VAGINAL DOUCHING is a common behavior among American, and particularly black, women. 1 In the most recent National Survey of Family Growth, 55% of black women and 21% of white women reported regular douching.
An association between douching and pelvic inflammatory disease (PID) has been suspected for decades. National surveys have shown that douching is common, and that a higher proportion of women with self-reported PID engaged in douching than those not reporting the condition. 1,2 Few studies have been conducted to assess 3–5 the independent association between douching and PID.
In the current case–control studies, after adjustment for confounding factors, one study involving predominantly white women reported a substantial elevation in risk (odds ratio [OR] 2.1 for douching in the previous 3 months), 4 and two studies separately analyzing black women and white women found the risk limited to white women. 3,5 However, this finding may have resulted from the limited sample sizes within race-specific strata or the relatively small number of black control women who douched. Moreover, several retrospective studies and a meta-analysis have related douching to infertility or ectopic pregnancy, both reproductive consequences from PID. 6–13
Douching may alter the number and type of microorganisms inhabiting the vagina. 14 An outstanding question is whether douching predisposes to PID through this mechanism. One prospective study of vaginal microecology showed that douching predisposes to bacterial vaginosis. 15 Douching also increased the risk of PID, but not PID associated with Neisseria gonorrhoeae or Chlamydia trachomatis in one study. 3
This article reports cross-sectional comparisons of douching patterns between women with clinical PID enrolled in the PID Evaluation and Clinical Health (PEACH) Study who were found to have endometritis and/or gonococcal or chlamydial upper genital tract infection (UGTI) and women who had neither endometritis nor UGTI. These data allow further examination of douching and endometritis among black women and women with bacterial vaginosis.
The conduct of the PEACH study has been described elsewhere. 16 Briefly, women 14 to 37 years of age were recruited from emergency departments, clinics, and STD units at seven primary and six secondary clinical sites located throughout the eastern, southern, and central regions of the United States between March, 1996 and February 1999. Human subjects approval was obtained at each participating institution, and all the participants signed an informed consent. Women were eligible for the PEACH study who, on the basis of general clinical criteria, had suspected PID. Specifically, all eligible women had (1) a history of pelvic discomfort lasting 30 days or less, (2) findings of pelvic organ tenderness (uterine or adnexal) on bimanual examination, and (3) leukorrhea and/or mucopurulent cervicitis and/or untreated but documented gonococcal or chlamydial cervicitis. Leukorrhea was defined as an excess number of white blood cells (greater than the number of epithelial cells) as viewed microscopically in lateral vaginal wall secretions admixed with normal saline. Mucopurulent cervicitis was defined as the presence of a grossly yellow or green exudate observed on a swab taken from the cervix.
Of the 2941 women screened for study entry, 346 (11.8%) did not meet the preceding inclusion criteria. An additional 36.7% of the women were excluded on the basis of a priori criteria: 141 of the women (4.8%) were pregnant; 246 (8.4%) had taken antimicrobials within the previous 7 days; 248 (8.4%) had undergone a prior hysterectomy or bilateral salpingectomy; 51 (1.7%) had undergone an abortion, delivery, or gynecologic surgery within the preceding 14 days; 191 (6.5%) had a suspected tuboovarian abscess or other condition requiring surgery; 163 (5.5%) had an allergy to the study medications; 29 (1%) were homeless; and 11 (0.4%) had experienced vomiting after a trial of antiemetic treatment.
Among the 1515 women eligible for the study, 864 (57%) agreed to participate, and 831 (54.9%) were reassessed as correctly eligible and randomized. Of the 831 randomized women, 654 (78.7%) underwent an endometrial biopsy by pipelle and thus became the focus of these analyses.
