Abstract
Purpose
Intimate care products may contain substances associated with increased risk of hormone-related cancers. The relationship between genital talc use and ovarian cancer, in particular, has been well-studied, but concerns about recall bias and exposure misclassification have precluded conclusions. We examined the association between intimate care products and female hormone-related cancers, accounting for potential biases, using data from a US-based cohort study.
Participants and Methods
The Sister Study enrolled 50,884 women who had a sister with breast cancer. Data on genital talc use and douching were collected at enrollment (2003–09) and follow-up (2017–19). We used Cox proportional hazards models to estimate hazard ratios (HRs) for associations between intimate care product use and breast, ovarian, and uterine cancer. To account for potential exposure misclassification and recall bias, we conducted quantitative bias analyses under various exposure re-assignment assumptions.
Results
Across considered scenarios, 41–64% of participants douched and 35–56% used genital talc. In models adjusted for exposure misclassification, genital talc use was positively associated with ovarian cancer (HR range 1.17–3.34) Frequent douching and douching during young adulthood were positively associated with ovarian cancer, but neither douching nor talc were associated with breast or uterine cancer. Differential reporting of talc use by cases and non-cases likely produces positive biases, but correcting for error still resulted in HRs above 1.0. For example, HR=1.40, CI: 1.04–1.89 when 25% of exposed cases and 10% of unexposed non-cases had talc status re-assigned.
Conclusions
While results show how differential recall would upwardly bias estimates, corrected results support a positive association between use of intimate care products, including genital talc, and ovarian cancer.
Keywords: intimate care products, genital talc, douching, ovarian cancer, breast cancer, uterine cancer, quantitative bias analysis
Intimate care products may contain endocrine disrupting chemicals, such as phthalates, parabens, and bisphenols,1,2 which can alter endogenous hormone levels and potentially impact the risk of developing hormone-related diseases such as breast, ovarian, or uterine cancer. They may also contain other known or suspected carcinogens such as volatile organic compounds,3,4 and asbestos.5,6
Douche and powder are two commonly used intimate care products. Douches are liquids inserted into the vagina using a device that produces an upward stream. Initially promoted as contraception,7 women also douche to improve perceived cleanliness and reduce odor, particularly after sexual intercourse or menstruation.8 Although some douches contain only water or vinegar, endocrine disrupting chemicals and volatile organic compounds have been detected in commercially available products,2,4 and users have elevated urinary phthalate levels.9 Douching has been linked to adverse reproductive health outcomes, including pelvic inflammatory disease, bacterial vaginosis, and ectopic pregnancy.10 It is also considered a risk factor for cervical cancer10,11 and possibly ovarian cancer.12,13
Powder consisting of talc or cornstarch may be applied to the genital area to prevent dampness and chafing, promote a feeling of cleanliness, or reduce odor.8 Genital powder use became a public health concern due to the natural co-occurrence of talc and asbestos.6 Though recent surveillance identified asbestos particles in certain talc products,5 the prevalence of asbestos contamination is unknown given the lack of routine monitoring. Use of powder in the genital area could plausibly promote carcinogenesis through mechanisms other than direct contact with asbestos, including exposure to other chemicals1,3 or irritation and inflammation of the reproductive tract.
When considering the potential carcinogenicity of intimate care products, the relationship between genital powder use and ovarian cancer has been especially well-studied, though without clear consensus.14 Initial case-control studies15–18 indicated a strong positive association that was not replicated in prospective studies,12,19,20 even when pooled.21 In studies with retrospective data collection, women with and without ovarian cancer may differentially report exposure, leading to recall bias.22,23 Though not affected by recall bias, prospective studies tend to have small case numbers and simplified exposure assessments, resulting in low statistical precision and increased likelihood of non-differential exposure misclassification.
Our main objective was to re-evaluate the associations between intimate care product use and incidence of hormone-related cancers, expanding on previous analyses,12,24 by incorporating newly diagnosed ovarian and uterine cancers, adding breast cancer as an outcome, and integrating new data on lifetime use of douche and genital talc. Because the newly-acquired exposure data was susceptible to differential missingness by cancer status, we employed quantitative bias analysis to estimate effects under several missingness assumptions. When examining the association between genital talc use and ovarian cancer, we additionally evaluated the potential impact of recall bias.
Methods
Study sample
The Sister Study consists of 50,884 women aged 35–74 who had a sister previously diagnosed with breast cancer, but who did not have breast cancer themselves at enrollment (2003–2009). All resided in the United States and provided written informed consent. The Sister Study is overseen by the institutional review board of the National Institutes of Health.
Participants completed an extensive computer-assisted telephone interview at enrollment, plus a separate personal and intimate care product use questionnaire.25 Trained examiners took anthropometric measurements. Major health events are reported annually, with extensive follow-up questionnaires completed every 3 years.
We excluded 5 women who withdrew their consent and 79 with pre-baseline breast cancer or unknown status (Figure 1). An additional 994 women were excluded from multivariable analyses due to missing data for key covariates, leaving 49,806. For ovarian cancer analyses, we additionally excluded 225 women with uncertain status or pre-baseline ovarian cancer and 8,753 women with prior bilateral oophorectomies. For uterine cancer analyses, we excluded 416 women with uncertain status or pre-baseline uterine cancer, as well as 15,202 women with prior hysterectomies. We did not exclude individuals missing data on intimate care product use.
