Bacterial vaginosis : prevalence , risk profile and association with sexually transmitted infections

Background and Objectives: Bacterial vaginosis (BV) is the most common cause of vaginal discharge in the world. The study aimed to estimate the prevalence and to identify risk factors associated with bacterial vaginosis. Methods: A cross-sectional study was conducted in Ouro Preto, Brazil, between February and December 2017. Three hundred and forty-one women aged 18 years or older, users of the Brazilian Unified Health System, participated in this study. Women who used oral or topical antibiotics in the four weeks prior to the sample collection and women who had undergone a total hysterectomy were excluded from the study. After signing the Informed Consent Form and filling out a questionnaire containing sociodemographic, behavioral and sexual data, the participants were directed to the collection room, where the nurse collected the samples for the preventive examination of the cervix and also two vaginal swabs. Vaginal swabs and cervical samples were analyzed for cytological abnormalities and BV using Gram staining and cytology. Pathogens causing sexually transmitted infections (STIs) were identified by Polymerase Chain Reaction (PCR). For the analysis of the data, statistical package STATA version 10.0 was used. This study was approved by the Research Ethics Committee of the Federal University of Ouro Preto (UFOP). Results: During the study, 341 women were evaluated. The prevalence of BV using Gram staining (32.5% [CI95% 27.737.7%]) and cytology (27.7% [CI95% 23.032.8%]) was similar, however, the sensitivity of cytology was lower (77.8%). Risk factors associated with BV were smoking (IRR 1.5 [CI95%: 1.1  2.1]), use of an intrauterine device (IRR 2.8 [CI95%: 1.2  6.5]), and past medical history of BV (IRR 1.5 [CI95%: 1.1  2.1]). Correlation between the presence of BV and Trichomonas vaginalis (TV) infection (r=0.24) was observed. Conclusion: The prevalence of BV was affected by life habits and was prevalent in women with TV. Thus, behavioral and social prevention approaches to women with diverse risk profiles may help mitigate TV/BV prevalence and recurrence of BV.


INTRODUCTION
Bacterial vaginosis (BV) is the leading vaginal disorder in women during their reproductive years and postmenopausal period, contributing to more than 60% of all alterations in the vaginal microbiota.Prevalence rates of BV vary according to the population studied and the Brazilian population presents similar prevalence rates to countries such as Canada (33.0%) 1 .
However, as Brazil has a mixed population, Brazilian regions may have different prevalence rates.
In the typical reproductive-aged vaginal environment, BV is categorized by a shift in the vaginal flora from aerobic to predominately pathogenic anaerobic bacteria.Moreover, the postmenopausal vaginal environment, including vaginal atrophy, may also contribute to a shift in the colonization of vaginal microbiota in older woman. 1 BV is characterized by an increase in the vaginal pH, reduction in Lactobacillus flora and an elevated number and type of facultative and anaerobic bacteria. 2e acidic environment of the normal vaginal flora inhibits colonization by pathogenic organisms; however, an alkaline environment facilitates the growth of sexually transmissible agents. 2 For the diagnosis of sexually transmitted infections (STIs), traditional methods are laborious and often not very sensitive.However, during recent years, molecular techniques have provided new approaches for the diagnosis of these pathogens. 3 is associated with an increased risk of preterm birth, spontaneous abortion, pelvic inflammatory disease, endometritis, and transmission of pathogens such as HIV.Given the range of reproductive consequences of BV, it is important to know the risk factors for BV among a population so that mechanisms of prevention can be developed. 4odemographic and behavioral factors have been associated with the development of BV, including number of sex partners, frequent use of scented soap, and irregular condom use. 5 Furthermore, studies have shown opposite conclusions between BV and cytological abnormalities and association between BV and STIs. 1 Gram staining Nugent scoring method is considered as a gold standard in the diagnosis of BV.However, Papanicolaou staining for cytological investigation is also a practical technique that demonstrates the presence of clue cells indicative of BV. 6 In Brazil, the cytological evaluation should be performed every three years if the first two annual tests were normal. 7is study aimed to estimate the prevalence and to identify risk factors and STIs associated with BV in a Brazilian subpopulation.Women who attended the Public Health Units in Ouro Preto city to perform the cervical screening test were invited to be part of the study (convenience sample).Thus, the participants were informed by the doctor or nurse of the health unit about the research objectives and were asked to sign an Informed Consent Form.