Two reference pathologists, masked to any clinical data, separately evaluated at least one section of hematoxylin and eosin–stained and one section of methyl green pyronine–stained endometrial biopsy tissue from each participant. For any case in which there was disagreement about the presence or absence of neutrophils and plasma cells, the pathologists reread the slides together and came to a consensus agreement. A standardized instrument was used to record elements of the reading including the presence of basal and functional layers, the phase of the endometrial cycle (proliferative, secretory, or menstrual), and the quantitation of plasma cells per ×120 field, glandular neutrophils per ×400 field, and stromal neutrophils per ×400 field. Histologic endometritis was defined according to a modification of the criteria proposed by Kiviat et al. 17 Endometritis was classified when at least five neutrophils were found in the endometrial surface epithelium in the absence of menstrual endometrium and/or at least two plasma cells in the endometrial stroma. This differs from the Kiviat criteria, in which endometritis is defined as at least five neutrophils and at least one plasma cell. The current authors have reported little difference in the positive predictive value using different criteria. 18
Upper Genital Tract Infection
Endometrial samples were cultured for N gonorrhoeae and tested for C trachomatis by polymerase chain reaction (PCR) (Roche Diagnostics, Indianapolis, IN) using standard manufacturer’s protocols. The presence of either or both of these pathogens in the endometrium was defined as UGTI. Endometritis represented inflammation, and UGTI represented infection in the upper genital tract consistent with salpingitis. Several studies have shown a strong correlation between salpingitis as visualized by laparoscopy, the gold standard for PID diagnosis, and endometritis. 17–19
Vaginal smears prepared from the lateral vaginal wall were available for 595 women in the analysis set. The smears were air dried and Gram stained. At a single laboratory, one of the authors (S. Hillier) read all the slides masked to clinical data. A score of 0 to 10 was assigned according to the relative proportions of large gram-positive rods (lactobacilli), small gram-negative or gram-variable rods (Bacteroides or Gardnerella), and curved gram-variable rods (Mobiluncus). 20 The results were interpreted by a standardized method used for the diagnosis of bacterial vaginosis as described by Nugent et al. 20 A score of 0 to 3 was interpreted as consistent with a normal flora; a score of 4 to 6, corresponding to a disturbed flora, was designated as intermediate, and a score of 7 to 10 corresponded to bacterial vaginosis. Relative to clinical signs, the scoring system has been shown to have a specificity of 83% and a sensitivity of 89% for bacterial vaginosis in nonpregnant women. 21
In a standardized 20-minute interview conducted by trained research staff at each center, women were asked whether they douched, and if so, about the following aspects of their douching behavior: any douching in the past month, frequency of douching in the past month, most recent episode of douching, reason for douching, and type of product used most often. For purposes of analysis, these data were categorized into frequency (none, once a month or less often, more than once a month), recency (14–30 days ago, 7–13 days ago, 0–6 days ago), reason (cleansing, abnormal discharge, health provider recommendation, pregnancy prevention), and type of product (homemade, Massengill, Summers Eve, other store brand).
Other potential risk factors.
Women were asked about demographic factors including age, race, and highest completed level of education. They reported past gynecologic and reproductive history including number of live births, pregnancies, miscarriages, and elective abortions; their history of PID and various sexually transmitted diseases; and their history of fertility problems. Women also were asked about sexual behaviors including the number of sexual partners, new partners, frequency of intercourse, anal and oral intercourse in the preceding 4 weeks, intercourse during the last menses, and tampon use. Women reported lifestyle factors including current tobacco, alcohol, and illicit drug (cocaine and marijuana) use.
The differences between groups were analyzed with Student’s t test for continuous variables, and the chi-square test was used to analyze categorical variables. The odds ratios and 95% CIs associated with frequency of douching, recency of douching, reason for douching, and type of douching product were calculated by comparing control women without endometritis or UGTI with women who had endometritis/UGTI. In secondary analyses, endometritis and UGTI were considered as separate outcomes. Also, analyses were stratified by race (black versus not black) and by Gram stain findings (categorized into normal or intermediate and bacterial vaginosis).
All the variables listed earlier as other potential risk factors that preliminary univariate analyses identified as possible confounders were forced into logistic regression models using the SPSS statistical package (SPSS, Inc., Chicago, IL). Separate multivariate models including this same set of covariates were run for each of the douching exposures: frequency, recency, reason, and type of product.
