Evaluating Chlamydia trachomatis and Neisseria gonorrhoeae screening and treatment among asymptomatic pregnant women to prevent preterm birth and low birthweight in Gaborone, Botswana: A secondary analysis from a non‐randomised, cluster‐controlled trial

To evaluate the impact of screening and treating asymptomatic pregnant women for Chlamydia (C.) trachomatis and Neisseria (N.) gonorrhoeae infections on the frequency of preterm birth or low birthweight infants in Botswana.


| I N TRODUC T ION
According to a United Nations report from 2021, 5 million children died before their fifth birthday and 56% of these deaths occurred in sub-Saharan Africa. 1 The leading cause of death before 5 years was preterm birth complications. 2 Low birthweight, primarily resulting from preterm birth or intrauterine growth restriction, is also associated with increased risk for neonatal death and morbidity. 3Many causes of preterm birth and low birthweight are preventable.
Common, curable sexually transmitted infections (STI) such as Chlamydia trachomatis and Neisseria gonorrhoeae are associated with preterm birth or low birthweight. 4,5A meta-analysis of 12 studies found that antenatal C. trachomatis infection increased the relative risk for preterm labour by 40%. 6A systematic review and meta-analysis found that N. gonorrhoeae infection increased the odds of preterm birth by 55% and low birthweight by 66%. 7lthough antenatal C. trachomatis and N. gonorrhoeae infections are associated with preterm birth or low birthweight, research is mixed on whether aetiological screening and treatment can reduce adverse outcomes. 8The majority of low-and middle-income countries do not screen pregnant women for those infections and continue to use the World Health Organization-endorsed syndromic approach for managing urogenital symptoms (e.g.vaginal discharge) caused by C. trachomatis and N. gonorrhoeae infections. 9Syndromic management allows for same-day treatment of symptoms; however, by only addressing symptomatic infections, asymptomatic infections are untreated. 10Hence, research is needed to evaluate whether routine screening and treatment of asymptomatic C. trachomatis and N. gonorrhoeae infections during pregnancy can reduce adverse birth outcomes.
This study evaluated the impact of C. trachomatis and N. gonorrhoeae screening and treatment on preterm birth or low birthweight outcomes compared with the standardof-care.Research was conducted among pregnant women living in Gaborone, Botswana who were asymptomatic for C. trachomatis and/or N. gonorrhoeae infections at their first antenatal care visit.

| Study population
The study population included pregnant women who participated in the parent 'Maduo study', which was a cluster-controlled, cross-over trial conducted in four antenatal care clinics in Gaborone, Botswana.The primary aim of the parent study was to evaluate the impact of asymptomatic screening on post-delivery prevalence and the vertical transmission of C. trachomatis or N. gonorrhoeae infections.Details of the Madua study have been previously described. 11In brief, starting in February 2021, consenting pregnant women in both arms were enrolled if they met the following eligibility criteria: aged 15 years or older, attending their first antenatal care visit, 27 weeks of gestation or less, not treated for an STI in the past 30 days and without signs or symptoms of an STI (most commonly vaginal discharge) and therefore not in need of syndromic management (i.e.standard-of-care).
Initially two clinics participated in the study, one intervention and one standard-of-care clinic (February 2021 to mid-September 2021).As previously described in the protocol for the parent study, 11 clinics were similar in terms of patient volume and services offered; however, one clinic had space readily available for the screening intervention and was designated the initial intervention clinic.However, COVID-19-related recruitment challenges resulted in expansion to an additional two clinics, assigned to the intervention arm, in mid-September 2021.In January 2022, the standardof-care clinic crossed-over and received the intervention and two intervention clinics crossed-over and received the standard-of-care.
Our primary analysis sample included participants with non-missing values for each of our key outcomes (low birthweight, preterm birth), and non-missing values for covariates.We excluded participants who experienced a miscarriage or stillbirth.We also restricted the sample to participants with singleton pregnancies, as non-singleton births are more likely to have lower birthweight and shorter gestation.

