Klebanoff, Mark A. MD, MPH*; Andrews, William W. PhD, MD†; Yu, Kai Fun PhD*; Brotman, Rebecca M. MPH*; Nansel, Tonja R. PhD*; Zhang, Jun MD, PhD*; Cliver, Suzanne P. MA†; Schwebke, Jane R. MD†
From the * National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland; and the † University of Alabama at Birmingham, Birmingham, Alabama
Supported by Contract NO-1-HD-8-3293 and by Intramural funds from the National Institute of Child Health and Human Development.
Correspondence: Mark A. Klebanoff, MD, MPH, DESPR, NICHD, NIH, 6100 Bldg., Room 7B05, Bethesda, MD 20892. E-mail: firstname.lastname@example.org
Received for publication September 8, 2005, and accepted January 31, 2006.
BACTERIAL VAGINOSIS (BV) is a condition in which the normal vaginal flora changes from a predominance of hydrogen peroxide-producing lactobacillus species to high concentrations of a variety of anaerobic organisms, Gardnerella vaginalis, and Mycoplasma hominis.1 In addition to being a source of minor morbidity, BV has been associated with a variety of adverse health outcomes, including HIV acquisition and preterm birth.2 Although the etiology of BV is unknown, increased occurrence of BV among women who douche have been observed in both cross-sectional3–5 and longitudinal studies.6 However, it is difficult to distinguish by observational research whether vaginal douching causes BV or whether subtle symptoms of BV7 might cause a woman to douche. Another approach would be to randomize women who do not douche to a douche or no-douche group. Because there are no known health benefits of douching and at least a theoretical risk,8 such a study might not be practical. Alternatively, one could begin with women who douche regularly and randomize them to “continue to douche” or “stop douching” groups. We report on a pilot study of such a design. This pilot study is intended to determine feasibility of a larger study, not to provide a definitive answer to the role of douching on the vaginal flora.
Materials and Methods
Women in this study were drawn from those who participated in the Longitudinal Study of Vaginal Flora, which has been described previously.7 In brief, the Flora Study recruited nonpregnant 15- to 44-year-old women who initially presented for a routine healthcare visit, not related to symptoms of genital tract disease, at one of 12 clinics in the Birmingham, Alabama, area from August 1999 to February 2002. Women with significant medical conditions (none of which were related to the presence or history of any sexually transmitted disease) or conditions that would preclude informed consent were ineligible. The purpose of the Flora Study was to evaluate the natural history of vaginal flora, including behaviors associated with the incidence, persistence, and/or remission of BV. Women were seen for a baseline visit and then for 4 follow-up visits at 3-month intervals. At every visit, the women underwent a pelvic examination; collection of genital tract specimens for gas-liquid chromatography; Gram stain; lactobacillus, trichomonas, and Neisseria gonorrhoeae cultures; Chlamydia trachomatis urinary ligase chain reaction; and a detailed interview.
After completion of their last Flora Study visit, women who had either normal vaginal flora (Nugent Gram stain9 score 0–3) or bacterial vaginosis (score 7–10) at this visit and who reported at one or more of their study visits douching at least weekly with either a homemade or commercial vinegar and water solution were invited to participate. Recruitment for this pilot study began in May 2002 and continued to April 2003. To increase enrollment, we also recontacted women whose last Flora Study visit had occurred in the past and invited them to participate. All interested and eligible women were seen in the research clinic, at which time another set of specimens was collected if they had not recently had their last study visit. Only women with normal (0–3) or BV (7–10) Gram stain scores were eligible. Those with a positive test for chlamydia, gonorrhea, or trichomoniasis at this examination were also ineligible for this pilot study, but women with a history of these diseases or a history of pelvic inflammatory disease were eligible. Both the Flora Study and this pilot study were approved by the Institutional Review Boards of the University of Alabama at Birmingham, the Jefferson County Health Department, and the National Institute of Child Health and Human Development; and all women provided separate written informed consent for each component.
Consenting women were randomized, using sealed envelopes previously prepared in blocks of 10 from a computer-generated randomization sequence, either to continue douching per their usual practice or to discontinue douching for 8 weeks. There were 2 randomization strata. one for women with Gram stain scores of 7 to 10 and the other for scores zero to 3 at the prior study visit. Participants completed weekly diaries on behaviors, symptoms, douching, menstruation, and medication use. They were given cotton swabs and glass microscope slides and instructed to swab their vaginas and roll the swab on a slide once weekly, on the same day each week, during weeks one to 3 and 5 to 7. Weekly, rather than daily, slides were requested because our interest was in long- term changes in flora, and we wanted to minimize burden on the women. On weeks 4 and 8, the women were seen in the clinic, where the slides and diaries were collected, a pelvic examination was done, and a Gram stain obtained. In addition, tests for gonorrhea, chlamydia, and trichomonas were obtained at the week 8 visit.