Characteristics significantly (P < 0.05) more common among patients with the primary study outcome of endometritis or UGTI than among control subjects who had neither of these findings were age of 19 years or younger (30.2% of cases, 19.9% of controls), black race (79.3% of cases, 62.6% of controls), self-reported current smoking (48.9% of cases, 40.3% of controls), and cocaine use (5.2% of cases, 1.5% of controls). Significantly less common among women with upper genital tract disease were education beyond high school (18% of cases, 30.4% of controls) and self-reported history of PID (25.5% of cases, 34.8% of controls).
More than 40% of the women enrolled douched in the month before study entry (Table 1). Overall, women with endometritis or UGTI were only 20% more likely to have douched in the month before enrollment. However, douching more than once a month versus not douching was associated with a significant 60% increase in the risk of endometritis or UGTI. Douching within 6 days of study enrollment versus not douching significantly elevated the risk of upper genital tract disease by 80%. Most of the women douched for cleansing, and those who douched for abnormal discharge (4%) did not have an elevated endometritis or UGTI risk. No particular product was associated with excess risk as compared with its nonuse. Among the women who had douched recently, frequency of douching had little impact on risk.
After adjustment for potentially important confounding factors (Table 2), frequent douching remained more common in women with endometritis or UGTI, but was no longer significantly more common. Recent douching remained statistically significantly as related to an elevated risk of upper genital tract disease. The subset of black women had elevated risks of a magnitude similar to that found among women enrolled in the study as a whole.
Douching was more common among women with bacterial vaginosis than among women with normal or intermediate vaginal flora. Among women with vaginal flora characterized as normal, intermediate, and bacterial vaginosis, 28.2%, 36.3%, and 50.8%, respectively, douched (P < 0.001). Also, women with endometritis or UGTI were more likely to have bacterial vaginosis than women with normal or intermediate vaginal flora (OR, 1.8; 95% CI, 1.3–2.5). However, it was among the women with normal or intermediate flora that the relation between douching and endometritis or UGTI was most apparent after adjustment for race and age (Table 3). Douching in the past month significantly increased the risk of endometritis or UGTI by 80%, and douching within 6 days of enrollment significantly increased the risk 2.3-fold. No statistically significant elevation in risk of endometritis/UGTI was associated with douching among women with bacterial vaginosis.
Similarly, douching was not related to upper genital tract disease in women with gonorrhea or chlamydia. However, among women without isolated cervical gonorrhea or chlamydia, douching in the past month (OR, 1.5; 95% CI, 1.0–2.3) and douching within 6 days of enrollment (OR, 1.5; 95% CI, 0.9–2.6) were related to endometritis or UGTI.
Women with endometritis but without UGTI (n = 133) were analyzed separately. The associations between douching and endometritis alone were similar to those reported earlier. Women with UGTI but without endometritis were uncommon (n = 17). To address the concern that douching may have come after PID symptoms, the relation between douching and endometritis or UGTI was analyzed among the 49.9% of women in the study who had symptoms for 0 to 6 days. For these women, the odds ratio (95% CI) for last douching in the preceding 14 to 30 days was 1 (0.5–2.0), in the preceding 7 to 13 days was 1.3 (0.6–2.6), and in the preceding 0 to 6 days was (2.4 (1.2–4.7). Therefore, even in women with symptoms of recent onset, douching in the preceding 6 days was related to an elevated risk of upper genital tract disease.
This large study, using an objective measure to define PID, showed that women with endometritis or UGTI were more likely to douche frequently (more than once per month) and recently (within the preceding 0–6 days) than they were never to douche. Three previous studies 11–13 showed that relatively frequent douching (more than once a month) was associated with a elevated risk for upper genital tract disease. Scholes et al 4 reported the adjusted risk of clinical PID from douching to be highest in women who engaged in the procedure four or more times a month (OR, 3.9; 95% CI, 1.4–10.9). 12 In the same study, women douching recently who did so one to three times a month were at a lower although still elevated risk (OR, 1.8; 95% CI, 1.0–3.4). These risk estimates were slightly modified downward after exclusion of women who douched for symptomatic infection.
