Key-words: Adverse pregnancy outcome - bacterial vaginosis - persistent infection
Introduction
Bacterial vaginosis (BV) is a polymicrobial vaginal infection due to an imbalance in the normal vaginal flora characterized by a reduction in the concentration of the dominant hydrogen peroxide-producing lactobacilli and an increase of anaerobic organisms in the vagina.[[1 ]],[[2 ]] BV is a common cause of vaginitis in both pregnant and nonpregnant women[[1 ]],[[2 ]],[[3 ]] and its prevalence among pregnant women in sub-Saharan Africa ranged from 20% to 50%.[[4 ]] Etiological organisms in BV include Gardnerella vaginalis, Mycoplasma hominis, Prevotella, and various species of Mobiluncus, Bacteroides, Fusobacterium, Veillonella, Propionibacterium, Bifidobacterium , and Eubacterium .[[2 ]],[[5 ]],[[6 ]] Hydrogen peroxide-producing lactobacilli appear to be important in preventing overgrowth of the vaginal anaerobes; however, an imbalance in the vaginal flora causes the anaerobes to produce large amounts of proteolytic carboxylase enzymes which break down vaginal peptides into amines that are volatile and malodourous. The resultant vaginal discharge in symptomatic cases present as malodourous, thin vaginal discharge which can be grayish to homogenous-white in appearance.[[6 ]] Available evidence suggests that BV in the lower genital tract is associated with infections of the upper genital tract and the microorganisms may ascend to the amniotic cavity resulting in inflammation of the decidua and chorioamnion.[[7 ]] This can result in sequelae such as premature rupture of membranes (PROM), preterm labor, preterm delivery, low birth weight, increased vertical transmission of HIV, and postpartum endometritis.[[1 ]],[[5 ]],[[7 ]],[[8 ]],[[9 ]],[[10 ]],[[11 ]] In addition, these adverse pregnancy outcomes persist despite treatment raising the question about the possible explanation for these events. Women of African descent are at a three-fold increased risk of BV;[[12 ]] however, there is paucity of data in many sub-Sahara Africa countries while there is no consensus on routine screening and treatment protocols due to insufficient evidence.[[13 ]] This study aimed at evaluating effect of treatment on pregnancy outcomes and the contribution of persistent BV infection to these pregnancy outcomes.
Materials and Methods
Study design
This was a prospective cohort study.
Study setting
This study was conducted in a tertiary facility in North-Central Nigeria.
Study participants
Participants were consenting, asymptomatic, antenatal clinic attendees receiving antenatal care services at the study site.
Inclusion criteria
Booked pregnant women in the second trimester of pregnancy without background medical disorders were included in the study.
Exclusion criteria
Unbooked women, those who used antibiotics within 2 weeks prior to recruitment, and those with underlying chronic medical disorders were excluded from the study.
Sampling method and sample size determination
Purposive sampling method was used and the sample size was calculated using the formula:[[14 ]]
where n = sample size
z = standard normal deviation (a constant which is 1.96 at 95% confidence interval [CI])
p = known prevalence of BV in pregnancy, i.e., 0.25 (25%)[[8 ]]
d = observed difference at 0.05 (5%) level of significance.
q = 1 − p = 1 − 0.25 = 0.75
Making provision for attrition rate of 10% (28), the total sample size was 316.
Study procedure
Antenatal clinic attendees were counseled about the study and those willing to participate were evaluated for eligibility using the inclusion and exclusion criteria. Informed consent was obtained from consenting eligible women after which the study questionnaire was administered. Thereafter, a sterile speculum examination was performed during which high vaginal swab samples were collected from the posterior fornix using a sterile swab sticks; a smear was made on a glass slide by rolling the swab stick on it and allowed to air dry followed by Gram staining and microscopy. The laboratory processing included wet sample preparations, microscopy, fixation with ethanol, and routine Gram staining followed by Nugent scoring.[[15 ]] A total Nugent score of 7 or more was diagnostic of BV, 4–6 was intermediate, while 0–3 was normal. All BV positive women were treated with a course of oral metronidazole 400 mg thrice daily for 7 days followed by a posttreatment repeat Gram staining for Nugent scoring 4 weeks after the treatment. All participants were followed up at the antenatal clinic until delivery.