| Intervention
Participants in the intervention clinics received C. trachomatis and N. gonorrhoeae screening (GeneXpert® CT/NG assay, Cepheid, Sunnyvale, CA, USA) at the first antenatal care visit, the third-trimester visit (approximately 36-38 weeks of gestation), and at the postnatal care visit, 6-8 weeks postdelivery.Vaginal samples were self-collected by participants and then transported to a central site for same-day processing according to the manufacturer's instructions.Women were provided with results in-person or by telephone within 24 hours.Participants testing positive for C. trachomatis and/ or N. gonorrhoeae were treated by a study nurse according to the US Centers for Disease Control and Prevention 2021 STI Treatment guidelines, 12 and were counselled on partner notification.Participants in the standard-of-care arm were tested only at their postnatal visit.If results were positive, they received similar case and partner management.
As reported previously, the parent study enrolled 500 participants (251 in the intervention arm and 249 in the standard-of-care arm).The prevalence of C. trachomatis and N. gonorrhoeae infections among the intervention group at enrolment was 23% (58/251), and 95% (57/58) who screened positive were treated.Among those treated, 91% (52/57) returned for a test-of-cure and all were cured. 13At the postnatal visit, the prevalence of C. trachomatis and N. gonorrhoeae infections in the intervention group was reduced to 1% (2/206).The postnatal prevalence among the standard-ofcare group was 13% (24/187).

| Data collection
At the first antenatal care visit, participants responded to an interviewer-administered questionnaire that collected data on sociodemographic characteristics, medical history, partner characteristics, condom use, HIV status and treatment, previous STI diagnoses and treatment, and alcohol use and smoking during pregnancy.Data were also abstracted from the participant's hand-held obstetric record, including obstetric and birth history, rapid HIV test results, other laboratory results (e.g.syphilis), weight and blood pressure.In Botswana, clinicians record obstetric and medical data, laboratory results, treatments and delivery information in the patient-held record.
At the third-trimester visit, participants were asked about new sex partners, STI symptoms, treatment and diagnoses; severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection testing and results; and alcohol and tobacco use.The patient-held obstetric record was also reviewed to confirm new diagnoses and record blood pressure and laboratory results.At the postnatal care visit, information was collected on the number of antenatal care visits, new STI symptoms and treatment, new sex partners, new test results; delivery information, including location, delivery type (spontaneous vaginal delivery, breech vaginal delivery, assisted vaginal delivery, elective caesarean section, emergency caesarean section), and singleton versus multiple births were also recorded.Neonatal information was collected, including gestational age at delivery, birthweight, length at birth, head circumference, and sex.All study data were entered directly into a Health Insurance Portability and Accountability Act (HIPAA)compliant REDCap electronic database hosted at the Botswana Harvard AIDS Institute Partnership (Research Electronic Data Capture, Vanderbilt University, Nashville, TN, USA). 14

| Outcomes
The primary outcome of interest for this secondary analysis was a composite measure of preterm birth or low birthweight.Preterm birth was defined as <37 weeks, 0 days gestation.Low birthweight was defined as <2500 g.Gestational age at delivery and birthweight were abstracted from patient-held obstetric records recorded by midwives.Gestational age was based on the estimated date of delivery that was calculated at the first antenatal care visit using the reported last menstrual period and confirmed by ultrasound when available.If the last menstrual period date was unknown or likely to be incorrect, symphysis-fundal height measurements were used by midwives to determine gestational age.Our study performed ultrasound scans on a small sample of 20 women (18 in our analytical sample) and found the intraclass correlation coefficient between the study and routine ultrasounds was 0.87 (95% confidence interval [CI] 0.67-1.00).

