Impact of preconception vaginal microbiota on women’s risk of spontaneous preterm birth: protocol for a prospective case-cohort study

Abstract

Introduction

Bacterial vaginosis (BV) and vaginal microbiota disruption during pregnancy are associated with increased risk of spontaneous preterm birth (SPTB), but clinical trials of BV treatment during pregnancy have shown little or no benefit. An alternative hypothesis is that vaginal bacteria present around conception may lead to SPTB by compromising the protective effects of cervical mucus, colonising the endometrial surface before fetal membrane development, and causing low-level inflammation in the decidua, placenta and fetal membranes. This protocol describes a prospective case-cohort study addressing this hypothesis.

Methods and analysis

HIV-seronegative Kenyan women with fertility intent are followed from preconception through pregnancy, delivery and early postpartum. Participants provide monthly vaginal specimens during the preconception period for vaginal microbiota assessment. Estimated date of delivery is determined by last menstrual period and first trimester obstetrical ultrasound. After delivery, a swab is collected from between the fetal membranes. Placenta and umbilical cord samples are collected for histopathology. Broad-range 16S rRNA gene PCR and deep sequencing of preconception vaginal specimens will assess species richness and diversity in women with SPTB versus term delivery. Concentrations of key bacterial species will be compared using quantitative PCR (qPCR). Taxon-directed qPCR will also be used to quantify bacteria from fetal membrane samples and evaluate the association between bacterial concentrations and histopathological evidence of inflammation in the fetal membranes, placenta and umbilical cord.

Ethics and dissemination

This study was approved by ethics committees at Kenyatta National Hospital and the University of Washington. Results will be disseminated to clinicians at study sites and partner institutions, presented at conferences and published in peer-reviewed journals. The findings of this study could shift the paradigm for thinking about the mechanisms linking vaginal microbiota and prematurity by focusing attention on the preconception vaginal microbiota as a mediator of SPTB.

Strengths and limitations of this study.

• This prospective case-cohort study enrols Kenyan women with fertility intent, enabling follow-up and exposure measurement from preconception through pregnancy and the early postpartum period.

• Monthly specimen collection during the preconception period allows for examination of the vaginal microbiota close to the time of conception.

• Fetal membrane swabs and placental samples are collected at delivery, enabling assessment of the association between detection and concentrations of bacteria in the fetal membranes, inflammation and preterm birth.

• A combination of broad-range 16S rRNA gene PCR with next-generation sequencing and quantitative PCR assays provide both the relative and absolute quantities of bacteria in vaginal secretions and in the fetal membranes.

• Because this is an observational study, it cannot definitively establish a causal relationship between vaginal bacteria and preterm birth.

Introduction

Globally, approximately 10% of births are preterm (<37 weeks of gestation), but the prevalence can be as high as 18% in low-resource countries.¹ Preterm birth and its sequelae are the leading causes of death among children under 5.² The majority of preterm deliveries are spontaneous preterm births (SPTB).³ The cause of SPTB is often unknown, but up to 40% may be associated with intrauterine infections.^{3 4} Other risk factors include other infections (urinary tract, sexually transmitted and systemic), sociodemographic characteristics (age, race and education),³ extremes of body mass index,⁵ periodontal disease,⁶ interpregnancy interval <6 months, prior SPTB, stress, depression and smoking.³ Further elucidation of the causes may provide insight into novel approaches for reducing SPTB.