Study outcome measures
The primary outcome measure was pregnancy outcome among the BV-positive and BV-negative women; the secondary outcomes were posttreatment laboratory testing result and comparative pregnancy outcome among women with resolution versus those with persistent BV infection.
Ethical issues and data management
Ethical approval was obtained from the ethical review committee of a teaching hospital (ERC/PAN/2016/01/1487) before the commencement of the study while informed written consent was obtained from each participant. Data analysis was performed using the Statistical Package for the Social Sciences software (SPSS version 21.0) (IBM, Armonk, NY, USA). Association between discrete variables was evaluated with Chi-square with level of significance set at P < 0.05 while structured t -test was used for continuous variables.
Results
Among the 316 women screened for BV in the study, 76 were BV positive with a prevalence of 24.1%. From [[Table 1 ]], the biosocial and obstetric parameters showed a statistically significant difference in the age (P = 0.001), type of marriage, and parity (P = 0.001) among the participants. However, the occupation of participants (P = 0.166), level of education (P = 0.384), gestational age at recruitment (P = 0.312), and the social class (P = 0.398) were not statistically significant.
Table 1: Comparison of biosocial and obstetric parameters among participants
From [[Table 2 ]], the observed significant clinical features for BV-positive compared to BV-negative women were vulva itching (odds ratio [OR]: 1.958, 95% CI: 1.161–6.463; P = 0.011) and vaginal douching (OR: 2.889, 95% CI: 1.658–15.503; P = 0.001). However, vaginal discharge (OR: 2.126, 95% CI: 0.861–5.253; P = 0.096), lower abdominal pain (OR: 1.176, 95% CI: 0.574–2.411; P = 0.657), and fever (OR: 0.539, 95% CI: 0.154–1.892; P = 0.479) were not significant.
Table 2: Clinical features observed at recruitment of participants
From [[Table 3 ]], comparison of pregnancy outcome in BV-positive to BV-negative women showed statistically significant occurrence of PROM (15 vs. 7, OR: 8.185, 95% CI: 3.196–20.962; P = 0.005), preterm delivery (18 vs. 3, OR: 24.517, 95% CI: 6.985–86.049; P < 0.001), birth weight <2500 g (18 vs. 11, OR: 6.460, 95% CI: 2.893–14.429; P = 0.005), and mean birth weight (2925.00 ± 503.09 vs. 3072.29 ± 396.55; P = 0.009). There was no statistical significant difference in Apgar score at the 1st min (OR: 2.486, 95% CI: 0.834–7.506; P = 0.172) as well as the need for active neonatal resuscitation (6 vs. 7, OR: 2.853, 95% CI: 0.928–8.768; P = 0.116).
Table 3: Pregnancy outcome among bacterial vaginosis-positive compared to bacterial vaginosis-negative women
From [[Table 4 ]], posttreatment BV persistence was 25.0% (19 out of 76). Women with posttreatment persistence of BV infection had significantly higher occurrences of PROM (11 vs. 4, OR: 18.21, 95% CI: 4.654–71.316; P = 0.001), preterm delivery (OR: 14.57, 95% CI: 4.138–51.308; P = 0.001), birth weight <2500 g (OR: 14.57; 95% CI: 4.138–51.308; P = 0.001), and low 1st min Apgar scores (OR: 7.33, 95% CI: 1.223–43.960; P = 0.049) compared to women with resolution of infection.
