Bidet toilet use is a sanitary habit of cleansing the vulva and anal area after urination or defecation. Habitual use of a bidet toilet may alter the vaginal flora and cause bacterial vaginosis, which usually produces minor symptoms such as excess discharge.1 During pregnancy, however, bacterial vaginosis can induce intrauterine infection, which may cause chorioamnionitis, an inflammation of the fetal membrane that accounts for 25–40% of preterm births.2 Very few studies have investigated the health risk of either the traditional nonelectric bidets or the modern electric bidet toilets with more functionalities. In Japan, 71.6% of households have electric bidet toilets installed,3 and bidet use has become a growing global trend.4 Given its widespread and growing use, health risk of bidet toilet use is an important public health issue. A recent Japanese study reported an association between bidet toilet use and aggravation of vaginal flora (decrease of lactobacillus species and increase of fecal bacteria)5 but did not account for confounding factors or reverse causality.
In this study, we focused on pregnant women as a population potentially vulnerable to bidet toilet use and estimated the association between bidet toilet use and preterm birth. Furthermore, microbes of the vaginal flora were analyzed to investigate the association between bidet toilet use and bacterial vaginosis. This article presents a preliminary recommendation of bidet toilet use for the potentially vulnerable population.
MATERIALS AND METHODS
The study targeted all women who gave birth between 2006 and 2010, at the Department of Obstetrics and Gynecology, Keio University Hospital in Tokyo, Japan. Excluding those who delivered before 22 weeks of gestation or at age younger than 20 years, the study population consisted of 2,545 women. A structured, 6-page questionnaire containing 27 questions about bidet toilet use and other lifestyle factors was sent to these women by postal delivery. To minimize nonresponse, we mailed another copy of the questionnaire to those who did not respond by the first deadline. Two hundred sixty-four women responded to this second batch and submitted their response. In total, 1,404 women returned responses. The final response rate was 64.1%, excluding from the denominator the 355 returned questionnaires with unknown or changed address. Furthermore, we excluded those who gave birth to twins and triplets to avoid the influence of multiple gestations (n=62), those who gave birth in 2011 (n=10), those with insufficient records about pregnancy and delivery (n=1), and those without information about bidet toilet use (n=39). Including one overlap in the excluded women, 1,293 total women were analyzed for the primary outcome of preterm birth.
Bacterial vaginosis based on prenatal microbiological examination was analyzed as the secondary outcome. This analysis focused on those who had a microbiological screening test conducted at approximately 35 weeks. From these 1,292 women, those who gave birth in 2011 (n=9), those who did not answer the question about bidet toilet use (n=32), those with insufficient delivery records (n=1), and those with invalid test results (n=124) were excluded. Those experiencing multiple gestations were included. A total of 1,126 women were analyzed for the secondary outcome of bacterial vaginosis.
Information on the exposure of interest, bidet toilet use before and during pregnancy, was collected by a self-report questionnaire. Questions about bidet toilet use included items on the duration, frequency, length of each use, and the reason for initiating bidet toilet use (Appendix 1, available online at http://links.lww.com/AOG/A377). These questions were repeated for each of the two different cleansing modes of the bidet toilet, “frontal cleansing” to wash the vulva and “rear cleansing” to wash the anus. A bidet toilet user was defined as a woman who was using bidet toilets less than 3 months before and during pregnancy, regardless of the frequency or length of each use.
The primary outcome of this study was the incidence of preterm birth. We collected information about pregnancy and delivery from the delivery records at the Department of Obstetrics and Gynecology at Keio University Hospital in Tokyo, Japan. Preterm birth was defined as delivery at less than 37 completed weeks of gestation, including spontaneous and indicated preterm birth. Within preterm birth, delivery before 33 weeks of gestation was defined as early preterm birth, and delivery between 34 and 36 weeks of gestation was defined as late preterm birth.
The secondary outcome was bacterial vaginosis estimated by the balance of lactobacilli and nonlactobacillus microbes. Information on lactobacillus was collected for all women as an index of normal microbial flora6 from the routine prenatal microbiological tests conducted by clinicians at the hospital to screen for the bacteria and fungi in cervicovaginal secretions during pregnancy. Information on nonlactobacillus microbes also was obtained from the same tests. The screening tests were conducted at the time point closest to the delivery for all women, approximately 35 weeks of gestation. The few exceptions were those who delivered preterm. For these women, the microbiological test conducted at the time closest to delivery was used for analysis. Of the 1,126 total women with microbiological test results, 982 (87.2%) had undergone the routine screening test that screens for any bacteria or fungi and 144 (12.8%) women had undergone a special screening test for the prevention of preterm birth and perinatal complications that screens for seven species of microbes most pertinent to perinatal complications.
Cervicovaginal specimens collected by the clinicians were immediately transported to the microbiological laboratory within the hospital and were cultured at 37°C for 72 hours. If no microbes were detected, incubation was discontinued. If any microbes were detected, incubation was extended to identify the species. The identified microbes from both the routine tests and the special tests were categorized into three groups as follows: intestinal bacteria (Escherichia coli, Enterococcus faecalis or other Enterococcus species, Klebsiella pneumonia, Staphylococcus aureus or other Staphylococcus species),7 bacterial vaginosis–related bacteria (Gardnerella vaginalis, Streptococcus agalactiae),8 and fungi (Candida albicans or other Candida species). The amounts of microbes were semi-quantitatively classified into the following five levels: none (-); small (<+); medium (+); medium-large (++); and large (+++).