Bacterial vaginosis (BV) is a common vaginal condition characterised by a shift from an optimal Lactobacillus-predominant vaginal microbiota to one characterised by high concentrations of diverse anaerobic species.⁷ Numerous studies indicate that BV during pregnancy is associated with increased risk of SPTB.⁸ However, a meta-analysis of clinical trials concluded that while antibiotic treatment prior to 20 weeks gestation successfully treated BV, the risk of SPTB was not substantially reduced.⁹

Molecular microbiology has transformed the understanding of vaginal microbiota to a broader spectrum of phenotypes ranging from low-diversity Lactobacillus-dominated bacterial communities to a heterogeneous group of high-diversity BV-associated communities.^{10 11} Studies exploring the relationship between the vaginal microbiota and SPTB have yielded conflicting findings. Some found significant associations between increased species diversity and preterm birth,^12–16 while others have not.^17–19 In addition, some studies have found that women with a low relative abundance of Lactobacillus species and higher relative abundance of BV-related taxa may be at higher risk of preterm birth.^{16 19–22} Others found no association between vaginal bacterial community type and preterm birth.^{12 14 17} Detection and higher concentrations of specific vaginal bacteria have also been associated with preterm birth including Leptotrichia/Sneathia,^{16 23} BV-associated bacterium 1 (BVAB1),^{16 23} Megasphaera,²³ Gardnerella vaginalis,^{24 25} Atopobium vaginae,²⁴ TM7-H1¹⁶ and some Prevotella species.¹⁶

A novel hypothesis to explain the lack of success of BV treatment during pregnancy on SPTB risk is that preconception vaginal microbiota may be a more important risk factor for SPTB than vaginal bacteria during pregnancy. Vaginal bacteria present around the time of conception could compromise the protective effects of cervical mucus,²⁶ gaining access to the endometrium before embryo implantation, formation of the cervical mucus plug and development of the fetal membrane. These bacteria could colonise and cause chronic, low-level inflammation in the decidua, placenta, fetal membranes or amniotic cavity.^{27 28} In this scenario, antibiotic treatment during pregnancy may be too late to influence a woman’s risk of SPTB associated with intrauterine bacteria.

The Microbiota and Preterm Birth Study (MPTB) enrols HIV-negative Kenyan women trying to become pregnant into a prospective case-cohort study to address the following aims:

1. Compare the species diversity and richness of the vaginal microbiota sampled close to the time of conception in women with SPTB versus term delivery using broad-range 16S rRNA gene PCR and next-generation sequencing.

2. Compare the presence and concentrations of select bacterial genera/species (based on published data²³ and results of Aim 1) using targeted quantitative PCR (qPCR) assays in vaginal specimens sampled close to the time of conception in women with SPTB versus term births.

3. Perform species-specific qPCR assays on samples collected from fetal membranes and histological examination of membranes and umbilical cord to determine if vaginal bacteria ascend to the upper genital tract and cause inflammation in the fetal membranes and umbilical cord.

Methods and analysis

Study design, setting and timeline

The MPTB Study is a prospective case-cohort study. Eligible women enrol prior to conception and are followed through preconception, pregnancy, delivery, and early postpartum. Participants will be selected into the case-cohort sample as detailed in the Statistical Analysis section. Study sites are located at Kenyatta National Hospital (KNH) in Nairobi and at Ganjoni Health Center and Coast Provincial General Hospital (CPGH) in Mombasa.

Enrolment began in Nairobi in April 2017 and Mombasa in April 2018. Preconception enrolment of participants will continue through approximately June 2019 with the last study deliveries occurring in 2021, approximately 18 months after the last enrolment.

Eligibility criteria and recruitment

The target population is HIV-negative women who are currently planning to become pregnant. Additional eligibility criteria include being ≤45 years old, having a menstrual period in the prior 3 months or recently discontinued contraceptive methods that induce amenorrhoea (eg, implant, hormonal intrauterine device), willing to comply with study procedures, and able to provide informed consent. Minors aged 14–17 are eligible if emancipated under Kenyan law. Exclusion criteria include current pregnancy, continuing contraception other than condoms for HIV/sexually transmitted infection (STI) prevention, having a depot medroxyprogesterone (DMPA) acetate injection in the last 3 months, history of cervical or uterine surgery (other than caesarean section), known autoimmune disease, antibiotic use in the prior 4 weeks, and history of infertility care-seeking. For women with known HIV-positive male partners, their partner must have an undetectable HIV viral load, or the participant must be taking pre-exposure prophylaxis.