Characteristics of the 2 groups were compared with the chi-squared or Student t test as appropriate. We used Poisson processes to approximate the behaviors and events reported for each woman during the weeks of follow up. To compare reported events and behaviors during the weeks of follow up, P values were computed from the uniformly most powerful unbiased tests for Poisson distributions.10 Separate comparisons were done for women with Gram stain scores zero to 3 and 7 to 10 at baseline.
We contacted 232 women who had participated in the Flora Study and who reported weekly douching during at least one study visit. There were 213 women who expressed possible interest in participating in the douching trial, 143 of whom had no conditions requiring exclusion. No longer douching at least once per week (n = 49) was the most common reason for exclusion. Of the 143 potentially eligible women, 113 agreed to come to clinic and to stop douching if so randomized, and 81 kept their clinic appointments. Seventy women were still eligible after the examination (women with intermediate flora were not eligible), and 48 agreed to be randomized. We enrolled 25 women with normal flora (12 assigned to continue douching, 13 to stop) and 23 with BV (10 assigned to continue douching, 13 to stop). Characteristics of the women are presented by douching group in Table 1. One woman reported a history of pelvic inflammatory disease, 5 reported a history of gonorrhea, and 7 a history of chlamydia.
Women assigned to continue douching reported douching in 77% of study weeks and those assigned to stop denied douching in 94% of study weeks. Behaviors and symptoms experienced by the women are presented in Table 2. The groups were similar on most measures. However, among women who had scores 7 to 10 at entry, receptive oral sex was reported in more person-weeks, and tampon use in fewer person-weeks, among women assigned to continue douching. Curiously, vaginal discharge was reported during more person-weeks among women with normal flora who were assigned to continue douching. Cultures at the 8-week visit from one woman in the continued douching group grew N. gonorrhoeae, and from 3 women (2 in the continued douching, one in the stop douching groups) grew T. vaginalis. Chlamydia ligase chain reaction from one woman in the continue douching group was positive (this woman also had the positive culture for gonorrhea).
The median number of slides each woman provided was 7.0 (interquartile range 4.5–8). Among women who had Gram stain scores of 7 to 10 at entry, the distribution of scores at each week according to douching assignment is shown in Figure 1. At nearly every study week, women assigned to continue douching were slightly less likely to continue to have BV and more likely to revert to normal flora than women assigned to stop, but in almost no case was the difference statistically significant. In general, prevalence of normal flora was low in either group.
The distribution of Gram stain scores at each study week among women who had normal flora at entry is shown in Figure 2. Among initially normal women, there were no consistent differences in BV, or in normal flora, between the 2 groups for the first 6 weeks of study. However, there was a slight indication that change to BV was more common, and persistence of normal flora less common, in weeks 7 and 8 in the group assigned to continue douching. However, the difference was not statistically significant (P = 0.38 and 0.11 for BV at weeks 7 and 8, respectively).
At the baseline physical examination, 100% of the women with Gram stain scores 7 to 10 also had BV by Amsel11 criteria, whereas only 6% of women with scores of zero to 3 had a clinical diagnosis of BV. Among women who had scores 7 to 10 at baseline, BV by Amsel criteria was slightly, but not significantly, less common at the 8-week examination among women randomized to continue douching (67%) than among women randomized to stop (80%). Among women with scores zero to 3 at baseline, BV by Amsel criteria was slightly, but not significantly, less common among women randomized to continue douching (25%) than among women randomized to stop (33%). Although the latter result disagrees with the Gram stain results (Fig. 2), the number of women with a clinical diagnosis of BV in this group was small (total of 6 women in the douching and stop douching groups).
We have demonstrated that a trial in which women who douche regularly are randomized to stop for 8 weeks is feasible. Although many such women need to be screened for entry, those who participate in such a study, once enrolled, will adhere acceptably well to their assigned behavior. There is no biomarker for douching to verify the woman's report, but studies have indicated that self-reported adherence to treatment is generally accurate12; and the women were compensated for participation regardless of adherence to treatment, thereby removing an incentive to misreport. Others have also demonstrated that in a randomized trial, women can be persuaded to give up douching.13
Our study was intended as a pilot to determine feasibility. It did not have statistical power to demonstrate either beneficial or harmful effect of douching cessation nor to analyze the impact of other behaviors on vaginal flora. Therefore, we do not claim that the absence of a dramatic difference in BV between the continue and stop douching groups means that douching does not cause BV. However, the absence of any dramatic beneficial or harmful effects argues for the ethical acceptability of a larger, definitive trial. Among women who had normal flora at entry, there was some evidence that the Gram stain scores began to diverge between the douching groups at week 7, but the number of women was small and differences were not statistically significant. This suggests that a definitive trial should last longer than 8 weeks, although how much longer is unknown.
