Koumans, Emilia H. MD, MPH*; Sternberg, Maya PhD*; Bruce, Carol BSN, MPH*; McQuillan, Geraldine PhD†; Kendrick, Juliette MD*; Sutton, Madeline MD, MPH*; Markowitz, Lauri E. MD*
BACTERIAL VAGINOSIS (BV), A COMMON cause of vaginal symptoms among reproductive-aged women,1 is characterized by the loss of normal vaginal flora, predominantly hydrogen peroxide-producing Lactobacillus spp., and the increase in the number and species of other bacteria in vaginal fluid.2 The decrease in lactobacilli and increase in numerous facultative and anaerobic bacteria, some of which have only been recently characterized,2 may lead to changes in the characteristics of vaginal fluid, such as thin discharge and odor. However, many women with BV do not have symptoms.
The diagnosis of BV can be made using clinical criteria3 or in the laboratory by scoring bacterial morphotypes from a Gram stain of vaginal fluid.4 A score of 0–3 is defined as normal, 4–6 as intermediate, and 7–10 as BV.4 Treatment is recommended for symptomatic women,5 but 1 month after therapy only 80% of women have regained normal flora; recurrences after 1 month are also common.5
The changes in vaginal microflora of BV are associated with numerous health problems.2,5 BV increases the risk of acquiring the human immunodeficiency virus (HIV),6 and among HIV infected women, BV increases HIV viral shedding.7,8 BV increases the risk of acquiring other sexually transmitted infections (STI) such as gonorrhea, trichomoniasis, and herpes simplex virus type 2 (HSV-2).9,10 In pregnant women, BV increases the risk of miscarriage,11 preterm labor, preterm delivery, and postpartum complications such as endometritis and wound infections.5,12,13
BV is not a reportable condition in the United States. We provide an in-depth examination of symptomatology and risk factors for BV using the 2001–2004 National Health and Nutrition Examination Survey (NHANES).
Study Population and Study Design
NHANES uses a complex, stratified, multistage probability sample design with unequal probabilities of selection to obtain a nationally representative sample of the US civilian noninstitutionalized population.14 Certain subgroups of people, such as adolescents, blacks, and Mexican Americans, are oversampled. All women aged 14–49 years selected for NHANES 2001–2004 were eligible for participation in this study. In 2001–2004 a total of 4646 women aged 14–49 years were interviewed, 4474 (96.3%) were examined in the mobile examination center, 3961 completed an additional private interview using an audio computerized self-interview (ACASI), and 3739 (80.5% of total or 83.6% of those examined) completed the examination and specimen collection.
Specimen Collection and Processing
Participating women were given oral and written instructions on using a cotton-tipped swab to collect their vaginal fluid while in a private bathroom. They were instructed to insert the swab at least 1 inch into the vagina, turn the swab against the walls of the vagina while counting to 10, and then put the swab into the collection tube. NHANES personnel applied the swab to narrow range pH paper to determine pH and then rolled the swab onto a glass slide to air dry. Slides were Gram stained and read using Nugent’s criteria3 at one central laboratory where readers were blind to any information about the women. As part of quality control, 2.5% of the BV samples were reread. There was 100% agreement between the two readings. For our analysis, a Gram stain score of 7–10 was defined as BV. We also analyzed the proportion of women with a Gram stain score of 7–10 and a vaginal pH of >4.5. Women were notified of their results by a password-protected call-in system used for STI.
