Effect of bacterial vaginosis on preterm birth: a meta-analysis

Bacterial vaginosis is a common genital tract disorder. It can lead to preterm birth, but its contribution is equivocal. Bacterial vaginosis is curable and, if diagnosed and appropriately treated, may reduce preterm births. The study desired to confirm the association between bacterial vaginosis and preterm birth. It was a meta-analysis. We included articles published from 2008 to 2022. The authors included studies that measured the association between bacterial vaginosis and preterm birth by relative risk/risk ratio (RR) or odds ratio (OR). We excluded studies with qualitative data. The study utilized five search engines, PubMed, Scopus, Google Scholar, Cochrane, and LILAC. We used the statistical package for social sciences (SPSS) to draw forest and funnel plots separately for RR and OR. After an extensive search, the study included 20 articles yielding 26 relevant results with a total of 290,397 observations. This meta-analysis proves that bacterial vaginosis is undoubtedly associated with preterm birth. The overall relative risk of preterm delivery is about two-fold as overall OR1.79 (95% Confidence Interval 1.32–2.43). The overall RR of preterm birth is 1.44 (95% Confidence Interval 1.19–1.73). Our study shows a significant association between bacterial vaginosis and preterm birth. The study concludes that investigation for bacterial vaginosis and management should be a part of the routine examination of a pregnant woman. The health system must initiate this strategy soon to reduce the prevalence of preterm births and consequent neonatal mortality.


Introduction
Bacterial vaginosis (BV) is the most common lower genital tract disorder among women of reproductive age. The imbalance between commensal and pathogenic bacteria leads to bacterial vaginosis. In a healthy lower genital tract, Lactobacillus crispatus constitutes 95% of the vaginal flora; among cases of bacterial vaginosis (BV), Lactobacillus crispatus is dramatically reduced, and an overgrowth of 1 3 various anaerobic bacteria exists [1][2][3]. In BV, there is an overgrowth of different gram-negative and/or anaerobic bacteria, such as Gardnerella vaginalis, Atopobium species., Megasphaera phylotype 1 species., Mobiluncus species., Ureaplasma urealyticum, Provetella, Peptostreptococcus, Mycoplasma hominus, and A. vaginae and/or G. vaginalis [1,4,5]. Bacterial vaginosis (BV) affects millions of women, is highly prevalent among low-income, urban pregnant women, and is frequently chronic [6]. The BV can be diagnosed by scoring bacterial morphotypes from a Gram stain of vaginal fluid, assigning a Nugent score between 0 and 10. A score of 0-3 represents "normal vaginal flora", 4-6 means "intermediate vaginal flora", 7-10 are considered diagnostic for bacterial vaginosis [4]. BV can be diagnosed clinically by the presence of three of the following four signs first described by Richard Amsel in 1983: 1. Presence of an adherent and homogeneous vaginal discharge. 2. A vaginal pH > 4.5.

Detection of clue cells (vaginal epithelial cells so
covered by bacteria as to render the borders indistinct) on saline wet mount. 4. An amine odor (positive 'whiff test') after adding the amine potassium The term vaginosis is preferred over vaginitis because of the absence of leukocytes in vaginal fluid. Bacterial vaginosis is associated with chorioamnionitis, postpartum endometritis, postsurgical infections, pelvic inflammatory disease, and preterm birth [7,8]. However, the present study focuses on preterm birth. These bacteria contribute to spontaneous preterm birth through localized inflammation of the endometrium creating an environment incompatible with proper placental formation and growth and/or by circulating cytokine production resulting in premature rupture of membranes, and many times, subsequent spontaneous preterm birth [9]. Preterm birth is defined as babies born before 37 weeks (259 days) of gestation. Based on gestational age, preterm is categorized into three groups. Despite the voluminous literature evaluating bacterial vaginosis and preterm birth, many doubts remain. Its association with the most severe neonatal sequelae and the different strengths of the association between bacterial vaginosis and preterm birth were observed in diverse populations [10]. A few meta-analyses studies are published, but most are more than 15 years old. Second, screening and treatment of BV have not yet been a uniformly accepted strategy.

Objective
To estimate the overall relative risk and odds ratio of preterm birth among women having bacterial vaginosis through Meta-analysis.