We elected to obtain weekly, rather than daily, smears and diaries to minimize burden on the participants. We also enrolled women based on a single Gram stain showing normal or BV flora. Both of these decisions, although simplifying the study, ignore daily variation in Gram stain scores.14 However, data in Figures 1 and 2 indicate that although there is some fluctuation between adjacent scores, fluctuation between BV (score 7–10) and normal (score 0–3) over a period of 1 or 2 weeks is not common. This suggests that weekly scores would be adequate in a larger trial. Similarly, we elected to recruit only women with normal or BV flora and to exclude women with intermediate flora. Intermediate flora is often unstable over time,15 and we may have excluded the very women whose flora might have been most affected by douching. Our interest was to determine the causal role of douching in the development of BV, and we believed that excluding women with intermediate flora would provide a clearer answer to this question. Nevertheless, it may be useful to include those with intermediate scores in a larger study.
Most reported symptoms and events occurred at similar frequency among women assigned to continue or stop douching, but there were a few differences. Although among women with BV, receptive oral sex was more frequent among women assigned to continue douching, this was mainly the result of 3 women, one who reported oral sex in every week, one during 4 different weeks, and a another during 1 week only. The remaining women with BV did not report it. Similarly, although among women with normal flora, discharge was reported in more weeks in those assigned to continue douching, this was the result of one woman reporting discharge in 5 weeks, another in 3 weeks, and a third in 2 weeks. Conversely, among women assigned to stop douching, discharge was reported by one woman during 1 week. Reported discharge was not more common when the weekly Gram stain score was 7 to 10 (odds ratio = 1.14) than when it was zero to 6. Although these results are intriguing, it appears that the differences noted in Table 2 are the result of a few women who reported the symptoms and behaviors on multiple occasions rather than a generalized difference in many women in the douching versus no douching groups. Therefore, these differences we observed in oral sex and discharge need to be confirmed in larger studies.
In summary, our pilot study of randomizing women who douche regularly either to stop or continue douching demonstrates that a larger, definitive study is both practically and ethically feasible. Although such a study should continue for longer than 8 weeks, the optimal length is unknown and will probably need to balance biologic plausibility with pragmatic concerns.
1. Hill GB. The microbiology of bacterial vaginosis. Am J Obstet Gynecol 1993; 169:450–454.
2. Marrazzo JM. Evolving issues in understanding and treating bacterial vaginosis. Expert Rev Anti Infect Ther 2004; 2:913–922.
3. Holzman CB, Leventhal JM, Qiu H, et al. Factors linked to bacterial vaginosis in nonpregnant women. Am J Public Health 2001; 91:1664–1670.
4. Ness RB, Hillier SL, Richter HE, et al. Douching in relation to bacterial vaginosis, lactobacilli, and facultative bacteria in the vagina. Obstet Gynecol 2002; 100:765–772.
5. Zhang J, Hatch M, Zhang D, et al. Frequency of douching and bacterial vaginosis in African-American women. Obstet Gynecol 2004; 104:756–760.
6. Hawes SE, Hillier SL, Benedetti J, et al. Hydrogen peroxide-producing lactobacilli and acquisition of vaginal infections. J Infect Dis 1996; 174:1058–1063.
7. Klebanoff MA, Schwebke JR, Zhang J, et al. Vulvovaginal symptoms in women with bacterial vaginosis. Obstet Gynecol 2004; 104:267–272.
8. Martino JL, Vermund SH. Vaginal douching: Evidence for risks or benefits to women's health. Epidemiol Rev 2002; 24:109–124.
9. 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.
10. Lehmann EL. Testing Statistical Hypotheses, 2nd ed. New York: Wiley Publishing Co, 1986:151–156.
11. Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, Holmes KK. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
12. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med 2005; 353:487–497.
13. Grimely DM, Oh MK, Desmond RA, Hook E, Vermund SH. An intervention to reduce vaginal douching among adolescent and young adult women: A randomized, controlled trial. Sex Transm Dis 2005; 32:752–758.
14. Schwebke JR, Morgan SC, Weiss HL. The use of sequential self-obtained vaginal smears for detecting changes in the vaginal flora. Sex Transm Dis 1997; 24:236–239.
15. Hillier SL, Krohn MA, Nugent RP, Gibbs RS. Characteristics of three vaginal flora patterns assessed by gram stain among pregnant women. Vaginal Infections and Prematurity Study Group. Am J Obstet Gynecol 1992; 166:938–944.