cohort, approximately 6% of participants were not included in analyses due to being lost to follow-up. 113 participants were eligible for the analysis of new-onset/newly-symptomatic BV. Demographic and behavioral characteristics among participants included in the analysis are listed in Table 1 . While these cohorts were largely similar, hormonal contraceptive use and a higher number of lifetime sexual partners were more frequently reported among lubricant-exposed participants. In contrast, thin, homogenous discharge at baseline, and condom use reported on daily diaries, were more common among lubricant-unexposed participants. There was a higher proportion of participants with no Amsel criteria at baseline in the lubricant-exposed versus the lubricant-unexposed cohort (67% versus 34%), and a Fishers exact test detected that the number of Amsel criteria present at baseline was significantly different between the cohorts (p<0.01). Among GALE participants, common indications for TVUS included pelvic pain (36%), assessment of pelvic mass (31%), abnormal uterine bleeding (26%), and IUD localization (19%). TVUS reports indicated that fibroids were visualized in 21% of participants, while 41% had no significant clinical findings. There was no significant difference in the distribution of self-reported race in each cohort for participants eligible for each outcome; however, 94% (17/18) and 88% (23/26) of all instances of new-onset BV and new-onset/newly-symptomatic BV occurred among Black participants, respectively. Across both studies, the risk for new-onset BV was 33% among Black participants and 3% among non-Black participants. The risk of any new-onset BV was 24% (11/46) in the lubricant-exposed cohort and 16% (7/43) in the lubricant-unexposed cohort. In the lubricant-exposed cohort, 73% (8/11) of new-onset BV was found at the post-TVUS follow-up visit in week 2. In the lubricant-unexposed cohort, 86% (6/7) of new-onset BV was diagnosed by the follow-up visit in week 5, with 2 participants returning for an early follow-up visit in week 2 for vulvovaginal symptoms. The risk of new-onset/newly-symptomatic BV was 27% (15/55) in the lubricant-exposed cohort and 19% (11/58) in the lubricant-unexposed cohort. 73% and 82% of new-onset/newly-symptomatic BV cases were found at the first follow-up visit in the lubricant-exposed and lubricant-unexposed cohorts, respectively. There was not strong evidence for an association between lubricant exposure and BV risk in adjusted models ( Table 2 ), though the risk for new-onset BV appeared to be higher after excluding participants reporting condom use ( Table S1 , aRR: 1.97, 95% CI: 0.77 – 5.03 versus Table 2 , aRR: 1.32, 95% CI: 0.58 – 3.01). Findings were also similar after excluding participants with 2 or more Amsel criteria present at baseline ( Table S2 , aRR: 1.71, 95% CI: 0.67 – 4.38). The risk for new-onset discharge was lower in the lubricant-exposed cohort (aRR: 0.52, 95% CI: 0.26 – 1.05), though this association did not hold in the sensitivity analysis excluding participants reporting condom use (aRR: 0.63, 95% CI: 0.27 – 1.48). There were no differences observed for the other Amsel outcomes (high pH, positive whiff test, or clue cells). Among participants reporting Black race, 38% (10/26) of those who were lubricant-exposed and 27% (7/26) of lubricant-unexposed developed any new-onset BV during the 10-week follow-up, and 38% (13/34) of lubricant-exposed and 26% (10/39) of lubricant-unexposed participants developed new-onset/newly-symptomatic BV. 76% (13/17) of the new-onset BV and 74% (17/23) of the new-onset/newly-symptomatic BV was observed by the first follow-up visit. Using modified Poisson regression, the risk for any new-onset BV was 2.2-fold higher (95% CI: 1.16 – 4.11) and the risk new-onset/newly-symptomatic BV was 1.9-fold higher (95% CI: 1.04 – 3.51) in the lubricant-exposed cohort, adjusting for recent sexual partners and age ( Table 3 ). We also observed increases in the risk for positive whiff test (aRR: 1.69, 95% CI: 0.92 – 3.10) and clue cells (aRR: 2.18, 95% CI: 1.17 – 4.06). Findings were similar after excluding participants reporting condom use ( Table S3 ) and excluding participants with 2 or more Amsel criteria present at baseline ( Table S2 ).

