Introduction

Bacterial vaginosis is a polymicrobial syndrome where the complex balance of microflora in the vagina is changed. There is a depletion of Lactobacilli and an overgrowth of facultative and anaerobic bacteria either singly or in combination. These bacteria include Gardnerella vaginalis, Bacteroides fragilis, Mobiluncus species, and many more others. This list continues to expand, and the specific causes of bacterial vaginosis are poorly understood1,2. It is an extremely common reproductive tract disorder worldwide accounting for one-third of vaginal infections3. It can occur in any age group but it is more prevalent in females of reproductive age groups worldwide4. It has been shown that nearly 5–10 million females every year seek gynecologic advice for vaginitis5.

The prevalence of bacterial vaginosis ranges between 8 and 75%6 with higher magnitudes in parts of Africa7. Commonly, the problem occurs among pregnant women, and studies have consistently shown BV to be a risk factor for adverse obstetric and gynecological outcomes such as pre-term labor and delivery8,9,10, premature rupture of membranes, and low birth weight11, spontaneous abortion12, postpartum infections such as endometritis13, and cesarean section wound infections14.

Regarding the associated factors, researchers reported that marital status15, residence2, multiparity15, history of abortion16,17, frequency of douching2,18 as well as multiple sexual partners15,16,17 are associated with bacterial vaginosis. However, the associated factors of bacterial vaginosis are poorly understood as reports have been conflicting. Some of the studies reported that socio-demographic, behavioral, and clinical characteristics of pregnant women are associated with bacterial vaginosis8,18,19,20,21, while another study reported no association with some of these factors22. Bacterial vaginosis is still prevalent among pregnant women. Since most cases remain asymptomatic, early detection is essential for timely treatment and prevention of related complications and poor pregnancy outcomes. Hence, screening of symptomatic and asymptomatic pregnant women by taking vaginal swabs plays a vital role15,23.

For many years, bacterial vaginosis has been given little attention. In recent years, however, the association of bacterial vaginosis with ascending genital tract infection on one hand and sexually transmitted diseases (STDs) on another has made the infection a major global problem as women with bacterial vaginosis are at increased risk for STDs24. Furthermore, it has been documented that bacterial vaginosis propagates viral replication25,26,27, and vaginal shedding of the HIV-1 and HSV-2 (herpes simplex virus-2)28 thereby further enhancing the spread of these viruses. In general, bacterial vaginosis has emerged as a public health problem due to its association with sexually transmitted infections3,29,30.

In low-income countries like Ethiopia, pregnant women on antenatal care are often screened for sexually transmitted infections (STIs) including syphilis and HIV but rarely other treatable STIs or BV. In some instances, pregnant women with vaginal discharge are treated for vaginal candidiasis without laboratory investigations2. To avoid pregnancy complications and poor outcomes due to BV, information on its prevalence and associated factors is important for treatment and prevention. Currently, there is limited information on the prevalence and associated factors of BV in Ethiopia; to our knowledge, a few studies were done among pregnant women16,17, and no study was done in West Shoa. Therefore, this study was conducted to determine the prevalence of bacterial vaginosis and associated factors among pregnant women receiving antenatal care at public hospitals in the West Shoa zone, Oromia region, Ethiopia.

Materials and methods

Study design and period

An institutional-based cross-sectional study was conducted from September 15 to December 14, 2021.

Description of study area

West Shoa is one of the 20 zones of the Oromia regional state which is located to the west of the capital city of Ethiopia, Addis Ababa. The study was conducted at selected four public hospitals (Ambo University Referral Hospital, Gedo General Hospital, Bako Primary Hospital, and Inchini Primary Hospital).

Source population

The source population for this study was all pregnant women attending ANC (antenatal care) clinics in the West Shoa zone.

Study population

The study population for this study was the selected pregnant women attending the ANC clinics of the selected hospitals in the west Shoa zone.

