This study used baseline data and biological specimens from the POWWeR Health Study (Periods: Optimizing Working Women’s Reproductive Health), an open-label, single arm trial assessing the impact of soft, disc-shaped menstrual cups on BV and adverse events (ClinicalTrials.gov NCT03051789) (Zulaika et al., 2024). The VMB, BV, and STIs, are measured among participants under usual practice conditions for one year, followed by provision of reusable soft disc-shaped menstrual cups that can be worn during sex, and then followed for VMB, BV, and STIs for another one year. A sample size of 402 participants was sought to be able to detect a 20% decrease in occurrence of BV between the control and intervention phases, and a 1% serious adverse event rate with 80% power (Zulaika et al., 2024).

To be eligible, participants had to be aged 15 to 35 years, having menstruated in the past two months, and having exchanged sex for money or basic necessities (rent/housing, food, medical care for self or children, tuition for self or children) within the past two months, and living and/or working in the Kisumu area. We excluded women who were currently pregnant (as determined by urine hCG test) or had been pregnant in the past six months (as they may still be lactating), or having an intrauterine device (IUD) in situ due to risk of expulsion with menstrual cup use. All participants provided written informed consent in their preferred language (English, DhoLuo, Kiswahili).

Participants underwent a tablet-based survey administered by female study staff in their language of choice to obtain information on socio-demographics and sexual and menstrual practices. Socio-demographic data included age and socioeconomic and financial indicators. Sexual practices were assessed with numerous variables relating to number of sexual partners, condom use, recency of sex, and payment for sex. Characteristics of the menstrual cycle were assessed as days since last menstrual period, duration of last menstrual period, and menstrual flow (asked as normal, light, or heavy). Phase of menstrual cycle was estimated by adding the days since the last period to the duration in days of the last menstrual period, and then assigned to estimated menstrual phase (Magoutas et al., 2025): 0–5 menstrual, 6–11 follicular, 12–16 ovulatory, 17–42 luteal. Those with missing days since last period or more than 42 days since in the estimated cycle were classified as uncertain. MHM was assessed with several questions including materials used to manage menses, intravaginal practices, and access to soap, water, and privacy at home and during sex work.

Use of sub-optimal menstrual materials was defined as a composite of any use of cloth, cotton balls, or tissue during the last menstrual period. Unsafe intravaginal practices examined: using cloth, tissue, paper, or cotton to wipe inside the vagina to remove fluids between clients, putting something inside the vagina before sexual intercourse to achieve a dry or tight sensation, putting something inside the vagina to keep it dry during menses, use of commercial douche product, and the frequency of wiping inside the vagina during menses, not during menses, and the difference between them (increased, decreased, or the same). We also examined MHM accounting for difficulty accessing water at home or at work, and difficulty accessing privacy at home or at work, in addition to using sub-optimal materials to manage menses.

All participants underwent a detailed medical history and physical examination by a study clinician. The clinician obtained four vaginal swabs. The first swab obtained was for 16S rRNA gene amplicon sequencing (microbiome), the second for BV, the third for detection of C. trachomatis (CT) and N. gonorrhoeae (NG), and the fourth for detection of T. vaginalis (TV). Vaginal swabs for amplicon sequencing were collected using OMNIgene Vaginal kits (OMR-130; DNA Genotek™), stored at -80 °C until shipment to Chicago for processing. Swabs for amplicon sequencing, CT/NG, and TV were taken immediately to the onsite UNIM Research and Training Laboratory for processing or storage.

Following manufacturer protocol, vaginal swabs were tested for CT/NG using the GeneXpert (Cepheid, Sunnydale, California, US). Swabs for TV were processed immediately upon receipt using the OSOM TV antigen detection assay (Sekisui, Lexington, MA, US). Air-dried smears were Gram stained and evaluated for BV according to Nugent’s criteria within 48 hours of receipt (Nugent RP and Hillier, 1991). HIV was assessed on a finger-stick blood sample using Determine rapid assay, following Kenyan national guidelines (Program NASC, 2022). Venous serum specimens were tested for HSV-2 antibody (Kalon HSV-2 immunoglobulin G enzyme-linked immunosorbent assay; Kalon Biological Limited Kingdom) using the manufacturer’s recommended cutoff.

