A cluster randomized controlled trial was performed in 96 Kenyan secondary schools, randomized (1:1:1:1) to control, menstrual cup, cash transfer, or menstrual cup plus cash transfer. This substudy assessing the impact of menstrual cups on BV, VMB, and STIs, included 6 schools from the control (3) and menstrual cup only (3) groups, both receiving BV and STI testing and treatment at each visit. Self-collected vaginal swabs were used to measure VMB (16S rRNA gene amplicon sequencing), BV (Nugent score), and STIs. STIs were a composite of Chlamydia trachomatis and Neisseria gonorrhoeae (nucleic acid amplification test) and Trichomonas vaginalis (rapid immunochromatographic assay). Participants were not masked and were followed for 30 months. The primary outcome was diagnosis of BV; secondary outcomes were VMB and STIs. Intention-to-treat blinded analyses used mixed effects generalized linear regressions, with random effects term for school. The study was conducted between May 2, 2018, and February 7, 2021. A total of 436 participants were included: 213 cup, 223 control. There were 289 BV diagnoses: 162 among control participants and 127 among intervention participants (odds ratio 0.76 [95% CI 0.59 to 0.98]; p = 0.038). The occurrence of Lactobacillus crispatus–dominated VMB was higher among cup group participants (odds ratio 1.37 [95% CI 1.06 to 1.75]), as was the mean relative abundance of L. crispatus (3.95% [95% CI 1.92 to 5.99]). There was no effect of intervention on STIs (relative risk 0.82 [95% CI 0.50 to 1.35]). The primary limitations of this study were insufficient power for subgroup analyses, and generalizability of findings to nonschool and other global settings.
Funding: This study was supported by the National Institutes of Health Eunice Shriver National Institute of Child Health and Human Development (R01-HD093780 to SDM), and the Joint Global Health Trials Initiative (UK-Medical Research Council/ Department for International Development/ Wellcome Trust/Department of Health and Social Care; MR/N006046/1 to PPH). The funders had no role in the design of the study, the collection, analysis, and interpretation of data, or in writing the manuscript.
Data Availability: Raw sequence data (FASTQ files) were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA), under BioProject identifier PRJNA746243. This study was conducted with approval from the Kenya Medical Research Institute (KEMRI) Scientific and Ethics Review Unit (SERU), which requires that data be released from any KEMRI-based Kenyan studies (including de-identified data) only after their written approval for additional analyses. In accordance, data for this study will be available upon request, after obtaining written approval for the proposed analysis from the KEMRI SERU. Their application forms and guidelines can be accessed at https://www.kemri.org/seru-overview . To request these data, please contact the KEMRI SERU at seru@kemri.org .
Menstrual cup use may hold promise as a multipurpose tool for BV and STI risk reduction by maintaining a Lactobacillus-dominated VMB. Our objective was to study the effects of menstrual cup use on the VMB, BV, and STIs in a longitudinal analysis nested within a cluster randomized controlled trial of 4,400 secondary school girls aged 14 to 25 in Siaya County, Kenya. We hypothesized menstrual cups would lead to increase in Lactobacillus crispatus and reduced BV and STIs.
Good menstrual hygiene management (MHM) includes using clean materials to absorb or collect menstrual blood, along with adequate education, sanitation facilities, and a conducive environment [ 7 ]. Lack of MHM materials can cause adolescent girls to miss school [ 7 ] and may place them at risk for coercive sex [ 8 – 9 ]. To address this intersection of lack of menstrual materials and increased risk of sexual exposures, a cluster randomized study of 644 girls aged 14 to 16 years old compared the effect of provision of reusable menstrual cups to controls provided puberty and hygiene training. After 1 year, girls receiving menstrual cups had a 35% lower prevalence of BV (p = 0.034) and a 52% lower prevalence of STIs (p = 0.039) compared to controls [ 10 ]. During menses, the structure of the vaginal microbiome (VMB) (collection of microorganisms in the vagina) often shifts toward lower relative abundance of Lactobacillus species and an increase in Gardnerella vaginalis, with increased detection of BV [ 11 ]. A prior study demonstrated that menstrual cup use was not associated with vaginal microflora destabilization during menses [ 12 ], which may explain the observed beneficial effect of menstrual cups on BV and STIs.