Inclusion and exclusion criteria

Pregnant women ≥ 18 years of age attending ANC clinics and willing to participate in the study were included. Pregnant women with vaginal bleeding, genital malignancy, who inserted some drugs in their vagina, have already ruptured membranes, with antepartum hemorrhage, and if for some reason(s) she was unable to consent despite the eligibility, and those on antibiotic treatment in the preceding two weeks were excluded.

Sample size determination and sampling techniques

For determining the sample size of pregnant women, a single population proportion formula was used with the following assumptions.

n = sample size;

Zα/2 = standard value for 95% confidence level of two sides normal = 1.96;

d = margin of error, 5% = 0.05;

P = prevalence rate = 19.4% = 0.194, was taken from the study conducted in Addis Ababa, Tikur Anbessa Specialized Hospital16;

$${\text{n}} = \frac{{\left( {{\text{z}}_{{{\upalpha }/2}} } \right)^{2} {\text{p}}\left( {1 - {\text{P}}} \right)}}{{{\text{d}}^{2} }} = \left( {1.96} \right)^{2} \times \frac{{0.194\left( {1 - 0.194} \right)}}{{0.05^{2} }} = 240.$$

Considering the 10% non-response rate, the final sample size became 264. The number of pregnant women was proportionally allocated to each hospital. Considering that the yearly attendants of ANC at the selected hospitals was 2044 (data from Zonal Health Department of 2012 E.C), and with the assumption that data collection would be completed within three months, the average attendants were estimated to become 511 pregnant women during data collection period where N = 511 and the calculated sample size, n = 264; giving K = 2. Hence, a systematic random sampling technique was used to include every other pregnant woman in the study (Fig. 1).

Fig. 1
figure 1

Proportional allocation of study participants and sampling technique.

Full size image

Data collection and laboratory investigations

Data were collected by midwives from the study participants using a pre-tested structured questionnaire that included socio-demographic characteristics, obstetric history and gynecological characteristics, hygienic behaviors/practices, sexual behaviors/practices, history of antibiotic use, history of previous medical conditions, and current situation of medical conditions. Upon admission to the study, midwives performed a clinical examination of each participant and recorded signs and symptoms of vaginal abnormalities such as the presence of vaginal discharge, vaginal itching, and vaginal burnings. The participants were also assessed by history for different medical conditions, like sexually transmitted diseases, urinary tract infections, hypertension, and diabetic mellitus.

After the interview, the midwives performed vaginal swabbing by instructing the participant to assume a lithotomy position, and a sterile disposable speculum was inserted into the vagina. A sterile cotton swab was used to swab the vaginal walls (lateral, anterior, and posterior fornices), and the swab was streaked on a clean grease-free slide for gram staining by trained Midwives and sent to the respective hospital laboratories for staining, microscopic examination, and interpretations of the results by Medical Laboratory Professionals.

For the diagnosis of bacterial vaginosis, the smeared vaginal slides were air dried, heat fixed, and then gram-stained and examined under an oil immersion objective (100× magnification), and it was graded as per standardized, quantitative, morphological classification method developed by Nugent et al.31. The method involves assigning a score between 0 and 10 based on the quantitative assessment of the Gram stain for three different bacterial morphotypes:

  • Large Gram-positive rods (indicative of Lactobacillus spp),

  • Small Gram-negative or variable rods (indicative of Gardnerella, Bacteroides, and other anaerobic bacteria), and

  • Curved, Gram-variable rods (indicative of Mobiluncus spp).

Each morphotype was quantified from 1 to 4 + concerning the number of morphotypes per oil immersion field (zero, no morphotypes; 1+, less than 1 morphotypes; 2+, 1 to 4 morphotypes; 3+, 5 to 30 morphotypes; 4+, 30 or more morphotypes). Then, the results of the three different morphotypes were added to give the final score. The scores between 0 and 3 represented normal vaginal flora; between 4 and 6, intermediate vaginal flora; and scores between 7 and 10 were considered diagnostic for bacterial vaginosis (BV). In this study, the microbiological definition of BV was a score of 7–10 according to the Nugent’s criteria. As previously described by Shayo and others22, two experienced medical laboratory professionals were recruited at each study hospital to score the vaginal smears independently, and in case of discrepancy, a third opinion was sought for confirmation and her/his opinion was the final.