Figure 1.
Flow chart (A) and timeline (B) describing characteristics and questionnaire data from Sister Study participants included in the quantitative bias analysis of intimate care product use and incidence of female hormone-related cancers.
Exposure assessment
The initial personal care product questionnaire focused on two specific time periods: ages 10–13 and the 12 months prior to enrollment. Women were asked how frequently they douched and how frequently they “applied talcum powder to a sanitary napkin, underwear, diaphragm, cervical cap, or directly to your vaginal area”. Participants responded whether they never, sometimes, or frequently used at ages 10–13, and whether they used never, <1 time/month, 1–3 times/month, 1–5 times/week, or >5 times/week in the year before enrollment. We considered frequent use to be at least once per month.
More detailed information about use of douche and genital talc was obtained in the fourth follow-up questionnaire (2017–19). Participants were asked if they ever used the products, and if yes, their age at first and most recent use, and their frequency of use during each decade. We primarily focused on ever versus never use of each product prior to enrollment, but we also examined frequency, duration, and timing of use.
Data on intimate care product use was sometimes contradictory or missing. Women who provided contradictory reports about intimate care product use in the year prior to enrollment8 were initially categorized based on their enrollment report, but we also considered models where a randomly-selected subset was re-classified (details in Supplementary Methods). We also used quantitative bias analyses to implement different approaches for imputing exposure in women who initially reported never use but did not complete the follow-up questionnaire. These comparisons were crucial for understanding potential biases, as women with incident cancer were over-represented in this “undefined” group.
Outcomes
Incident cases were women who self-reported cancer of the breast, ovary (including fallopian tubes or peritoneum26) or uterus after enrollment. These were verified via medical reports, when possible, with some fatal cases confirmed through the National Death Index or death certificates. Data are complete through September 2021 (data release 11.1).
For most breast cancer analyses, we analyzed pre- and post-menopausal person-time separately to account for differential associations with body mass index (BMI).27,28 We additionally examined subtype-specific associations based on estrogen receptor (ER) status. We separately assessed medically-confirmed serous29 and non-serous ovarian cancer and medically-confirmed endometrial cancers (type I or II).30
Covariates
The following covariates were considered potential confounders based on their previously established relationships with intimate care products and hormone-related cancers (all self-reported at enrollment, unless otherwise specified): race/ethnicity, attained education, examiner-measured BMI, BMI ages 30–39, age at menarche, duration of hormonal birth control use, parity, menopausal status, hormone therapy use, geographic region of residence, smoking status, and alcohol use. Other covariates included in imputation models due to possible associations with intimate care product use were childhood household income level, adult household income, highest attained education level in the household when the participant was 13, marital status, and weight relative to peers in teen years.
Statistical Analysis
We first compared covariate distributions across categories of intimate care product use. After excluding women missing key covariates and imputing missing exposure, we used multivariable Cox proportional hazards models with age as the time scale to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for potential confounders. Follow-up began at enrollment and continued until age at diagnosis, with censoring at end of follow-up (9/2021), loss to follow-up, or death. For assessments of ovarian and uterine cancer, women were censored at age of bilateral oophorectomy or hysterectomy, respectively.
In addition to estimating HRs for ever versus never use, we considered the effects of frequent use, long-term use, and use during specific age periods (teens, 20s, 30s, 12 months prior to enrollment). Subtype-specific analyses were limited to comparisons of ever versus never use.
We separately estimated HRs for ever versus never use based only on product use as reported at enrollment (prospective) or follow-up (retrospective). Additionally, we considered any product use before the fourth detailed follow-up in relation to cancers occurring after that time (prospective with left truncation). Lastly, we conducted sensitivity analyses specific to ovarian cancer, estimating HRs for douching and talc use combined, and for genital talc use according to reproductive tract patency at time of use.
Quantitative Bias Analysis
For ever versus never use analyses, we compared four possible scenarios using quantitative bias analysis: (1) no correction; (2) contradictory data correction; (3) contradictory data correction plus categorizing missing or undefined as exposed; and (4) contradictory data correction with multiple imputation of missing or undefined data.
Under no correction, we prioritized reporting at enrollment, categorizing women in the undefined category as non-users. Women missing both questionnaires were assumed to be missing completely at random. The second scenario added a correction for contradictory data, as described in Supplementary Methods. We did this for each of 10 copies of the data, summarizing HRs and 95% CIs using Rubin’s rules.31 Results were consistent across different initiation seeds. For the third scenario, we included the contradictory data correction and categorized all women in the undefined category as users. This contrasted with Scenario 2, where the same undefined women were considered never users. Together scenarios 2 and 3 demonstrate the range of results defined by how women in the undefined category are classified, with the true exposure distribution falling somewhere between the two extremes.
For the fourth scenario we used multiple imputation with chained equations (MICE; PROC MI, SAS v9.3), to generate covariate-informed probabilistic imputations of the exposure status of participants who were undefined or missing. This scenario should account for covariate-dependent missingness, and we consider it our best estimate of the true association in the absence of recall or other unknown biases. We ran 10 iterations on the data set that already included 10 copies of the data corrected for contradictory responses, then summarized effect estimates across all 100 imputed data sets. We included all of the previously described confounders and covariates in the imputation model, as well as each of the cancer outcomes and crude cumulative hazard estimates,32 which corresponded to the hazard of the earliest of the 3 cancer events. To ensure that responses regarding ever, frequency, duration, and timing of product use were internally consistent and accounted for co-exposure, all talc and douching-related variables were imputed concurrently.
