Introduction

Bacterial vaginosis (BV), also called vaginal bacteriosis, is the principal reason for vaginal discharge, irritation, and odor in childbearing-aged women1,2. However, in most cases (about 50%), BV is often asymptomatic3. The global estimated prevalence of BV ranges from 23 to 29%4. In non-pregnant and pregnant Iranian women, BV prevalence is 28% and 16%, respectively5. A dysbiosis of the vaginal flora from the dominant Lactobacillus spp. to a mix of G. vaginalis, Bacteroides spp., Mobiluncus spp., and Mycoplasma hominis, characterizes BV6. BV may persist after treatment, and symptoms return in 50% of women within 12 months7. BV infection promotes the increased chance of pelvic inflammatory disorder, preterm childbirth, postpartum endometritis, and increased risk of other infections like Trichomonas vaginalis, Chlamydia trachomatis, and Vulvovaginal candidiasis8. The specific cause of BV is not clear yet, but several risk factors have been characterized, including smoking, vaginal douching, recent antibiotics usage, using intrauterine devices, and frequent sexual contacts9. Despite the adverse reproductive health outcomes of BV, there is little information about the relation between BV and dietary factors.

Vaginal flora imbalances may occur as a result of dietary intakes. Previous studies have suggested a potential relationship between BV and nutrients10,11,12,13,14. In Neggers et al. study, dietary fat was connected with an increased risk of BV11. According to Thoma et al., the dietary glycemic load was correlated with BV development and persistence10. Dietary intake is a multidimensional exposure. When it comes to addressing the multicollinearity of foods and understanding the relationship of dietary intakes to chronic disease, researchers increasingly look at the diet as a whole15. Therefore, using factor analysis to determine dietary patterns is a functional dietary assessment approach16. The 2015 Dietary Guidelines Advisory Committee defined dietary patterns as follows: "the quantities, proportions, variety or combination of different foods and drinks in diets and the frequency with which they are consumed"17.

To our knowledge, no earlier observational research has examined the association between dietary patterns and BV. Since considering foods alone does not show the synergistic effects of multiple diet components18, this research intended to examine the link between dietary patterns and BV.

Methods

Ethical considerations

This hospital-based case–control study recruited 148 women with bacterial vaginosis as cases and 153 healthy women from November 2020 until June 2021. The study was reviewed and authorized by the National Nutrition and Food Technology Research Institute, Shahid Beheshti University of Medical Sciences, Tehran, Iran. Participants signed the informed consent before inclusion. All procedures were conducted according to the latest version of the Helsinki Declaration (The ethics committee code: IR.SBMU.NNFTRI.REC.1399.054).

Sample size calculation

According to the primary objective, dietary pattern exposure was used to compute the required sample size. The available data indicated that 73% of Iranian adults had adherence to unhealthy dietary patterns19. The odds ratio (OR) for BV was postulated 2.5 among women with unhealthy dietary patterns relative to non-adherence. A total of 125 BV-affected women (cases) and 125 healthy women (controls) were calculated to attain 80% statistical power for such effect size at a 5% level of significance20. To compensate for over- and underestimation of energy intake or withdrawal, 316 participants were recruited.

Study population

Eligible participants were selected by convenience sampling method among the women referred to the gynecology clinic at Imam Hossein Hospital in Tehran, Iran. A gynecologist examined all patients to evaluate for bacterial vaginosis. Incident cases were diagnosed with a first BV according to the Amsel criteria with at least three of the four determining criteria: a homogeneous and dilute vaginal discharge, vaginal pH greater than 4.5, fish odor after adding 10% potassium hydroxide to the discharge slide, and the presence of 20% of clue cells under saline microscopy21.

Eligible participants were 18–45 years old, non-lactating, non-pregnant, not menopause and not suffering from systemic immunity diseases, chronic infection, chronic diet-related diseases (cancer, diabetes, cardiovascular disease, etc.), or any disease in the uterine cavity such as polyp and fibroids as well as lack of hysterectomy. Also, they did not consume vaginal douches, antibiotics, hormonal contraceptives, probiotics, and immunosuppressive drugs. Participants were included in the control group upon the absence of ongoing or previous BV. The exclusion criteria for both BV-affected and healthy women included those who had reported energy intakes outside of the range of ± 3 standard deviation (SD) from the average energy intakes of the study population and were unable to respond to the questions.

