Article Text

Original article
Intermediate vaginal flora and bacterial vaginosis are associated with the same factors: findings from an exploratory analysis among female sex workers in Africa and India
  1. Fernand A Guédou1,2,
  2. Lut Van Damme3,4,
  3. Jennifer Deese3,
  4. Tania Crucitti5,
  5. Florence Mirembe6,
  6. Suniti Solomon7,
  7. Marissa Becker8,
  8. Michel Alary1,2
  1. 1URESP, Centre de recherche FRSQ du CHU de Québec, Quebec, Canada
  2. 2Département de Médecine Sociale et Préventive, Université Laval, Québec, Canada
  3. 3FHI360, Durham, North Carolina, USA
  4. 4CONRAD, Arlington, Virginia, USA
  5. 5Department of Microbiology, Institute of Tropical Medicine, Antwerp, Belgium
  6. 6Makerere University College of Health Sciences, Kampala, Uganda
  7. 7Y.R Gaitonde Center for AIDS Research and Education, Chennai, India
  8. 8Centre for Global Public Health, University of Manitoba, Winnipeg, Manitoba, Canada
  1. Correspondence to Dr Michel Alary, URESP, Centre de recherche FRSQ du CHA Universitaire de Québec, Hôpital Saint-Sacrement, 1050, Chemin Sainte Foy, Québec, Canada G1S 4L8; malary{at}


Objectives Several recent studies suggest that intermediate vaginal flora (IVF) is associated with similar adverse health outcomes as bacterial vaginosis (BV). Yet, it is still unknown if IVF and BV share the same correlates. We conducted a cross-sectional and exploratory analysis of data from women screened prior to enrolment in a microbicide trial to estimate BV and IVF prevalence and examine their respective correlates.

Methods Participants were interviewed, examined and provided blood and genital samples for the diagnosis of IVF and BV (using Nugent score) and other reproductive tract infections. Polytomous logistic regressions were used in estimating respective ORs of IVF and BV, in relation to each potential risk factor.

Results Among 1367 women, BV and IVF prevalences were 47.6% (95% CI 45.0% to 50.3%) and 19.2% (95% CI 17.1% to 21.2%), respectively. Multivariate polytomous analysis of IVF and BV showed that they were generally associated with the same factors. The respective adjusted ORs were for HIV 1.98 (95% CI 1.37 to 2.86) and 1.62 (95% CI 1.20 to 2.20) (p=0.2248), for gonorrhoea 1.25 (95% CI 0.64 to 2.4) and 2.01 (95% CI 1.19 to 3.49) (p=0.0906), for trichomoniasis 3.26 (95% CI 1.71 to 6.31) and 2.39 (95% CI 1.37 to 4.33) (p=0.2630), for candidiasis 0.52 (95% CI 0.36 to 0.75) and 0.59 (95% CI 0.44 to 0.78) (p=0.5288), and for hormonal contraception 0.65 (95% CI 0.40 to 1.04) and 0.62 (95% CI 0.43 to 0.90) (p=0.8819). In addition, the association between vaginal flora abnormalities and factors such as younger age, HIV, gonorrhoea trichomoniasis and candidiasis were modified by the study site (all p for interaction ≤0.05).

Conclusions IVF has almost the same correlates as BV. The relationship between some factors and vaginal flora abnormalities may be site-specific.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Bacterial vaginosis (BV) is a polymicrobial infection characterised by a depletion of lactobacilli, accompanied by intense overgrowth of commensal vaginal anaerobic bacteria.1 The current ‘gold standard’ for BV diagnosis is the Nugent's scoring system, grading vaginal flora as normal with Nugent score (NS)=0–3, intermediate vaginal flora (IVF) with NS=4–6 or BV with NS=7–10.2

Adverse health outcomes associated with BV include upper genital tract infections,3 unfavourable pregnancy outcomes and sexually transmitted infections including HIV.4 Some recent studies revealed that IVF may also be associated (and even as strongly as BV) with some of these adverse health outcomes.5 ,6

Though there is a significant amount of data on BV risk factors, very little is known about the risk factors for IVF and whether they are similar to those of BV.

We used data from female sex workers (FSWs) screened at two African and two Indian sites prior to their enrolment in the phase III microbicide trial on the vaginal gel of Cellulose sulfate 6% sponsored by CONRAD, Virginia, USA,7 to compare correlates for BV and IVF.