Recall bias
We additionally investigated the potential impact of recall bias on the association between genital talc use and ovarian cancer. The initial correction and imputation procedures were identical to scenario 4. Additional details are available in the Supplementary Methods, but briefly, we considered: 1) recoding a proportion (range 10–90%) of ovarian cancer cases classified as talc users to be non-users; 2) recoding a proportion (range 10–90%) of ovarian cancer cases classified as non-frequent and short-term talc users to be non-users; and 3) re-coding a proportion (range 5–25%) of non-cases classified as non-talc users to be infrequent and short-term talc users.
We also generated a single recall bias-corrected estimate that simultaneously corrected cases and non-cases. We assumed 25% of ovarian cancer cases initially categorized as infrequent and short-term users were re-assigned to be non-users and 10% of the women without ovarian cancer initially categorized as never users were re-assigned to be infrequent, short-term users. Estimates based on this correction are included in the results as examples of plausible, yet cautious, estimates of the association between genital talc use and ovarian cancer after correcting for case-differential recall.
Role of the funding source
This work was funded by the National Institutes of Health, which had no role in the design, conduct or interpretation of the study.
Results
In uncorrected data, 41% of women reported ever douching and 35% reported ever using genital talc prior to enrollment (Table 1). Compared to women who did not douche, women who douched had higher BMI and were more likely to be non-Hispanic Black, live in the South, have a lower attained education, have had gynecological surgery, have used unopposed estrogen hormone therapy, and to have smoked. Similarly, genital talc users had higher BMIs than non-users, and were more likely to be non-Hispanic Black, live in the South, have lower attained education, and have had gynecological surgery.
Table 1.
Covariate distributions at enrollment by history of genital talc use and douchinga, Sister Study (2003–2009); n=50,800
| Douching | Genital Talc Use | |||||
|---|---|---|---|---|---|---|
| Never n=29,549 (59%) | Ever n=20,542 (41%) | Missing n=709 | Never n=32,541 (65%) | Ever n=17,560 (35%) | Missing n=697 | |
|
|
||||||
| Age; Mean (SD) | 55.6 (9.3) | 55.7 (8.5) | 53.1 (9.0) | 55.6 (9.2) | 55.8 (8.6) | 53.0 (8.9) |
| Follow-up time; Mean (SD) | 12.7 (3.4) | 13.5 (2.5) | 9.4 (4.8) | 12.9 (3.2) | 13.2 (2.9) | 9.4 (4.9) |
| Age at menarche; Mean (SD) | 12.7 (1.5) | 12.6 (1.5) | 12.6 (1.7) | 12.7 (1.5) | 12.5 (1.5) | 12.6 (1.7) |
| Age at menopause b ; Mean (SD) | 50.0 (6.1) | 49.6 (6.6) | 48.2 (7.4) | 49.9 (6.2) | 49.7 (6.4) | 48.1 (7.3) |
| Baseline BMI (kg/m2); Mean (SD) | 27.4 (6.1) | 28.3 (6.4) | 30.2 (6.7) | 27.3 (6.0) | 28.8 (6.6) | 30.1 (6.6) |
| Self-reported BMI in 30s; Mean (SD) | 23.2 (3.8) | 23.4 (4.2) | 24.4 (4.7) | 23.0 (3.8) | 23.7 (4.3) | 24.5 (4.6) |
| Self-reported race/ethnicity; N (%) | ||||||
| Non-Hispanic White | 25,726 (87) | 16,399 (80) | 361 (51) | 27,792 (85) | 14,337 (82) | 357 (51) |
| Non-Hispanic Black | 1,500 (5) | 2,745 (13) | 214 (30) | 2,115 (7) | 2,134 (12) | 210 (30) |
| Hispanic/Latina | 1,490 (5) | 911 (4) | 109 (15) | 1,703 (5) | 697 (4) | 110 (16) |
| Other | 824 (3) | 482 (2) | 24 (3) | 921 (3) | 389 (2) | 20 (3) |
| Attained education; N (%) | ||||||
| High school equivalent or less | 4,138 (14) | 3,543 (17) | 109 (15) | 4,978 (15) | 2,708 (15) | 104 (15) |
| Some College | 9,179 (31) | 7,689 (37) | 294 (42) | 10,602 (33) | 6,267 (36) | 293 (42) |
| Bachelor’s Degree | 8,541 (29) | 4,973 (24) | 176 (25) | 9,016 (28) | 4,499 (26) | 175 (25) |
| Graduate Degree | 7,685 (26) | 4,332 (21) | 129 (18) | 7,937 (24) | 4,084 (23) | 125 (18) |
| Census region; N (%) | ||||||
| Northeast | 5,495 (19) | 2,903 (14) | 101 (14) | 5,154 (16) | 3,242 (18) | 103 (15) |
| Midwest | 8,162 (28) | 5,298 (26) | 139 (20) | 8,940 (27) | 4,522 (26) | 137 (20) |
| South | 9,306 (31) | 8,033 (39) | 348 (49) | 10,821 (33) | 6,532 (37) | 334 (48) |
| West | 6,586 (22) | 4,308 (21) | 121 (17) | 7,626 (23) | 3,264 (19) | 125 (18) |