Wolner-Hanssen et al 3 reported that douching one to two times a month elevated the adjusted risk for endometrititis or salpingitis 1.6- to 4.7-fold, depending on the control group used, and that douching three or more times elevated the adjusted risk 3.4- to 3.9-fold. Jossens showed similarly elevated risks (OR, 2.0–2.2 in unadjusted analyses) for clinical PID among women who douched in the past month and in the past week, as compared with women who never douched. 5 The current data are the only statistics other than those of Wolner-Hanssen et al 3 that relate douching to a biomarker of actual upper genital tract infection or inflammation. The current estimates of risk are more modest than those reported by Wolner-Hanssen et al, 3 but they are in keeping with the notion of an increase in risk.
The current study also showed that recent (0–6 days before enrollment) douching is associated with the greatest elevation in risk of upper genital tract disease. Previous studies did not specifically analyze data about douching recency within the preceding month. The current observations also were consistent for the subgroup of black women, which represented the majority of women enrolled in the current study. Previous studies have implied that the risk may be less for black than for white women. 3,5 However, one study showed that douching is related significantly to ectopic pregnancy, a complication of PID, in a group of black women. 8
This study showed that douching was more common among women with bacterial vaginosis than among women with normal or intermediate flora, an observation that had been made previously. 15 Surprisingly, the current study found that douching was related to endometritis or UGTI in women with normal or intermediate vaginal flora, but not in women with bacterial vaginosis. Although previous reports examined the issue of chlamydial PID and douching, the current study was the first to examine directly the relation between douching and PID by type of vaginal flora including the bacterial vaginosis type.
The current finding is inconsistent with the hypothesis that bacterial vaginosis mediates the association between douching and upper genital tract disease. If that were the case, douching women with bacterial vaginosis would be more likely to have upper genital tract findings. However, an explanation could be that douching initiates endometritis by increasing the ascension of vaginal bacteria into the upper genital tract. Women with bacterial vaginosis may already have high rates of bacterial movement into the endometrium, but women with normal or intermediate vaginal flora may not. Therefore, inciting upward bacterial movement may initiate endometritis in women with normal or intermediate flora, but may not alter the rates of endometritis in women who already have bacterial vaginosis. Similarly, women with gonococcal or chlamydial cervicitis may get upper genital tract disease independent of douching, whereas in women without these pathogens, douching may initiate nongonococcal, nonchlamydial upper genital tract disease.
Strengths of this study included the consistent entry criteria, the use of histologic and microbiologic markers for upper genital tract disease, and the large number of women studied. Weaknesses included the lack of a purely unaffected control group (i.e., a group of women without clinical symptoms of PID) and the lack of a laparoscopic diagnosis for PID. Women with objective evidence of PID were compared with women in the same cohort who had no evidence of the disorder, all of whom had signs and symptoms consistent with PID. This made it unlikely that women would be selected in a biased fashion (i.e., by both douching patterns and outcome) for this study because the outcome was actual upper genital tract disease. Furthermore, in the previous report by Wolner-Hanssen et al, 3 both control subjects such as those used in the current study and asymptomatic “external” control subjects gave similar results. Therefore, the authors think it unlikely that their results, comparing clinically similar women, are spurious. The authors also regarded laparoscopic data as not compulsory because the histologic finding of endometritis has been well correlated with salpingitis in previous studies. 17–19
Another weakness was the fact that for many of the significant findings, the lower bounds of the confidence intervals included, or came close to including, 1. This may well reflect small sample sizes within stratified comparisons. Finally, by necessity, this study relied on reported douching behavior. The focus on douching within the previous month may have minimized, but probably not eliminated, all the problems with exposure classification based on recall.
This large study of predominantly black women with symptoms of PID showed a relation between douching and endometritis or UGTI, particularly for women who had douched more frequently and more recently. These data add to the growing literature suggesting that douching might relate to PID.
1. National Survey of Family Growth. From Vital and Health Statistics. Data from the National Survey of Family Growth. Hyattsville, MD: U.S. Department of Health & Human Services, Public Health Service, National Center for Health Statistics, 1995.