Data quality assurance and management

Data quality was ensured through the use of pre-tested tools for data collection, proper data collection, and processing of all activities. Training was given to midwives on specimen collection, and smear preparation. Every activity in the laboratory was done in adherence to standard operational procedures. To ensure good and uniform practice in the preparation and reading of gram stains, Medical laboratory technologists were trained on Nugent scoring before the study.

Data processing and analysis

After data collection, each questionnaire was checked for completeness, and missing, and edited for other errors. Data was entered into an Excel spreadsheet and exported to STATA-14 for analysis. Data were organized and presented using tables and graphs. Binary and multivariable logistic regressions were performed to examine the presence of association between a dependent variable and independent variables. Data were presented using Odds ratios (OR) and their 95% confidence intervals (CIs). Variable with a P value ≤ 0.25 at the bivariate logistic regression model were entered into a multivariate logistic regression model. Finally, a multivariate logistic regression model was carried out to identify predictors of bacterial vaginosis. AOR was used to explain the strength of the association, and a P value ≤ 0.05 was considered statistically significant.

Results

Socio-demographic characteristics of study subjects

This study was conducted in four hospitals; two primary, one general, and one referral hospital. The response rate was 98.5% (260 out of 264); four participants were excluded because of poor smear quality, and hence, data presentation, analysis, and interpretations were performed on a total of 260 pregnant women attending antenatal care. The mean age of the participants was 25.32 years (standard deviation = 0.33). The majority of the study participants, 54.6% (142/260), were in the age category of 21–29 years. By religion, Protestants were dominating, 64.23% (167/260). Most of the participants, 60% (156/260) were housewives by occupation. Rural and urban residents accounted for 51.54% and 48.46%, respectively (Table 1).

Table 1 Socio-demographic characteristics of study subjects.
Full size table

Obstetric history and gynecological characteristics of study subjects

The majority of the study participants, 65% (169/260) were multigravida. Regarding parity, over half of the participants, 56.92% (148/260), were with low multiparity (1–4 deliveries). By gestational age, study participants in the 2nd trimester accounted for 45.77% (119/260) followed by the 3rd trimester, which accounted for 39.23% (102/260). Only 15.38% (40/260) of the participants had a history of abortion; of which 80% (32/40) encountered it only once while the rest 20% (8/40) had it twice. Stillbirth was reported by 7.31% (19/260) of the participants (Table 2).

Table 2 Obstetric history and gynecological characteristics of study subjects.
Full size table

Hygienic behaviors and practices of study subjects

More than three-fourths of the study participants, 76.54% (199/260), reported douching their vagina more than once a day. Of the participants, 42.31% (110/260) used soap for douching. Fifteen percent of the study subjects reported using disinfectants for washing their pants, while the majority of them (70%; 182/260) wore only a single pant for a week. Regarding frequency of showering, 55.38% (144/260) of the subjects took only once per week (Table 3).

Table 3 Hygienic behaviors and practices of study subjects.
Full size table

Sexual behaviors and practices of study subjects

The sexual behaviors and practices of study participants are depicted in Table 4. Among the study participants, a few, 2.31% (6/260), had two or more sexual partners in the last 12 months. Almost, two-thirds of the participants (65.38%) have had only one sexual partner in their lifetime. Regarding condom use, 77.69%, 17.69%, and 4.62% of the participants had not used, used sometimes, and used usually, respectively.

Table 4 Sexual behaviors and practices of study subjects.
Full size table

Previous and current medical conditions of study subjects

Among the study participants, only 11.15%, 11.54%, and 2.69% reported having sexually transmitted diseases, urinary tract infections, and diabetes mellitus, respectively. History of antibiotic use was reported in 7.31% of the participants. During the study, the presence of vaginal discharge, itching, and burning were reported by 28.46%, 17.69%, and 18.08% of the participants, respectively. Of the study participants, 11.54% were presented with symptomatic STD, 8.08% with hypertension, and 2.69% with diabetic mellitus during the study. The majority of the participants knew their HIV serostatus; 83.85% (218/260) were non-reactive and 3.85% (10/260) were reactive for HIV, while 12.31% (32/260) were with unknown serostatus (Table 5).