| Ever hormonal birth control use; N (%) | ||||||
| Never use | 4,919 (17) | 2,432 (12) | 117 (17) | 4,962 (15) | 2,395 (14) | 111 (16) |
| Used, 0–5 years | 11,673 (40) | 8,178 (40) | 288 (14) | 12,841 (40) | 7,016 (40) | 282 (41) |
| Used, >5 years | 12,793 (44) | 9,828 (48) | 297 (42) | 14,550 (45) | 3,069 (46) | 299 (43) |
| Parity; N (%) | ||||||
| No births | 5,295 (18) | 3,778 (18) | 121 (17) | 5,797 (18) | 3,279 (19) | 118 (17) |
| 1 birth | 3,900 (13) | 3,331 (16) | 102 (14) | 4,713 (15) | 2,521 (14) | 99 (14) |
| 2 births | 10,822 (37) | 7,590 (37) | 246 (35) | 11,894 (37) | 6,514 (37) | 250 (36) |
| ≥3 births | 9,520 (32) | 5,822 (28) | 238 (34) | 10,119 (31) | 5,231 (30) | 230 (33) |
| Menopausal status; N (%) | 19,463 (66) | 13,928 (68) | 400 (56) | 21,353 (66) | 12,045 (69) | 393 (56) |
| Hysterectomy; N (%) | 8,622 (29) | 7,115 (35) | 224 (32) | 9,745 (30) | 6,001 (34) | 215 (31) |
| Bilateral oophorectomy; N (%) | 5,077 (17) | 3,963 (19) | 118 (17) | 5,613 (17) | 3,432 (20) | 113 (16) |
| Tubal ligation; N (%) | 8,004 (27) | 6,748 (33) | 243 (34) | 9,351 (29) | 5,399 (31) | 245 (35) |
| Patent reproductive tract c ; N (%) | 15,636 (53) | 9,128 (44) | 328 (46) | 16,581 (51) | 8,190 (47) | 321 (46) |
| Hormone therapy use; N (%) | ||||||
| Never | 17,419 (59) | 11,222 (55) | 500 (71) | 18,928 (58) | 9,722 (56) | 491 (70) |
| Unopposed Estrogen | 5,390 (18) | 4,530 (22) | 121 (17) | 6,169 (19) | 3,755 (21) | 117 (17) |
| Estrogen plus Progestin | 6,663 (23) | 4,725 (23) | 87 (12) | 7,364 (23) | 4,021 (23) | 90 (13) |
| Smoking; N (%) | ||||||
| Never | 17,340 (59) | 10,780 (53) | 391 (55) | 18,497 (57) | 9,634 (55) | 380 (55) |
| Former | 9,985 (34) | 7,897 (38) | 226 (32) | 11,395 (35) | 6,489 (37) | 224 (32) |
| Current | 2,215 (8) | 1,859 (9) | 91 (13) | 2,641 (8) | 1,431 (8) | 93 (13) |
| Alcohol use; N (%) | ||||||
| Never or former | 5,557 (19) | 3,932 (19) | 169 (24) | 6,074 (19) | 3,421 (20) | 163 (23) |
| Current, <7 drinks/week | 19,898 (67) | 13,832 (67) | 466 (66) | 21,789 (67) | 11,948 (68) | 459 (66) |
| Current, ≥7 drinks/week | 4,033 (14) | 2,743 (13) | 71 (10) | 4,610 (14) | 2,164 (12) | 73 (11) |
Abbreviations: BMI = body mass index
Exclusions: 5 withdrawals and 79 women diagnosed with breast cancer before completing enrollment.
Covariate data was missing as follow (did not use douche [ND], used douche [D], missing douche status [MD], did not use talc [NT], used talc [T], missing talc status [MT]: age at menarche (n=46 [23 ND, 22 D, 1 MD; 29 NT, 16 T, 1 MT); BMI (n=17 [12 ND, 5 D; 10 NT, 7 T]), BMI in 30s (n=408 [247 ND, 153 D, 8 MD; 296 NT, 106 T, 6 MT]), race/ethnicity (n=15 [9 ND, 5 D, 1 MD; 10 NT, 3 T, 2 MT]), attained education (n=12 [6 ND, 5 ND, 1 MD; 8 NT, 2 T, 2 MT]), hormonal birth control use (n=275 [164 ND, 104 D, 7 MD; 188 NT, 80 T, 7 MT]), parity (n=35 [12 ND, 21 D, 2 MD; 18 NT, 15 T, 2 MT]), hysterectomy (n=5 [2 ND, 3 D; 3 NT, 2 T]), bilateral oophorectomy (n=77 [44 ND, 33 D; 47 NT, 30 T]), tubal ligation (n=16 [12 ND, 4 D; 13 NT, 3 T]), patency (n=1 [1 ND; 1 T]), hormone therapy (n=143 [77 ND, 65 D, 1 MD; 80 NT, 62 T, 1 MT]), smoking (n=16 [9 ND, 6 D, 1 MD; 8 NT, 6 T, 2 MT]), or alcohol (n=99 [61 ND, 35 D, 3 MD; 68 NT, 27 T, 4 MT])
If baseline and follow-up data are contradictory or the latter is missing, assign status provided at baseline; if missing baseline and not missing follow-up, assign status based on follow-up
Among post-menopausal women
No hysterectomy and no tubal ligation
The estimated HRs for the uncorrected models where women in the uncertain category (i.e., non-users at enrollment, missing follow-up) were categorized as non-users (Table 2, Scenario 1) indicated inverse or weakly positive associations between both types of intimate care products and all cancers of interest. Introducing probabilistic corrections for contradictory responses slightly increased exposure prevalence estimates (Scenario 2; 43% douching, 37% talc), but HRs were similar to Scenario 1. Under the assumption that all women in the uncertain category were users (Scenario 3), prevalence estimates were higher (64% douching, 56% talc), and all estimated HRs were greater than 1.00.