| Primary exposure and covariates
Our primary exposure was assignment to either the intervention or standard-of-care arms of the study.Covariates of interest were factors found to be associated with preterm birth 15 or low birthweight 16 that could be imbalanced between the arms.Initial covariates examined were age (≤25 years versus >25 years), relationship status (married/cohabitating versus not), education level (primary level or less versus secondary level or higher), income (≤2000, >2000 Pula [~US$ 160]/ month, or unknown), nationality (Motswana versus other), HIV-infection status (infected versus uninfected), partner's HIV-infection status (positive, negative or unknown), parity (previous live birth versus not), recorded history of previous preterm birth or low birthweight outcome among normal singleton births (ever had a preterm birth or low birthweight infant versus no history of preterm birth or low birthweight births among normal singleton births), proportion of World Health Organization-recommended antenatal care visits achieved by gestational age at delivery, 17 delivery mode (spontaneous vaginal delivery versus other), sex of the baby (male or female), and hypertension during pregnancy (defined as either a systolic pressure of ≥140 mmHg or diastolic pressure of ≥90 mmHg recorded at first antenatal care visit or during the third trimester.We also examined any hypertension versus none, as well as a three-category variable ('early' hypertension at or before 20 weeks, 'late' hypertension after 20 weeks, or no hypertension), alcohol use during pregnancy (using the Alcohol Use Disorders Identification Test [AUDIT-C] of any use [AUDIT-C score >0] versus none, and harmful use [AUDIT-C score ≥3] versus none and non-harmful), 18 any tobacco use during pregnancy, clinic site where care was received, and enrolment period (February 2021-January 2022 versus February 2022-December 2022).At the third trimester and postnatal visits, participants were asked if they had received any treatment for urogenital symptoms since their last study visit.Study staff also went through the obstetric record with participants and abstracted the antibiotic treatment that was received as part of routine care.(e.g.ceftriaxone, azithromycin, erythromycin).We created a dichotomous variable for antibiotic treatment outside the study.

| Analysis
All analyses were conducted using STATA 18 (StataCorp, College Station, TX, USA).We assessed whether the proportions of participants who reported miscarriages or stillbirths were similar across study arms using chi-square or Fisher's exact tests.
We provided descriptive statistics of the study sample by intervention and standard-of-care arm assignment and assessed the balance between the arms using chi-square or Student's t tests.To address potential clustering and unmeasured confounding, we assessed results from three statistical models.First, we used a multivariable logistic regression model.Next, we estimated a fixed effects model in an attempt to purge some effects of unobserved cliniclevel characteristics that may influence birth outcomes. 19s observations may have been clustered within clinics (the intraclass correlation coefficient was 0.0017, suggesting that only 0.17% of the variance in the outcome was due to variations across clinics), we also estimated a random effects model, with the clinic specified as a random effect.For each model, the 'disjunctive cause criterion' 20 was used to select the initial confounders based on their potential causal relationship with the exposure (imbalanced between the arms) or outcome or both, while seeking to discard instrumental variables.Next, we employed a 'purposeful selection of variables' strategy, introduced by Hosmer and Lemeshaw 21,22 where subject matter significance, relationships among the independent variables (e.g.correlations and interactions) and statistical significance were considered.In all statistical tests, significance was determined based on an α level of 0.05, except for interactions, where a more liberal α level of 0.10 was used.Models were run for the primary outcome (preterm birth or low birthweight), as well as preterm birth only (includes low birthweight if preterm) and low birthweight only (includes preterm if low birthweight) individually.When the clinic site variable was included, we assessed the models with and without the clinic that did not cross-over.
Further, we conducted post-estimation predictive margins analysis to estimate the predicted prevalence of the composite outcome of low birthweight or preterm birth in the intervention and standard-of-care arms as well as the adjusted risk ratios.In other words, following the logistic regression, all participants were set to both exposure values (standard-of-care and intervention) and the logistic regression coefficients were used to calculate the predicted prevalence for everyone at their observed confounder pattern and assigned exposure.Confidence intervals were calculated using the delta method. 23Thereafter, the effect of intervention versus the standard-of-care was calculated as an adjusted risk ratio.In another attempt to balance confounding, we conducted post-hoc analysis that stratified the sample by nulliparous and multiparous women.
Further, we examined patterns of missing data.First, we generated dichotomous indicator variables for each variable in the model to indicate whether it was missing.Next, we ran logit models to test if baseline participant characteristics differed by missing status.For our main intervention evaluation, we conducted complete case analysis.In response to a reviewer-suggested sensitivity analysis, we also imputed missing data using the multivariable normal model 24 and chained equations approach to see if our results held.

| Ethics approval
The Maduo Study was approved by the Botswana Health Research Development Committee of the Botswana Ministry of Health and Wellness (HRDC #00881) and in accordance with the National Institutes of Health single institutional review board policy, a reliance agreement was approved by the University of Southern California (HS-21-00245) and the University of California, San Diego.