2. Aral SO, Mosher WD, Cates W. Vaginal douching among women of reproductive age in the United States: 1988. Am J Pub Health 1992; 82: 210–214.
3. Wolner-Hanssen P, Eschenbach DA, Paavonen J, et al. Association between vaginal douching and acute pelvic inflammatory disease. JAMA 1990; 263: 1936–1941.
4. Scholes D, Daling JR, Stergachis A, Weiss NS, Wang SP, Grayston JT. Vaginal douching as a risk factor for acute pelvic inflammatory disease. Obstet Gynecol 1993; 81: 601–606.
5. Jossens MOR, Eskenazi B, Schachter J, Sweet RL. Risk factors for pelvic inflammatory disease: a case–control study. Sex Transm Dis 1996; 23: 239–247.
6. Mueller BA, Luz-Jimenez M, Daling JR, Moore DE, McKnight B, Weiss NS. Risk factors for tubal infertility: influence of history of prior pelvic inflammatory disease. Sex Transm Dis 1992; 19: 28–34.
7. Baird DD, Weinberg CR, Voigt LF, Daling JR. Vaginal douching and reduced fertility. Am J Public Health 1996; 86: 844–850.
8. Chow WH, Daling JR, Weiss NS, Moore DE, Soderstrom R. Vaginal douching as a potential risk factor for tubal ectopic pregnancy. Am J Obstet Gynecol 1985; 153: 727–729.
9. Chow JM, Yonekura ML, Richwald GA, Greenland S, Sweet RL, Schachter J. The association between Chlamydia trachomatis
and ectopic pregnancy: A matched-pair, case–control study. JAMA 1990; 263: 3164–3167.
10. Daling JR, Weiss NS, Schwartz SM, et al. Vaginal douching and the risk of tubal pregnancy. Epidemiology 1991; 2: 40–48.
11. Kendrick JS, Atrash HK, Strauss PM, Gargiullo PM, Ahn YW. Vaginal douching and the risk of ectopic pregnancy among black women. Am J Obstet Gynecol 1997; 176: 991–997.
12. Zhang J, Thomas G, Leybovich E. Vaginal douching and adverse health effects: a meta-analysis. Am J Public Health 1997; 87: 1207–1211.
13. Phillips RS, Tuomala RE, Feldblum PJ, Schachter J, Rosenberg MJ, Aronson MD. The effect of cigarette smoking, Chlamydia trachomatis
infection, and vaginal douching on ectopic pregnancy. Obstet Gynecol 1992; 79: 85–90.
14. Onderdonk AB, Delaney ML, Hinkson PL, Dubois AM. Quantitative and qualitative effects of douche preparations on vaginal microflora. Obstet Gynecol 1992; 80: 333–338.
15. Hawes SE, Hillier SL, Benedetti J, et al. Hydrogen peroxide–producing lactobacilli and acquisition of vaginal infections. J Infect Dis 1996; 174: 1058–1063.
16. Ness RB, Soper DE, Peipert J, et al. Design of the PID Evaluation and Clinical Health (PEACH) Study. Control Clin Trials 1998; 19: 499–514.
17. Kiviat NB, Wolner-Hanssen P, Eschenbach DA, et al. Endometrial histopathology in patients with culture-proved upper genital tract infection and laparoscopically diagnosed acute salpingitis. Am J Surg Pathol 1990; 14: 167–175.
18. Ness RB, Keder LM, Soper DE, et al. Oral contraception and the recognition of endometritis. Am J Obstet Gynecol 1997; 176: 580–585.
19. Wasserheit J, Bell TA, Kiviat N, et al. Microbial causes of proven pelvic inflammatory disease and efficacy of clindamycin and tobramycin. Ann Intern Med 1986; 104: 187–193.
20. 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.
21. Schwebke JR, Hillier SL, Sobel JD, McGregor JA, Sweet RL. Validity of the vaginal gram stain for the diagnosis of bacterial vaginosis. Obstet Gyn 1996; 88: 573–576.