Demographic and Behavioral Data
Race and ethnicity were self-reported and classified as non-Hispanic black, non-Hispanic white, and Mexican American. Women not reporting one of these categories were classified as “other.” Poverty index ratio was calculated by dividing total family income by the poverty threshold index adjusted for family size at year of interview.15 Assessment of educational attainment was limited to those aged 18 and older. Population estimates were generated by multiplying 2002 census estimates of civilian noninstitutionalized women aged 14–49 by the weighted prevalence estimate. Symptom and behavioral data were self-reported. Having sex was explained to participants as including vaginal, oral, or anal sex. Using the ACASI, participants were asked “Have you ever had sex?” The date of onset of the most recent menstrual cycle and the date of the exam were used to calculate day of menstrual cycle.16
Statistical analyses were conducted using SAS (version 9.1, SAS Institute, Cary NC) and SAS callable SUDAAN (RTI, Research triangle Park NC), and account for the complex survey design by incorporating the survey weights and using a Taylor series linearization to calculate variance estimates. The 2001–2004 medical examination survey weights provided by NHANES were used to account for the unequal probabilities of selection and adjustment for nonresponse. A prevalence estimate was considered unreliable if the relative standard error was ≥30% of the prevalence estimate and those estimates are not presented. The sampling design for 2001–2004 yields 30 degrees of freedom for variance estimation and model building.
Confidence intervals were calculated using a log transformation with the standard error of the log prevalence based on the delta method and applying SUDAAN estimated standard errors.17 Significance tests for associations between BV and other variables were based on a Wald χ2 statistic. A 0.05 level was used to define statistical significance.
For the multivariate model we used SUDAAN for logistic regression and considered all variables that had a P value of less than or equal to 0.10 in bivariate analysis. Because of missing data in the poverty index ratio, we used a variable with more complete data: family income greater than or equal to $20,000 or below $20,000. Using a backwards elimination approach, we removed variables in order of least statistical significance. When all variables in the model were statistically significant at 0.05 level, all pairwise interactions in the model were explored. When an interaction was entered into the model the main effects variables were always retained. A pairwise interaction was retained only if the overall P value for the interaction was <0.05.
The prevalence of BV was 29.2% (95% confidence interval 27.2%–31.3%); corresponding to 21.4 million women with BV. The proportion of women with a Nugent’s score of 7–10 and a pH >4.5 was 22.3% (95% CI 20.7%–24.1%). In general, for the variables mentioned below, prevalence of BV decreased by 25% when we added the criterion of pH >4.5. The sensitivity and specificity of pH >4.5 for detecting BV was 76.7% (95% CI 71.4%–81.1%) and 70.3% (95% CI 67.9%–72.7%).
BV prevalence varied significantly by sociodemographic characteristics, earlier initiation of sexual activity, increasing numbers of sex partners, and a previous female sex partner (Table 1). There was a significant association between increasing numbers of lifetime sex partners and BV among non-Hispanic whites and non-Hispanic blacks, but not among Mexican Americans (Fig. 1). Women who reported more than one sex partner in the previous 12 months (recent) and more than one lifetime sex partner had a prevalence of 39.6%, those who reported no or one recent sex partner and more than one lifetime partner had a prevalence of 29.1%, those who reported no or one sex partner in the last 12 months and one lifetime partner had a prevalence of 22.4%, and those who reported never having had sex had a prevalence of 18.8% (P = 0.0001). Among the 163 women with BV who reported never having had sex, 90% were aged 14–19 years, two were infected with Chlamydia trachomatis, one with Neisseria gonorrhoeae, and four had serologic evidence of previous infection with HSV-2. Fifty-six women with BV who reported never having had sex were married or cohabitating. After excluding women with another STD and those who were married or cohabiting, prevalence among 14–19 year-old women was 18.5% (95% CI 13.5%–25.5%).
BV was also associated with having had a prior pregnancy, being single, smoking, and having a high BMI; current oral contraceptive use was inversely associated with BV prevalence (Table 1). Current pregnancy or day in menstrual cycle was not associated with BV prevalence.
BV prevalence varied by use of douches and other over-the-counter feminine hygiene products (Table 2). Overall, 22.0% (95% CI 19.5%–24.9%) of women had douched in the previous 6 months. Race/ethnicity, age, education, smoking status, income, poverty, number of lifetime sex partners, young age at first sex, BMI, and being born in the United States were all significantly associated with douching. However, interactions between these variables and douching were not significantly associated with BV prevalence.