Table 2.
Quantitative bias analysis of the association between pre-enrollment use of douche or genital talc use and female reproductive system-related cancers (n=49,806)
| Scenario 1: No corrections, assume unexposed if unexposed at enrollment + missing at FU, fill in missing at randomb | Scenario 2: Correct contradictory datac, assume unexposed if unexposed at enrollment + missing at FU | Scenario 3: Correct contradictory datac, assume exposed if unexposed at enrollment + missing at FU | Scenario 4: Correct contradictory datac + Multiple imputationd | |
|---|---|---|---|---|
|
|
||||
| Douching | ||||
| % of cohort exposed | 41% | 43% | 64% | 53% |
| Premeno Breast Cancer (n=613) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.76 (0.63, 0.91) | 0.73 (0.61, 0.88) | 1.18 (0.99, 1.41) | 0.87 (0.71, 1.06) |
| Postmeno Breast Cancer (n=3,927) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.85 (0.80, 0.91) | 0.84 (0.78, 0.89) | 1.16 (1.08, 1.24) | 0.96 (0.89, 1.03) |
| Ovarian Cancer (n=292) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.69 (0.54, 0.88) | 0.67 (0.52, 0.86) | 1.86 (1.41, 2.45) | 1.03 (0.78, 1.37) |
| Uterine Cancer (n=433) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.76 (0.62, 0.93) | 0.79 (0.64, 0.97) | 1.11 (0.89, 1.37) | 0.86 (0.68, 1.08) |
| Genital Talc Use | ||||
| % of cohort exposed | 35% | 37% | 56% | 40% |
| Premeno Breast Cancer (n=613) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.91 (0.76, 1.10) | 0.91 (0.75, 1.09) | 1.36 (1.14, 1.61) | 0.98 (0.81, 1.19) |
| Postmeno Breast Cancer (n=3,927) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.92 (0.86, 0.98) | 0.92 (0.86, 0.99) | 1.18 (1.11, 1.26) | 0.96 (0.90, 1.03) |
| Ovarian Cancer (n=292) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 1.07 (0.84, 1.35) | 1.17 (0.92, 1.49) | 3.34 (2.51, 4.44) | 1.82 (1.36, 2.43) |
| Uterine Cancer (n=433) | ||||
| Never Use | 1.00 | 1.00 | 1.00 | 1.00 |
| Ever Use | 0.98 (0.80, 1.20) | 0.98 (0.80, 1.21) | 1.28 (1.04, 1.58) | 1.01 (0.82, 1.25) |
Adjusted for race/ethnicity (non-Hispanic White, non-Hispanic Black / African American, Hispanic/Latina, other), attained education (high school equivalent or less, some college, college graduate, graduate degree), examiner-measured body mass index at enrollment (continuous, kg/m2), self-reported body mass index ages 30–39 (continuous, kg/m2), age at menarche (continuous), hormonal birth control use (never, 0–5 years, >5 years), parity (0, 1, 2, ≥3 births), menopausal status (pre or postmenopausal), hormone therapy use (never, unopposed estrogen only, estrogen plus progestin), smoking status (never, former, current), and alcohol use (never or former, current <7 drinks/week, current ≥7 drinks/week), geographic region (Northeast, Midwest, South, West). Women missing one or more of these potential confounders were excluded (n=994). An additional 8,978 women excluded from the ovarian cancer analysis due to uncertain ovarian cancer status or pre-baseline ovarian cancer (n=225) or pre-baseline bilateral oophorectomy (n=8,753); 15,618 women excluded from uterine cancer analysis due to uncertain uterine cancer status or pre-baseline uterine cancer (n=416) or pre-baseline hysterectomy (n=15,202).
If baseline and follow-up data are contradictory or the latter is missing, assign status provided at baseline; if missing baseline and not missing follow-up, assign status based on follow-up; if missing both time points randomly assigned exposure status based on overall distribution (assumed missing completely at random).
If report unexposed at baseline but exposed at follow-up and ages contradictory, assume 80% truly exposed; if report exposed at baseline but never exposed at follow-up assume 90% of those truly exposed
Multiple imputation models included covariates considered confounders for the multivariable analysis in addition to: childhood household income level (well off, middle income, low income, poor, missing), adult household income (<$50,000/year, $50–99,999/year, ≥$100,000/year), highest attained education level of the head of the household when the participant was 13 (<high school, high school or GED, some college, college graduate, missing), marital status (never married, divorced/ widowed/ separated, married / living as married, missing), weight relative to peers in teen years (lighter, same, heavier, missing), breast cancer (prebasline/never/incident), ovarian cancer (prebasline/never/incident), uterine cancer (prebasline/never/incident), and cumulative hazard values for time to any of the cancer events of interest.