Genomic DNA was extracted using a Chemagic 360 device (Revvity, Hamburg, Germany) and processed for sequencing using a two-stage PCR protocol (Naqib et al., 2018). Full-length 16S rRNA gene sequences were PCR amplified using primers 27F and 1492R, processed using PacBio Kinnex chemistry, and sequenced on a PacBio Revio instrument (Verma et al., 2025) (Supplementary Materials 1). Bioinformatics analysis was performed using QIIME2 2023.2 (Supplementary Materials 1) (Bolyen et al., 2019). Alpha‐diversity metrics (observed features (Faith, 1992), Shannon Index (Shannon, 1948), Simpson’s Index (Simpson, 1949), and Pielou’s Evenness (Pielou, 1966)) and beta diversity metrics were calculated using q2‐diversity. Taxonomy was assigned to ASVs against multiple databases to find the best match and reported accordingly. Contaminants were identified using the decontam program via ASVs in the reagent negative blank controls (Davis et al., 2018). Community state types (CSTs) were identified in a reference dataset using nearest centroid classification (VAginaL community state typE Nearest CentroId classifier, (VALENCIA) (France et al., 2020). Raw sequence data (FASTQ files) were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA), under BioProject identifier PRJNA1279642.

Of the 407 participants, 406 had a sample submitted for VMB assessment, with median sequence depth of 31,220 reads (interquartile range 23,362 – 41,185). Among 406 observations, n=11 (2.5%) had fewer than 2,500 sequence reads and were excluded from VMB analyses. Prior to analyses, data were filtered to retain taxa that contributed at least 0.01% of the total sequence reads, resulting in retention of 111 of 821 taxa. These 111 taxa included 16 “uncultured” and “unidentified” taxa with no phylogeny, which were excluded from inferential analysis as being non-informative, along with seven sparse taxa (having fewer than 10 non-zero observations). The resulting dataset comprised N = 88 taxa.

The primary exposure of interest was sub-optimal menstrual materials, defined as using cotton balls, cloth, or tissues to manage the last menstrual period occurring within the past 6 weeks. To identify factors associated with sub-optimal MM, we conducted multivariable Poisson regression with robust variance to estimate prevalence ratios. Next, we examined sub-optimal MM in relation to VMB composition in targeted and non-targeted analyses. Our a priori hypothesis was that number of sex partners, types of sex work venue, and sociodemographic indicators could be confounders, being associated with both VMB composition (primary outcome) and sub-optimal menstrual materials (primary exposure).

From February 8^th through September 25^th, 2023, 473 women were screened, and 415 (87.8%) were enrolled. Women could be ineligible for more than one reason; the results of screening and eligibility are detailed in Supplementary Materials 3. After informed consenting, there were 8 screening failures: 3 subsequently refused speculum examination, 2 subsequently disclosed being outside the eligible age range, 3 subsequently disclosed being amenorrheic. Thus, the final cohort proceeded with 407 women, with 395 evaluable in relation to VMB.

Participants were median age 27 years, and 59.2% with primary education or less (Table 1). Half (49.4%) of participants reported past month income of < 10,000 KES (~$77 USD), and 11.6% reported missing a meal in the past 7 days due to not having enough money. In the past 30 days, participants reported a median of 25 (IQR 15-50) clients and median payment at last sex act was 700 KES (median ~$5.50 USD). HIV was common, with 24.7% of women reporting themselves to be infected (no new infections were detected with testing), of whom 84% reported taking ARVs. By laboratory testing, 62.8% were HSV-2 seropositive, 42.2% had BV, and 21.9% had an STI (CT (12.8%), NG (5.4%), and/or TV (7.4%)). The combinations of infections are shown in Figure 1. Overall, 20.2% of participants were negative for HIV, BV, and all STIs. Mixed infections were common, most frequently in combinations with HSV-2 seropositivity and BV.