Adolescent girls and young women account for more than 60% of new HIV infections in sub-Saharan Africa (SSA), with an estimated 750 new infections occurring daily [ 1 ]. In western Kenya, HIV prevalence among 15- to 19-year-old adolescent girls is estimated at 3.8%, rising to 9.5% among young women aged 20 to 24 years old [ 2 ]. The sexually transmitted infection (STI) epidemic runs parallel to the HIV epidemic, with prevalences of chlamydia and gonorrhea among adolescents and young adults ranging from 7% to 17% [ 3 – 4 ]. Bacterial vaginosis (BV) affects 20% to 50% of the general population of women in SSA and Kenya [ 5 ] and increases the risk of HIV acquisition 1.6-fold, accounting for up to 15% of HIV infections worldwide given its high prevalence [ 6 ]. For adolescent girls, the HIV/STI epidemic overlaps with broader reproductive health concerns such as high risk of teenage pregnancy, with poor maternal outcomes, and higher risk of school dropout.
Methods
Ethics statement This study was approved by the institutional review boards of the Kenya Medical Research Institutes (KEMRI, SERU #3215), Liverpool School of Tropical Medicine (LSTM, #15–005), and University of Illinois at Chicago (UIC, #2017–1301). Written informed parental consent and written informed assent from minors was obtained for all participants.
Study setting, design, and participants Cups and Community Health (CaCHe, pronounced “Cash-Ay”) was a prospective analysis of secondary school girls in Siaya County, Kenya, nested within the CCG trial (ClinicalTrials.gov NCT03051789; S1 CONSORT Checklist) [13]. The CCG trial was an open-label, 4-arm, school-cluster randomized controlled superiority trial. Schools were allocated into 4 arms (1,1,1,1) via block randomization: (1) provision of menstrual cups with training on safe cup use and care; (2) conditional cash transfer (CCT) based on ≥80% school attendance in previous term; (3) menstrual cup and CCT; and (4) usual practice. All girls received puberty and hygiene education. Girls in control and cup only arms received hand wash soap. Blinding was not possible due to the nature of the interventions (menstrual cups and cash transfer). For the CaCHe study, nested within the CCG trial, we aimed to enroll 20% of girls in the cup only and control arms of the CCG trial. Because CCG was randomized, with participants in 2 of the 4 arms included in CaCHe, it would not have been feasible, ethical, or statistically necessary to again randomize rather than maintain the randomized groups. The cup and control participants under CaCHe differed from CCG in that they all received BV and STI testing and treatment; thus, the comparison was cups plus BV and STI testing and treatment versus control plus STI testing and treatment. All cups and control schools for CaCHe were enrolled in one of the 5 wards of Rarieda subcounty. There were 16 schools screened for inclusion in Rarieda, 4 of which were excluded due to small size. Of the 12 schools, 3 schools were randomized into each arm of the CCG trial, leading to the 3 menstrual cup arm schools and 3 control arms included in CaCHe. Eligibility for CaCHe mirrored eligibility for CCG: attendance at a participating school, being a resident of the study area, provision of assent and parental/guardian consent, and girls had to report established menses (≥3 times). Girls were excluded at baseline if they declared pregnancy (n = 1), had not reached menarche (n = 4), or were under age 14 (n = 2) (Fig 1. Participant flow diagram). Four participants (n = 2 in the control arm; n = 2 in the menstrual cup arm) did not provide specimens for BV, VMB, or STI testing subsequent to baseline and are not included in outcome analyses. After baseline, follow-up visits occurred every 6 months through 30 months, with the 24-month visit (to have taken place April through June 2020) not conducted due to the Coronavirus Disease 2019 (COVID-19) pandemic. The cancelled 24-month visit was agreed upon by the Principal Investigator and all Co-Investigators, and aligned with IRB and ethical review requirements. Missed visits are identified in Fig 1 under “loss-to-follow-up”; the number of participants with missed visit at any time point ranged from 8 to 22 in the intervention arm and from 4 to 20 in the control arm. All participants with “missed visit” were included in analyses (only those n = 4 with no measurement of outcome over follow-up were excluded, as described above). PPT PowerPoint slide
PNG larger image
TIFF original image Download: Fig 1. Participant flow diagram for the progress of clusters and individuals through the CaCHe substudy of the Cups for Cash for Girls (CCG) cluster randomized controlled trial. https://doi.org/10.1371/journal.pmed.1004258.g001
Data collection Participants self-completed a tablet-based survey in their language of choice (English or DhoLuo) to obtain sociodemographic information and to assess MHM and sexual practices. Study nurses and counsellors trained in research and survey administration provided assistance as needed. Sociodemographic data included age and detailed socioeconomic measures. Socioeconomic status (SES) was measured using questions from the health and demographic surveillance system (HDSS) household survey. The SES score was constructed using the absolute index method [14], and quintiles were dichotomized as quintiles 1 to 2 and quintiles 3 to 5. School-level WASH (water, sanitation, and hygiene) was assessed as a score ranging 0 to 3, reflecting 1 point each for direct observation of the following: any water for handwashing, soap, and acceptable girl-to-latrine ratio (0 to 30 girls per latrine) for latrines that were clean, without offensive smell, without holes in walls, and with door, roof, and stable floor [15]. A series of questions assessed whether girls were sexually active, and whether they were forced or threatened to have sex (referred to as coerced sex), or whether they were given money, items, or favors in exchange for sex (referred to as transactional sex). Although current pregnancy was an exclusion criterion in eligibility assessment, 3 girls who did not declare pregnancy at eligibility screening reported being pregnant after enrolment and were maintained in analyses (1 in control arm, 2 in cups arm).
