VAGINAL DOUCHING IS A common practice among women in the United States, and over one-quarter of reproductive-aged women report douching regularly.1 Douching has been associated with a number of adverse outcomes including pelvic inflammatory disease (PID), bacterial vaginosis (BV), and preterm birth2–4; however, controversy exists about the extent of risk and the likelihood that the association is causal.
Since douching and BV have been linked to the same adverse outcomes, the development of BV has been suggested as an intermediate step toward the development of adverse sequelae following douching.4,5 Cross-sectional studies consistently indicate that douching is associated with BV.6–11 However, it is difficult to determine the temporal and causal associations between douching and BV; that is, do women douche as a result of symptoms associated with BV, or does douching cause an imbalance in flora leading to the development of BV? Using longitudinal data, we previously reported that among a group of women with normal vaginal microflora, douching 2 or more times per month was not associated with the development of BV.12 However, we did not investigate how preexisting BV or already disturbed flora might differentially impact any effect of douching. In addition, in previous cross-sectional studies, douching has been associated with several BV-associated organisms4; we did not explore this in our previous longitudinal analysis, and it remains unclear how douching influences the acquisition of specific vaginal flora.
In this analysis, we employed data from the GYN Infections Follow-Through (GIFT) study to explore the effect of douching on the development of BV conditional upon previous experience of abnormal vaginal flora; we also explored the relationship between douching and the acquisition of a variety of BV-associated organisms.
Materials and Methods
The methods used for subject enrollment, data collection, and follow-up have been reported in detail elsewhere.4,13 Briefly, women 13 to 36 years of age were recruited into the GIFT Study from 5 US sites between May 1999 and June 2001. Human subjects’ approval was obtained at each institution, and all women signed informed consent. Women enrolled were at high risk for chlamydial cervicitis14 and had to have a score of at least 3 points on a algorithm wherein points were 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 STD, including N. gonorrhoeae, C. trachomatis, and T. vaginalis = 2. Women were excluded if they were pregnant, married, virginal, or on antibiotics at baseline. Among the 1628 women who were eligible for the study, 1193 (73.3%) completed a baseline questionnaire and are the focus of these analyses.
At baseline and at 6, 12, 24, and 36 months thereafter, each subject obtained her own vaginal specimens with a cotton swab.15 Smears from these swabs were Gram stained, and a microscopy score of 0 to 10 was assigned by laboratory staff using the standardized method described by Nugent et al.16 A score of 0 to 3 was interpreted as normal vaginal flora; a score of 4 to 6 was designated as intermediate; and a score of 7 to 10 was considered to be BV. Swabs were also characterized for the following: Lactobaccillus species, anaerobic Gram-negative rods, Gardnerella vaginalis, group B streptococcus, Enterococcus species, Escherichia coli, Candida species, Mycoplasma hominis, and Ureaplasma urealyticum. The amount of growth for each of these microorganisms was recorded on a semiquantitative scale from 0 to 4. N. gonorrhoeae and C. trachomatis were identified using a strand displacement DNA Amplification Assay (Becton Dickinson, Sparks, MD).
At baseline, women were asked about demographic factors, including age, race, education, income, pregnancy history, and smoking, and relevant lifestyle behaviors such as sexual activity and douching practices. Women were also asked to recall whether they had a history of vaginal infections (including BV, PID, gonorrhea, and chlamydia). Questions about pregnancy history, sexual activity, contraceptive use, recent diagnoses with STDs (self-report), and douching were repeated every 6 months during follow-up.
About douching, women were asked about their frequency, reason for douching, timing of douching, and product used. Eighty-seven percent of those who douched reported using 2 major brands consisting of purified water, sodium citrate, citric acid, vinegar, diazolidinyl urea, octoxynol-9, cetylpyridinium chloride, and edentate disodium, or purified water, vinegar, benzoic acid, “lemon mist,” octoxynol-9, cetylpyridinium chloride, sodium benzoate, disodium ethylenediaminetetra-acetic acid, and fragrance. For analyses, douching was categorized into frequency (never, <1 time per month, >1 times per month) and reason for douching according to previously published work.4,9,17 Reason for douching was created using hierarchically mutually exclusive categories of (1) abnormal symptoms (abnormal vaginal discharge, to reduce odor, and bleeding between menses), (2) before or after sex, and (3) for hygiene (general cleansing, after menses, because “it’s normal to douche,” and to prevent pregnancy).
