Bacterial vaginosis is characterized by complex alterations of the normal vaginal flora,1 in which hydrogen-peroxide-producing Lactobacillus species (which normally maintain a healthy microenvironment) are overgrown by endogenous facultative and anaerobic microorganisms such as Gardnerella vaginalis, Mycoplasma hominis, and anaerobic gram-negative rods (Bacteroides and Prevotella spp.).2,3 Bacterial vaginosis is one of the most common vaginal infections with prevalence estimates ranging from less than 5% to over 40%.4 Moreover, bacterial vaginosis is associated with a number of adverse sequelae, including preterm delivery, intrapartum and postpartum infections, postabortion endometritis, pelvic inflammatory disease (PID), and the acquisition of chlamydia, gonorrhea, and HIV.5,6
Among the risk factors associated with bacterial vaginosis are behaviors related to sexual activity. These include multiple sexual partners, new sexual partners, early age at first intercourse, sex during menses, and a history of sexually transmitted infections.7,8 These risk factors resemble those of classic sexually transmitted infections, and have raised the possibility that bacterial vaginosis may be sexually transmitted. However, other factors, such as the presence of bacterial vaginosis in adolescent virgins, suggest mechanisms other than sexual transmission.9 Moreover, partner treatment has generally proven to be an ineffective means for preventing bacterial vaginosis in women.10
Condoms can prevent the transmission of sexually transmitted infections by providing a mechanical barrier against pathogens.11 However, cross-sectional analyses have yielded mixed results regarding the efficacy of condoms for reducing bacterial sexually transmitted infections.12 A recent study by Warner et al,13 using a case-crossover design to reduce potential bias and confounding, found that condoms significantly reduced the risk of incident chlamydial infection. The case-crossover design compares exposure histories (eg, condom use) within the same participant at “case” times immediately preceding the outcome of interest (eg, vaginal infection) with crossover (“control”) periods of time that are assumed to be etiologically unrelated to the outcome of interest. Thus, time-independent potential confounders, whether measured or unmeasured, are automatically controlled for in the design. The case-crossover design is a particularly powerful method for assessing the influence of time-varying exposures and behaviors, including the potential protection afforded by condom use. We conducted case-crossover analyses of the relationships among condom use, bacterial vaginosis (ascertained by gram stain), and the growth of vaginal microorganisms associated with bacterial vaginosis.
The methods used for subject enrollment, data collection, and follow-up have been reported in detail elsewhere.14,15 Briefly, women 13–36 years of age were recruited into the GYN Infections Follow-Through (GIFT) Study from family planning clinics, university health clinics, gynecology clinics, and sexually transmitted disease units at each of 5 US sites between May 1999 and June 2001. Human subjects’ approval was obtained at each participating institution, and all women signed informed consent. Women were eligible for the study if they were not specifically seeking care for a sexually transmitted disease, but were nonetheless at high risk for such infection based upon a risk stratification algorithm for chlamydial cervicitis.16 Specifically, to be enrolled a women had to have a score of 3 or more points derived as follows: age 24 or less = 1; black race = 2; never pregnant = 1; 2 or more sexual partners = 1; douches at least once per month = 2; and any prior sexually transmitted infection (including N. gonorrhoeae, C. trachomatis, and Trichomonas vaginalis) = 2. Of the 2740 women screened for study entry, 853 (31%) did not meet these inclusion criteria. An additional 259 (9%) women were excluded because they were pregnant, married, or virginal, or on antibiotics at baseline. Among the 1628 women who were eligible for the study, 1199 (74%) completed a baseline questionnaire. Most women were seeking regular gynecologic care or birth control upon study entry. Of these 1199 women, 871 (73%) experienced a crossover in status from the absence of bacterial vaginosis or associated organisms to the presence of bacterial vaginosis or associated organisms. These 871 women provided the data for the case-crossover analyses.
At baseline and at 6, 12, 24, and 36 months thereafter, each woman obtained her own vaginal secretions with a cotton swab.17 Smears from these swabs were gram-stained and a microscopy score of 0–10 was assigned by laboratory staff using the standardized method described by Nugent et al.18 A score of 0–3 was interpreted as consistent with normal vaginal microflora; a score of 4–6, corresponding to disturbed microflora, was designated as intermediate; and a score of 7–10 was considered to be bacterial vaginosis.