Table 4: Comparison of posttreatment pregnancy outcome among participants with persistence or resolution of BV infection
Discussion
From this study, the prevalence of BV in pregnancy was 24.1% among apparently asymptomatic women. Some BV-positive women did not have clinical features while other BV-negative women had clinical features. Despite treatment, adverse pregnancy outcomes were higher among BV-positive women with significant PROM, preterm delivery, birth weight <2500 g, and lower mean birth weight compared to BV-negative women. Posttreatment BV persistence was 25.0% while adverse pregnancy outcomes were significantly higher among women with persistence compared to those with resolution of infection.
The prevalence of BV in pregnancy varies widely with values of 25%[[8 ]] from Maiduguri and 26%[[16 ]] from Lagos both in Nigeria, 20.5%[[7 ]] from India, and 28.1%[[11 ]] from Brazil which compares to this study report. However, lower figures of 19.4% from Ethiopia[[9 ]] and 16%[[10 ]] from Denmark and a higher report of 37.3%[[17 ]] from South Africa have also been reported. However, a meta-analysis estimated 20%–50% prevalence for BV in pregnancy for sub-Saharan Africa.[[14 ]] This variation may be due to the differences in methodology across the studies ranging from the sociodemographic characteristics of the participants to the method of laboratory testing.[[5 ]],[[7 ]],[[8 ]],[[9 ]],[[10 ]],[[15 ]],[[16 ]]
There is no consensus on the approach to BV in pregnancy; the divide has been the choice to focus on asymptomatic or symptomatic women. A report indicated a high burden of asymptomatic infection among HIV uninfected women[[4 ]] while another showed that the use of clinical approach will result in asymptomatic women going untreated.[[18 ]] This forms the bedrock of the suggestion for a universal testing for BV among pregnant women, especially since asymptomatic disease poses similar risks as symptomatic disease. Many studies on BV which recruited apparently asymptomatic women documented clinical features of which the women were either unaware or did not consider worthy of reporting. In addition, there were women without these features who tested positive while a number with the features tested negative. This observation limits the reliance on the symptom-based approach to BV testing in pregnancy. Although this study recruited apparently asymptomatic women, evaluation showed that vulvar itching, vaginal discharge, fever, lower abdominal pain, and vaginal douching were present in a number of both BV-positive and BV-negative women similar to previous reports. Vulvar itching was recorded among 17%[[5 ]] and 17.2%[[16 ]] of asymptomatic women in two previous BV studies. Douching has been shown to cause disequilibrium in the vaginal microbiota or to induce inflammation through physical or chemical irritation which predisposes women to BV.[[19 ]] It was reported to be a significant factor for BV similar to a report from Canada.[[20 ]] In a study among women who presented with vaginal infections at a facility in Egypt, participating women viewed douching as a religious obligation (88.9%) or a form of personal cleanliness (80.6%), but women who douched were reported to be predisposed to reproductive health hazards including pelvic inflammatory diseases.[[21 ]] Another report among asymptomatic women in South Africa[[17 ]] showed that 52.9% of participants had vaginal discharge on examination similar to this study. This further raises a concern about the clinical approach to BV management in pregnancy which is based on reported symptoms by the parturient woman before laboratory testing. The clinical approach therefore has the potential to miss out asymptomatic women who are at risk of possible negative pregnancy outcome.
A randomized control trial showed the benefit of treatment with significantly fewer miscarriages and preterm deliveries among women who received antibiotic treatment as against those that received placebos for BV.[[22 ]] However, the need to screen, the necessity of posttreatment evaluation, and management of posttreatment persistent BV remain without consensus. Oral or vaginal metronidazole or clindamycin has been recommended for treatment of BV,[[23 ]] but the CDC recommended them for only symptomatic women.[[24 ]] However, no emphasis was placed on posttreatment microbiology testing for persistent infection and its management. Metronidazole has been shown to be active in vivo with suggestions that BV-related organisms may be sensitive to the hydroxymetabolite of metronidazole or the drug may exert an indirect action through synergism by killing anaerobes that provide substrates to BV-related organisms such as G. vaginalis or Atopobium vaginae .[[25 ]] Metronidazole was used to treat smear positive participants in this study, but a posttreatment repeat smear showed a 25% persistence rate. The available data on posttreatment BV testing were among nonpregnant women, and it showed persistence rates of 23%, 43%, and 58% at 1, 3, and 12 months posttreatment, respectively.[[25 ]] While researchers have tried to address recurrent BV infection, the literature is sparse of studies on persistent BV infection. In a case series with women described as having refractory BV, a report showed that prebiotic lactoferrin therapy significantly improved the vaginal bacteria flora restoring Lactobacillus as the dominant bacteria at 1 month after treatment with subsequent term live births.[[26 ]] However, while experience with prebiotic lactoferrin in pregnancy is limited, this further heightens the need to address this category of patients as well as conduct further research aimed at formulating a management guideline.