Of the 1193 subjects, 27 (2.3%) had a baseline visit only. Among the remaining 1166 study participants, the median length of follow-up was 3.0 years (interquartile range: 2.4-3.4 years). Eighty-eight percent of the women had 4 or more visits, and 72% of the women had 3 or more vaginal swab samples.
To evaluate the association between douching and BV, other BV-associated organisms (G. vaginalis, M. hominis, anaerobic Gram-negative pigmented and nonpigmented rods), and lack of normal H2O2-producing lactobacilli, we conducted both cross-sectional and prospective analyses. Throughout follow-up, douching was evaluated concurrently with BV diagnosis and captured practices during the prior 2 months; however, douching may or may not have occurred before BV infection (i.e., depending on duration of BV). Thus, we initially conducted cross-sectional analyses using generalized estimating equations (GEEs) to estimate the adjusted odds ratios of BV throughout follow-up in relation to frequency and reason for douching. This method accounted for the correlation of multiple observations (i.e., clinic visits) per subject. Stepwise methods were used to determine final multivariate models, with consideration to biologic plausibility and magnitude of effect on the primary independent variable (douching). The final models were adjusted for age, race (black vs. white/other), smoking status (current vs. prior/never), hormonal conceptive use (yes/no), ever pregnant, condom use (proportion of sexual encounters condoms were used: 100%, <100%, and not sexually active), number of sex partners (>1, 1, and not sexually active), and clinical site. Additionally, the impact of douching frequency on BV was considered within the following subgroups: history of BV (yes/no) and vaginal flora status at the immediately preceding visit (normal flora, intermediate flora, BV).
To prospectively assess the relationship between douching and the acquisition of BV, women who had at least one follow-up visit were evaluated. Subsets of women were separately assessed for the development of BV who at baseline had (1) normal flora, and (2) intermediate flora. Similarly, women were evaluated for the development of BV-associated organisms by baseline presence or absence of: (1) G. vaginalis, (2) M. hominis, (3) anaerobic Gram-negative rods (pigmented and nonpigmented). Loss of H2O2-producing lactobacilli was determined for women who had H2O2-producing lactobacilli at baseline. Since vaginal flora was routinely assessed at fixed semiannual and annual visits, discrete time hazard models were fit using the complementary log-log link. This model parallels the continuous-time proportional hazards model while accommodating interval-censored data in which information about event occurrence is restricted to discrete time intervals.12,18,19 Time-varying douching status, which was ascertained at the immediately preceding visit to Gram-stain assessment (average length between intervals = 7 months; interquartile range = 5–8 months), and baseline status of douching were assessed in separate models. For all analyses, women could only contribute one “event” for analysis to focus on acquisition of BV rather than chronic or recurrent BV. Potential confounders were identified as described above. All models were adjusted for race, number of sex partners, baseline Gram-stain flora score, history of chlamydial/gonococcal infection, history of BV, current chlamydial/gonococcal cervical infection (vaginal swab assessment), and high school education. Analyses were conducted using the SAS System for Windows, version 8.02, Cary, NC.
Participants were predominantly aged 19 to 24 (66.1%) and black (75%). At baseline, 425 (36.2%) reported a history of BV, 164 (13.8%) reported a history of pelvic inflammatory disease, and 464 (39.2%) reported a history of chlamydial infection. Four hundred seventy women (39.8%) entered the study with BV as determined by Gram stain. Forty-four percent of the women reported douching, and of the women who douched, 464 (87.7%) douched at least once per month.
Cross-Sectional Analyses of Douching and BV, Stratified by Prior Experience of BV
In cross-sectional analyses, BV was more common among women who douched at least once per month (adjusted odds ratio 1.45, 95% CI = 1.22, 1.74) (Table 1). In analyses stratified by a woman’s prior vaginal flora status, douching 1 or more times per month was significantly associated with BV among women who had BV at the immediately preceding visit (adjusted OR 1.89, 95% CI = 1.31, 2.72), but not among women with intermediate flora (adjusted OR 1.22, 95% CI = 0.78, 1.90) or normal flora (adjusted OR 1.33, 95% CI = 0.86, 2.07) at the prior visit. The frequency of douching times prior vaginal flora status test of interaction approached statistical significance (P = 0.08). In more detailed assessment of the impact of douching, the significant association between douching and BV was consistently observed among the subset of women with BV at the immediately preceding visit regardless of the reason for douching (abnormal symptoms: adj. OR = 1.94; before of after sex: adj. OR = 2.01; hygiene: adj. OR = 1.89).