METHODS
Women aged over 18 years old were considered for inclusion.The exclusion criteria included women who were using or who took oral or topical antibiotics within four weeks prior to sample collection and women who had undergone a total hysterectomy.The sample size calculation was performed based on a population of 6,200 women aged ≥18 years old in three Public Health Units according to the Municipality Health Service.
Prevalence of BV estimated at 20%, 95% confidence level, and estimated accuracy of 4% were considered.Thus, it was established that the sample size should be 362 women.
Vaginal and cervical samples were collected from 341 women who filled out a questionnaire 8,9 addressing sociodemographic (educational level, marital status, ethnic origin and average monthly household income of their family), behavioral (smoking, use of alcohol and use of illicit drugs), and sexual (age of first intercourse, history of female sexual partner, pregnancy, abortion, IST, contraceptive use, use of intrauterine device, condom use, history of bacterial vaginosis and age of first menstruation) characteristics.The interviews were conducted individually in a private room.
Cervical samples were collected by nurses from each Basic Health Unit.For collection of the material, a speculum was introduced into the vagina.After the visual inspection of the interior of vagina and cervix, the nurse performed the collection of biological specimens from the external and internal surface of the cervix with an Ayre's spatula and a cervical brush.
Collected cells were then placed on a slide, fixed in 95% alcohol and sent to the Exfoliative Cytology of Uterine Cervix Sector at the Pilot Laboratory of Clinical Analysis (LAPAC).The slides were stained by the standard Papanicolaou method 10 and analyzed according to Claus, 1992. 11 Vaginal samples were also collected by nurses from each Basic Health Unit for Gram stain and nucleic acids amplification tests.With a speculum inserted into the vaginal canal, two sterile cotton swabs (Labor Swab, São Paulo, Brazil) were used to collect vaginal samples.The appearance and type of vaginal secretions were recorded.One swab was used for the preparation of slides that, after drying, were heat-fixed and gram stained and another swab was stored in tubes containing 0.5 mL of 0.9% NaCl solution at -20 ºC for further molecular testing.
Briefly, each stained smear was observed for bacterial morphotypes under oil immersion objective at 100 X magnification and graded as per the standardized quantitative morphological classification method developed by Nugent (1991) 12 , which assigns a score between 0 and 10 based on the following bacterial morphotypes: large Gram-positive rods (Lactobacillus morphotypes), small Gram-variable rods (G.vaginalis morphotypes), small Gram-negative rods (Bacteroides spp.morphotypes), curved Gram-variable rods (Mobiluncus spp.morphotypes), and Gram-positive cocci. 2,11.
In the cervical samples, the diagnosis of BV was performed according to the Bethesda System nomenclature 13 .[16] Briefly, each stained smear was observed for bacterial morphotypes under oil immersion objective at 100 X magnification and graded as per the standardized quantitative morphological classification method developed by Nugent (1991) 12 , which assigns a score between 0 and 10 based on the following bacterial morphotypes: large Gram-positive rods (Lactobacillus morphotypes), small Gram-variable rods (G.vaginalis morphotypes), small Gram-negative rods (Bacteroides spp.morphotypes), curved Gram-variable rods (Mobiluncus spp.morphotypes), and Gram-positive cocci. 2 , 12 Samples from forty women with BV detected by the Gram staining and forty women without BV were randomly selected and submitted for Trichomonas vaginalis (T.vaginalis), Neisseria gonorrhoeae (N.gonorrhoeae), Chlamydia trachomatis (C.trachomatis) and Human Papilloma Virus (HPV) detection by Polymerase Chain Reaction (PCR). 3,17,18 Dbases were generated using EpiData version 3.2 (EpiDataAssociation, Odense, Denmark) by double entry of the results, and they were subsequently compared, corrected, and analyzed using STATA version 14.0 software (Stata Corp., College Station, TX, USA).
The results of the Gram stain and cytological evaluation were used to calculate the prevalence of BV.Gram stain was used as reference to calculate diagnostic performance of the cytological evaluation.Sensitivity, specificity, accuracy, positive predictive value and negative predictive value of the cytological method were calculated using the Gram stain as a gold standard.Besides that, agreement between these methods was evaluated using agreement percentage and the Kappa coefficient.Interpretation of the results of Kappa coefficient was done according to the following scale: 1.000.81excellent; 0.800.61good; 0.610.40moderate; 0.400.21weak; 0.200.0absence of agreement.
Factors associated with BV (detected by Gram stain) were determined using a univariate analysis performed previously with Poisson regression with estimated robust variance and all the variables of the questionnaire were tested.Variables with p value < 0.25 in univariate analysis were selected for multivariate analysis.
Multivariate analysis was based on a complete model, with all the variables selected in univariate analysis.Non-significant variables were discarded in a step-by-step manner (backward selection).In this step, significance level of 0.05 was used, that is, the variables that presented p > 0.05 were withdrawn from the step-by-step model.The final model consisted only of variables with significance level < 0.05.
The association between the presence of cytological abnormalities and the presence of bacterial vaginosis by Gram staining was evaluated through the chi-square test followed by the univariate analysis using logistic regression.Besides that, the association between noncommensal pathogens (T.vaginalis, N. gonorrhoeae, C. trachomatis and HPV) and presence of vaginosis was evaluated similarly.It was considered as presence of association when the p value was lower than 0.05.This study was approved by the Committee of Ethics in Research, protocol no.