Participants are recruited by study staff or referred by healthcare providers at sites providing reproductive or maternal health services. Women discontinuing contraception for the purpose of becoming pregnant are a key recruitment population.

Study visits and procedures

Study participation occurs in six phases including screening/enrolment, periodontal examination, preconception, pregnancy, delivery and postpartum (figure 1). Specimen collection, laboratory testing and other clinical procedures are summarised in table 1.

Figure 1.

Microbiota and Preterm Birth Study Phases—additional participant pathways: (1) Participants who discontinued DMPA injectable contraception within 6 months of study enrolment are eligible for 9 months of preconception follow-up. (2) Participants who miscarry are eligible to re-enter preconception follow-up for a second pregnancy attempt. DMPA, depot medroxyprogesterone.

Table 1.

Specimen collection, laboratory testing and other procedures by study visit

Procedure type	Screening	Enrolment	Periodontal examination	Preconception	9–12 weeks gestation	Delivery*	Postpartum
Specimen collection
Urine pregnancy test	X	X		X
Genital swabs
APTIMA swab		X
Push-off swabs†		X		X	X
Cotton and Dacron swabs		X		X	X
Sialidase test swab		X		X
Fetal membrane swab						X
Placental punch						X
Fetal membrane roll						X
Umbilical cord						X
Laboratory testing
HIV rapid test	X
Neisseria gonorrhoeae, Chlamydia trachomatis and Trichomonas vaginalis		X
Cervical Gram stain		X
Vaginal pH		X
Vaginal wet mount		X
Vaginal Gram stain		X		X	X
Lactobacillus culture‡		X		X	X
Sialidase test		X		X
PSA test		X		X	X
Participant Questionnaires
Eligibility screen	X
Demographics		X
Reproductive history		X
Medical history and current symptoms		X		X	X
Sexual behaviour		X		X	X
Substance use		X		X	X
Dental health history			X
Neonatal heath							X
Obstetric ultrasound					X

*Delivery follows standard obstetrical procedure and is not conducted by study staff.

†Storing for vaginal microbiota and vaginal inflammatory response testing.

‡Rogosa agar, followed by subculture for hydrogen peroxide production on tetramethylbenzidine agar containing horseradish peroxidase.

PSA, prostate specific antigen.

Screening and enrolment

Following written informed consent for screening, study staff perform urine pregnancy testing and rapid HIV testing according to Kenyan guidelines,²⁹ and conduct an eligibility interview. Eligible women can enrol immediately on consent. If enrolment occurs at a later date, pregnancy testing is repeated to reconfirm eligibility.

Enrollees complete a structured face-to-face interview regarding demographics, sexual behaviour, substance use, depression symptoms (Patient Health Questionnaire-9), and reproductive and medical history. A study clinician performs a physical examination and speculum-assisted pelvic examination. If a woman is menstruating, the examination is deferred to avoid sampling during this interval when the vaginal microbiota undergoes rapid changes.^{30 31} Two vaginal fluid specimens are collected by rolling push-off Dacron swabs (FitzCo) three rotations against the lateral vaginal wall; these are stored for vaginal microbiota and inflammatory response evaluation. Additional genital specimens are collected for STI diagnosis (Neisseria gonorrhoeae, Chlamydia trachomatis and Trichomonas vaginalis by nucleic acid amplification testing (NAAT)), vaginal and cervical Gram stains, detection of prostate-specific antigen (PSA) and elevated sialidase using a point-of-care diagnostic test for BV (Diagnosit BVBLUE, Gryphus Diagnostics). A vaginal specimen is inoculated onto Rogosa agar for detection of Lactobacillus. Syndromic management is provided as indicated for genital syndromes including vaginal discharge.³² Additional therapy is provided at the first preconception visit based on STI NAAT results.