BV prevalence did not vary by vaginal symptoms (Table 2); 13.9% (95% CI 12.7–15.1) of all women, 15.7% (95% CI 13.1–18.8) of women with BV, and 13.2% (95% CI 11.8–14.6) of women without BV reported vaginal symptoms in the previous month (P = 0.11). The prevalence of symptomatic BV was 4.4%. Among women reporting vaginal symptoms the overall prevalence of BV was 32% (95% CI 27.3–37.1). Among these women, those who douched had a prevalence of 38% (95% CI 29.8–47.0) and those who did not douche had a prevalence of 28.5% (95% CI 22.4–35.5). (P = 0.11). Among women reporting no vaginal symptoms the overall prevalence of BV was 27.7% (95% CI 25.5–29.9). Among these women, the prevalence of BV was higher among those who douched, 46.4% (95% CI 42.3–50.5) versus 23.1% (95% CI 20.9–25.4). (P <0.0001).
In the multivariate model, independent risk factors for BV were race/ethnicity, frequency of douching, number of lifetime sex partners, birth control pill use, and educational attainment (Table 3). There were no significant interactions between these variables.
We found the prevalence of BV among 14 to 49-year-old US women to be 29.2%, similar to another estimate.18 This corresponds to 21.7 million women with BV; however, most (84%) reported no symptoms. The proportion of women with a Nugent’s score of 7–10 and a pH >4.5 was 22.3%, (corresponding to 16.6 million women), and the prevalence of symptomatic BV was 4.4% (3.3 million women). Reported BV prevalence varies by clinical setting, from 17% to 19% in family planning clinics to 24%–47% in sexually transmitted disease clinics.1,3,10 The two largest studies, both among women in urban prenatal care, used Gram stain and pH criteria and found prevalences of 34.4% (n = 15,864)19 and 18.5% (n = 13,747).20 Our estimate is similar to results obtained from a population of young women entering the military.21
We also found substantial differences in BV prevalence by race (for non-Hispanic blacks prevalence 51%, OR 2.75, 95% CI 2.2–3.5) and ethnicity (for Mexican Americans prevalence 32%, OR1.60, 95% CI 1.3–2.1), compared with non-Hispanic whites (prevalence 23%), after controlling for other risk factors.19,20,21 One study that controlled for an extensive list of behavioral and demographic risk factors found that women reporting black race had an OR of 2.2 (95% CI 15–31)22 for BV or intermediate flora. The prevalence among 14 to 19-year-old non-Hispanic blacks of 40% (Table 1) and among non-Hispanic blacks before sexual activity of 28.7% (Fig. 1) suggests that there may be other factors contributing to BV in addition to sexual activity.
The role of sexual activity in the acquisition of BV is not clear. We found the BV prevalence among women who never reported having had sex was 18.8%. Some women with BV who reported never having had sex had evidence of other STI or were married or cohabiting. Therefore, some women may have misstated their previous sexual activity or misinterpreted the question about sexual activity. Nonetheless, many women with BV who did not report sexual activity had no other evidence indicating they had had sex. These data do not support a uniquely causal role of sexual activity. Few studies have reported BV prevalence, incidence, or risk factors conducted among women who have never been sexually active. Our estimate corresponds with the prevalence found among sexually naïve young women entering the military, 18%.21 Another study among school girls who reported never having had sex found a prevalence of 12%23 using the criteria of four of the following five for the definition of BV: vaginal pH >4.5; abnormal discharge; clue cells; positive sniff test; or succinate/lactate ratio greater than or equal to 0.4 on gas-liquid chromatography of a vaginal specimen. These first four criteria (“Amsel” criteria3) may underestimate the prevalence of BV.24
In support of a role of sexual transmission, previous studies have found that multiple or new sex partners increased the risk of acquiring BV by a factor of 1.6–2.5.16,21,22,23,25 These studies also suggested that condom use may be protective. In our study, BV prevalence increased with number of lifetime sexual partners among non-Hispanic white and non-Hispanic black women (Fig. 1), and increased by a factor of 1.4–2.4 with increasing numbers of sex partners in multivariate analysis. However, the NHANES questionnaire did not assess recent sex act-specific condom use, so we were unable to assess the relationship between recent condom use and BV prevalence. Also supporting sexual transmission are data from a study among lesbian couples; BV was associated with sharing of sex toys, and couples had greater than 90% concordance in BV status26 suggesting transmission of an infectious agent. We found that women who had ever had sexual activity with another woman had a higher prevalence of BV. Ongoing research into novel bacteria may shed light on the cause and transmission of BV.2
Many previous studies16,22,25 have found that douching was associated with higher BV prevalence by a factor of 1.4–2.1. We found increasing douching frequency was independently associated with BV prevalence. However, because this was a cross-sectional study, we could not evaluate the temporal relationship, if any, of symptoms, the menstrual cycle, and douching to BV. Our data showed no differences in BV prevalence among symptomatic women on the basis of douching status; however, among asymptomatic women, those who douched were more likely to have BV than those who did not. The interpretation of this observation is unclear. We could not capture the dynamic nature of BV; it may appear and resolve spontaneously over days to weeks16. Of note, prevalence did not vary significantly by timing in menstrual cycle in this sample of women.