Among women with a history of BV at baseline, douching 1 or more times per month was associated with an estimated 72% increase in risk of BV. This excess risk of BV was particularly apparent when the reported reason for douching was for abnormal symptoms (adjusted OR 2.41, 95% CI = 1.60, 3.63). The frequency of douching times reason for douching test of interaction approached statistical significance (P = 0.10). Among women without a history of BV at baseline, a modest statistically significant association was observed between douching one or more times per month and BV (adjusted OR 1.32, 95% CI = 1.02, 1.63). The frequency of douching times history of BV test of interaction did not achieve statistical significance (P = 0.28).
Douching and Incident Development of BV by Baseline Flora Status
Among women whose baseline Gram-stain score was normal (0–3), douching at least once per month at the immediately preceding visit during follow-up was not associated with the development of BV (adjusted HR = 1.01) (Table 2). In contrast, among women who had intermediate vaginal flora at baseline, douching at least once per month at the preceding visit during follow-up was significantly associated with the development of BV (adjusted HR 1.51, 95% CI = 1.08–2.10). The frequency of douching times prior vaginal flora status test of interaction did not achieve statistical significance (P = 0.23). The elevated risk for developing BV in relation to douching at the prior visit among women with intermediate vaginal flora at baseline was consistent across reasons for douching (hygiene: adj. OR = 1.61 (95% CI = 1.10, 2.36); abnormal symptoms: adj. OR = 1.31; before or after sex: adj. OR = 1.20).
Douching ascertainment at baseline (i.e., remote from BV acquisition) was not associated with future development of BV in women with normal or intermediate flora at baseline (Table 2). In addition, neither douching at the immediately preceding visit nor at baseline increased the risk of acquiring any of the specific microflora studied, including G. vaginalis, M. hominis, anaerobic Gram-negative rods, and absence of H2O2-producing lactobacilli among women who previously lacked these microorganisms (results not shown).
Previous findings regarding the relationship between douching and BV are inconsistent. Cross-sectional studies have generally linked douching to BV6–11 with risk estimates as high as 6-fold.11 Few prospective studies have been conducted, and of the 3 studies published,12,20,21 2 found that douching is not related to the development of BV.12,20 Some authors have suggested that the confusion arises because women douche following the development of BV in response to abnormal symptoms; however, in cross-sectional analyses, douching for reasons not associated with symptoms remains associated with BV.4,9 Douching may also differentially affect vaginal flora depending on whether a woman has normal or already disturbed flora, yet no previous study, to our knowledge, has assessed this possibility.
In the current study we found that the presence of already-altered flora appeared to impact the association between douching and BV. Associations between douching and BV were more apparent among women with abnormal vaginal flora status at the immediately preceding visit. In contrast, douching was not associated with BV among women who lacked a previous diagnosis of BV or evidence of abnormal vaginal flora.
Similarly, while formal tests of interaction were not statistically significant, we found that douching was associated with incident development of BV among women with intermediate flora at baseline but not among women with normal flora at baseline. Our results were consistent across the BV-associated organisms evaluated; that is, douching did not significantly increase the risk of acquisition of these vaginal organisms among women who previously lacked them. Although we lacked statistical power to formally demonstrate interaction, consistent differences in the observed magnitude of association across strata suggest that douching may disrupt already-imbalanced flora sufficiently to create BV; however, douching may not induce de novo BV or BV-associated microorganism acquisition.
To our knowledge, this is the first study to evaluate how abnormal vaginal flora may impact the association between douching and BV. This study was also the first to discriminate between the acquisition of BV-associated vaginal microorganisms versus the progression of vaginal microflora imbalance in relation to douching practices. The consistency of our findings across vaginal microorganisms lends credence to the interpretation that douching is not associated with new acquisition of vaginal microflora among women who were previously free of them. However, our analysis was dependent upon a limited spectrum of cultivable organisms, and evaluation of a broader spectrum of BV-associated organisms is warranted to fully understand the impact of douching on the acquisition of the BV and BV-associated organisms.