| R E SU LTS
Among 500 participants, 436 (87%) had single, non-missing, live birth outcomes.We excluded 20 participants who experienced miscarriages, 11 who had stillbirths, seven with twin births, and 26 participants who were lost to follow up (Figure 1).The distribution of miscarriages and stillbirths was similar across study arms.In our analysis of missing data, we determined that the missingness was not at random.
As seen in Table 1, the median age in our analytical sample was 27 years, the median gestational age at first antenatal care visit was 17 weeks, about half of the participants were married or cohabiting, most women were citizens of Botswana, 7% had a recorded measure of hypertension, 17% were living  with HIV and the median number of antenatal care visits was six.Self-reported SARS-CoV-2 testing rates were low (28%), and the positivity rate among those tested was high (26%).Characteristics that differed statistically between the intervention and standard-of-care arms included education level, income, citizenship, HIV status, first pregnancy, partner's HIV status, clinic and study period (p ≤ 0.05 for all).
The overall prevalence was 13.3% for the composite outcome of preterm birth and/or low birthweight, 9.4% for preterm birth and 8.5% for low birthweight.Without controlling for potential group imbalance, the prevalence of preterm birth or low birthweight was not significantly difbetween the intervention (14%) and standard-of-care (13%) arms.Abbreviations: IQR, Interquartile range; WHO, World Health Organization.a Among those in our analytical sample who received an ultrasound during antenatal care, 18 received an ultrasound through the study (6 in the standard-of-care and 12 in the intervention arm) and the remainder through routine care.
b Received antibiotic treatment as part of routine care.(e.g.Ceftriaxone, Azithromycin, Erythromycin).

T A B L E 1 (Continued)
Results from the multivariable logistic regression can be found in Table 2.In all tested models, exposure to the intervention arm was associated with reduced odds of the composite outcome of preterm birth or low birthweight after controlling for parity, antenatal care visits, hypertension and clinic site; however, the confidence intervals crossed one.Further, odds ratios were similar when limiting the sample to participants in cross-over clinics that received the intervention and standard-of-care assignment over time.
In the post-hoc analysis, the stratification reduced some of the differences between the intervention and standardof-care groups.Among nulliparous participants, the groups were balanced on nationality, HIV status and time period; however, the groups continued to differ on education level, income and partner's HIV status.Although the HIV prevalence was 5% among nulliparous participants, 18% of participants said their partners were living with HIV and 29% did not know their partner's HIV status.In the multiparous group, the intervention and standard-of-care arms differed on income level, nationality and partner's HIV status.The HIV prevalence among multiparous women was 24%, and 20% reported that their partners were living with HIV and 31% did not know.Table 2 also provides results from the multivariable logistic regressions stratified by previous live birth.After controlling for hypertension and clinic, the intervention was associated with a reduced odds of the composite outcome, preterm birth and low birthweight, among the nulliparous group.However, no effect was found among multiparous women.
Through our analysis of missing data, we found that those missing birth outcomes were similar in terms of age, relationship status, education, income, HIV status, partner's HIV status, parity, history of a preterm birth or low birthweight outcome, hypertension, nationality and alcohol use.However, the standard-of-care group was more likely to have missing birth outcome data (8.7%)compared with the intervention group (2.6%).Our study team was able to contact 11 of the 26 participants missing birth outcome data by phone and determined that they were now living outside Gaborone.However, those missing data on antenatal care visits were more likely to be younger, have had hypertension during pregnancy, and had previous preterm birth or low birthweight outcome.We imputed data for those missing the number of antenatal visits using multiple imputation with chained equations that included age, gestational age at first antenatal care visit, hypertension during pregnancy, education level, parity, nationality and clinic as variables.We included the imputed variable in our final model and results were similar to the complete case analysis with a predicted prevalence of preterm birth or low birthweight at 16% in the standard-of-care arm and 12% in the intervention arm.

| Main findings
We evaluated the impact of C. trachomatis and N. gonorrhoeae screening and treatment on preterm birth or low birthweight among asymptomatic pregnant woman in Gaborone, Botswana.The intervention arm had a lower predicted prevalence of a composite outcome of preterm birth or low birthweight compared with the T A B L E 2 Results from logistic regression models assessing the relationship between intervention arm assignment and adverse birth outcomes.Note: The covariates included in the full analytical sample were parity, antenatal care visits, hypertension during pregnancy, and clinic.The covariates in the sub-group analysis included hypertension during pregnancy and clinic.