Socio-demographic assessment

A data collection form was developed to gather data on age, history of medication and supplementation, BV family history, education level, occupational status, smoking, number of sex partners, and monthly family income. Due to religious and cultural beliefs in Iranian society, questions regarding alcohol and opium were not determined.

Non-dietary exposure assessment

Anthropometric measures, including weight, height, and waist circumference (WC) were performed following standard protocol by the one qualified examiner for all participants to avoid random observer error. Bodyweight (in kg to the nearest 100 g) was measured with a reliable scale while participants wore light clothing and without shoes. To assess central adiposity, WC (in cm to the nearest 0.1 cm) was measured by an unstretched tape measure at the umbilical site, at the abdominal level, on light clothing, in a standing position, without any pressure on the body surface. Body height (in cm to the nearest 1 mm) was measured by standing and straight tape. The participants were asked to stand straight while the shoulders were in a normal position, without shoes, and with the heels together. Next, the body mass index (BMI) was computed by dividing weight by height squared (square meters). BMI and WC were categorized according to World Health Organization cut-off criteria22 and the first report of the Iranian National Committee on Obesity23, respectively. Physical activity assessment was based on the International Physical Activity Questionnaire (IPAQ), and its validity and stability have been examined in a previous study in Iran24.

Dietary intake assessment

Individuals' typical diets were gathered with a valid and reliable semi-quantitative food frequency questionnaire (FFQ) developed and validated in the Iranian population25. Participants were asked about their usual intakes over the past year before diagnosing BV for cases and the previous year before the interview in controls. This FFQ has been designed according to the Willet method18, including 168 food items with a standard serving size for each food item. The average size of each food item was explained to the participants in the interview. Then, they reported the frequency of consumption of each food item given serving on a daily, weekly, or monthly basis. The mentioned values of each food were converted to grams using the household scale guide. The mean daily intakes of energy and nutrients were determined using the Iranian food composition table26 and the USDA food composition table27.

Statistical analysis

Data were analyzed using IBM SPSS Software version 20.0 (SPSS, Inc.). All hypothesis tests were 2-tailed, with a P value less than 0.05 considered significant. Normal distribution of continuous variables was checked using Q–Q plots, histogram charts, and Kolmogorov–Smirnov test. General quantitative and categorical characteristics were expressed as median (interquartile range (IQR)) and frequency and percentages. A chi-squared test analyzed differences in the distribution of categorical variables (e.g., familial history of BV). Mann–Whitney test was applied to check differences in the distribution of continuous variables (e.g., physical activity). Factor analysis with the principal component analysis (PCA) method was used to simplify the factor structure to extract dietary patterns. FFQ Food items were classified into 30 groups based on similarity in nutrients, shown in Table 1. Kaiser–Meyer–Olkin (KMO = 0.634) and Bartlett's tests (P < 0.001) were used to endorse the sample size adequacy and data suitability for factor analysis. According to eigenvalue greater than 1.6, the scree plot and percentage of variance greater than 5%, factors (dietary patterns) were retained. Varimax rotation (orthogonal) was used to examine the relation between variables and factors. Food groups with a factor loading of more than 0.4 remained, and factors were specified based on them. Factor scores were measured for each participant, and they were categorized into tertiles. The first tertile was assumed as a reference. The base and multivariable-adjusted odds ratios (OR) with 95% confidence intervals (CIs) were determined by the logistic regression to expound the association between dietary patterns and odds of BV. Adjusted ORs were computed by adjusting for familial history of BV, BMI (Kg/m2), WC (cm), cigarette smoking, energy intake (Kcal/d), calcium supplement, and physical activity (MET/h/d).