Although the cause of BV remains unclear, BV has been associated with numerous adverse health outcomes, such as increased vaginal HIV shedding.7,8,27 In prospective studies, BV increased the risk of acquiring HIV by a factor of 2,6,9 HSV-2 by 2,10 gonorrhea by 1.7,9 and trichomoniasis by 1.8.9 Increased susceptibility to these STI may be related to the decrease in protective lactobacilli,9 the increase in abnormal flora,9 and the reduction in vaginal secretory leukocyte protease inhibitor.28 In our study, the prevalence of BV among non-Hispanic black women was more than twice that of non-Hispanic white women. This higher prevalence may in part contribute to higher prevalence of HIV, HSV, trichomoniasis, and gonorrhea among non-Hispanic black women. A recent trial has shown that treating asymptomatic women who have BV with a suppressive treatment regimen reduced the incidence of acquiring a new sexually transmitted infection (Jane Schwebke, MD, personal communication).
The higher prevalence of BV among non-Hispanic black women may also relate to racial disparities in the rate of preterm birth.29 In this sample, women who had delivered a preterm and low-birth weight infant had a higher prevalence of BV than either women who had not had these pregnancy outcomes or had never been pregnant. BV in pregnant women is associated with preterm birth and other complications.5,13–15 In this sample, women who had delivered a preterm and low birth weight infant had a higher prevalence of BV than women who had not had these pregnancy outcomes or had never been pregnant. Racial disparities in the rate of preterm birth29 may be related to the higher prevalence of BV among non-Hispanic black women. An intervention targeting BV-related preterm birth may play a role in reducing these disparities.
This study provides additional information on the correlates of BV prevalence. Most women were asymptomatic. Current methods for diagnosing BV do not discriminate between possibly different polymicrobial2 vaginal syndromes that could have different clinical outcomes. Further research should clarify the cause of and risk factors for BV, therapies that have high and long-lasting effectiveness, methods to identify those at risk, and methods to reduce associated adverse health outcomes.
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the agency.
1. Sobel JD. Vaginitis. N Engl J Med 1997; 337:1896–1903.
2. Fredricks DN, Fiedler TN, Marrazzo JM. Molecular identification of bacteria associated with bacterial vaginosis. N Engl J Med 2005; 353:1899–1911.
3. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
4. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991; 29:297–301.
5. Centers for Disease Control and Prevention. Sexually transmitted diseases treatment guidelines 2006. MMWR 2006; 55(RR–11): 50–54.
6. Myer L, Denny L, Telerant R, et al. Bacterial vaginosis and susceptibility to HIV infection in South African women: A nested case–control study. J Infect Dis 2005; 1372–1380.
7. Cu-Uvin S, Hogan J, Caliendo AM, et al; for the HIV Epidemiology Research Study. Association between bacterial vaginosis and expression of human immunodeficiency virus type 1 RNA in the female genital tract. Clin Infect Dis 2001; 33:894–896.
8. Cohn JA, Hahemi FB, Camarca M, et al. HIV-inducing factor in cervicovaginal secretions is associated with bacterial vaginosis in HIV-1-infected women. J Acquir Immune Defic Syndr 2005; 39:340–346.