After the examination, women receive counselling on healthy behaviours during preconception and pregnancy, including smoking cessation, refraining from vaginal washing and a healthy diet. Study clinicians discuss participants’ menstrual cycle and identify the probable fertile window using calendar-based methods. Ovulation is estimated to occur 14 days prior to the first day of the next predicted menses, with the most fertile days emphasised as the 5 days before and day of ovulation.³³ All participants receive prenatal vitamins.

Periodontal examination

Periodontitis has been associated with SPTB⁶ so participants undergo a periodontal examination. Since pregnancy increases gingival inflammation,³⁴ examinations occur within 4 weeks of enrolment. The oral examination includes assessment for periodontitis using periodontal pocket depth and clinical attachment measurements, and the Decay-Missing-Filled Index and Gingival Index assessments.³⁵ Presence and severity of periodontitis are defined using the 2007 Centers for Disease Control and Prevention/American Academy of Periodontology case definitions.³⁶ Participants also complete a modified version of WHO’s oral health questionnaire.³⁵

Monthly preconception visits

Participants return for study visits at 1-month intervals while trying to become pregnant. Retention staff call participants in advance of each appointment. A structured interview is conducted to update sexual behaviour and medical history. Women self-collect vaginal swabs for microbiota and inflammatory response analysis, vaginal Gram stain, PSA detection, Lactobacillus culture and sialidase detection. Participants with genital symptoms receive syndromic management.³²

A urine pregnancy test is performed, and participants with a positive test are scheduled for a first trimester visit between 9 and 12 weeks of gestation. Most women who remain non-pregnant after 6 months exit the study; those who discontinued DMPA less than 6 months prior to enrolment are eligible for 9 months of preconception ‘trying time’ due to delayed return to fertility after DMPA discontinuation.^{37 38}

First trimester visit (9–12 weeks of gestation)

At the first trimester visit, a structured interview is conducted to update information on sexual behaviour and medical history. Women self-collect vaginal swabs for microbiota and inflammatory response analysis, vaginal Gram stain, PSA detection and Lactobacillus culture. Women then undergo an obstetrical ultrasound to confirm gestational age. The ultrasound is conducted by sonographers at KNH in Nairobi and at a private radiology facility in Mombasa (training and quality control described in table 2). The American College of Obstetricians and Gynecologists’s 2014 guidelines are used to estimate gestational age if the ultrasound derived estimate differs from that calculated using the last menstrual period by >7 days before 16 weeks.³⁹ If a later ultrasound is obtained, the sonographic dates are used if they differ from last menstrual period dates by >10 days between 16 and 22 weeks, >14 days between 22 and 28 weeks, and >21 days after 28 weeks.

Table 2.

Periodontal examination, first trimester obstetric ultrasound, and fetal membrane sample training and continuous quality control procedures

Procedure	Initial training	Continuous quality control procedures
Periodontal examination	The UoN periodontist conducted an in-person training with the CPGH dentist that included teaching, observation and calibration.	Interexaminer calibration biannually—The UoN periodontist and a second UoN experienced periodontist conduct periodontal exams on a series of patients to compare results. The UoN periodontist serves as the gold standard comparator for the CPGH dentist. Intraexaminer calibration once per 50 examinations—Examination is repeated on one participant to compare results between the first and second examination.
First trimester obstetric ultrasound	All sonographers received a standardised refresher training on ultrasound determination of gestational age, early pregnancy complications and MPTB study specific procedures by the Kenyan study radiologist.	For each participant with a first trimester ultrasound, an image of the crown rump length is saved and quality is assessed by the study radiologist according to criteria developed for the International Fetal and Newborn Growth Consortium for the 21st century.59 (For participants with a crown rump length measurement of >84 mm, gestational age is assessed using biometric measurements.) Sonographers access the results of quality assessments for ongoing feedback. Annually, sonographers in Nairobi meet with the study radiologist to review quality metrics and receive refresher training on study procedures. The radiologist provides ongoing one-on-one support to the Mombasa sonographers by phone.
Fetal membrane sample	Initial training of labour and delivery nurses and KNH-based Kenyan obstetrician was performed by SL, who developed the fetal membrane collection technique.41 Trainees were required to demonstrate proficient sample collection technique and verbalise study procedures.	Subsequent trainings are conducted as needed and led by a Kenyan obstetrician-gynaecologist based at KNH who participated in the initial training.