Characteristics of the population
From 341 women, the mean age was 40.1±14.0years old, ranging from 18 to 83 years old.The majority of the women had completed high school (48.7%), were married or lived with a partner (53,4%), self-declared Asian-indigenous or brown (39.4%), had a family income of less than two minimum wages (55,6%) (Table 1).The majority of women were non-smokers (83.0%) or never smoked (74.3%) and 55.6% reported drinking alcoholic beverages and not using illicit drugs (98.5%), oral hormonal contraception (51.6%), injectable hormonal contraception (67.6%), intrauterine device (IUD) (72.7%) or condoms (72.6%)(Table 2).were not in menopause, 72.9% had a history of pregnancy, 77.6% had an active sex life, 96.0% had only one sexual partner and 97.9% did not have a history of a female sexual partner (Table 3).The mean age and standard deviation (SD) of first menstruation and the first sexual intercourse were 12.8 (SD: 1.8) and 18.3 (SD: 3.5) years old, respectively.

Factors associated with bacterial vaginosis
In order to identify the factors associated with BV, women with and without BV were compared according to Gram staining.From the output of the multivariate model, the variables associated with vaginosis were smoking (PR 1.5 [CI95%: 1.1  2.1]), use of IUD (PR 2.8 [CI95%: 1.2  6.5]) and previous history of BV (PR 1.5 [CI95%: 1.1  2.1]).These results show smoking and previous history of bacterial vaginosis increased the prevalence of BV by 1.5 times when compared to non-smokers and to women who do not have previous history of BV, respectively.Use of IUD increased the prevalence of BV by 2.8 times in comparison with women who do not use this contraceptive method (Table 4).* The non-applicable category, which is composed of menopausal women, was also not significant in the multivariate analysis.