This secondary analysis utilized data from women enrolled in two longitudinal cohorts. Lubricant-exposed and lubricant-unexposed participants were drawn from the Gynecology and Lubricant Effects (GALE) study 33s and the University of Maryland Baltimore Human Microbiome Project (UMB-HMP) study, respectively. 34s The GALE study evaluated the association between a single use of hyperosmolal lubricant (GLIDE lubricating jelly, Athena Medical Products: ~2,400 mOsm/kg, pH=5.5) on vaginal health in non-pregnant women aged 18 and older referred for TVUS at the University of Maryland Medical Center between July 2017 and February 2020. 33s Participants over the age of 45 were excluded from this analysis to match the UMB-HMP definition of reproductive-age. Participants attended three clinical visits: baseline (week one), post-TVUS follow-up (week two, approximately 2–5 days after TVUS), and final (week 10). The UMB-HMP study evaluated the longitudinal dynamics of the vaginal microbiota in non-pregnant, reproductive-age women aged 18 to 45 years old recruited at the University of Alabama at Birmingham between September 2009 and July 2010. 34s Participants who reported lubricant use in the two months prior to enrollment were excluded from this analysis. Participants attended three clinical visits: baseline (week one), follow-up (week five), and final (week 10), but could return for additional clinical visits if they were experiencing vaginal symptoms. In both studies, exclusion criteria included self-reported diabetes mellitus, a positive test for gonorrhea, chlamydia, trichomoniasis, HIV, or syphilis at baseline, and antibiotic or antifungal use at baseline for symptomatic BV or vulvovaginal candidiasis. The UMB-HMP study excluded those with antibiotic or antifungal use within four weeks prior to baseline, while the GALE study excluded those with antibiotic or antifungal use within four weeks prior to TVUS. The GALE study additionally excluded those self-reporting immunosuppression or lubricant use within a week of TVUS. All participants filled out daily health and behavior diaries. During study visits, clinicians evaluated participants for BV using Amsel criteria (thin homogenous discharge, vaginal pH > 4.5, positive whiff test, and presence of clue cells). The speculum was lubricated with water. To diagnose BV, at least three of four Amsel criteria needed to be present. 13 Participants who met the criteria for BV and who reported symptoms were considered to have symptomatic BV and were treated with antibiotics following CDC guidelines. 35s IRB approvals were obtained from the University of Alabama Birmingham and the University of Maryland, Baltimore for the UMB-HMP cohort, and from the University of Maryland, Baltimore for the GALE cohort. All participants provided written informed consent. Modified Poisson regression with robust standard errors 36s was used to model the cumulative incidence of BV (new-onset and new-onset/newly-symptomatic) and individual Amsel criteria within 10 weeks, comparing the lubricant-exposed (GALE) and lubricant-unexposed (UMB-HMP) cohorts. Participants had to have no BV at baseline to be at risk for developing new-onset BV, which was defined as a diagnosis of any BV (symptomatic or asymptomatic) at follow-up or final visit. Participants with asymptomatic BV at baseline were eligible for the analysis of new-onset/newly-symptomatic BV ( Figure S1 ). Participants who did not have any individual Amsel criteria present at baseline were eligible for the analysis of new-onset Amsel criteria ( Figure S1 ). Most instances of incident BV in both cohorts occurred in participants self-reporting Black race; given prior work has documented strong associations between race/ethnicity and the frequency of certain types of vaginal microbiota, 37s we repeated the analysis among Black participants. For each model, participants were excluded if they reported use of antibiotics, antifungals, or commercial vaginal lubricants prior to experiencing the outcome of interest. Participants who experienced the outcome of interest at follow-up but later used lubricants or were lost to follow-up before the final visit were included. As a sensitivity analysis, because condoms may utilize water-based lubricants, we excluded participants reporting condom use in the study and repeated the analysis. Due to differences between the two cohorts in baseline Amsel characteristics, we also excluded participants with 2 or more Amsel criteria present at baseline and repeated the analysis of new-onset BV. We examined bivariate associations between potential confounders (selected a priori based on previously identified risk factors for BV) and both lubricant use and each outcome, including age, race, hormonal contraception use, menses (any during follow-up or concurrent with outcome assessment), as well as sexual or personal hygiene behaviors prior to and during study enrollment. Variables were included when they were associated with both lubricant exposure and the outcome of interest.