Table 5 Previous and current medical conditions of study subjects.
Full size table

Prevalence of bacterial vaginosis among the study subjects

According to the Nugent scoring criteria, 22.3% (58/260) of the participants were with depleted vaginal flora (score of 7–10), 25% (65/260) were with intermediate vaginal flora (score of 4–6) and 52.7% (137/260) were with normal vaginal flora (score of 0–3). Both intermediate and normal vaginal floras were interpreted as negative for bacterial vaginosis according to the Nugent scoring system. The prevalence of bacterial vaginosis was, therefore, 22.3% (95% CI 17.4 to 27.9%) (Fig. 2).

Fig. 2
figure 2

Prevalence of bacterial vaginosis among study subjects.

Full size image

Factors associated with bacterial vaginosis

For the statistical analysis using logistic regression models, variables were segregated according to their relatedness: socio-demographic variables, obstetric characteristics, hygienic behaviors and practices, sexual behaviors, and practices as well as previous and current medical conditions. In each category, binary logistic regression analysis was performed, and variables with a P value of ≤ 0.25 were selected as candidates for multiple logistic regression analysis (Table 6).

Table 6 Binary and multiple logistic regression analysis results of factors associated with bacterial vaginosis among pregnant women.
Full size table

In the binary logistic regression analysis of socio-demographic characteristics, the age category of ≥ 30 years old (P = 0.095), marital status other than married (P = 0.161), educational level of ‘able to read and write’ (P = 0.053), occupation category of trader (P = 0.244), and residence (P = 0.081) were found candidates for multiple logistic regression analysis. In the multiple logistic regressions, marital status other than married women and residence were found predictors of bacterial vaginosis. The odds of having bacterial vaginosis among marital status other than married were reduced by 74% (AOR = 0.260, 95% CI 0.068 to 0.9995; P = 0.05) as compared with married pregnant women. The odds of having bacterial vaginosis among rural residents was 2.1 (AOR: 2.1 (95% CI 1.05 to 4.24); P = 0.036) times the odds of having bacterial vaginosis among urban residents.

Concerning the obstetric and gynecologic characteristics, gravidity (multigravida, P = 0.052), parity (low multiparity, P = 0.207; High multiparity, P = 0.204), and history of abortion (P = 0.207) were found candidates for multiple logistic regressions in binary logistic regressions, but none of them were predictors of bacterial vaginosis.

In the binary logistic regression of hygienic behaviors and practices, vaginal douching (more than once per day, P = 0.060), washing pants with disinfectants (P = 0.076), and number of pants worn per week (one pant per week, P = 0.038; two or more pants per week, P = 0.003) were candidates for multiple logistic regression model. In the multiple logistic regression analysis, the number of pants was found predictor for bacterial vaginosis. The odds of having bacterial vaginosis among those wearing one pant per week was 2.7 (AOR = 2.7, 95% CI 1.04 to 7.20; P = 0.041) times the odds of having bacterial vaginosis among those not wearing pants. Those wearing two or more pants per week were 4.5 more likely to develop bacterial vaginosis as compared with those not wearing pants (AOR = 4.5, 95% CI 1.49 to 16.57; P = 0.009).

Of the sexual behavior and practice variables, number of lifetime sexual partners (2–4 partners in life, P = 0.235 and 5 or more partners in life, P = 0.071), and condom use (not using at all, P = 0.03 and using sometimes, P = 0.040) were found candidates for multiple logistic regression analysis. However, these variables were found insignificant predictors of bacterial vaginosis. Again, none of the variables related to the previous and current medical conditions were found significant in the multiple logistic regressions although history of urinary tract infection, history of antibiotics use, presence of vaginal discharge, and hypertension were candidates in the binary logistic regression analysis with P values of 0.128, 0.121, 0.250, and 0.211, respectively.