Detection of pathogens causing sexually transmissible diseases
The search for the pathogens responsible for STIs was carried out in 80 samples.C. trachomatis was the most commonly detected pathogen (20/80 cases), with 11 cases in women without BV and 9 in women with BV.HPV (17/80) was observed in 12 cases in women with BV and 5 in women without BV, followed by T. vaginalis (13/80) with 10 cases detected in women with BV and 3 in women without BV.N. gonorrhoeae (2/80) was observed in 1 case in a woman with BV and 1 in a woman without BV.The pathogens vaginalis and HPV were detected with higher frequency in women with BV (25.0 and 30.0%respectively).An association was observed between the presence of BV and the detection of T. vaginalis (OR:

DISCUSSION
The results of this study demonstrate a high prevalence of BV.Prevalence rates of vaginosis vary according to the population studied and the method used for diagnosis.In the present study, conducted in a population of women aged ≥18 years old, the prevalence of BV using Gram staining was 32.5%.Studies performed in Lithuania and Ethiopia with women in the same age group and using the same diagnostic method showed prevalence rates of 24.4 and 48.6%, respectively 2,19 .Other countries such as Canada (33.0%) and the United States (30.1%) showed prevalence rates of 33.0% and 30.1%, respectively 1,4 , exemplifying that different populations have different prevalence rates .Variability in the composition of the vaginal flora has been pointed out as an important factor for the difference in BV prevalence among populations.The higher prevalence of BV in African women, for example, can be explained by the low occurrence of Lactobacillus species (lactobacillary grade III) producing hydrogen peroxide, which are less frequent in black women and have pathogenic defense activity. 20rvical cytology by Papanicolaou smears was also used to evaluate BV.The prevalence was 27.7%, which is lower than that detected by Gram staining.In a population in Brazil (Olinda, Pernambuco), Peres et al. (2015) 15, using the same analysis method, similar prevalence (28.1%) to the present study, showing that populations within the same country may have similar prevalence rates.Thus, this population's prevalence rate could represent the country's prevalence rate.Although there was a difference in the prevalence rate between the cytological evaluation and the Gram staining, good agreement between them was found (Kappa = 0.77).Both methods can be used to diagnose BV 21 , however, it is important to emphasize that the cytological evaluation is less sensitive.In the present study, a sensitivity of 77.8% was observed when the cytological technique was used, that is, the ability to correctly diagnose bacterial vaginosis is lower than that of Gram staining.
Gram staining using the Nugent score, despite being considered the gold standard in the diagnosis of BV and having been used in our study, has limitations as the punctuation system can generate confusion in the presence of a more complex vaginal flora, classifying it into only one category. 22Conversely, cytological evaluation has an important role in the recognition of infectious and inflammatory changes of the female genital tract, designated as reactive cellular changes.Moreover, Anand et al. (2020) 23 have shown that Pap-stained smears can be an effective instrument for BV diagnosis, in addition to being important for the detection of cervical cancer.Papanicolaou is performed routinely for women and can be important in diagnosis because it can lead to an early detection of BV. 24 Despite this, considering the frequency of BV detection, our study showed that the Gram staining method may be more reliable than the Papanicolaou method.Thus, although more studies are needed, we suggest that the combination of the Gram method and cytological analysis can be a good strategy for the correct diagnosis of bacterial vaginosis.
Although no association was found between the presence of BV and the detection of HPV, this pathogen was detected more frequently in women who had BV.Mongelos et al.
(2015) 25 showed that a large variety of HPV genotypes was detected in indigenous Paraguayan women with BV.Enzymes produced by anaerobic bacteria involved in the pathogenesis of BV can potentially alter immune signals and promote the degradation of host factors, rendering women more susceptible to acquiring HPV. 26 HPV screening in this study was performed on vaginal swabs, unlike most studies where cervical specimens were used.Coorevits et al.
(2018) 27 compared the sensitivity of HPV detection in cervical and vaginal samples and observed that these samples are equivalent to the detection of this pathogen.Thus, vaginal swabs can be used for the detection of HPV in screening strategies cervical cancer.
Regarding the factors associated with BV, it was observed that smoking is a factor that increases the prevalence of BV. 28 This association can be explained by the fact that the level of protective Lactobacillus flora in smoker females is lower when compared to that of nonsmokers.In addition, some authors have shown that the cigarette has an antiestrogenic effect, which in addition to amine accumulation in the vaginal epithelium, predisposes the women to develop BV. 28 An association was also found between the use of IUD and the presence of BV.It is believed that the IUD can modify the vaginal flora, favoring the colonization of bacteria associated with BV.The presence of this device could facilitate the rise of cervicovaginal microorganisms, predisposing women to vaginosis and pelvic inflammatory disease. 29e recurrence of BV was another factor associated.The previous history of BV associated with new cases was also observed in other studies. 4, 25 Recurrence episodes are often associated with the failure of antibiotic treatment.This is mainly due to biofilm formation in the vaginal mucosa.It is possible that Gardnerella vaginalis, through its metabolic pathways and ability to form a biofilm, lowers the reduction-oxidation potential in the vaginal microbiome. 25 this study, an association was observed between BV and T. vaginalis infection.A study has shown that women with BV can present a higher risk of STIs, especially Chlamydia trachomatis and HPV infection. 1Reasons for this increase are unclear but are believed to be due to elevated vaginal pH and the presence of inflammatory mediators and enzymes such as mucinase, which would provide a favorable environment for the establishment of pathogens.In fact, T. vaginalis grows at high pH, an environment provided when the woman has vaginosis. 1sides being the pathogen most frequently detected in this study, C. trachomatis was also detected simultaneously with the other pathogens studied and the co-infection with HPV was the most frequent.The presence of C. trachomatis may increase the risk of HPV acquisition for two reasons: the presence of these bacteria is associated with disruption of the epithelial barrier, leading to the inflammatory response and C. trachomatis may interfere with the immune response by decreasing the number of antigen-presenting cells and reducing the cell-mediated immunity, thus allowing the persistence of HPV. 30 We acknowledge several limitations to this study.Only the population assisted in the public health service has participated in the study.However, we believe the associated factors described by our study are apparently not influenced by the socioeconomic profile.Another limitation was the use of a non-probabilistic sample evaluating only three basic health units, despite of the fact these units represent 49% of the female population aged ≥18 years old, living in the urban area of Ouro Preto.Thus, the internal validity of the study was possibly a little impaired.
In conclusion, the present study showed that both Gram staining and cytological evaluation can be used in the diagnosis of BV, despite the lower sensitivity of the cytological method.It is important that physicians and nurses from the Basic Health Units pay particular attention to smoking women, users of IUD and those with a previous history of BV since they have an increased risk for BV.The association found between BV and the detection of T.
vaginalis signals that socio-educational measures should be taken to reduce the number of cases of BV and STIs.