BV in pregnancy is a risk to both term and preterm pregnancies and has been associated with adverse obstetric outcomes in many studies[[5 ]],[[6 ]],[[7 ]],[[8 ]],[[9 ]],[[10 ]],[[11 ]] similar to this study. This study corroborated previous significant reports of increased PROM[[6 ]],[[16 ]] and preterm delivery[[6 ]],[[10 ]],[[16 ]] from BV. The mean gestational age at delivery in this study was not statistically significant unlike a similar study in Bangladesh which reported a statistical significant reduction in the mean gestational age at delivery following BV infection.[[27 ]] This may be due to difference in methodology as the Bangladeshi study comprised symptomatic women with complaints of abnormal vaginal discharge at recruitment. Low birth weight has been documented to be high among babies delivered by women with BV in pregnancy in previous[[8 ]],[[16 ]] and index study. This may particularly be due to the higher preterm delivery rate among women with BV reported by all these studies.
It has been shown that despite antibiotic treatment, pregnancy outcome among BV-positive women remained worse compared to uninfected women. While the reason for the worse outcome has not been clarified, posttreatment persistence infection in pregnancy has remained unexplored. After laboratory diagnosis and appropriate antibiotic treatment for BV, this study conducted a repeat posttreatment microbiological testing which showed a 25% persistence infection. A subanalysis of the women treated with BV in this study into those with persistence (posttreatment smear positive) and resolution (posttreatment smear negative) of infection showed worse pregnancy outcome among women with persistent infection. Women with posttreatment persistent BV infection had significantly worse pregnancy outcomes characterized by higher occurrence of PROM, preterm delivery, low birth weight, and low 1st min Apgar scores at birth compared to women with resolution of infection. This suggests that the adverse pregnancy outcomes of BV despite treatment are most likely a resultant effect of women with persistent posttreatment infection. This brings to the fore a suggestion for posttreatment screening for BV in pregnancy to identify persistent infections coupled with further research toward formulation of a treatment guideline for such women. This may need to explore the possible roles of repeat treatment or the use of other agents aimed at achieving resolution of the persistent infection. It may be reasonably expected that the eradication of the persistence infection may improve the pregnancy outcome among such women, thereby reducing the adverse effect of BV on pregnancy outcome.
Conclusion
This study concludes that the symptom-based approach to BV screening in pregnancy excludes asymptomatic women. Persistent posttreatment BV infection is not uncommon and appears to be a major contributor to the adverse pregnancy outcomes associated with BV infection. Based on the results of this study, routine BV screening in pregnancy, antimicrobial treatment of BV-positive women, posttreatment repeat testing and repeat treatment for persistent infections are recommended to improve pregnancy outcome. However, further research activities are recommended to obtain further evidence to enable the formulation of an evidence-based management guideline for the management of persistent posttreatment BV infection in pregnancy.
Strengths and limitations of the study
The recruitment of asymptomatic women, pre- and post-treatment laboratory testing for BV, and the comparative analysis of pregnancy outcome among women with resolution to those with persistent infection sought to provide additional information on the relationship of BV to adverse pregnancy outcome. The study was however limited by the sample size and restricted geographical coverage.