Sample size CaCHe was designed to estimate the effect of menstrual cups on girls’ risk of BV as the primary outcome, with an anticipated cumulative event rate of 30% to 40% among controls occurring over 30 months. In a design of 6 repeated measurements having AR (1) covariance structure, correlation between observations on the same subject ranging 0.25 to 0.4, and accounting for 20% loss to follow-up, group sample sizes of 220 in cup arm and 220 in control arm would achieve >80% power to detect 25% reduced relative prevalence of BV for the cup arm compared to control arm when BV prevalence is 30%, and 97% power when prevalence is 40% (p = 0.05 two-sided test, 2 proportions in a repeated measures design; PASS v15).
Randomization and masking As detailed previously, randomization assignments were computer generated by the trial statisticians, using SAS (version 9.3), with blinded arm allocation achieved through community randomization ceremonies [13]. Participants were informed of group allocation on screening day, after which time masking the assignment was not possible. Due to the nature of the intervention, research assistants collecting specimens were not masked. Lab staff in Kenya and the United States were masked as specimens were received with only study numbers.
Specimen collection At baseline and each follow-up visit, girls were asked to take self-collected vaginal swabs. The first swab obtained was for 16S rRNA gene amplicon sequencing (microbiome) using the OMNIgene vaginal (OMR-130, DNA Genotek, Ontario, Canada). A cotton tipped swab was collected to test for BV. The third swab was collected for detection of Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) using the Cepheid GeneXpert Vaginal/Endocervical Specimen Collection Kit. The fourth swab collected was used for detection of T. vaginalis (TV) using rayon tipped swabs. Participants were instructed verbally and with visual aids on how to collect the swabs, as detailed elsewhere [16]. Nurses and research assistants prepared smears for BV immediately after collection. Swabs for amplicon sequencing, CT/NG, and TV were immediately placed on ice packs in coolers for transport to the lab in Kisumu. Upon receipt at the lab, specimens for amplicon sequencing were placed at −80°C until shipment to Chicago for processing.
Detection and treatment of bacterial vaginosis and sexually transmitted infections At baseline, swabs for CT/NG were shipped weekly for processing at the University of Nairobi Institute for Tropical and Infectious Diseases; from the 12-month visit, onward swabs for CT/NG were processed in Kisumu at Nyanza Reproductive Health Society. All swabs for TV and BV were processed in Kisumu at Nyanza Reproductive Health Society for the duration of the study. Following manufacturer protocol, vaginal swabs were tested for CT/NG using the GeneXpert (Cepheid, Sunnydale, California, United States). 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 according to Nugent’s criteria within 48 hours of receipt; a score of 7 to 10 was defined as BV [17]. For quality control, each slide was confirmed by a second technician, with discrepancies adjudicated via discussion between the 2 technicians prior to reporting. Fingerstick whole blood collected in EDTA tubes were tested for HIV according to Kenyan national guidelines. HIV–positive girls were linked to care. CT, NG, and TV were treated following Kenyan national guidelines. BV was treated with 2 g of tinidazole once daily for 2 days, for lower occurrence of gastrointestinal symptoms, likely higher adherence, and equivalent efficacy to longer duration regimens or other topical or oral antimicrobials [18–19]. All girls with Nugent score 7 to 10 were treated due to lack of correlation of reported symptoms with BV [16].