Two swabs, placed in an anaerobic transport vial, were also shipped to the microbiology laboratory for characterization of Lactobaccillus species, anaerobic gram-negative rods, Gardnerella vaginalis, group B streptococcus, Enterococcus species, Escherichia coli, Candida species, Mycoplasma hominis, and Ureaplasma urealyticum. Lactobacilli were identified at the genus level on the basis of gram stain morphology and production of lactic acid. The amount of growth for each of these microorganisms was recorded on a semiquantative scale from 0 to 4.
DNA amplification for N. gonorrhoeae and C. trachomatis was performed using a strand displacement DNA amplification assay (Becton Dickinson, Sparks, MD) from the self-obtained vaginal swabs. All positive test results for gonococcal or chlamydial infection were reported to the clinical sites within 1 week of enrollment.
Women were asked at baseline about demographic factors including age, race, education, income, pregnancy history, smoking, alcohol use, and drug use. They also reported relevant lifestyle behaviors such as number of sexual partners in the past 2 months, acquisition of a new partner in the past 2 months, contraceptive use, sex during menses, and douching practices. Women were further requested to recall past episodes of sexually transmitted infections, including PID and gonococcal or chlamydial genital infections. Questions about pregnancy history, sexual activity, sexually transmitted infections, and douching were repeated during follow-up.
For women who reported being sexually active during the 2 months prior to the interview, we categorized condom use (during the prior 2 months) as yes/no and as average consistency of use (never, 5 or fewer times out of each 10 acts of vaginal intercourse, 6–9 times out of 10 acts of vaginal intercourse, and 10 out of 10 acts of vaginal intercourse).
Among the 871 study participants eligible for this analysis, the median length of follow-up was 3.0 years (interquartile range: 2.4–3.4 years). The median number of follow-up visits was 6 (interquartile range: 5–7), and the median number of vaginal swab samples was 4 (interquartile range: 3–5). Ninety-four percent of the women had 4 or more visits, and 90% of the women had 3 or more vaginal swab samples.
To determine the effect of condom use on the presence of bacterial vaginosis, we conducted case-crossover analyses. Only intervals that had both interview data and microbiologic data were considered for analyses. Exposure history intervals were designated as the periods between each of the 6-month follow-up visits. Case intervals were those in which bacterial vaginosis was diagnosed at the end of the interval. Control intervals were those in which women were diagnosed with normal or intermediate vaginal microflora at the end of the interval.
The majority of women had more than 1 interval with a diagnosis of bacterial vaginosis. Because of the complexity of bacterial vaginosis infection over both short and long periods of time, we limited our selection of case intervals. Bacterial vaginosis, with or without treatment, can occur as an incident infection, reinfection, recurrent infection, persistent infection, or as natural variability associated with the menstrual cycle. It is not possible to ascertain the type of infection for any given interval. Moreover, inclusion of women with multiple or persistent infections may attenuate or obscure the identification of a negative association between condom use and the acquisition of bacterial vaginosis. However, exclusion of these women or intervals may limit our understanding of bacterial vaginosis and its association with condom use, particularly among women with recurrent (rather than persistent) infections. Thus, we conducted 2 separate analyses, 1 including both incident and recurrent case intervals, and the other including only first incident case intervals. The “incident/recurrent incident” analysis evaluated all incident or recurrent incident case intervals irrespective of whether or not a woman had bacterial vaginosis at baseline. This analysis captured only forward crossovers (ie, lacking bacterial vaginosis to obtaining bacterial vaginosis), and women were allowed to have multiple case intervals if a resolution of bacterial vaginosis and its reacquisition occurred multiple times during follow-up. All control intervals preceding the occurrence of the incident or recurrent case period were used for this analysis. Additionally, we separately considered control intervals where (1) normal flora was diagnosed at the end of the interval, or (2) either normal flora or intermediate flora was diagnosed at the end of the interval. Furthermore, all analyses were repeated using a single control interval that occurred immediately prior to the case interval (within 6 months); however, these results were similar to those including all control intervals and are not reported.