Inclusion criteria
All types of studies, i.e., case-control, cohort and clinical trials, systematic review, and meta-analysis conducted to assess the association between bacterial vaginosis and preterm birth by deriving relative risk/risk ratio or odds ratio. We grouped the studies into the relative risk/risk ratio and odds ratio categories. Studies whose free full text is available. Article published from 2008 to 2022.

Exclusion criteria
Studies with qualitative data. Studies in which original data were not available. Clinical trial in which all participants received medications.

Information sources
We utilized PubMed, Scopus, Cochrane, Google Scholar, and LILACS search engines up to 15th August 2022.

Search strategy
We selected English and Spanish language (for LILACS database and then translated the articles) for five search engines. In PubMed, the keywords [Bacterial vaginosis] AND [Preterm birth] were searched as MeSh. We searched other search engines using a combination of keywords; "bacterial vaginosis" AND "preterm birth and/or preterm delivery" AND "relative risk OR odds ratio." Similar articles mentioned in the search engines and references included in the selected articles were also searched and screened.
The preterm was considered birth/delivery before 37 completed weeks (up to 259 days).

Selection process
We first reviewed the titles and authors of identified articles. Then duplicates were removed. The authors screened all available abstracts of remaining articles for relevance to BV and PTB and deleted the articles if there was no mention of preterm birth, relative risk/risk ratio or odds ratio. The authors then read the available full text of withheld articles for inclusion or exclusion in meta-analysis as per predecided criteria.

Data collection process
All authors independently collected data from different search engines. The study compiled information about the authors, year of publication, type of study design, number of participants, method of diagnosis of BV, gestational period of screening, definition of preterm birth, relative risk or odds ratio and if available adjusted statistics, the 95% confidence intervals. We also collected original data for calculation of overall statistics and for drawing plots. Authors calculated relative risk or odds ratio if not mentioned in the article.
The PRISMA flow diagram in Fig. 1 provides an overview of the study selection process.

Effect measures
Relative risk/risk ratio or odds ratio of preterm birth/delivery among women having bacterial vaginosis. We noted both crude statistics as well as adjusted statistics with confidence intervals.

Data synthesis and analysis
The collected data were coded and entered into a Microsoft Excel sheet. Each study summary effect (the result), either RR or OR were noted with a 95% confidence interval from individual studies' result section. We then calculated overall estimates from individual studies as a weighted average following the fixed effect model for meta-analysis. We drew forest and funnel plots using SPSS (version 25).
The authors obtained Institutional Ethics Committee's approval before starting the study.