After multiple imputation (Table 2, Scenario 4), 53% of participants were categorized as ever douchers and 40% were categorized as ever genital talc users. Douching was not strongly associated with any of the examined outcomes. Ever genital talc use was positively associated with ovarian cancer (HR=1.82, CI: 1.36–2.43), but showed no evidence of an association with pre- or post-menopausal breast cancer, or uterine cancer.
If ovarian cancer cases over-reported their exposure to genital talc, the HR estimates would be biased upwards and away from 1.00 (Figure 2, Supplementary Table 2). For example, the HR dropped to 1.41 (CI: 1.06–1.87) if 25% of cases who only reported being exposed on the post-diagnosis questionnaire were truly unexposed. However, the impact of recall bias was greatly reduced when only short-term, infrequent users were re-assigned, with HRs indicating a positive association even when 90% were re-classified (HR=1.25, CI: 0.95–1.65).
Figure 2:
Forest plots showing covariate-adjusted hazard ratios (HRs) and 95% confidence intervals for the association between ever genital talc use and incident ovarian cancer across different possible scenarios of recall bias induced by exposure misclassification that differed by ovarian cancer status
If some non-cases reporting never exposure were truly users (albeit infrequent and short-term ones), HR estimates would also be biased up and away from the null. However, the positive association held when we assumed 15% of non-cases (HR=1.40, CI: 1.04–1.88) had misreported. In a model assuming moderate corrections for both cases (re-assign 25% of infrequent/short-term users to be non-users) and non-cases (re-assign 10% of non-users to be infrequent/short-term users), the estimated prevalence of ever talc use changed from 40% (55% in cases) to 45% (54% in cases) and the HR estimate was 1.40 (CI: 1.04–1.89).
The association between genital talc use and ovarian cancer was higher for frequent (recall bias corrected-HR [HRrb]=1.81, CI: 1.29–2.53) and long-term users (HRrb=2.01, CI: 1.39–2.91), compared to never users (both p-for-trend=0.001; Table 3). Genital talc use during a woman’s 20s and 30s was positively associated with incident ovarian cancer, but HRs were near null for teen use. Frequent douching (HR=1.50, CI: 1.02–2.19) and douching during a woman’s 20s (HR=1.35, 1.02–1.78) and 30s (HR=1.48, 1.11–1.99) were also positively associated with ovarian cancer.
Table 3.
Covariate-adjusteda hazard ratios (HR) and 95% confidence intervals (CIs) for the associations between douching and genital talc use by frequency, duration and timing of use based on models with contradictory data correctionsb and multiple imputationc
| % exp. In full cohortd | Premenopausal Breast Cancer n=613 | Postmenopausal Breast Cancer n=3,927 | Ovarian Cancer n=292 No recall bias correction | Ovarian Cancer n=292 Recall bias correctione | Uterine Cancer n=433 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
|||||||||||
| % exp | HR (95% CI) | % exp | HR (95% CI) | % exp | HR (95% CI) | % exp | HR (95% CI) | % exp | HR (95% CI) | ||
| Douching | |||||||||||
| Ever vs. Never Use | 41/53 | 44 | 0.87 (0.71, 1.06) | 53 | 0.96 (0.89, 1.03) | 56 | 1.03 (0.78, 1.36) | -- | -- | 48 | 0.86 (0.68, 1.08) |
| Never Use | 59/47 | 56 | 1.00 | 47 | 1.00 | 44 | 1.00 | -- | -- | 52 | 1.00 |
| Sometimes Use | 31/40 | 38 | 0.90 (0.73, 1.10) | 38 | 0.95 (0.87, 1.02) | 35 | 0.88 (0.65, 1.21) | -- | -- | 36 | 0.86 (0.67, 1.09) |
| Frequent Use | 10/14 | 6 | 0.69 (0.46, 1.05) | 15 | 0.98 (0.88, 1.10) | 21 | 1.50 (1.02, 2.19) | -- | -- | 12 | 0.87 (0.60, 1.26) |
| p-trend=0.08 | p-trend=0.48 | p-trend=0.15 | -- | p-trend=0.29 | |||||||