Our second “incident” analyses evaluated the association of condom use on first incident infection. This analysis focused only on the first incident case interval among women without bacterial vaginosis at baseline, and captured only forward crossovers. This analysis included only intervals up until the first incident interval for women who lacked bacterial vaginosis at baseline and developed bacterial vaginosis during follow-up; thus all control intervals that preceded the first occurrence and case interval for bacterial vaginosis were included.
To evaluate the association between condom use and the presence of bacterial vaginosis-associated vaginal flora or chlamydial/gonococcal infection, we repeated both the incident/recurrent incident and first incident case-crossover analyses described above for each of the following outcomes: (1) G. vaginalis, (2) M. hominis, (3) anaerobic gram-negative rods (pigmented), (4) anaerobic gram-negative rods (nonpigmented), and (5) chlamydial and/or gonococcal infection. Loss of hydrogen peroxide-producing lactobacilli was also evaluated among women who had growth of H2O2-producing lactobacilli at baseline or during follow-up.
Given that time-independent factors (both measured and unmeasured) are controlled for with the case-crossover method, only time-dependent variables were considered for confounding adjustment. Potential confounders were entered into conditional logistic regression models based on biologic plausibility and univariate analyses that indicated associations with the main outcome variable (P < 0.10). Variables were eliminated if the P value was consistently greater than 0.10 and did not alter the effect size of the primary independent variable (condom use). All factors associated with sexual activity (number of partners, new partners, and use of spermicide) were retained for the assessment of vaginal flora because they were variables of particular interest. Self-reported symptoms associated with bacterial vaginosis (abnormal discharge and odor) were evaluated as both potential confounders and effect modifiers; however, neither met our criteria for inclusion in any of the models. Final models for the evaluation of bacterial vaginosis and its associated vaginal flora included number of sex partners (>1/1), new partner (yes/no), use of spermicide (yes/no), recent douching (yes/no), and use of hormonal contraception (yes/no). Use of spermicide and hormonal contraception were not retained in the assessment of chlamydial/gonococcal infection.
Conditional logistic regression analyses were conducted using Stata, Version 8, College Station, TX.
Overall, the 871 women had a total of 5402 visits. Condom use (yes/no) information was collected in 4495 (83%) of the visits; in 55% of these visits, women reported having used condoms in the past 2 months. Women who smoked, had ever been pregnant, had sex during menses, had sex 2 or more times per week, and douched once or more per month were less likely to consistently use condoms (results not shown). Women who had 2 or more sex partners or had a new sex partner were more likely to consistently use condoms.
Microbiology samples were scheduled to be obtained for 3069 of the 4495 visits in which condom use was reported. Of these 3069 visits, samples were obtained in 2806 visits (91%) and women were diagnosed with bacterial vaginosis in 1059 (38%). The prevalence of bacterial vaginosis ranged from 33% at the 6-month follow-up to 43% at the 36-month follow-up (results not shown).
In case-crossover analyses of condom use and relationship to bacterial vaginosis, we first considered all incident and recurrent incident case and associated control intervals (Table 1). Condoms were used in 572 (58%) of the intervals in which women crossed over to bacterial vaginosis. Women who consistently used condoms had a 45% decreased risk of bacterial vaginosis (vs. normal/intermediate microflora) compared with women who did not use condoms (condom use in 10/10 sex acts: adjusted OR = 0.55, 95% CI = 0.35–0.88, P for trend = 0.02). A more profound risk reduction from condom use was noted for bacterial vaginosis when contrasted with normal vaginal microflora (condom use in 10/10 sex acts: 0.37 (0.20–0.70), P for trend = 0.004).
We repeated our analyses to specifically evaluate acquisition of bacterial vaginosis by focusing on the first incident case-interval (Table 1). Because the number of observations was reduced by more than half, these analyses were less powerful. However, estimates were very similar to the analyses with inclusion of recurrent episodes of bacterial vaginosis. We found that condoms were protective against acquisition of bacterial vaginosis when compared with normal/intermediate microflora (0.57 [0.32–1.02], P for trend = 0.07). Additionally, similar to analyses with recurrent episodes of bacterial vaginosis, a more profound risk reduction from condom use was noted for bacterial vaginosis as compared with normal vaginal microflora (condom use in 10/10 sex acts: 0.41 [0.15–0.86], P for trend = 0.01).