Table 1 Food groups and food items of FFQ used in the analysis.
Full size table

Ethics approval

The study was reviewed and approved by the National Nutrition and Food Technology Research Institute, Shahid Beheshti University of Medical Sciences, Tehran, Iran. The signed informed consent form was obtained from each participant before inclusion. All procedures were conducted according to the latest version of the Helsinki Declaration. The ethics committee code was IR.SBMU.NNFTRI.REC.1399.054.

Results

After estimating the participants' calorie intake, 4 BV-affected women and 2 healthy women whose log scale of total energy intake was either >  + 3SD or < -3SD from the mean were excluded from the mean statistical analysis (Fig. 1). Therefore, participation rates were 97.3% and 98.7% among BV-affected and healthy women, respectively.

Figure 1
figure 1

Study flow diagram.

Full size image

Baseline characteristics of the BV-affected women (n = 144) and healthy women (n = 151) are demonstrated in Table 2. The median (IQR) age of participants was 30 (25–33) and 32 (24–37) years in BV-affected and healthy women, respectively. Compared to the healthy women, BV-affected women tended to have higher levels of obesity (P = 0.016), higher levels of abdominal obesity (P = 0.05), higher frequency of familial BV (P < 0.001), higher smoker (P < 0.001), and lower calcium supplement use (P = 0.016). There was no statistically significant difference concerning other general characteristics between BV-affected and healthy women.

Table 2 Baseline characteristics of participants in case and control groups.
Full size table

Table 3 shows factor loading values of food groups for dietary patterns. Five principal dietary patterns were derived, representing 38.7% of the total variance. A "Healthy diet" pattern, represented 9.9% of the total variance, comprised high fruits (all kinds), poultry, and other vegetables consumption (Garlic, Onion, Eggplant, Tomato, Mushrooms). An "Unhealthy diet" pattern represented 8.6% of the variance and comprised sugar, solid oils, sweets and desserts, red meat, fried potato, refined grains, visceral meat, and sweet drinks. An "Ovo-vegetarian diet" pattern represented 8.4% of the variance, comprised all kinds of vegetables, beans, whole grains, and eggs. A "Pseudo-​Mediterranean diet" pattern represented 6.3% of the variance, comprised nuts, fish, olive, and olive oil. A "Western diet" pattern represented the slightest variance (5.6%) and comprised high intakes of processed meat, snacks, and pickles.

Table 3 Factor loading values of food groups for dietary patterns.
Full size table

Table 4 presents the base and adjusted OR for odds of BV through the emerged dietary patterns' tertiles. In the base model, being in the last tertile of the "Pseudo-Mediterranean diet" was negatively associated with BV odds (OR = 0.53; 95%CI: 0.29, 0.95; Ptrend: 0.042). However, the association was not statistically significant after adjustment for potential confounders. In the full model, after controlling for the confounders, women in the second and the third tertiles of the "unhealthy diet" pattern had a 2.04 (95% CI: 1.02, 4.09; Ptrend: 0.04) and 3.35 (95%CI: 1.41, 7.94; Ptrend: 0.006) times, respectively, a higher chance for BV in comparison to women in the first tertile. The chance to experience BV was 84% (95%CI: 0.07, 0.34; Ptrend < 0.001) lower in the third tertiles of the "Ovo-vegetarian diet" pattern, in comparison to women in the first tertile. There was no significant relationship between BV odds and other dietary patterns.

Table 4 Association between tertiles of dietary patterns and odds of bacterial vaginosis among the participant.
Full size table

Discussion

This is the first case–control study to assess the association between dietary patterns and BV to the best of our knowledge. In this research, five principal dietary patterns, including "Healthy diet," "Unhealthy diet," "Ovo-vegetarian diet," "Pseudo-Mediterranean diet," and "Western diet," were identified. Among the dietary patterns obtained, the "Unhealthy diet" pattern, defined by high loading of sugar, solid oils, sweets and desserts, red meat, fried potato, refined grains, visceral meat, and sweet drinks, was substantially associated with a higher BV odds. In contrast, a considerable association was found between the "Ovo-vegetarian diet" pattern high in all kinds of vegetables, beans, whole grains, and egg and BV in the multivariable-adjusted model. In addition, women with the highest adherence to the "Pseudo-Mediterranean diet" pattern high in nuts, and fish, olive, and olive oil also had lower odds for BV in the base model. No clear association was found in relation to the "Healthy diet" and the "Western diet."