Results
In PubMed 370 articles were found in the last fifteen years, out of which 15 results had RR in their abstract and nine had OR. The authors' search in Scopus for the reference period yielded 55 articles for relative risk and odds ratio. For full review one included in relative risk category and four in odds ratio list. In Cochrane, the keyword bacterial vaginosis and preterm birth was searched. Only seven review articles and 14 clinical trials have been published, none was eligible. In Google Scholar, keywords "bacterial vaginosis" AND "preterm birth" yielded 283 results, out of which only 131 were full-text articles. Only twelve were eligible having RR or OR for screening, and ten were repeated. They were found in PubMed as well. Rest articles were either review articles, letters to editors, or qualitative data articles. In LILAC, 25 results were available. All articles were in Spanish with abstract in English. Only five of these 25 articles were suitable as 20 did not mention their results, so the required data could not be extracted. Out of five two full texts were not available. After google translation, we extracted results from three articles. Twenty articles were also received from feeds from Mendeley with the exact keywords. Only one article was eligible; nine were duplicates, six were review articles, and four had qualitative data. The authors also reviewed cross references and included the articles if not considered earlier. So, finally, 20 articles were eligible for this study. LILAC 1; Google scholar 2; Mendeley mentioned similar article 1; cross references 5; PubMed 18 (Directly 12, Mendeley 1, Cross references 5) exclusively provided article. All articles identified by Scopus were also available  [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30].
Five studies had more than one effect size; all the respective effect sizes have been considered from those studies, for analysis. We allotted a number to each result of these studies. Eight results from six articles have RR as their effect size, and 18 results from 14 studies have OR as their effect size.
Four study results were extracted for outcomes of preterm delivery at < 34 weeks. One result had the delivery outcome at 34-37 weeks, and rest had their results extracted for delivery outcomes at < 37 weeks. Two results were extracted separately from a single study for preterm birth at ≤ 34 weeks and ≤ 37 weeks. Three studies had very early preterm birth/miscarriage along with PTB as primary outcomes. In one study, three different statistics were calculated for spontaneous preterm birth at < 34 weeks, 34-37 weeks, and < 37 weeks. In one retrospective study, the primary outcomes were bacterial vaginosis, abnormal vaginal flora, and aerobic vaginitis. In two retrospective studies, the association was assessed for other five outcomes.
Many studies assessed women at their first antenatal visit. Some studies assessed bacterial vaginosis among high-risk pregnancy or low-risk pregnancy, and few assessed bacterial vaginosis among women after giving birth, asymptomatic pregnant women with prior risk factor or history of preterm delivery.
Ten studies used the standard Nugent score for the diagnosis of bacterial vaginosis. One study used Nugent score and real-time quantitative polymerase chain reaction assays, another study used tumor necrosis factor α. The diagnosis of three studies was based on lactobacillary grades and modification of Schröder's classification. Two studies used Amsel criteria for the diagnosis of bacterial vaginosis. Many studies assessed the effect of different types of organisms.
The Forest plot (Fig. 2) representing all the eligible results shows overall risk ratio of pregnant women with bacterial vaginosis having a preterm birth was 1.44 (95% C.I. 1.19-1.73). The overall statistic indicates a significant association between preterm birth and bacterial vaginosis. Individually only two results were significant, first screened women during late gestational period and second study did not mention the period.
The overall odd ratio is 1.79 (95% C.I. 1.32-2.43) as depicted by Forest plot of all eligible studies (Fig. 3). Among studies which screened BV during early gestational period four had significant OR. While among studies that screened women late in gestational period seven had significant OR (Fig. 4).
Funnel plots 4 and 5 depict publication bias assessment of RR and OR, respectively (Fig. 5).