In incident/recurrent case-crossover analyses assessing specific vaginal microflora (Table 2), consistent condom use was associated with a decrease in the risk for anaerobic gram-negative pigmented rods (0.58 [0.36–0.94]). Reductions in risk were not observed for chlamydia/gonococcal infection. In the less powerful first incident case-crossover analyses, consistent condom use was associated with risk reductions for all vaginal microflora (OR’s = 0.54–0.75), which was most apparent for M. hominis (P for trend = 0.07) and anaerobic gram-negative rods pigmented (P for trend = 0.06). First incident case-crossover analyses did not show a clear association with chlamydial/gonococcal infection (0.81 [0.33–1.97]).
Consistent condom use reduced the risk of both incident and recurrent bacterial vaginosis in this large cohort of women at high risk for sexually transmitted infections. The association was strongest when assessing cases representing a shift from normal microflora to bacterial vaginosis, suggesting a protective effect against the acquisition of bacterial vaginosis. Additionally, our results were strongest when condoms were most consistently used, which has been observed in previous studies on condom effectiveness.13,19 Condoms are often used incorrectly, and reported consistency may exceed true consistency.20 Thus our estimate of condom effectiveness may underestimate the true association with both incident and recurrent bacterial vaginosis.
Women in our study had acute, chronic, and recurrent bacterial vaginosis, which complicated the analysis. Persistent bacterial vaginosis would obscure any protective effect on acquisition afforded by condom use, resulting in a bias towards the null. We limited our analyses to incident and recurrent incident case intervals, but even with this restriction it was not possible to distinguish among newly acquired infections, natural variation, and recurrent infections—particularly among women with multiple infections throughout follow-up. However, whether or not we assessed only incident case intervals, we observed trends in keeping with a protective effect. Even among women with unknown treatment status, condom use provided protection against incident and recurrent bacterial vaginosis. Thus, condoms may prevent the acquisition of bacterial vaginosis, in addition to potentially protecting the vaginal environment from disruptions or recurrence after a prior abnormality.
We were able to test the robustness of our findings by assessing the relationship between condom use and bacterial vaginosis -associated microorganisms. The gold standard for diagnosis of bacterial vaginosis is Nugent’s criteria based upon the gram stain, which assesses the proportion of Lactobacilli species, G. vaginalis, and curved gram-variable rods or Mobiluncus species.18 We found a risk reduction for H2O2-producing lactobacilli and G. vaginalis. Of interest was the substantial reduction in anaerobic gram-negative rods, which are commonly seen in the mix of microorganisms categorized as bacterial vaginosis.21,22 A recent study by Ness et al23 found that PID was more strongly predicted by a cluster of bacterial vaginosis -associated microflora including anaerobic gram-negative rods than by the traditional Nugent’s criteria, suggesting a strong role for anaerobic gram-negative rods in the development and adverse consequences of bacterial vaginosis.
Our study is the first to assess condom use in relation to the specific microorganisms associated with bacterial vaginosis. In cases of bacterial vaginosis acquisition via sexual contact, 1 or more of the associated organisms may be transmissible, and others not. Alternatively, another as-yet-unidentified microorganism may be the transmissible agent. Recent studies have isolated numerous previously-unidentified microorganisms associated with bacterial vaginosis, and further study of these may improve our understanding of acquisition and development of this infection.24,25
Previous studies have been inconsistent in showing a relationship between condom use and bacterial vaginosis. Several cross-sectional studies have shown condom use (and particularly consistent condom use) to be protective.26–28 However, other studies have shown no evidence that condom use protects against bacterial vaginosis.29–32 To date, only 1 longitudinal cross-sectional study and 3 cohort studies have assessed condom use in relation to bacterial vaginosis, and results were mixed. Ahmed et al29 followed a cohort of 17,264 women in rural Uganda for approximately 4 years at 10-month intervals. The risk of bacterial vaginosis among consistent condom users versus nonusers during the last year was only slightly reduced (OR = 0.89 [95% CI = 0.74–1.07]). Similarly, a small study by Hawes et al,33 which followed a cohort of 182 women recruited from an STD clinic for 2 years, found that barrier methods of contraception slightly decreased the risk for acquisition of bacterial vaginosis (HR = 0.8 [95% CI = 0.3–1.7]). Baeten et al34 followed a cohort of 948 female sex workers in Kenya for a median of 421 days and found a slight decrease in the risk for bacterial vaginosis (HR = 0.9 [95% CI = 0.7–1.0]). Schwebke and Desmond35 found that always using condoms was somewhat protective against the development of bacterial vaginosis, but only in occasional partners (RR = 0.80 [95% CI = 0.67–0.98]).