CPGH, Coast Provincial General Hospital; KNH, Kenyatta National Hospital; MPTB, Microbiota and Preterm Birth Study; UoN, University of Nairobi, Nairobi.

Pregnant participants are referred to routine antenatal care (ANC) and enrolled into a short message service (SMS) programme to support retention (see the Retention during pregnancy section). Routinely collected antenatal data (eg, syphilis test results, blood pressure) are abstracted from participants’ ANC facility records and the Mother and Child Health Booklet provided to all pregnant Kenyan women.

If a participant suffers a miscarriage, management of the pregnancy loss is conducted by non-study clinicians according to standard of care. Women who miscarry at <20 weeks of gestation may remain in the study for six additional monthly preconception visits if they would like to try for another pregnancy.

Retention during pregnancy

To improve retention, participants are offered enrolment into an adaptation of a two-way SMS programme that was initially designed to support HIV-positive Kenyan women during pregnancy and postpartum.⁴⁰ Messages are sent at 16, 20 and 24 weeks gestation, biweekly beginning at 28 weeks, and weekly from 38 weeks through 6 weeks postpartum. In addition, study nurses call participants weekly starting at week 35 to confirm pregnancy status and planned delivery location. Participants are instructed to call at the onset of labour so study staff can coordinate collection of delivery samples for deliveries at KNH or CPGH, and to assist with identification of the delivery date for births occurring elsewhere.

Delivery procedures

Deliveries follow standard obstetrical procedures and are not conducted by study staff. Swabs from between the fetal membranes and placental samples are collected for births occurring at KNH and CPGH by trained labour and delivery nurses (training and quality control described in table 2). On vaginal or caesarean delivery of the placenta, it is placed in a sterile container. Samples are collected as soon as feasible and within 2 hours of delivery. Using sterile technique, a pair of nurses collects samples from between the amnion and chorion.⁴¹ A sterile push-off swab (FitzCo) is collected from between the membranes, placed in a cryovial and temporarily stored in a −4°C (Nairobi) or −20°C (Mombasa) freezer. The placenta is placed in 10% neutral-buffered formalin. Laboratory staff transport the fetal membrane swab to a −80°C freezer in study laboratories at KNH and CPGH within one working day. Laboratory staff also collect placental samples for histopathology. These include a 4 cm fetal membrane roll from the ruptured edge of the membranes to the edge of the placental disc, a 1 cm section of umbilical cord starting 3 cm from the placental disk, and a 1 cm block of placenta with overlying membranes adjacent to the site of cord insertion. All pathological specimens are stored in 10% neutral-buffered formalin.

Study staff abstract data from delivery records, including type of delivery (ie, vaginal or caesarean; laboured or did not labour; spontaneous or induced labour), complications, live birth or stillbirth, and birth weight.

Postpartum visit

At the 6-week postpartum visit, clinicians abstract any ANC and delivery details not previously captured from the participant’s delivery discharge report and Mother and Child Health Booklet. Participants complete an interview about the infant’s heath and immunisation status.

Incentives

Participants receive KSH300 (about US$3.00) at each study visit. This amount is consistent with other studies in Kenya and is provided to cover transportation costs. Women who become pregnant receive a free obstetrical ultrasound. An incentive of KSH1000 is provided for deliveries at KNH or CPGH.