9. Martin HL, Richardson BA, Nyange PM, et al. Vaginal lactobacilli, microbial flora, and risk of human immunodeficiency virus type 1 and sexually transmitted disease acquisition. J Infect Dis 1999; 180:1863–1868.
10. Cherpes TL, Meyn LA, Krohn MA, et al. Association between acquisition of herpes simplex virus type 2 in women and bacterial vaginosis. Clin Infect Dis 2003; 37:319–325.
11. Hay PE, Lamont RF, Taylor-Robinson D, et al. Abnormal bacterial colonization of the genital tract and subsequent preterm delivery and late miscarriage. BMJ 1994; 308:295–298.
12. Minkoff H, Grunebaum AN, Schwarz RH, et al. Risk factors for prematurity and premature rupture of membranes: A prospective study of the vaginal flora in pregnancy. Am J Obstet Gynecol 1984; 150:965–972.
13. Gravett MG, Nelson HP, DeRouen T, et al. Independent associations of bacterial vaginosis and Chlamydia trachomatis
infection with adverse pregnancy outcome. JAMA 1986; 256:1899– 1903.
15. Bureau of the Census. Poverty in the United States: 1990. In: Bureau of the Census, ed. Current Population Reports, Series P-60. Washington, DC: U.S. Government Printing Office, 1991.
16. Schwebke JR, Richey C, Weiss H. Correlation of behaviors with microbiological changes in vaginal flora. J Infect Dis 1999; 180:1632–1636.
17. Korn EL, Graubard BI. Analysis of Health Surveys. Wiley: New York, 1999:66.
18. Allsworth JE, Peipert JF. Prevalence of bacterial vaginosis: 2001–2004 National Health and Nutrition Examination Survey data. Obstet Gynecol 2007; 109:114–120.
19. Klebanoff MA, Hillier SL, Nugent RP, et al. Is bacterial vaginosis a stronger risk factor for preterm birth when it is diagnosed earlier in gestation? Am J Obstet Gynecol 2005; 192:470–477.
20. Goldenberg RL, Klebanoff MA, Nugent R, et al. Bacterial colonization of the vagina during pregnancy in four ethnic groups. Vaginal Infections and Prematurity Study Group. Am J Obstet Gynecol 1996; 174:1618–1621.
21. Yen S, Shafer M-A, Moncada J, et al. Bacterial vaginosis in sexually experienced and non-sexually experienced young women entering the military. Obstet Gynecol 2003; 102:927–933.
22. Ness RB, Hillier S, Richter HE, et al. Can known risk factors explain racial differences in the occurrence of bacterial vaginosis? J Natl Med Assoc 2003; 95:201–212.
23. Bump RC, Buesching WJ III. Bacterial vaginosis in virginal and sexually active adolescent females: Evidence against exclusive sexual transmission. Am J Obstet Gynecol 1988; 158:935–939.
24. Schwebke JR, Hillier SL, Sobel JD, et al. Validity of the vaginal gram stain for the diagnosis of bacterial vaginosis. Obstet Gynecol 1996; 88:573–576.
25. Hawes SE, Hillier SL, Benedetti J, et al. Hydrogen peroxide-producing lactobacilli and acquisition of vaginal infections. J Infect Dis 1996; 174:1058–1063.
26. Marrazzo JM, Koutsky LA, Eschenbach DA, et al. Characterization of vaginal flora and bacterial vaginosis in women who have sex with women. J Infect Dis 2002; 185:1307–1313.
27. Sha BE, Zariffard MR, Wang QJ, et al. Female genital-tract HIV load correlates inversely with Lactobacillus
species but positively with bacterial vaginosis and Mycoplasma hominis
. J Infect Dis 2005; 191:25–32.
28. Draper DL, Landers DV, Krohn MA, et al. Levels of vaginal secretory leukocyte protease inhibitor are decreased in women with lower reproductive tract infections. Am J Obstet Gynecol 2000; 183:1243–1248.
29. Hoyert DL, Matthews TJ, Menacker F, et al. Annual summary of vital statistics: 2004. Pediatrics 2006; 117:168–183.