Overall, 97.8% (n=398/407) of participants reported menstruating in the past 6 weeks (Table 2). Nearly all (95.0%) women reported using disposable pads to manage menses, with a minority (5.3%) reporting reusable pads. Use of sub-optimal materials was common, with 42% reporting using any of the following: cotton wool (34.0%), tissues (14.1%), or cloth (8.1%). Difficulty accessing water was reported more frequently during sex work (23.8%) than at home (16.2%). Women reported they carried their own soap when doing sex work (57.5%) or that someone else provided soap (67.1%). Having difficulty for personal privacy was experienced by 12.5% of women while at home and by 18.7% during sex work. Over half of participants (57.0%) reported having fewer clients during menses compared to when they are not having menses. Overall, 56.0% of women wiped inside their vagina more frequently during menses than not during menses (on average 2.5 times per day vs. 1.9 times per day, p<0.001 sign rank test). The majority (92.9%) reported wiping inside the vagina with a cloth, tissue, or cotton to remove fluids between clients: always (76.9%), often (13.5%), sometimes (2.5%), never (7.1%).

In multivariable regression, factors associated with using sub-optimal MM included older age, lower educational attainment, and indicators of economic strain (missing a meal due to lack of money, not having income outside of sex work) (Figure 2; Supplementary Table 1). Difficulty having privacy at sex work, meeting clients at specific types of sex work venues (sex den/brothel, lodge/hotel/guest house), and having 20 or more clients in the past 30 days were also associated with increased likelihood of using sub-optimal MM.

Among the 395 participants included in VMB analysis, CST-IV (diverse) was the most common, identified in 251 (63.4%) women, primarily of sub-type IV-B (n=243). CST-III (L. iners dominated) was the next most common (28.3%), followed by CST-I (7.8%, L, crispatus dominated). The top 10 taxa with the highest relative abundance (Figure 3) accounted for 73.3% of sequence reads on average.

A small number of participants had CST-I, reducing power in this aspect of the analysis, and thus few factors associated with CST reached statistical significance at the p<0.05 level. In our non-targeted analysis, we observed that the taxa selected were not limited to the most common or abundant; however, taxa with low relative abundance may have lower variation in sub-samples of stability selection, smaller effect size, and therefore lower selection frequency across sub-samples. We measured menstrual materials used at most recent menses for ease of recall. Though our findings were in keeping with measures of menstrual management in the region (Phillips-Howard et al., 2015; UNFPA, 2025), it is possible this is not representative of women’s ongoing or typical practices. Nevertheless, the directions of the associations observed were in keeping with prior literature or had rational biological basis. Socioeconomic indicators were statistically significant in our analyses and associations are supported in the literature; however, we lacked a standardized index that would support direct comparability of socioeconomic status to other settings. Despite the high frequency of reported condom use, STIs and HIV were common, indicating that use may be over-reported or insufficiently consistent, reflecting an interventional gap. Ulcerative and non-ulcerative STIs have bidirectional effects on the VMB and in this cross-sectional analysis we do not know temporality; however, with longitudinal follow-up of this cohort and repeated STI testing, subsequent analyses will enable us to estimate changes in the VMB in response to incident STI and subsequent treatment. Our nested analysis within a cluster randomized study among adolescent girls found that sub-optimal MM was associated with increased likelihood of non-optimal VMB (Mehta et al., 2021) and that provision of menstrual cups had beneficial effects for the VMB (Mehta SD et al., 2023). Other observational studies have similarly found beneficial association between menstrual cups and VMB (Lebeer et al., 2023; Haahr et al., 2025). To test the generalizability of this finding, we intentionally expanded study of MM and VMB to a population with very high rates of sexual exposures and reproductive tract infections. Our findings add to the limited knowledge of menstrual materials and relation to VMB and reproductive tract health, and similar studies should be undertaken in a variety of settings to better identify avenues for optimizing menstrual health management and reproductive health benefit.