Laboratory methods

Microscopy, Lactobacillus culture, STI testing and detection of sialidase and PSA in vaginal fluids

Vaginal Gram stained slides are evaluated for BV using the criteria developed by Nugent and Hiller.⁴² Saline and potassium hydroxide wet mounts are examined for the presence of motile trichomonads, clue cells, yeast and sperm. Endocervical Gram stained slides are scanned at low power, and polymorphonuclear leucocytes in three nonadjacent oil immersion fields are counted and averaged to evaluate cervical inflammation. Clinicians inoculate vaginal specimens directly on Rogosa agar for detection of cultivable Lactobacillus and store the plate in a candle jar until transportation to the laboratory within 4 hours.⁴³ Hydrogen peroxide production is evaluated by subculture of Lactobacillus isolates on tetramethylbenzidine agar containing horseradish peroxidase.⁴⁴ A vaginal specimen is tested for N. gonorrhoeae, C. trachomatis and T. vaginalis by NAAT (Aptima Combo-2 CT/NG Detection System, Aptima Trichomonas vaginalis assay; Hologic). One vaginal swab is used for detection of sialidase using a commercially available BV diagnostic test (Diagnosit BVBLUE; Gryphus Diagnostics). Testing for PSA in vaginal samples is performed with a commercially available assay (ABAcard, Abacus Diagnostics), which can detect semen for 24–48 hours after condomless sex.⁴⁵

Molecular methods for identification of bacteria in vaginal and fetal membrane samples

Stored vaginal and fetal membrane swabs for bacterial PCR will be transported on dry ice to the Fredricks Laboratory at the Fred Hutchinson Cancer Research Center in Seattle, Washington, USA. Qiagen QIAamp BiOstic Bacteremia DNA Isolation Kits (Qiagen, Germantown, Maryland, USA) will be used to extract DNA from vaginal swabs. This protocol uses bead beating and chaotropic lysis to break apart bacterial cells and recover DNA that is free of PCR inhibitors. Swabs that have not contacted a human surface will be processed in parallel to serve as sham DNA extraction (negative) controls. Extracted DNA will be subjected to broad-range 16S rRNA gene PCR using primers that anneal with highly conserved regions of the small subunit rRNA gene, amplifying a 470 base pair segment that contains a highly variable sequence useful for species identification. Bar coded primers will be used to multiplex samples.⁴⁶ Libraries of 16S rRNA gene amplicons will be mixed for sequencing on the Illumina MiSeq platform using 300 bp paired-end reads. The assembled reads will be binned into individual study samples using the nucleic acid bar codes. Approximately 10 000–30 000 sequence reads will be generated per sample, providing robust detection of minority species. Raw sequence reads will be demultiplexed using Illumina’s on-board bcl2fastq conversion software V.1.8.4 with zero mismatches in either forward or reverse indices. The DADA2 software package will be used to quality control, filter, pair and cluster the amplicon sequence reads.⁴⁷ Sequence variants will be assigned taxonomy using the phylogenetic placement tool pplacer⁴⁸ and a custom vaginal reference set.¹¹ Sham extraction controls will also be processed in parallel to exclude bacterial contamination of reagents. Bacterium-specific qPCR assays will also be performed. This study focuses on species hypothesised to be associated with increased risk of SPTB, such as BVAB1, Megasphaera, and Sneathia. The final set of species/genera tested will be fine-tuned following analysis of the deep sequencing data from Aim 1, selecting additional bacteria based on a comparison of the relative abundance of individual taxa in women with and without SPTB. A standard exogenous jellyfish qPCR amplification control will be used to assess for PCR inhibitors.⁴⁹ A broad-range bacterial 16S rRNA gene qPCR assay will be used to measure total bacterial load in each sample.

Histopathological examination of fetal membranes, placenta, and umbilical cord samples

H&E staining of fetal membrane rolls, placental samples, and umbilical cord sections will be graded according to published guidelines by an experienced pathologist (KM).⁵⁰ Acute and subacute inflammatory lesions will be graded and staged for both maternal and fetal components. Chronic inflammatory lesions will be characterised including any observation of chronic deciduitis or the presence of decidual plasma cells.