| Never Use | 59/47 | 56 | 1.00 | 47 | 1.00 | 44 | 1.00 | -- | -- | 52 | 1.00 |
| Short-term use (1 decade only) | 17/21 | 18 | 0.81 (0.62, 1.05) | 21 | 0.94 (0.86, 1.03) | 16 | 0.72 (0.48, 1.08) | -- | -- | 19 | 0.80 (0.61, 1.07) |
| Long-term use (≥2 decades) | 24/32 | 26 | 0.93 (0.73, 1.17) | 32 | 0.96 (0.89, 1.05) | 40 | 1.27 (0.94, 1.72) | -- | -- | 29 | 0.86 (0.65, 1.13) |
| p-trend=0.40 | p-trend=0.40 | p-trend=0.16 | p-trend=0.35 | ||||||||
| Ever vs. Never use, teens | 12/15 | 16 | 0.92 (0.72, 1.18) | 12 | 0.86 (0.78, 0.96) | 12 | 0.76 (0.50 1.15) | -- | -- | 10 | 0.73 (0.52 1.03) |
| Ever vs. Never use, 20s | 31/42 | 34 | 0.89 (0.72, 1.09) | 42 | 0.99 (0.92, 1.06) | 50 | 1.35 (1.02, 1.78) | -- | -- | 38 | 0.89 (0.70, 1.13) |
| Ever vs. Never use, 30s | 22/30 | 24 | 1.12 (0.90, 1.41) | 31 | 1.04 (0.96, 1.13) | 39 | 1.48 (1.11, 1.99) | -- | -- | 27 | 0.92 (0.71, 1.19) |
| Ever vs. Never use, year prior to baseline | 14/14 | 16 | 1.07 (0.85, 1.35) | 13 | 1.05 (0.95, 1.16) | 17 | 1.34 (0.97, 1.83) | -- | -- | 12 | 0.98 (0.73, 1.33) |
| Genital Talc | |||||||||||
| Ever vs. Never Use | 35/40 | 36 | 0.98 (0.81, 1.19) | 40 | 0.96 (0.90, 1.03) | 55 | 1.82 (1.36, 2.43) | 54 | 1.40 (1.04, 1.89) | 42 | 1.01 (0.82, 1.25) |
| Never Use | 65/60 | 65 | 1.00 | 60 | 1.00 | 46 | 1.00 | 45 | 1.00 | 58 | 1.00 |
| Sometimes Use | 17/19 | 16 | 0.89 (0.69, 1.14) | 19 | 0.97 (0.88, 1.06) | 23 | 1.56 (1.09, 2.22) | 24 | 1.18 (0.83, 1.69) | 19 | 0.99 (0.75, 1.29) |
| Frequent Use | 18/20 | 19 | 1.10 (0.87, 1.39) | 21 | 0.96 (0.88, 1.05) | 31 | 1.99 (1.43, 2.78) | 31 | 1.81 (1.29, 2.53) | 23 | 1.03 (0.79, 1.33) |
| p-trend=0.68 | p-trend=0.35 | p-trend<0.001 | p-trend=0.001 | p-trend=0.88 | |||||||
| Never Use | 65/60 | 65 | 1.00 | 60 | 1.00 | 46 | 1.00 | 45 | 1.00 | 58 | 1.00 |
| Short-term use (1 decade only) | 22/23 | 19 | 0.91 (0.72, 1.13) | 23 | 0.93 (0.86, 1.01) | 26 | 1.48 (1.06, 2.06) | 28 | 1.17 (0.84, 1.62) | 24 | 0.98 (0.77 1.26) |
| Long-term use (≥2 decades) | 13/17 | 16 | 1.13 (0.87, 1.47) | 17 | 1.01 (0.92, 1.11) | 27 | 2.20 (1.52, 3.19) | 27 | 2.01 (1.39, 2.91) | 18 | 1.08 (0.80, 1.45) |
| p-trend=0.67 | p-trend=0.76 | p-trend<0.001 | p-trend=0.001 | p-trend=0.76 | |||||||
| Ever vs. Never use, teens | 15/15 | 14 | 1.06 (0.82, 1.36) | 14 | 0.95 (0.87, 1.05) | 17 | 1.17 (0.84, 1.63) | 18 | 0.98 (0.71, 1.37) | 14 | 0.92 (0.69, 1.23) |
| Ever vs. Never use, 20s | 15/19 | 16 | 1.02 (0.79, 1.31) | 18 | 1.00 (0.92, 1.10) | 31 | 2.03 (1.49, 2.77) | 31 | 1.88 (1.37, 2.57) | 18 | 1.01 (0.77, 1.33) |
| Ever vs. Never use, 30s | 11/15 | 16 | 1.30 (1.00, 1.68) | 16 | 1.11 (1.01, 1.22) | 26 | 2.12 (1.53, 2.96) | 26 | 2.08 (1.50, 2.89) | 15 | 1.07 (0.80, 1.44) |
| Ever vs. Never use, year prior to baseline | 18/18 | 15 | 0.87 (0.69, 1.10) | 18 | 0.92 (0.84, 1.00) | 17 | 0.91 (0.66, 1.23) | 18 | 0.83 (0.61, 1.14) | 21 | 1.09 (0.86, 1.39) |
Adjusted for race/ethnicity (non-Hispanic White, non-Hispanic Black / African American, Hispanic/Latina, other), attained education (high school equivalent or less, some college, college graduate, graduate degree), examiner-measured body mass index at enrollment (continuous, kg/m2), self-reported body mass index ages 30–39 (continuous, kg/m2), age at menarche (continuous), hormonal birth control use (never, 0–5 years, >5 years), parity (0, 1, 2, ≥3 births), menopausal status (pre or postmenopausal), hormone therapy use (never, unopposed estrogen only, estrogen plus progestin), smoking status (never, former, current), alcohol use (never or former, current <7 drinks/week, current ≥7 drinks/week) and geographic region (Northeast, Midwest, South, West).