Outcome
Abbreviations: AOR, Adjusted odds ratio.a 18 participants were missing data on number of antenatal care visits.
standard-of-care, after adjustment for parity, antenatal care visits, hypertension during pregnancy and clinic site; however, the confidence intervals were wide.In posthoc analysis among nulliparous women, the intervention was associated with reduced odds of preterm birth or low birthweight outcomes, necessitating further research into reasons for a differential effect.

| Strengths and limitations
A strength of this study was the prospective design that allowed us to compare STI screening and treatment results with birth outcomes.The majority of previous observational studies that compared C. trachomatis or N. gonorrhoeae infections during pregnancy with birth outcomes were unable to determine receipt of treatment. 7,25Further, this study enrolled a unique cohort of pregnant women that would have been missed by the syndromic approach for management of STIs.The World Health Organization recommends that countries enhance syndromic management by gradually scaling-up laboratory testing, 26 but the guidelines for management of asymptomatic C. trachomatis or N. gonorrhoeae infections are currently being developed and more research is needed.Finally, because lowand middle-income countries rely on syndromic management during pregnancy, 9 there is a lack of data on the prevalence and impact of asymptomatic, curable STIs.
Our study is subject to several limitations.First, neither participants nor clinics were randomised and thus unmeasured confounding between intervention assignment and birth outcomes is likely.For example, SARS-CoV-2 testing rates were low and we were unable to control for test results in our model.Although research assessing the association between SARS-CoV-2 infection and preterm birth and low birthweight has mixed results, 27,28 if SARS-CoV-2 infection prevalence differed by intervention assignment, our results may be biased.We attempted to address threats to internal validity through the cross-over design, controlling for imbalanced participant characteristics, and estimating a fixed effects model that purges clinic-level unobserved variables.Second, birth outcome data were missing not at random.If those missing data in the standard-of-care arm were more likely to have an adverse outcome, then our study results were attenuated.However, we made every effort to contact participants with missing data and determined that almost half were missing because they had moved away from Gaborone.Third, estimates of gestational age and birthweight were subject to mismeasurement.Birthweight measurement procedures and scale accuracy were not validated by the study.Although gestational age was not measured by the study, a sub-sample of women with a study ultrasound received estimates similar to the clinic-recorded results, and 68% of participants had access to routine ultrasound before their third-trimester visit.Finally, the sample size for this study was determined to find a difference in postdelivery prevalence and vertical transmission of C. trachomatis and N. gonorrhoeae and was underpowered to find a difference in preterm birth or low birthweight frequency.

| Interpretation
Our post-hoc finding that C. trachomatis and N. gonorrhoeae screening and treatment reduced preterm birth or low birthweight among nulliparous women needs further exploration.Although this finding is subject to Type I error, it is similar to a post-hoc result in the AuTop trial, 29 which found that bacterial vaginosis screening and treatment reduced preterm birth among nulliparous, but not multiparous low-risk pregnant women in France.Authors of the AuTop trial hypothesised that the differential effect may be the result of nulliparous women having an unknown risk for preterm birth, whereas multiparous participants had a low risk as women with a history of preterm birth or late miscarriage were excluded from enrolment.However, the Maduo study did not exclude participants based on risk factors and 13% of multiparous women had a documented history of preterm birth or low birthweight outcomes.The AuTop authors also speculated that the results could be due to higher treatment uptake among nulliparous women compared with multiparous women (85% versus 79%).In Maduo, the prevalence of C. trachomatis and N. gonorrhoeae infections and treatment uptake were similar by parity.Another reason for the differential impact in our study could be that nulliparous women were less likely to be living with HIV compared with multiparous women (5% versus 24%), suggesting a lower risk of previous exposure to C. trachomatis and N. gonorrhoeae infections.Some research has found that previous chlamydial infection could confer a degree of protective immunity, which would make the screening and treatment intervention less effective at reducing adverse outcomes. 30ur finding that STI screening and treatment may reduce adverse birth outcomes compared with syndromic management adds support to previous systematic reviews and meta-analyses.Adachi et al., 31 identified 13 of 15 studies that found antenatal C. trachomatis screening and treatment reduced adverse birth outcomes. 8Tang et al. found that C. trachomatis infection increased the odds of preterm labour by 1.29 (95% CI 1.11-1.50;n = 11 studies). 25hmadi found that women with C. trachomatis infections were 2.28 times more likely to deliver preterm (95% CI 1.64-3.16;n = 24 studies), compared with those without C. trachomatis infection.He et al. found that among 15 case-control and 15 cohort studies, C. trachomatis infection increased the odds of preterm birth by 1.73 (95% CI 1.34-2.23). 32However, most included studies were observational and did not control for confounding, and receipt of treatment was not reported.
A recent study investigated the relationship between C. trachomatis, N. gonorrhoeae and Trichomonas vaginalis screening and treatment among pregnant women living with HIV in South Africa 33 and found no association between screening and preterm birth or low birthweight.However, this study was limited to women living with HIV, did not exclude symptomatic participants and found that the testing group was less likely to be on anti-retroviral therapy; also, participants were treated for T. vaginalis.Further, participants had high levels of persistent infections as 27% were positive for C. trachomatis at repeat testing. 34iven the equipoise around the impact of STI screening on adverse pregnancy outcomes, future research may consider individual randomisation or other methods for addressing confounding, measurement bias, and lack of power.As discussed by Low in a recent commentary, 35 observational studies face major threats to internal validity because many factors are associated with both STIs and pregnancy outcomes.Case-control and cohort studies are subject to measurement bias when receipt of treatment is not recorded because treatment probably reduces the risk of the outcome. 35Cluster trials, such as the current study and the upcoming WANTAIM Trial 36 seek to reduce confounding and measurement bias; however, confounding can still occur through unmeasured baseline imbalance and clustering has a negative impact on the power to detect an effect. 37