Due to the wide variability in study design among studies, the varying assessments of both bacterial vaginosis (gram stain vs. Amsel’s criteria) and condom use, and the inherent difficulty with accurately measuring condom effectiveness, some inconsistency in risk across studies is not surprising. To our knowledge, our study is the first to assess the association between condom use and bacterial vaginosis using a case-crossover study design. While our study had similar limitations with regard to accuracy of condom assessment, the case-crossover design limited the overall bias by accounting for all time-independent confounders. It is possible that residual confounding from other time-dependent factors associated with a generally healthy lifestyle may have contributed to the observed protective effect. However, we did not find any indication of such self-selected condom behavior in that condom use appeared to be associated with less healthy behavior in general (ie, multiple sex partners and new sex partners). Use of condoms with more risky behavior has also been observed in other studies of condom use,36 and not accounting for these factors likely results in an underestimation of effect rather than an overestimation.13
There are a number of limitations with our study. Condom use did not clearly reduce chlamydial or gonococcal infection in the case-crossover analyses, although the effect was in the direction of risk reduction. Only 113 women acquired incident C. trachomatis and N. gonorrhoeae, so we had limited power for detecting associations. Factors that affect the apparent effectiveness of condoms, such as inconsistent condom use, incorrect use, and infection status of partner, are difficult variables to measure and control for in studies.20,37 Effectiveness according to the number of unprotected sex acts may be a more precise measurement of condom effectiveness than consistency of use.13 A further study limitation was the length (6 months) of intervals between vaginal microbiologic assessments; these allowed only a relatively crude assessment of bacterial vaginosis infection status. Bacterial vaginosis is variable over short periods of time, and incident versus prevalent or recurrent infections, cannot be accurately assessed with long intervals, as previously mentioned. Finally, women were not treated for bacterial vaginosis or referred for treatment as part of the study, and information on treatment of bacterial vaginosis was not ascertained for this study.
Our internally consistent findings suggest that, in this cohort at high risk for bacterial sexually transmitted infections, condom use reduced the risk of bacterial vaginosis. This lends some support to the theory that bacterial vaginosis is sexually transmitted, and provides a further rationale for recommending that women use condoms to reduce the risk of bacterial vaginosis.
1. Koumans E, Markowitz L, Hogan V. Indications for therapy and treatment recommendations for bacterial vaginosis in nonpregnant and pregnant women: a synthesis of data. Clin Infect Dis
. 2002;35(suppl 2):S152–S172.
