Prevalence, risk factors and adverse pregnancy outcomes of second trimester bacterial vaginosis among pregnant women in Bukavu, Democratic Republic of the Congo

Background Bacterial vaginosis (BV) is the most common gynecological condition in women of reproductive age and associated with adverse pregnancy outcomes. In the Democratic Republic of the Congo (DRC), neonatal mortality rate is as high as 2.8 percent with preterm birth (PTB) and low birth weight (LBW) as leading causes. Because no studies have addressed BV in DRC, we aimed to investigate the prevalence of BV, the risk factors and the association between BV and adverse pregnancy outcomes in a population of pregnant women from Bukavu, DRC. Methods A total of 533 pregnant women in the second trimester of pregnancy were recruited in the Provincial Reference Hospital of Bukavu, DRC, between January and October 2017, and followed until delivery. Clinical and sociodemographic data of mother and newborn, and data on (vaginal) hygiene practices, sexual behavior and reproductive history were collected. BV was diagnosed by Nugent scoring of Gram-stained vaginal smears. Two multivariate regression models were built to identify risk factors for BV and to investigate BV as a risk factor for adverse pregnancy outcomes. Results The prevalence of BV was 26.3% and approximately half of the women with BV were asymptomatic. Independent risk factors for BV were the use of alternatives to water for intravaginal washing, concurrent partners, unemployed status, the presence of vaginal Candida and clay consumption. BV was independently associated with both LBW and PTB of an infant with LBW. Conclusion The prevalence of BV in Bukavu is high but in line with the global average. BV was associated with adverse pregnancy outcomes in our study population. Hence, research on modifiable risk factor-based interventions to reduce the prevalence of BV, and on screening/ treatment of BV during antenatal care should be explored to reduce neonatal mortality and morbidity. This study was ethically approved by the internal review board of the Catholic University of Bukavu (reference number UCB/CIE/NC/016/2016), by the Ministry of Public Health of DRC (reference number 062/CD/DPS/SK/2017) and by the Ethical Committee of Ghent University Hospital (reference number PA2014/003). Women who were found eligible for the study were orally informed in detail on the study and were asked to sign an informed consent form. The current study is part of the AVEONS study, an acronym of Angamiza Vizuri (Swahili for “break well” or “stop”) Early Onset Neonatal Sepsis. The AVEONS study has the overall aim to describe the prevalence of early onset neonatal sepsis (EONS) in Bukavu (South Kivu, DRC), to document the pathogens causing EONS and their antibiotic susceptibility patterns, and to investigate the role of maternal infections during pregnancy on APOs. The province of South Kivu is located in the eastern part of DRC, a region characterized by high social, economic and political instability [20, 21]. Thousands of people have been forced to flee their homes and seek refuge either in larger cities such as Bukavu or in neighboring countries (Tanzania and Burundi) [22, 23]. In South Kivu, the prevalence of contraception use by married women and unmarried sexually active women has been estimated to be respectively 64.8% and 94.5% [19]. The fertility rate in South Kivu has been appraised at 7.6%, the access to at least one antenatal visit for pregnant women 93.5% and the home delivery rate 8.9% [19]. The AVEONS study was a prospective observational study where pregnant women were seen between 16 and 20 weeks (visit 1 (V1) = recruitment visit), between 36 and 38 weeks (V2) and at delivery. Newborns were seen at delivery and during the first week of life. This first manuscript of the AVEONS study reports on a cross sectional analysis of women seen at V1 and pregnancy outcomes. We recruited study participants among pregnant women seeking antenatal care at the Provincial Referral Hospital of Bukavu (PRHB) between January and October 2017. To increase awareness of the AVEONS study, church announcements, radio and TV spots and posters were used. Women were also informed on the study during community leaders’ meetings. Pregnant women who were interested in participating were briefed individually in detail. If women were still interested after this information session, they were screened for inclusion applying the inclusion and exclusion criteria. Women were considered for inclusion if they were between 16–20 weeks of gestational age, agreed to receive antenatal care only at PRHB, accepted to deliver at PRHB and were willing to be contacted by phone (to increase adherence by phone call and text message reminders). Women were not considered eligible if they planned to move out of Bukavu during their pregnancy, had twin pregnancies, suffered from genital bleeding and/or if they used antibiotics in the two weeks before V1. Pregnant women were reimbursed for their transport to and from PRHB. At V1, a general physical examination including anthropometric measurements (including height, weight, body mass index (BMI) and mid-upper arm circumference (MUAC)) was performed. A gynecological examination was performed, and the vaginal mucosa and cervix were inspected for sores and tumors. In case women had abnormal vaginal discharge, an abnormal foul smell, experienced itching and/or a burning sensation after intercourse, a vaginal infection was diagnosed according to the syndromic-based protocol for the management of pregnancy issued by the Ministry of Public Health of DRC [24]. These symptomatic women were treated empirically according to the local protocol (daily clotrimazole (200 mg) against candidiasis and clindamycin (100 mg) against BV for six days). In case of allergy against clindamycin, a daily metronidazole ovule was prescribed for six days. During the gynecological examination, the vaginal pH was determined by means of an indicator pH paper (Hilindicator pH paper). An ultrasound examination was performed at V1 to assess the viability of the fetus and the cervical length. At 24 weeks of gestation, participants were offered a single dose of mebendazole (500 mg) against soil-transmitted helminths and a single dose of sulfadiazine-pyrimethamine (500 mg) against malaria as recommended in the pregnancy protocol. Participants further followed antenatal care as usual. At V1, five mL of blood was collected in a tube without anticoagulant. Subsequently, the serum was tested for HIV, malaria and hemoglobin using rapid tests (Alere Determine™ HIV-1/2 (Abbott), Malaria AG P.f/pan (Bioline) and Hemocue Hb201+ (Hemocue AB), respectively). A urine sample was collected in a sterile container and tested for the presence of white blood cells and nitrite (indicating a urinary tract infection or bacteriuria) by Multistix® dipsticks (Siemens). A vaginal swab (COPAN) was taken by gently rolling the top of the swab against the midportion of the vaginal wall. The swab was rolled on three glass slides, one smear was used for routine wet mount microscopy and two were used for Gram-staining (Nugent score). The wet mount was examined microscopically immediately after collection of the vaginal swab for the presence of clue cells (one of the four Amsel criteria, used for the diagnosis of BV in clinical practice), Candida cells and/or hyphae and Trichomonas vaginalis. At delivery, the labor was followed, and appropriate data such as obstetrical parameters (uterine height, fetal presentation, status of the fetal membranes, number of vaginal examinations, maternal temperature, the use of the childbirth kit) were collected by nurses and the senior assistant. Neonates were subjected to a general examination and anthropometric parameters (length, weight and head circumference) were assessed by a pediatrician. The presence of an early neonatal infection was assessed based on the WHO criteria, i.e., the presence of at least one of the following signs: temperature instability, lethargy, feeding intolerance, respiratory distress, hemodynamic instability, convulsion, hypotonia, irritability or bleeding diathesis [25]. Slides for Gram-staining were heat fixated, stored at room temperature and shipped to the Laboratory Bacteriology Research (Ghent University, Ghent, Belgium), where Gram-staining was carried out with an automated PolyStainer (IUL). All Gram-stained slides were scored according to the Nugent scoring system [26] for the laboratory-based diagnosis of BV. Briefly, five microscopic fields per Gram-stained slide were scored on a 1000x magnification for the presence and quantity of Lactobacillus (Gram-positive rods), Gardnerella vaginalis/Bacteroides (Gram-variable coccobacilli) and Mobiluncus (Gram-negative curved rods) cell types. The smear was then categorized as representing a healthy vaginal microbiota (Nugent score 0–3), an intermediate vaginal microbiota (Nugent score 4–6) or BV (Nugent score 7–10). Furthermore, during the Nugent scoring, we also documented the presence of Candida cells and/or hyphae. All slides were scored single-blinded by two independent readers. In case of a discrepancy in categorization, the slide was reassessed by the two reviewers and discussed. If no consensus was obtained, a third person assessed the slide as a tie breaker. At V1, data on the sociodemographic characteristics, reproductive health history, sexual behavior, vaginal practices and complaints of the pregnant women were obtained in a confidential way using a questionnaire (S1 File). Women were questioned individually in Swahili by a gynecologist or gynecologist in training, who was also responsible for the patient’s care. The interview took about 20 minutes and answers were noted on printed questionnaires. All clinical data forms, laboratory forms and questionnaires of each participant were scanned and stored as a single file. The raw data were then captured in the CSPRO software by means of double data entry. The ‘compare data tool’ of the CSPRO software was used to compare the two data sets. In case of discrepancies, the raw data were consulted, and discrepancies were solved. The two datasets were further checked for discrepancies in STATA 14 (Stata Corp, College Station, Texas, USA) before making one final locked dataset for analysis. Categorical variables were summarised into frequencies and proportions, continuous variables into median and interquartile range (IQR). First, we determined the prevalence of BV, defined based the Nugent score, and reported this prevalence with a 95% confidence interval (CI). A Kappa test was calculated to evaluate the agreement in categorization based on the Nugent score between the two readers. The agreement was considered as poor (<0.00), slight (0.00–0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80) or almost perfect (0.81–1.00) [27]. Second, to determine independent risk factors for BV (defined as a Nugent score of 7–10), we built a multivariate model. Here, plausible explanatory variables were sociodemographic characteristics, sexual behavioral and vaginal hygiene characteristics and laboratory findings. Logistic regression was applied to assess the association between the various risk factors and BV. Independent variables with a p-value <0.10 in the univariate analysis were incorporated into a multiple regression model. In this model, independent variables were considered as statistically significantly associated with BV if the p-value was <0.05. Third, to determine whether BV itself was a risk factor for one or several APOs, we estimated unadjusted odds ratios and adjusted odds ratios applying logistic regression models. We considered PTB, LBW, preterm birth of a LBW infant (PTB-LBW), premature rupture of membranes (PROM), and suspicion of EONS as dependent variables. Here, BV was considered as the independent variable. PTB was defined as delivery before 37 completed weeks of gestation, LBW as birth weight <2,500 g at delivery, PTB-LBW as delivery occurring before 37 weeks of gestation with a birth weight < 2,500 g. Suspicion of EONS was defined on the basis of WHO criteria, i.e., a neonate with minimum one of the following symptoms: temperature instability, lethargy, feeding intolerance, respiratory distress, hemodynamic instability, convulsion, hypotonia, irritability or bleeding diathesis [25]. We did not assess the association between BV and second trimester pregnancy loss because the number of cases were not sufficient to draw statistically significant conclusions. A modified Poisson regression model with robust standard error was built using generalized linear regression equations to model the link. APOs with a significant association with BV in the univariate analysis were selected for the multivariate model. In this model, to better control the effect of BV on APOs, we included a priori covariates for APOs that were expected to be relevant, based on literature [15, 28–30]. These covariates were vaginal Candida colonization (as assessed by means of microscopic examination of Gram-stained slides), anemia, MUAC, cervix length, parity, BMI, maternal age, education level, previous PTB and high diastolic blood pressure at V1 (≥ 90 mm Hg).