A
cross-sectional study was conducted between February and December 2017 in three different Public Health Units in Ouro Preto city, Minas Gerais State, Brazil, during the evaluation of the Extension Program Ambar: Desafios e Ações em Saúde da Mulher that mainly aims to expand the structure of care for women, focusing on pharmaceutical attention and quality of life in climacteric.Ouro Preto has nine Public Health Units in its urban area; three units were selected to perform the study.The choice of Public Health Units was based on low to medium socio-economic conditions of different regions in the urban area of the city and for logistical and operational reasons.This study was conducted in collaboration with the Municipality Health Service.
1.907.198,CAAE: 62883616.8.0000.5150.All procedures in this study were in accordance to the protocol number 466/2012 of the National Health Council.Participants were informed of the research objectives and were required to sign an Informed Consent Form before sample and data collection.

Table 2 .
Distribution of the behavioral characteristics of the women interviewed in three different Public Health Units, Ouro Preto, Minas Gerais, Brazil, 2017 * missing "did not answer" ** Not applicable: concerning menopausal women Most women were not pregnant at the time of the interview (96.8%) and did not have a history of abortion (79.2%), or STIs (92.0%).Furthermore, 50.8% reported previous BV, 73.3%

Table 3 .
Distribution of the sexual behavioral characteristics of the women interviewed in three different Public Health Units, Ouro Preto, Minas Gerais, Brazil 2017 * missing "did not answer"

Table 4 .
Risk factors for the development of bacterial vaginosis, Brazil 2018