| CONCLUSION
In conclusion, the screening and treatment intervention among asymptomatic pregnant women may have reduced the frequency of preterm birth or low birthweight infants compared with the standard-of-care among nulliparous women; however, the effect size among multiparous women and the entire sample was uncertain.As prematurity and low birthweight are leading causes of preventable death for children under 5 years, more research designed to detect differences in birth outcomes is needed for evaluating the effectiveness and cost-effectiveness of antenatal C. trachomatis and N. gonorrhoeae screening interventions.

AU T HOR C ON T R I BU T ION S
JDK and CM were the principal investigators of this study.JDK, CM and AW wrote the study protocol with technical expertise from DR-M.KR, LT, NN, EH and RR implemented the study and collected data.AM managed the data collection and quality assurance processes.Study implementation was supervised by JDK, CM, AW, RR, CMB and AM.AW and MLW performed the data analysis.AW wrote the draft version of the manuscript; all authors contributed to the final version.

AC K NOW L E D GE M E N T S
Two GeneXpert® machines were loaned by Cepheid for the duration of the study.The authors would like to acknowledge the Greater Gaborone District Health Management Team and clinic managers, who gave permission to conduct the study at the respective sites, and the facility staff who accommodated our study teams.

F U N DI NG I N FOR M AT ION
This study was funded through The Eunice Kennedy Shriver Institute of Child Health and Human Development, National Institutes of Health, award R21HD100821 to JDK and CM.AW was supported by a National Institute on Alcohol Abuse and Alcoholism K01 Career Development Award (K01AA027733).This work was supported in part by Open Philanthropy.The funders had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
Dr Klausner reports personal fees from Cepheid, during the conduct of the study, and personal fees from Danaher, outside the submitted work.All other authors: none declared.

F I G U R E 1
Flow of participants in the Maduo study, February 2021-December 2022, Gaborone, Botswana.LBW, low birthweight; PTB, preterm birth; SoC, Standard-of-Care.T A B L E 1 Characteristics of participants in the Maduo study, February 2021-December 2022, by Chlamydia trachomatis and Neisseria gonorrhoeae screening and treatment intervention assignment, Gaborone, Botswana.

Intervention Standard of care AOR (95% CI) p-Value
Predicted prevalence of preterm birth/low birthweight, preterm birth and low birthweight by intervention arm assignment.Intv, Intervention assignment; LBW, low birthweight; PTB, preterm birth; SoC, Standard-of-Care.Predicted prevalence using post-estimation predictive margins analysis resulting from multivariable logistic regression controlling for parity, antenatal care visits, hypertension and clinic.