Statistical analysis

Sample size estimate and case-cohort population generating

Aim 2 requires the largest sample size and was used to guide sample size estimation for this case cohort study.⁵¹ The prospective case-cohort analysis set will include three women who delivered at term for each one with SPTB. Assuming three primary species/genera of interest, Simes’ methodology was used to fix a type-1 error rate adjusted for three tests.⁵² A sample of 80 SPTB and 240 term births has >80% power to detect a statistically significant ≥2.8 fold difference in the odds of detecting a preconception vaginal bacterial species/genus in women with SPTB versus term delivery, assuming >10% prevalence of the organism at preconception visits for term deliveries.

To accrue 80 SPTB cases including women with spontaneous preterm labour or preterm premature rupture of membranes, a cohort of approximately 1100 women will be enrolled. Once 80 SPTB cases occur, the sample for the case-cohort will be defined. First, cases of spontaneous abortion (<20 weeks gestation) and preterm births without spontaneous labour or preterm premature rupture of membranes will be excluded. Next, a random sample of the remaining pregnancies will be selected such that when added to the remaining SPTB cases, the case-cohort sample will have a term birth to SPTB ratio of 3:1. A sampling fraction f, with f solved using the formula: nf + (1 f)*80=320, where n is the total number of women (cases and non-cases) with delivery data, will be used to select the random sample. Lastly, all of the remaining SPTB cases will be added to the random sample, creating the full case-cohort sample.

Statistical analysis plan

The goal of aim 1 is to characterise and compare preconception vaginal species diversity and richness between women with SPTB vs term birth. All women with SPTB and a random sample of the same number of women with term birth from the case-cohort sample will be included. To describe the overall frequency and relative abundance of species, cumulative rank abundance plots will be generated for each group.⁵³ We will compare the cumulative distribution of preconception vaginal bacterial taxa between women with SPTB versus term birth using the Kolmogorov-Smirnov test. Rarefaction curves will be used to evaluate species richness (number of taxa at a 97% sequence similarity cut-off defining an operational taxonomic unit) in women with SPTB vs term birth. Finally, we will assess species diversity (Shannon Diversity Index⁵⁴) and species richness (Chao1 richness estimator⁵⁵) by comparing the mean values between women with SPTB versus term birth. We will perform logistic regression with SPTB status as the outcome and species rank abundance percentage for each species. We will first determine the score statistics (SPTB status as the outcome) for each variable, then rank the variables from largest to smallest score statistic. Next, we will perform the logistic modelling on each of these in rank order of score statistic until reaching a p value of 0.2 in univariate logistic regressions. These data will be examined to refine targets for the primary hypothesis test in aim 2.

For aim 2, based on the literature and informed by aim 1 results, bacterial species/genera will be selected for evaluation using qPCR assays, comparing their presence and concentrations in vaginal specimens sampled before conception. These analyses will use data from the full case-cohort sample and will be weighted to account for the sampling scheme. For each bacterial taxon, we will perform unadjusted logistic regression to examine the association between the presence of that taxon and the risk of SPTB. Multivariable logistic regression analysis will be used to determine the independent contributions of bacterial species to the risk of SPTB. Species associated with SPTB in univariate analyses (p<0.10) will be included in the multivariable model after addressing collinearity. A manual forward stepwise model building approach will be used to address confounding. Potential confounding factors are listed in table 3. For the three species selected for the primary hypothesis test, p<0.033 will be considered statistically significant, using the Simes’ correction for multiple comparisons. Other bacterial taxa may be evaluated as exploratory analyses. Additional exploratory analyses will be performed to investigate the relationship between preconception quantities of specific bacteria and risk of SPTB.

Table 3.