2. Pybus V, Onderdonk A. Microbial interactions in the vaginal ecosystem, with emphasis on the pathogenesis of bacterial vaginosis. Microbes Infect
3. Forsum U, Holst E, Larsson P, et al. Bacterial vaginosis—a microbiological and immunological enigma. APMIS
4. Priestly C, Kinghorn G. Bacterial vaginosis. Br J Clin Pract
5. Koumans E, Kendrick J. Preventing adverse sequelae of bacterial vaginosis. Sex Transm Dis
6. Schwebke J. Gynecologic consequences of bacterial vaginosis. Obstet Gynecol Clin North Am
7. Morris M, Nicoll A, Simms I, et al. Bacterial vaginosis: a public health review. BJOG
8. Nelson D, Macones G. Bacterial vaginosis in pregnancy: current findings and future directions. Epidemiol Rev
9. Bump R, Buesching W. Bacterial vaginosis in virginal and sexually active adolescent females: evidence against exclusive sexual transmission. Am J Obstet Gynecol
10. Larsson P, Bergstrom M, Forsum U, et al. Bacterial vaginosis: transmission, role in genital tract infection and pregnancy outcome: an enigma. APMIS
11. Shlay J, McClung M, Patnaik J, et al. Comparison of sexually transmitted disease prevalence by reported level of condom use among patients attending an urban sexually transmitted disease clinic. Sex Transm Dis
12. Holmes K, Levine R, Weaver M. Effectiveness of condoms in preventing sexually transmitted infections. Bull World Health Organ
13. Warner L, Macaluso M, Austin H, et al. Application of the case-crossover design to reduce unmeasured confounding in studies of condom effectiveness. Am J Epidemiol
14. Ness R, Hillier S, Richter H, et al. Douching in relation to bacterial vaginosis, lactobacilli, and facultative bacteria in the vagina. Obstet Gynecol
15. Ness R, Hillier S, Kip K, et al. Bacterial vaginosis and risk of pelvic inflammatory disease. Obstet Gynecol
16. Stergachis A, Scholes D, Heidrich F. Selective screening for Chlamydia trachomatis
infection in a primary care population of women. Am J Epidemiol
17. Schwebke J, Hillier S, Sobel J, et al. Validity of the vaginal gram stain for the diagnosis of bacterial vaginosis. Obstet Gynecol
18. Nugent R, Krohn M, Hillier S. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol
19. Warner L, Stone K, Macaluso M, et al. Condom use and risk of Gonorrhea
: a systematic review of design and measurement factors assessed in epidemiologic studies. Sex Transm Dis
20. Devine O, Aral S. The impact of inaccurate reporting of condom use and imperfect diagnosis of sexually transmitted disease infection in studies of condom effectiveness: a simulation-based assessment. Sex Transm Dis
21. Hill G. The microbiology of bacterial vaginosis. Am J Obstet Gynecol
22. Holst E, Wathne B, Hovelius B, et al. Bacterial vaginosis: microbiological and clinical findings. Eur J Clin Microbiol
23. Ness R, Kip K, Hillier S, et al. A cluster analysis of bacterial vaginosis-associated microflora and pelvic inflammatory disease. Am J Epidemiol
24. Fredricks D, Fiedler T, Marrazzo J. Molecular identification of bacteria associated with bacterial vaginosis. N Engl J Med
25. Verstraelen H, Verhelst R, Claeys G, et al. Culture-independent analysis of vaginal microflora: the unrecognized association of Atopobium vaginae
with bacterial vaginosis. Am J Obstet Gynecol
26. Smart S, Singal A, Mindel A. Social and sexual risk factors for bacterial vaginosis. Sex Transm Infect
27. Calzolari E, Masciangelo R, Milite V, et al. Bacterial vaginosis and contraceptive methods. Int J Gynecol Obstet
28. Shoubnikova M, Hellberg D, Nilsson S, et al. Contraceptive use in women with bacterial vaginosis. Contraception
29. Ahmed S, Lutalo T, Wawer M, et al. HIV incidence and sexually transmitted disease prevalence associated with condom use: a population study in Rakai, Uganda. AIDS
30. Bradshaw C, Morton A, Garland S, et al. Higher-risk behavorial practices associated with bacterial vaginosis compared with vaginal candidiasis. Obstet Gynecol
31. Chiaffarino F, Parazzini F, Besi P, et al. Risk factors for bacterial vaginosis. Eur J Obstet Gynecol Reprod Biol
32. Schwebke J, Desmond R, Oh M. Predictors of bacterial vaginosis in adolescent women who douche. Sex Transm Dis
33. Hawes S, Hillier S, Benedietti J, et al. Hydrogen peroxide-producing lactobacilli and acquisition of vaginal infections. J Infect Dis
34. Baeten J, Nyange P, Richardson B, et al. Hormonal contraception and risk sexually transmitted disease acquisition: results from a prospective study. Am J Obstet Gynecol
35. Schwebke J, Desmond R. Risk factors for bacterial vaginosis in women at high risk for sexually transmitted diseases. Sex Transm Dis
36. Macaluso M, Demand MJ, Artz LM, et al. Partner type and condom use. AIDS
37. Crosby R, DiClemente R, Holtgrave D, et al. Design, measurement, and analytical considerations for testing hypotheses relative to condom effectiveness against non-viral STIs. Sex Transm Infect