Based on the categorization of the species and the classification of their amounts, bacterial vaginosis was defined as meeting either of the following two criteria: none (-) or small (<+) amount of lactobacillus coexistent with any amount of intestinal bacteria or bacterial vaginosis–related bacteria; or medium (+) amount of lactobacillus coexistent with medium to large (+, ++, +++) amount of intestinal bacteria or bacterial vaginosis–related bacteria.
Information on confounding factors based on the questionnaire and delivery record included maternal age at delivery, height, weight, body mass index, history, systemic complications (respiratory, cardiac, renal, autoimmune, psychological, infectious, musculoskeletal, and other diseases), gestational complications (uterus malformation, myoma uteri, adnexa abnormality, cervical incompetence, placenta abnormality, amniotic fluid abnormality, high blood pressure), implementation of infertility treatment, cesarean delivery, gestational diabetes, parity, number of children (singleton, twins, or triplets), antibiotics (oral antibacterial drugs administered for more than 3 days and topical antifungal drugs) and immunosuppressive medications used during pregnancy, smoking, alcohol consumption, sanitary habits (bathing, showering, vaginal douching), sexual intercourse during pregnancy, and educational background.
Characteristics of the women were analyzed using the χ2 test and unpaired t test. For the primary outcome of preterm birth, logistic regression model was used to estimate whether bidet toilet use was associated with incidence of preterm birth. Power analysis showed that the minimum odds ratio (OR) that can be detected at alpha equal to 0.05 and with statistical power of 0.80 is 1.35 with 1,293 women (819 users of bidet toilets and 474 nonusers) when the preterm birth rate was 0.16 in the nonusers group and no covariates were included in a logistic regression model.
Crude OR was first calculated, followed by multivariable analysis adjusting for the following confounding factors selected based on literature review and univariable analyses: maternal age at delivery; parity; cesarean delivery; gestational complications; gestational diabetes; sexual intercourse during pregnancy; oral antibiotics use for 3 days or more during pregnancy; shower use to wash vulva; and educational background. Use of oral antibiotics, an important risk factor of vaginitis,9 was included in the multivariable model irrespective of the crude association. All other potential confounding variables not independently associated with the incidence of preterm birth (P<.05) were not included in the multivariable model. Subanalyses were conducted by excluding the women who used vaginal douching more than once per week (n=17), and also by excluding those who used oral immunosuppressive agents (n=1) and oral adenocorticosteroids more than 5 mg/d for more than 2 weeks (n=7).
For the secondary outcome of bacterial vaginosis, logistic regression model also was used to determine whether bidet toilet use is associated with bacterial vaginosis. The covariates in the multivariable model were the same as those included in the preterm birth analysis model, except for two variables, number of children was included and cesarean delivery was excluded. The association between prescription rate of topical antifungal agents and bidet toilet use also was estimated using the same multivariable model.
Further microbe-related analysis was conducted to determine whether bidet toilet use has an effect on fungi in the cervicovaginal secretions during pregnancy. The same multivariable model from the bacterial vaginosis analysis was used. The condition of fungi was dichotomized into none (-) compared with any amount ranging from small to large (<+, +, ++, +++).
In addition, the relationship between frequency and length of bidet toilet use and preterm birth was examined. For these analyses, the use of bidet was broken down into rear and front cleansing modes, as well as by five frequency categories and five categories of length of bidet usage. Trends of association were examined using a logistic regression model, which assigned scores to the level of the independent variable (frequency or length of time for each mode).
All statistical analyses were performed using SAS 9.2. All reported P values are two-tailed and were considered statistically significant at P<.05. This study was approved by the Ethical Committee of Keio University School of Medicine (No. 2010-194 and No. 2010-194-2). Participants were informed about the study before answering the questionnaire, and responding to the questionnaire was regarded as consent for study participation.
Table 1 shows the characteristics of the analyzed women. Of 1,293 women, 63.3% were users of the bidet toilets. The incidence of preterm birth was 15.8% among bidet users and 16.0% among nonusers. Incidence was 9.8% for late preterm birth and 6.0% for early preterm birth. No difference was observed between the users and nonusers with regard to educational background, a proxy for socioeconomic status.
Although 64.1% is a modest response rate, comparison of 1,404 responders and 1,141 nonresponders showed no significant differences regarding preterm birth incidence and vaginal flora. The incidence of preterm birth was 19.2% among the responders and 21.2% among the nonresponders, indicating that the selection bias was small (Appendix 2, available online at http://links.lww.com/AOG/A378). To clarify, the preterm birth incidence of the analyzed women shown in Table 1 was 15.8% as opposed to 19.2% because the women with multiple gestations were excluded to eliminate its possible association with higher risk of premature delivery.