Discussion

In the current study, the prevalence of bacterial vaginosis among pregnant women using the Nugent scoring system was 22.3% (95% CI 17.4 to 27.9%). This result is comparable with the prevalence of bacterial vaginosis recently reported from Harar City, Eastern Ethiopia, 20.6%17. It is also in line with the previous report with the prevalence of 19.4% for bacterial vaginosis from Addis Ababa16. Comparable prevalences of bacterial vaginosis with our recent result were also reported from other African countries, such as Kenya, 19.3%32, Northeastern Nigeria, 17.3%33, South Africa, 17.7%34, and Cameroon, 26%2. The consistency of our findings with these reports may be due to the consideration of similar study participants. Our current result is also comparable with the report from India, 23%23. The similarities could be suggested that there may be related lifestyles, and hygienic behaviors/practices among Ethiopian and Indian pregnant women.

The detection rate of bacterial vaginosis among pregnant women was 49.8% in Sudan35. This result is much higher than the result of our current study though similar study subjects were included, and the same study design was employed. The variations could suggest that there might be differences in hygienic behaviors and practices. A higher prevalence of bacterial vaginosis (30.9%) was also reported in Ghana36 compared with our current result (22.3%). The inconsistencies could be due to the utilization of a larger sample size with convenient sampling techniques in the study from Ghana while a smaller sample size with systematic random sampling was employed in our study. As a study from Southeastern Nigeria indicated, a higher prevalence was reported, 38%37; this could be due to a smaller sample of the study from southeastern Nigeria. Compared with our current result, higher magnitudes of bacterial vaginosis were also reported from Tanzania and Zimbabwe, 28.5% and 32.6%, respectively22,38. The differences might be the inclusion of study subjects during their 3rd trimester only in the study from Zimbabwe and during delivery in the case of Tanzania while our study included all gestational ages of women during antenatal follow-up.

In contrast to our current study, a slightly smaller prevalence of bacterial vaginosis, 16%, was reported from southwestern Nigeria15. This variation might be due to the consecutive recruitment of a larger number of study subjects in the case of southwestern Nigeria while a relatively smaller sample size was employed using systematic random sampling in the current study, in addition to geographical and/or hygienic variations. The prevalence of bacterial vaginosis in Portugal was 3.88%39, which is much smaller than our report. The inconsistent results might be population and/or geographical differences.

Our study revealed that the marital status of pregnant women other than married were at reduced risk of bacterial vaginosis (AOR = 0.260, 95% CI 0.068 to 0.9995; P = 0.05) as compared with married pregnant women. In agreement with our finding, a statistically significant association was found between marital status and bacterial vaginosis according to a report from Southwestern Nigeria15. However, contrasting findings were reported from eastern Ethiopia17, Addis Ababa18, Ghana36, Cameroon2 and Tanzania22. Our result also showed that rural residents had 2.1 (AOR = 2.1, 95% CI 1.05 to 4.24; P = 0.036) times more risk of developing bacterial vaginosis as compared with urban residents. In line with this, a study from Cameroon indicated that bacterial vaginosis was significantly higher in women from rural parts than those from urban areas2. On the contrary, residence was not associated with bacterial vaginosis as reported from eastern Ethiopia17 and Tanzania22. The variations might be due to the differences in the lifestyle of the participants from different areas.