Our study has a number of limitations. Relatively long intervals separated vaginal microbiologic assessments. BV can be extremely variable over short periods of time, so infrequent assessment is necessarily incomplete and only a gross assessment of variability. We also did not ascertain information on variations in vaginal flora due to the menstrual cycle, which may have introduced short-term variability that we were unable to adjust for. Prospective measurement of douching involved assessment at the immediately preceding visit, which was approximately 6 months prior; this, too, may also have been too distant to be optimally meaningful. However, we did observe a significant adverse effect associated with douching 6 months prior among women with intermediate flora. Also, we cannot rule out that unmeasured confounding (e.g., unmeasured partner behavior) may have influenced our results. Additionally, treatment of BV was not ascertained in this study and treatment may influence the effect of douching on vaginal flora. Finally, a focus on high-risk women may limit the generalizability of our results; it is unknown whether douching may differentially affect women based upon their risk for STDs.
In summary, among predominantly young, black women, we found that frequency of douching and reason for douching were not associated with progression to BV or the acquisition of BV-associated organisms among women with previously normal flora. However, douching was associated with the development of BV among women with already-disturbed vaginal flora. As women are not screened for normality of vaginal flora before douching, the women at risk for BV (i.e., those with prior intermediate flora) are not identified as such. Moreover, the occurrence of intermediate vaginal flora is common. Thus, to avoid progression to BV, douching is best avoided.
1. Abma J, Chandra A, Mosher W, et al. Fertility, family planning, and women’s health: New data from the 1995 National Survey of Family Growth. Vital Health Stat 1995; 23:1–114.
2. Martino J, Vermund S. Vaginal douching: Evidence for risks and benefits to women’s health. Epidemiol Rev 2002; 24:109–124.
3. Zhang J, Thomas G, Leybovich E. Vaginal douching and adverse health effects: A meta-analysis. Am J Public Health 1997; 87:1207–1211.
4. Ness R, Hillier S, Richter H, et al. Douching in relation to bacterial vaginosis, lactobacilli, and facultative bacteria in the vagina. Obstet Gynecol 2002; 100:765–772.
5. Rosenberg M, Phillips R, Holmes M. Vaginal douching: Who and why? J Reprod Med 1991; 36:753–758.
6. Schwebke J, Desmond R, Oh M. Predictors of bacterial vaginosis in adolescent women who douche. Sex Transm Dis 2004; 31:433–436.
7. Chiaffarino F, Parazzini F, Besi P, et al. Risk factors for bacterial vaginosis. Eur J Obstet Gynecol Reprod Biol 2004; 117:222–226.
8. Fonck K, Kaul R, Keli F, et al. Sexually transmitted infections and vaginal douching in a population of female sex workers in Nairobi, Kenya. Sex Transm Infect 2001; 77:271–275.
9. Holzman C, Leventhal J, Qiu H, et al. Factors linked to bacterial vaginosis in nonpregnant women. Am J Public Health 2001; 91:1664–1670.
10. Newton E, Piper J, Shain R, et al. Predictors of vaginal microflora. Am J Obstet Gynecol 2001; 184:845–855.
11. Rajamanoharan S, Low N, Jones S, et al. Bacterial vaginosis, ethnicity, and the use of genital cleaning agents: A case control study. Sex Transm Dis 1999; 26:404–409.
12. Ness R, Kip K, Soper D, et al. Variability of bacterial vaginosis over 6–12 month intervals. Sex Transm Dis 2006; 33:1–5.
13. Ness R, Hillier S, Kip K, et al. Bacterial vaginosis and risk of pelvic inflammatory disease. Obstet Gynecol 2004; 104:761–769.
14. Stergachis A, Scholes D, Heidrich F. Selective screening for Chlamydia trachomatis
infection in a primary care population of women. Am J Epidemol 1993; 138:143–153.
15. Schwebke J, Hillier S, Sobel J, et al. Validity of the vaginal Gram stain for the diagnosis of bacterial vaginosis. Obstet Gynecol 1996; 88:573–576.
16. Nugent R, Krohn M, Hillier S. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991; 29:297–301.
17. Scholes D, Stergachis A, Ichikawa L, et al. Vaginal douching as a risk factor for cervical Chlamydia trachomatis
infection. Obstet Gynecol 1998; 91:993–997.
18. Hosmer D, Lemeshow S. Applied Logistic Regression, 2nd ed. New York, NY: Wiley; 2000.
19. Allison P. Survival Analysis Using the SAS System: A Practical Guide. Cary, NC: SAS Institute; 1995.
20. Schwebke J, Desmond R. Risk factors for bacterial vaginosis in women at high risk for sexually transmitted diseases. Sex Transm Dis 2005; 32:654–658.
21. Hawes S, Hillier S, Benedietti J, et al. Hydrogen peroxide-producing lactobacilli and acquisition of vaginal infections. J Infect Dis 1996; 174:1058–1063.