Characterization of vaginal microbiome DNA extraction and PCR amplification of microbial 16S rRNA genes employed a two-stage PCR protocol using primers 341F and 806R (V3-V4 variable region) [20]. Amplicons were sequenced on an Illumina MiSeq instrument, implementing V3 chemistry (600 cycles). DNA extraction, library preparation, and sequencing were performed by the Genome Research Core (GRC) at the University of Illinois Chicago. Forward and reverse reads were merged using the software package PEAR [21]. Quality filtered and primer trimmed sequence data were then processed using a standard bioinformatics pipeline for chimera removal, annotation, and community state types (CSTs) typing was conducted by University of Maryland Institute for Genomic Science [22]. Subsequently, a biological observation matrix (BIOM) was generated at the lowest taxonomic level identifiable. Vaginal CSTs were identified in a reference dataset using nearest centroid classification (VAginaL community state typE Nearest CentroId clAssifier (VALENCIA)) [23]. Putative contaminants were identified and removed following application of decontam program in R (version 4.1.3) [24]. There were 23 observations with <5,000 sequence reads, which were excluded from analyses of VMB. Prior to relative abundance estimation, data were filtered to retain taxa that contributed at least 0.01% of the total sequence reads, resulting in retention of 62 of 1,448 taxa. Taxa with highest mean relative abundance by CST and BV and STI status are summarized in S1 Fig. Raw sequence data (FASTQ files) were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA), under BioProject identifier PRJNA746243.
Statistical analysis The data analyst (RB) was blinded to group assignment. All statistical analyses were conducted using Stata/SE v17 software package. All analyses presented are intention to treat (ITT), according to assigned study arm. The study design and statistical analysis plan did not account for multiple hypothesis testing, and, therefore, results of analyses of secondary outcomes are reported as point estimates and 95% confidence intervals.
Primary outcome The primary outcome was BV, dichotomized as positive (Nugent score 7 to 10) versus negative (Nugent score 0 to 6). Participants in whom treatment for BV was not documented at baseline (n = 1) or subsequent visits (n = 4) were excluded from further analysis as it would be unknown whether subsequent infections represented new infections (S1 Table). Those with documented antibiotic treatment were maintained in analyses (i.e., could have multiple infections contributing to analysis; S2 Table). Generalized linear mixed models (GLMMs) were fitted to allow for the hierarchical structure of the study, with missing at random assumption for complete case data. The GLMM with binomial distribution and logit link function included treatment arm as fixed effect, and participant and cluster as random effects, to estimate odds ratios and 95% confidence intervals, with odds ratios providing an average estimate across all time points. We estimated odds ratios for BV rather than relative risks given the relatively low efficacy of antibiotic treatment of BV, with failure in up to 50% at 6 months [25]. We applied robust variance estimate due to the small number of clusters (n = 6). Baseline BV and STI status were part of the longitudinal data, and, therefore, estimations account for baseline BV and STI status in both the crude and multivariable adjusted models. Secondary outcomes were as follows: (1) VMB CST, dichotomized as CST-I (L. crispatus dominated) versus other CST (CST-II, CST-III, CST-IV, CST-V); (2) relative abundance of L. crispatus (continuous); and (3) incident STI, dichotomized as positive (composite of infection with CT, NG, and/or TV) versus negative for all three. For analysis of CST, while CST-II (L. jesenii dominated) and CST-V (Lactobacillus gasseri dominated) are generally not associated with adverse outcomes [26], they comprised a small proportion of observations and as defined a priori, the desired outcome was CST-I. Methods for analyzing CST-I followed those as described for BV. The GLMM with Gaussian distribution was applied for analysis of relative abundance of L. crispatus, including treatment arm as fixed effect, and participant and cluster as random effects, with 95% confidence intervals estimated via robust variance estimate. We estimated relative risk of STI given the high efficacy of antibiotic treatment for CT and NG [27–28]. The GLMM with Poisson distribution was applied for analysis of STI (as log binomial did not converge), following methods specified as above. Participants with documented antibiotic treatment for BV or STI detected at baseline or follow-up visits could contribute multiple infections (S1 and S2 Tables); one participant for whom baseline STI treatment was not documented was excluded from incidence analyses.
Adjusted analyses and subgroup analyses Adjusted ITT analyses were performed on the primary and secondary outcomes. Baseline a priori confounders were specified as follows: school WASH score, HIV status, age, ever sexually active, and SES. At baseline, few participants were HIV infected (n = 7), and due to sparsity, we did not adjust for this. ITT subgroup analyses were performed for the primary outcome (BV). The a priori specified subgroup variables were baseline values of the following: age, ever sexually active, SES, WASH score, coerced sex, transactional sex, and STI status. Subgroup analyses are presented visually as coefficient plots with 95% confidence intervals (95% CI).