Methods and materials

Settings, subjects and procedures

The methods and materials used have been described in detail elsewhere.7 Briefly, we used data from participants screened at four of the five sites of the trial.7

All participants were aged 18 years or older and provided written informed consent for the trial screening visit. At screening, a questionnaire was administered by trained health providers. Trained clinicians then performed a speculum and bimanual pelvic examination, collected two endocervical swabs and one high vaginal swab for the diagnosis of genital tract infections including BV. Swabs were immediately transferred to laboratory where a trained lab technician immediately (within 10 min) performed wet mount while fixed smears were Gram-stained to be examined at the end of the day. Vaginal flora abnormalities (BV and IVF) were described according to the Nugent's scoring system after Gram-stain.2 Venous blood was taken for syphilis and HIV antibody testing.

All assays were conducted according to the manufacturers’ instructions. The study was approved by ethics committees of the Eastern Virginia School of Medicine (USA) and of each collaborating centre.

 Statistical methods

Continuous variables were described by their medians along with the IQR. For categorical variables, proportions and their 95% CIs were computed.

The measure of association used was the OR, which was estimated using vaginal flora abnormalities as the dependent variable and data on socio-demographics, sexual behaviour, medical history and current genital tract infections as independent variables. To generate respective ORs for IVF versus normal vaginal flora (NVF) and BV versus NVF, we constructed a polytomous multinomial logistic regression with the outcome (vaginal flora abnormalities) classified into three categories: BV, IVF and NVF (reference category). Bivariate analyses (controlling for the study site) were first performed; covariates that were statistically significant at p value <0.20 were included in the multivariate model. Backward elimination based on Wald χ2 tests was done and variables with p values <0.05 were retained in the final model. To test the significance of the difference between the ORs obtained for IVF versus NVF and that obtained for BV versus NVF, we tested the significance of the OR for BV versus IVF (using IVF as reference in the same previous model). Finally, we test for interaction between the study sites and the covariates of the final multivariate model by including, each at once, the product terms combining the study sites and each of the other covariates.

Variables with p value <0.05 were considered as significantly associated with the considered vaginal flora abnormality.

Statistical tests were two-tailed. Data were analysed with SAS software V.9.2 (SAS Institute, Cary, North Carolina, USA).


Characteristics of the study population

From July 2005 to January 2007, 1491 FSWs were screened at the four study sites. Data on vaginal flora abnormalities were available for 1367 (91.7%) of them and these were thus included in the present analysis. Their socio-demographic, behavioural and medical characteristics were presented elsewhere.6 BV and IVF prevalences were 47.6% (95% CI 45.0% to 50.3%) and 19.2% (95% CI 17.1% to 21.2%), respectively. The median number of sexual acts in the week preceding the study entry was 8 (IQR=4–20).

Factors associated with vaginal flora abnormalities in multivariate analyses

Results from the final multivariate polytomous model are presented in table 1. Factors which were positively and significantly associated with both IVF and BV included the site of Cotonou (compared with that of Kampala), HIV and trichomoniasis. For gonorrhoea, the association was significant only for BV. Factors which were negatively associated with both IVF and BV comprised candidiasis, hormonal contraception, anal sex in the past 30 days and intrauterine device. For candidiasis, the association was significant for both IVF and BV.

Table 1

Final model for the multivariate analysis of the correlates of vaginal flora abnormalities (modelled as polytomous response) among 1367 female sex workers screened at two African and two Indian sites

Younger age, high number of sexual acts in the last 7 days and the site of Mudhol/Jamkhandi were positively associated with BV but negatively with IVF. However, none of these associations was statistically significant. For none of the studied factors (except for the site of Mudhol/Jamkhandi and high number of sexual acts), the ORs obtained respectively for IVF and BV were significantly different (all p values >0.07).

Finally, there were significant associations between vaginal flora abnormalities and the product-terms combining study sites and respectively younger age, HIV, gonorrhoea, trichomoniasis and candidiasis (table 2).