The association between bidet toilet use and the incidence of preterm birth is shown in Table 2. Adjusting for the covariates, no association was found (adjusted OR 1.04, 95% confidence interval [CI] 0.72–1.48). Among the covariates, multiparous women, gestational complications, and gestational diabetes were significantly associated with higher incidence of preterm birth. Older age and more education were significantly associated with lower incidence of preterm birth.
No association was observed between bidet toilet use and bacterial vaginosis (adjusted OR 0.96, 95% CI 0.70–1.33; Table 3). Further analyses of microbe categories also did not find any association between bidet toilet use and intestinal bacteria (adjusted OR 0.97, 95% CI 0.68–1.38; data not shown) or between bidet toilet use and bacterial vaginosis–related bacteria (adjusted OR 1.00, 95% CI 0.73–1.36; data not shown).
However, the detection rate of fungi was significantly higher among bidet toilet users (adjusted OR 1.68, 95% CI 1.14–2.48; Table 4). Older maternal age was associated with lower OR of fungi detection. Less habitual washing of the vulva in the shower also was associated with lower OR of fungi detection, but no linear trend was observed between decreasing frequency of shower use and lower OR of fungi detection.
Although fungus detection was more frequent among bidet toilet users, no relationship was observed between bidet toilet use and prescription of topical antifungal medication (Table 5). Because Candida vaginitis usually is thought to have no effect on preterm birth,10,11 these results are consistent with the existent knowledge. Information about antifungal medication prescription was obtained from each woman's monthly routine check-up when any clinically obvious fungal vaginitis would be prescribed an antifungal medication.
The strength of this study lies in the abundance of information on confounding factors. Focusing on pregnant women eliminated the need to consider the woman's menstrual cycle, a major confounding factor for the vaginal flora condition. In addition, combining the detailed information collected from the medical records and questionnaire responses allowed us to closely examine the potential confounders before finalizing the multivariable model.
The downside of choosing pregnant women as the study population is the low external validity. The women in this study experienced a considerably higher incidence of preterm birth (15.9%) compared with that of Japan overall (5.7%).12 This is most likely because the data were collected at the university hospital known for treating infertility and gestational complications. From the perspective of generalizability, our study population was not ideal. That said, the purpose of this study was to focus on the vulnerable population to assess the risk of bidet use, and the study population at least would not be more resistant to preterm birth and infectious diseases than average pregnant populations. Therefore, the nonassociation of bidet toilet use and preterm birth in this population implies that the same result would apply to average pregnant women.
Retrospective design of the study connotes the possibility of reverse causality, especially regarding vaginal flora. Women with clinical symptoms of bacterial vaginosis may be prone to use the bidet toilet. Analyzing the reasons for initiating bidet toilet use indicated that a relatively small portion (10.7%) of bidet toilet users started using bidet toilets to mitigate gynecologic discomfort. Excluding these women did not substantially change the significant relationship between bidet toilet use and fungi detection (adjusted OR 1.53, 95% CI 1.02–2.30). These subanalyses indicate minimal effect of reverse causality on the present study.
Recall bias is also a limitation of retrospective studies. Mothers who experienced preterm birth may be keen on finding an explanation and may give more precise answers, leading to false-positive associations. In fact, subanalysis of detailed questions about bidet use showed significantly less missing values among those who experienced premature birth (Appendix 3, available online at http://links.lww.com/AOG/A379). Also, although the statistically significant dose-response relationship was not observed between premature birth and frequency or length of bidet use, the highest category of frequency and length of each use showed a noticeably high OR, as high as 1.79. However, adjusting for this recall bias would shift the association toward the negative and would not alter the study result in the direction of bidet toilets having harmful effects.
Microbiologic culture–based categorization may have missed some bacterial vaginosis cases, which would have lowered the bacterial vaginosis prevalence. However, bacterial vaginosis prevalence of the entire targeted population (n=2,545) was 20.1%. This falls on the higher end of the 15–20% bacterial vaginosis prevalence range of the overall Japanese pregnant population.13,14 Also, underdiagnosis of bacterial vaginosis would be nondifferentiating with respect to bidet toilet use and would have had little effect on the main results.
By excluding women who delivered before 22 weeks of gestation, we may have missed an association with spontaneous abortion or very early preterm delivery. However, delivery before 22 weeks of gestation implies a high neonatal death rate as well as cases of induced abortion. Sending a questionnaire about bidet use to these women with sensitive prenatal memories did not seem ethical.
Compared with the past study that did not adjust for confounders or take reverse causality into consideration,5 our study presents a reliable report of bidet toilet use not posing a health risk to pregnant women with regard to preterm birth. However, there may be other female populations with compromised immune systems who are more vulnerable to vaginal infections, such as chemotherapy patients or postmenopausal women.15 Further studies should be conducted with such women. In addition, the supposed benefits of the bidet toilet use such as preventing urinary tract infections16 should be properly investigated.
In summary, we conclude that normal use of bidet toilets by pregnant women does not pose any clinical health risk with regard to preterm birth. We also recommend that other potentially vulnerable populations and potential benefactors of bidet toilet use be studied in the future.