The main characteristic of the "Unhealthy diet" pattern is a higher intake of saturated fat sources such as solid oils, red meat, fried potato, visceral meat. Similar to our finding, Neggers et al. have shown that saturated fat increases BV risk11. Although the mechanism for the role of saturated fat consumption in BV occurrence remains unknown, a high saturated fat consumption may lead to an elevated vaginal pH and alter the vaginal microflora3,11. Moreover, a high-saturated fat diet may also affect the mucosal immune system11. Another main characteristic of the "Unhealthy diet" pattern is high dietary carbohydrate content with a high glycemic index and glycemic load, including sugar, sweets, desserts, refined grains, and sweet drinks. Consistent with our finding, Thoma et al. have demonstrated a direct link between dietary glycemic load and BV progression and persistence10. High-GI/GL foods overconsumption as an unhealthy diet may be related to the pathogenesis of BV and could influence host response to bacterial colonization through oxidative damage and impaired immune response28,29. In return, participants adhering to a diet composed mainly of plant-based foods had a lower likelihood of BV in our study. The main components of the "Ovo-vegetarian diet" pattern include high-fiber and starchy foods, including vegetables, beans, and whole grains. In support, according to Shivakoti et al., higher fiber intake was inversely associated with the odds of molecular-BV30. Diets rich in fiber may reduce the risk of bacterial infections related to BV by affecting the microflora through more Lactobacillus-dominant profiles and positively impacting vaginal health30. Also, three previous studies showed a higher prevalence of BV among non-vegetarians than vegetarians, in line with our results3,31,32. The practical consequence of the "Ovo-vegetarian diet" pattern on BV can explain by creating a suitable environment for Lactobacillus33, lowering vaginal pH, and decreasing oxidative stress34. The vaginal micro ecological environment is affected by diets rich in starch due to high glycogen levels in the genital fluid35. Neggers et al. have also demonstrated an inverse association between severe BV and folate intakes11. Eggs, vegetables, and whole grains have been identified to represent the primary sources of folate. This result reveals the possibility that high intakes of folate sources may improve the immune system and are associated with decreased risk of BV11.

The current study has several potencies. The dietary pattern-based investigation was used to assess the total dietary intakes rather than individual dietary intakes. Since some foods may act synergistically, the dietary pattern approach provided better information about the correlation between dietary intake and diet-related diseases36. Dietary patterns derived from a validated-FFQ. Participants who under- or over-report their energy intakes were excluded. Newly diagnosed BV patients were included. Incident case selection would support a causal interpretation and control the recall bias37,38. High participation rates were obtained in both cases and controls. Information on several potential covariates was available to adjust in regression models. One person in the hospital laboratory performed a diagnostic test in both groups to control information bias. Besides, a trained dietician completed the questionnaires and was blinded to the diagnostic results during the interview. However, the current study has several limitations. This study did not use the Nugent score for BV diagnosis, although it is the gold standard for BV assessment. Although researchers tried to minimize bias as much as possible, selection bias, measurement bias, and recall bias might result in misleading findings in a case–control design. Different types of bacteria causing the BV were not considered in the current study. The cultural and religious taboo on alcohol and opium in Iran prevented the collection of data on these variables. Besides, the current study did not consider percentages of body fat, and the only criteria to assess obesity was the body mass index.

Conclusion

To conclude, the results revealed that the "Ovo-vegetarian diet" characterized by all kinds of vegetables, beans, whole grains, and eggs decreases the odds of BV in this sample of Iranian women. However, the "Unhealthy diet" pattern defined by high sugar, solid oils, sweets and desserts, red meat, fried potato, refined grains, visceral meat, and sweet drinks increases the odds. Therefore, it is recommended that these findings should be considered in nutritional education and dietary recommendation to prevent BV. The findings need to be confirmed by further longitudinal dietary research to clarify whether dietary change is feasible and affects BV outcomes.