Table 2

Comparison of the strength of the association between some factors and vaginal flora abnormalities (modelled as polytomous response) according to the study sites, among 1367 female sex workers screened at two African and two Indian sites


In this cross-sectional study involving FSWs from two African and two Indian sites, we found overall prevalence rates of 47.6% and 19.2% for BV and IVF, respectively. From the multivariate analysis, HIV, gonorrhoea, trichomoniasis, hormonal contraception, anal sex and candidiasis were associated with both IVF and BV. More interestingly, for none of these factors, the strengths of the respective associations with IVF and BV were significantly different, suggesting that IVF and BV share these correlates. Though being younger than 20 years was positively associated with BV and negatively associated with IVF, the two ORs were not significantly different (p=0.0796). In addition, neither the strength nor the statistical significance of the inversed association with IVF is substantial, suggesting that the latter may merely be due to an uncontrolled bias.

The strength of the associations between some factors (being younger than 20 years, having HIV or gonorrhoea or trichomoniasis or candidiasis) and the vaginal flora abnormalities significantly varied across study sites (as shown in table 2 with p values for interaction <0.05). This finding reveals the need to study the relationship between these factors and vaginal flora abnormalities by site and not as overall. This modification of the relationship by the site may partly explain the diversity observed in the data reported on BV risk factors.

Several recent studies confirmed the relationship between vaginal flora abnormalities and HIV and showed that IVF may even be associated with HIV as strongly as BV.5 ,6 Brotman et al8 found that BV was positively associated with incident genital gonorrhoea, chlamydia and trichomoniasis. We found that BV was more common in women younger than 20 years. Similarly, McClelland et al4 reported a decrease of BV risk with increasing age.

Having had 35 or greater sexual intercourses in the last 7 days was significantly associated with BV, while the number of sexual partners in the last 3 months was not. This is in agreement with Verstraelen et al9 who found that sex frequency, and not number of sexual partners, is the critical risk factor for BV.

There was an inverse association between hormonal contraception and BV, as reported in other studies.4 The inverse association between vaginal candidiasis and abnormal vaginal flora is the most consistently reported throughout the literature.4

Besides its cross-sectional design, our study presents some other limitations. Because of the high frequency (98%) of self-reported intravaginal cleansing practice in our study population, we were not able to ascertain whether this factor was associated with BV or IVF. There were several self-reported data (number of sexual partners, sex frequency, condom use), suggesting the possibility of residual confounding for some associations.

Nonetheless, the study presents several strengths including the large sample size, the use of the current gold standard for BV and IVF diagnosis (NS) and the high level of the quality assurance system conferred by the clinical trial setting. A great particularity of the present paper is that it contributes to delineating the role of IVF in the determination of BV risk factors. To our best knowledge, this is the first paper reporting the use of logistic regression with polytomous outcome to explore correlates of vaginal flora abnormalities.

In summary, this study confirmed the high prevalence of BV among FSWs and found that IVF has almost the same correlates as BV. It also showed that the relationship between some factors and vaginal flora abnormalities may be modified by the study site and this might explain, at least partly, the inconsistency in some findings on BV risk factors. Appropriately designed prospective studies would help better disentangle risk factors of vaginal flora abnormalities.

Key messages

  • Intermediate vaginal flora (IVF) has almost the same correlates as bacterial vaginosis (BV).

  • Researchers looking at risk factors for vaginal flora abnormalities may learn more from comparing normal vaginal flora with both IVF and BV than considering BV alone.

  • The relationship between some factors and vaginal flora abnormalities may be site-specific.


The authors thank the research teams from all the four study sites for their hard work in collecting and capturing data, the monitoring staff for their assistance in assuring data quality, Doug Taylor from FHI360 for statistical advice and Dr Thurman AR from CONRAD for reviewing early drafts of this article. The authors are particularly indebted to participants without whom this study would not have been possible.


Supplementary materials

  • Abstract in French

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Jackie A Cassell

  • Contributors All authors were involved in the parent multicentre microbicide clinical trial that generated the data. For the present manuscript, FAG and MA conducted the statistical analyses and interpreted results. FAG wrote the first draft of the manuscript and MA revised it before further revision by other coauthors. In addition, JD and MB edited the text. All authors revised and approved the present version of the manuscript.

  • Funding The Cellulose sulfate clinical trial was sponsored by CONRAD (Virginia, USA) and cofunded by the USA Agency for International Development (USAID) through agreement No HRN-A-00-98-00020-00 and the Bill & Melinda Gates Foundation (BMGF) through grant No 655000.

  • Competing interests None.

  • Ethics approval The study was approved by ethics committees of the Eastern Virginia School of Medicine (USA) and of each collaborating centre.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The database relevant to each study site has been made available to the corresponding collaborating centre.