Discussion
There is a lack of knowledge regarding many aspects of bacterial vaginosis and preterm birth. Information regarding the optimal time for diagnosing bacterial vaginosis and its association with obstetric and neonatal sequelae is also unclear. The only answer is a meta-analysis to assess the association between bacterial vaginosis and preterm birth.
This meta-analysis generates convincing evidence through different results extracted from some studies.
This meta-analysis shows a strong association between bacterial vaginosis and preterm birth. Preterm is a significant health problem in India. It is among the ten countries with the most significant number of preterm birth (351,900) [31]. Preterm birth is an important cause of death in the critical first month of life, and it is the second-leading cause of death in children under 5 years of age. Preterm birth is mainly caused due to multiple pregnancies, infections, and chronic conditions such as diabetes and high blood pressure [32]. Among all infections, bacterial vaginosis is important, which can cause adverse obstetric and fetal outcomes.
A few reasons have been incriminated for bacterial vaginosis, like unfamiliar or multiple sex partners, cervical douching, etc. These behaviors can disturb the balance of vaginal microbiota [33]. Bacterial vaginosis may elevate the risk of getting other STDs. Pregnant women with bacterial vaginosis can give rise to preterm birth, implying that most of those babies have low birth weight. Many cases are asymptomatic, hence diagnosis during the antenatal period is crucial. The screening should be done during early antenatal visits. Diagnosis is mainly made via Nugent score or Amsel criteria. If an affected pregnant woman is not treated, it will increase adverse fetal outcomes, especially preterm birth. Therefore, these two variables (bacterial vaginosis and preterm birth) have been chosen to do a meta-analysis and estimate their association.
Difficulties in conducting cohort studies have been reflected in this meta-analysis by the number of results RR and OR categories. The inherent challenges for comparison were the ununiform gestation period of diagnosis, and the investigators have studied outcomes at various gestations. Moreover, many institutions' local protocol advocates treatment after diagnosis, and we found many studies ineligible in this context [34,35]. All the results considered for RR in this meta-analysis endorse the association between bacterial vaginosis and preterm birth. But only two showed significant association. The OR results reveal that two studies showed some protection but were not significant. The remaining 16 show some association, and 11 of them are significant.
This meta-analysis proves that bacterial vaginosis is significantly associated with preterm birth. The risk of preterm delivery is about twofolds in women with bacterial vaginosis OR 1.79 (95% Confidence Interval 1.32-2.43). Bacterial vaginosis late in pregnancy is a much stronger risk factor for preterm delivery than bacterial vaginosis infection in early pregnancy. The risk of bacterial vaginosis causing preterm delivery may be eight times in late pregnancy [25]. The compilation and comparison from meta-analysis studies become difficult because studies investigated BV at different gestational periods; some studies included varied outcomes like premature rupture of membrane and low birthweight [36,37]. Some studies assessed the association with periodontal infection [16,38]. Some studies focused on the organisms [5,14,16,17,28,39,[40][41][42]. Change in vaginal PH is one of the constituents of diagnosis of BV, and the association between vaginal PH and BV is known [43]. Hence, some studies exclusively used PH including by self-swab taking by the women also [44][45][46]. In a metaanalysis, Harald Leitch et al. calculated higher risks for women screened for bacterial vaginosis at < 16 weeks of gestation (OR 7.55; 95% CI 1.80-31.65) or at < 20 weeks of gestation (OR 4.20; 95% CI 2. 11-8.39). That meta-analysis also had a significantly higher risk of spontaneous abortion (OR 9.91; 95% CI 1.99-49.34). The overall OR of present meta-analysis is similar to the present study (OR 2.19; 95% CI 1.54-3.12). That meta-analysis included 18 studies with 20,232 patients. They concluded that bacterial vaginosis early in pregnancy is a strong risk factor for preterm delivery and spontaneous abortion [47].
Another meta-analysis on treating abnormal vaginal flora in early pregnancy with clindamycin to prevent spontaneous preterm birth calculated the separate relative risk for spontaneous preterm birth at < 37 weeks and < 33 weeks. In both gestational ages, it showed low risk. (RR at < 37 weeks = 0.60; 95% CI 0.42-0.86 and RR at < 33 weeks = 0.44; 95% CI 0.14-1.41). They included five trials with 2346 women as study participants. They concluded that administration of clindamycin at < 22 weeks of gestation was associated with a significantly reduced risk of preterm birth [48].
One meta-analysis included 19 studies and the summary OR was 1.85, 95% CI 1.62-2.44. They also concluded that bacterial vaginosis was a risk factor for low birth weight (OR 1.57; 95% CI 1.32-1.87), preterm premature rupture of membranes (OR 1.83; 95% CI 1.39-2.44), and preterm labor (OR 2.19; 95% CI 1.73-2.76). Pooling adjusted ORs gave a 60% increased risk of preterm delivery in the presence of bacterial vaginosis [49]. The references included in these meta-analyses were not a part of our study because they did not fit our inclusion criteria, as they were old studies.
A recent systematic review and meta-analysis from 37 studies and results from 25,435 women showed that the overall risk of PTB among women having BV may be almost double (OR 1.98; CI 1.55-2.54) [37].
Another systematic review (14 results from nine studies) from Sub Saharan countries although observed that most of the studies showed association between BV and preterm birth but there were few studies which did not find any association [36].
Five studies showed significant OR out of seven that screened BV at the time of labor. It implies that untreated BV continues and results in preterm birth. The study reconfirmed a definite relationship between bacterial vaginosis and preterm birth. There is lack of knowledge management of bacterial vaginosis after diagnosis, measures to prevent it, and prevent reinfection during the rest of the pregnancy. Regular screening of symptomatic and asymptomatic and proper treatment for reducing preterm deliveries/low birth weight babies has already been advised, [50] but universally followed. There is a question of the availability of effective treatment. Various study results about antibiotic use are ambiguous.
What should be the ideal time for screening for bacterial vaginosis? Should only pregnant women be targeted, what about non-pregnant women of reproductive age? These are the issues generated through this meta-analysis.

Limitations
There are a few limitations to this meta-analysis. We did not use some acknowledged search engines (e.g., Embase) as they require substantial payment. Many articles have inadequate data; hence the authors did not consider them in the meta-analysis. Every meta-analysis has the limitation of non-inclusion of unpublished data, which applies to our study also. We did not assess the risk of bias in individual studies. The studies we included did not explore the effect of treatment of BV on the primary outcome.

Conclusion
Our study shows a significant association between bacterial vaginosis and preterm birth. We conclude that investigation for bacterial vaginosis and appropriate treatment should be a part of the routine examination of a pregnant woman.