Among all participants, we did not observe a statistically significant increase in the risk for new-onset or newly-symptomatic BV comparing the lubricant-exposed to the lubricant-unexposed cohorts over 10 weeks. In contrast, Black participants, who accounted for 94% of the incident BV cases in the study, were twice as likely to have new-onset or newly-symptomatic BV after a single lubricant exposure during TVUS and were more likely to develop clue cells and a positive whiff test. This heightened risk of BV in Black participants is notable, as it may reflect a lower threshold for developing BV. It is particularly relevant given that Black women exhibit the highest prevalence of uterine fibroids, which are typically diagnosed and monitored through TVUS. 38s As noted by Swilley-Martinez et al, 39s stratified or restricted analyses are important when an intervention may be harmful in a specific subgroup. We conducted such an analysis to avoid relying solely on pooled estimates adjusting for race, which can obscure meaningful differences and may not represent the association between lubricant and BV in either population. We were unable to evaluate whether lubricant exposure had different effects on BV risk in Black versus other participants due to the low number of non-Black participants with incident BV; however, we could determine that the risk for incident BV over 10 weeks was significantly higher among all Black participants (17/52, 33%) versus all non-Black (1/37, 3%) participants (p=0.01, Fisher’s exact test). It is possible that if we had a larger sample size, we may have observed a similar increased BV risk in non-Black lubricant-exposed versus non-Black lubricant-unexposed participants, despite the overall lower rate of BV in this population. Regarding new-onset Amsel criteria among all participants, there was a decreased risk for vaginal discharge in lubricant-exposed participants (p=0.07). However, there were significant differences in the prevalence of discharge at baseline between the cohorts ( Table 1 ). This suggests that there may be differences in how clinicians in each study were recording abnormal discharge, which could ultimately affect incidence measures. These findings build on evidence from prior analyses in this cohort. Mature superficial vaginal epithelial cells help maintain the structural integrity of the vaginal epithelium. 40s We used differential fluorescence staining of cytoplasm and nuclei to measure vaginal epithelium maturity index and found that reproductive-age participants with a maturity index greater than three prior to TVUS shed more immature vaginal epithelial cells up to two weeks after TVUS, suggesting compromised epithelial barrier integrity. 33s We also observed an increased prevalence of Amsel-BV and positive whiff test in the days following TVUS compared to the week before in reproductive-age participants. 41s Lastly, an analysis using all 10 weeks of follow-up data found that participants with a prior history of BV and peri/post-menopausal participants were more likely to have a vaginal microbiota which was not dominated by Lactobacillus spp. after TVUS versus before. 12 Because these analyses utilized a single-crossover design, the findings may reflect typical fluctuations in the vaginal microbiota or BV status over time rather than an effect of lubricant. The present analysis utilized a separate lubricant-unexposed cohort with the same length of follow-up and therefore controls for the effect of time. There are limitations to this analysis. BV status was initially assessed before the lubricant exposure (median: 7 days later), and we assume no changes in BV status during that period. The UMB-HMP study provided a convenience sample from which we could construct a lubricant-unexposed cohort. However, the two cohorts differed in the timing of their first follow-up visit, which prevented us from conducting a traditional time-to-event analysis. Furthermore, due to these differences, we have an unequal opportunity in each cohort to capture instances of new-onset BV or Amsel criteria, which were followed by spontaneous resolution. Data on short-term spontaneous resolution of asymptomatic BV are limited, with one study reporting 11% (3/27) within 21 days of receiving placebo gel compared to 36% (9/25) in those receiving metronidazole gel treatment. 42s In our study, 0% (0/15) of UMB-HMP participants and 10% (1/10) of GALE participants with asymptomatic BV at baseline had no BV at their next follow-up visit, suggesting spontaneous resolution was uncommon within 2–5 weeks and is unlikely to contribute to a biased outcome assessment between the two cohorts. Additionally, UMB-HMP participants could return for an interim visit before week 5 if they experienced vulvovaginal symptoms. As this is an observational study, the participants in the UMB-HMP cohort who chose not to use personal lubricants (lubricant-unexposed) may differ in unmeasured demographic and behavioral characteristics from GALE participants referred for TVUS (lubricant-exposed). Some of these indications, such as evaluation of uterine fibroids, may also be associated with risk for BV. 43s However, the prevalence of fibroids among GALE participants (21% overall, 29% among Black women) was similar to that of asymptomatic women (15% overall, 27% among Black women) in another study, 44s suggesting our two cohorts may be similar with regard to fibroid prevalence. The two cohorts also differed in the number of Amsel criteria present at baseline, which could accelerate a tipping toward BV; however, we still observed an increased risk of new-onset BV among Black lubricant-exposed participants with 0–1 Amsel criteria present at baseline versus Black lubricant-unexposed participants with 0–1 Amsel criteria present at baseline ( Table S2 ). Other limitations include the overall small number of events and sample size despite high cumulative incidence of BV among Black participants, and that the GALE study conditions may not mimic typical lubricant use in personal practice. Future studies, particularly clinical trials, could explore the effects of lubricants on the vaginal microbiota and risk for vaginitis and reproductive tract infections. These trials should standardize lubricant exposure and include iso-osmolal lubricants for comparison. While some women with a less resilient vaginal microbiota may be affected by a single instance of lubricant use, 12 it would be of interest to determine whether there are similar effects on BV incidence with more frequent use. Researchers should consider the sample size needed to explore whether effects are more pronounced in certain subgroups. Where clinical trials are not feasible, observational studies may document specifics such as brand, ingredients, osmolality, and frequency of lubricant use.