Potential key covariates in analysis of the association between preconception vaginal microbiota and spontaneous preterm birth

Variable	Timing	Method
Demographics
Maternal age	Baseline	FTFI
Educational level	Baseline	FTFI
Socioeconomic status	Baseline	FTFI
Obstetrical history (eg, no of pregnancies, time since last pregnancy, prior preterm birth)	Baseline	FTFI
Prior contraceptive history	Baseline	FTFI
Current pregnancy
Maternal haemoglobin level, blood pressure	Throughout pregnancy	ANC records
Fetal health (eg, Apgar scores, spontaneous breathing, birth weight, gender)	Delivery	Delivery records, FTFI or phone interview
Maternal infection	Throughout pregnancy	ANC and delivery records, FTFI or phone interview
Maternal pregnancy complications	Throughout pregnancy	ANC and delivery records, FTFI or phone interview
General health status
Body mass index	Baseline	Measure weight and height
Smoking	Baseline and follow-up	FTFI
Alcohol use	Baseline and follow-up	Alcohol use disorders identification test (AUDIT)
Depressive symptoms	Baseline	Patient Health Questionnaire-9 (PHQ-9)
Periodontitis	Within 4 weeks of baseline	Oral examination
Genital tract infection/STI	Baseline	Detection of gonorrhoea, chlamydia and Trichomonas vaginalis by NAAT (Hologic). Wet prep, vaginal and cervical Gram stain.
Lactobacilli	Baseline and follow-up	Culture on Rogosa and TMB agar
Unprotected sex	Baseline and follow-up	FTFI
PSA	Baseline and follow-up	ABACard PSA detection (Abacus Diagnostics)
Intravaginal practices	Baseline and follow-up	FTFI

ANC, antenatal clinic; FTFI, face-to-face interview; NAAT, nucleic acid amplification test; PSA, prostate specific antigen; STI, sexually transmitted infections; TMB, tetramethylbenzidine.

Aim 3 explores the association between bacteria detected in vaginal preconception samples and fetal membrane samples to further support the role of ascending vaginal bacteria in SPTB. This analysis will also evaluate the association between bacteria identified in the fetal membranes and histological evidence of deciduitis, chorioamnionitis and funisitis. The analysis will use data from women in the case-cohort sample with delivery samples. The exposures are preconception detection of each of the three vaginal species tested for the primary hypothesis in aim 2. Outcomes include detection of the same bacterial species in fetal membranes, deciduitis, chorioamnionitis and funisitis. ORs will be estimated for each exposure using unadjusted logistic regression models, and multivariable logistic regression analyses will be used to adjust for potential confounding factors (table 3).

Patient and public involvement

Neither patients nor the public were involved in the design of this research. Healthcare providers providing reproductive/maternal health services and study participants can refer potentially eligible women to the study.

Ethics and dissemination

There are inherent risks to women and fetuses during pregnancy and delivery, but participation in this observational study does not increase these risks. Study participation risks include discomfort associated with the fingerstick for HIV testing, sensitive questions about sexual behaviour, and the pelvic examination; stress associated with HIV or STI diagnosis; and breach of confidentiality. To minimise these risks, experienced Kenyan research staff fluent in both English and Kiswahili conduct a robust consent process. At screening, the screening procedures are explained and written informed consent is obtained with forms available in English and Kiswahili. Eligible women who opt to participate in the study undergo a separate consenting process for study enrolment, including verbal explanation of the procedures and written informed consent. Participants are counselled on potential risks and are informed that they can withdraw from the study at any time. The risk of breach of confidentiality is minimised by following Good Clinical Practice procedures for data security. Signed consents are secured in locked cabinets in a restricted area. Study identification numbers are assigned to deidentify hard copies of the case report forms, which are locked in a restricted area. Data are entered into a password-protected database using encrypted computers. Results will be disseminated to clinicians at study sites and partner institutions, presented at conferences and published in peer-reviewed journals.

Discussion

Most cases of SPTB occur without a known cause.³ BV during pregnancy is associated with increased risk of SPTB, but clinical trials of BV treatment in pregnancy have shown minimal or no reduction in SPTB.⁹ The MPTB Study addresses the hypothesis that the vaginal microbiota detected during the preconception period may be linked to SPTB. The findings could shift the paradigm for thinking about the mechanisms linking vaginal microbiota and preterm birth, and could be used to guide the development and evaluation of interventions aimed at lowering the risk of SPTB by identifying and eradicating high-risk vaginal bacteria prior to conception.^56–58