Obstetric and gynecologic variables considered in our study, such as gravidity, parity, gestational age, history of abortion, and history of stillbirth were not significantly associated with bacterial vaginosis. In line with our findings, no significant associations were revealed between gravidity and the prevalence of bacterial vaginosis from India40 and Cameroon2. But, a study from Ghana reported that high multigravidas were at reduced risk of bacterial vaginosis36. In agreement with the current study, no association of parity with bacterial vaginosis was reported from Cameroon2. On the other hand, a study from southwestern Nigeria reported that multiparity was significantly associated with bacterial vaginosis15. Regarding the gestational age, contrasting results were reported to our finding. From Ghana, the study reported that 3rd trimester was at a reduced risk of bacterial vaginosis36, while the study from South Africa revealed that both 2nd and 3rd trimesters were associated with a reduced risk of the disease as compared with the 1st trimester34. Conflicting with these studies34,36, other studies from Cameroon2 and Northeastern Nigeria33, reported that 2nd trimester was at increased risk of bacterial vaginosis. Consistent with our recent study, a history of abortion was not significantly associated with bacterial vaginosis as reported by Bitew and others from Addis Ababa18, and Ghana36. However, another study from Addis Ababa16 and eastern Ethiopia17 indicated that a history of abortion was significantly associated with bacterial vaginosis.

In the current study, hygienic behaviors and practices of pregnant women, such as frequency of vaginal douching per day, use of soup for douching, washing pants with disinfectants, and showering frequency per week were not significantly associated with bacterial vaginosis. However, the number of pants worn per week was found significantly associated with bacterial vaginosis. Wearing one pant for a week was 2.7 (AOR = 2.7, 95% CI 1.04 to 7.20; P = 0.041) times more associated with developing bacterial vaginosis as compared with those not wearing. Again, the risk of developing bacterial vaginosis also increased 5 times (AOR = 4.96, 95% CI 1.49 to 16.57; P = 0.009) among those wearing two or more pants per week as compared with those not wearing. This could be explained that the chemical ingredients of cleansing agents used to wash the pants may affect the vaginal flora as frequently exposed to chemical residues. In agreement with our result, douching was not significantly associated with bacterial vaginosis according to a study from Ghana36. Inconsistent with our finding, a study from Cameroon reported significantly higher bacterial vaginosis in those practicing douching2, and frequent douching (≥ 4 per day) was significantly associated with bacterial vaginosis as a report from Addis Ababa, Ethiopia18. Like our result, a Cameroon study reported that the use of soap for douching was not significantly associated with bacterial vaginosis2. In contrast to our study, participants who did not wash their pants with disinfectants were significantly infected with bacterial vaginosis as reported from Cameroon2.

In our study, the sexual behavior/practice variables, such as the number of sexual partners in the last 12 months, lifetime sexual partners, and use of condoms were found insignificant predictors of bacterial vaginosis. Similar to our finding, a study from Addis Ababa by Bitew and colleagues reported that the number of lifetime sexual partners was not significantly associated with bacterial vaginosis18. However, other studies from Ethiopia, Addis Ababa16, and Harar City17, reported that multiple sexual partners were significantly associated with bacterial vaginosis, and a study from southwestern Nigeria15 also supported these findings opposing our result. This could be suggested that bacterial vaginosis may be enhanced through frequent sexual activities as it can disturb the vaginal flora.

As any study may not be free from limitation(s), our study has a limitation. Clinical diagnosis of bacterial vaginosis using ‘Amsel’s Criteria’ was not performed because a kit for determining vaginal pH was not found in the market during the purchasing process of materials and reagents for this research project. However, it had little effect on our work since the ‘Nugent scoring criteria’, a gold standard method for the diagnosis of bacterial vaginosis, was used, which was one of the strengths of our study. Hence, the objectives of our study have been answered in line with the plan. The other strength was that our study was a multicenter study, conducted in four hospitals which were selected from eight hospitals found in the West Shoa zone. Furthermore, a random sampling technique was employed to recruit study participants allowing the generalizability of the findings of our study.

Conclusions

The current study revealed a high prevalence of bacterial vaginosis among pregnant women attending antenatal care in the study settings. Pregnant women in rural areas were at increased risk of bacterial vaginosis. Marital status other than married was significantly associated with reduced bacterial vaginosis. Considering the high prevalence of bacterial vaginosis, we recommend that pregnant women attending antenatal care should be screened and treated to prevent possible adverse outcomes. Hence, screening for the disease should be integrated into the recommended basic laboratory investigations during antenatal visits.