Over half of women report ever using personal vaginal lubricants. 1 These products are often used to enhance sexual satisfaction or combat vaginal dryness, a concern particularly prevalent among older women. 2 Lubricants are also used to reduce patient discomfort during exams like transvaginal ultrasound (TVUS). Lubricants used in clinical settings are highly similar in composition to personal lubricants. 3 Water-based lubricants are cost-effective, compatible with all condoms, and are associated with less vulvovaginal irritation compared to silicone-based lubricants. 4 However, there have been concerns regarding their use. Most vaginal lubricants are hyperosmolal relative to semen and vaginal secretions due to glycols such as glycerin, propylene glycol, and polyethylene glycol which act as humectants. 5 In vitro studies indicate hyperosmolality can damage the vaginal epithelium and increase cellular shedding. 3 , 6 , 7 Other factors that may contribute to lubricant cytotoxicity include preservatives, microbicides, and pH above 4.5. 5 Lubricant cytotoxicity is concerning because the vaginal epithelium and its microbiota serve as the first line of defense against reproductive tract infections. Lactobacillus spp. produce lactic acid, which in turn promotes the production of anti-inflammatory cytokines and inhibits infection by urogenital pathogens including HIV. 8 Some studies have suggested certain hyperosmolal lubricants affect the viability of Lactobacillus spp, 7 , 9 – 11 and the dominance of Lactobacillus spp. in the vaginal microbiota among older women and those with a prior history of bacterial vaginosis (BV). 12 BV is a clinical syndrome common in reproductive-age women and is marked by low levels of Lactobacillus spp. and a wide array of strict and facultative anaerobic bacteria. BV can be diagnosed when a patient presents with at least 3 of 4 Amsel clinical criteria, and is often defined in research studies by Nugent scoring of Gram-stained vaginal smears. 13 , 14 Although BV is the most common cause of vaginal symptoms among reproductive-age women, over half of women with BV are asymptomatic (40–84%). 15 , 16 Beyond personal discomfort, BV is a public health concern as it is associated with increased risk for STI acquisition and transmission, 17 – 19 and preterm delivery. 20 Some established risk factors for BV include douching, sexual activity, a greater number of sexual partners, cigarette smoking, menstruation, and not using hormonal contraception. 21 – 23 Black women are twice as likely as their white counterparts to experience BV; 15 research on the mechanisms underlying these differences is ongoing, but may include psychosocial and sexual network factors. 24 , 25 Some observational studies have found a greater BV prevalence among women reporting vaginal lubricant use, 26 – 28 while other studies found no association. 29 , 30s, 31s There is significant heterogeneity between studies regarding the type of lubricant evaluated, how recently lubricant use was reported (within 1 to 90 days), and how BV is defined (Amsel-BV versus Nugent-BV 32s ). Personal lubricants vary in composition and characteristics, including differences in osmolality and pH. 5 Therefore, it is important to evaluate lubricant exposure in a standardized manner, considering the product and frequency of use. We sought to evaluate the risk of Amsel-BV and individual Amsel criteria over 10 weeks in a cohort of reproductive-age participants exposed once to a standardized dose of a hyperosmolal lubricant during TVUS and compared these risks to a separate cohort of reproductive-age participants who reported no recent lubricant exposure.