If report unexposed at baseline but exposed at follow-up and ages contradictory, assume 80% truly exposed; if report exposed at baseline but never exposed at follow-up assume 90% are truly exposed
Exposure values imputed if missing both time points or non-user at baseline and missing follow-up. Multiple imputation models included covariates considered confounders for the multivariable analysis in addition to: childhood household income level (well off, middle income, low income, poor, missing), adult household income (<$50,000/year, $50–99,999/year, ≥$100,000/year), highest attained education level of the head of the household when the participant was 13 (<high school, high school or GED, some college, college graduate, missing), marital status (never married, divorced/ widowed/ separated, married / living as married, missing), weight relative to peers in teen years (lighter, same, heavier, missing), geographic region of residence (Northeast, Midwest, South, West, missing), breast cancer (prebasline/never/incident), ovarian cancer (prebasline/never/incident), uterine cancer (prebasline/never/incident), and cumulative hazard values for time to any of the cancer events of interest.
(% before imputation / % after imputation); % of users in comparisons of ever versus never use
Assume 25% of non-frequent, short-term users with ovarian cancer misreport their exposure and that 10% of non-cases who report no use are actually short-term and infrequent users. Recall bias corrections not made for douching status.
Results from analyses limited to medically-confirmed cancers were similar, and there were no clear subtype differences (Table 4). Estimates based only on exposure status reported at enrollment were mostly null (Supplementary Table 3), except for a possible positive association between douching and ovarian cancer. Cancer cases were under-represented in analyses relying on follow-up data only, and most HRs were less than 1.0. The exception was ovarian cancer and genital talc use, where an estimated HR of 2.65 (CI: 1.91–3.70) could indicate some recall bias. Analyses considering person-time accrued since follow-up questionnaire completion were not subject to recall bias, but had reduced sample size; estimates of the genital talc and ovarian cancer association were consistent with a positive association (HR=1.84, 95% CI: 0.90–3.77).
Table 4.
Adjusted hazard ratios (HRs)a and 95% confidence intervals (CIs) estimating the association between history of douching, genital talc use, and female reproductive system-related cancers by subtype (n=49,806), based on multiple imputation modelsb with contradictory data correctionsc
| N cases | Ever Douching HR (95% CI) | Ever Genital Talc Use, no recall bias correction HR (95% CI) | Ever Genital Talc Use, corrected for recall biasd HR (95% CI) | |
|---|---|---|---|---|
|
|
||||
| Breast Cancer | 4,540 | 0.94 (0.88, 1.00) | 0.96 (0.90, 1.03) | -- |
| Estrogen Receptor Positive | 3,272 | 0.93 (0.86, 1.01) | 0.96 (0.88, 1.03) | -- |
| Estrogen Receptor Negative | 574 | 0.99 (0.82, 1.21) | 1.00 (0.84, 1.20) | -- |
| Ovarian Cancer | 292 | 1.03 (0.78, 1.36) | 1.82 (1.36, 2.43) | 1.40 (1.04, 1.89) |
| Medically-confirmed | 226 | 0.95 (0.69, 1.31) | 1.89 (1.37, 2.62) | 1.46 (1.06, 2.02) |
| Serous | 126 | 1.05 (0.69, 1.59) | 2.12 (1.38, 3.26) | 1.62 (1.06, 2.48) |
| Non-Serous | 100 | 0.83 (0.52, 1.34) | 1.64 (1.02, 2.65) | 1.29 (0.79, 2.09) |
| Uterine Cancer | 433 | 0.86 (0.68, 1.06) | 1.01 (0.82, 1.25) | -- |
| Medically-confirmed | 338 | 0.91 (0.70, 1.17) | 1.05 (0.83, 1.33) | -- |
| Endometrial Cancer | 317 | 0.90 (0.68, 1.17) | 1.03 (0.81, 1.32) | -- |
| Type I Endometrial Cancer | 257 | 0.91 (0.68, 1.23) | 0.99 (0.75, 1.30) | -- |
| Type II Endometrial Cancer | 45 | 0.80 (0.41, 1.54) | 1.51 (0.77, 2.95) | -- |
Adjusted for race/ethnicity (non-Hispanic White, non-Hispanic Black / African American, Hispanic/Latina, other), attained education (high school equivalent or less, some college, college graduate, graduate degree), examiner-measured body mass index at enrollment (continuous, kg/m2), self-reported body mass index ages 30–39 (continuous, kg/m2), age at menarche (continuous), hormonal birth control use (never, 0–5 years, >5 years), parity (0, 1, 2, ≥3 births), menopausal status (pre or postmenopausal), hormone therapy use (never, unopposed estrogen only, estrogen plus progestin), smoking status (never, former, current), alcohol use (never or former, current <7 drinks/week, current ≥7 drinks/week), geographic region (Northeast, Midwest, South, West), and an interaction term for BMI and menopausal status at enrollment. Women missing one or more of these potential confounders were excluded (n=603). An additional 8,952 women excluded from the ovarian cancer analysis due to pre-baseline ovarian cancer or pre-baseline oophorectomy; 15,473 women excluded from uterine cancer analysis due to pre-baseline uterine cancer or pre-baseline hysterectomy.
Multiple imputation models included covariates considered confounders for the multivariable analysis in addition to childhood household income level (well off, middle income, low income, poor, missing), adult household income (<$50,000/year, $50–99,999/year, ≥$100,000/year), highest attained education level of the head of the household when the participant was 13 (<high school, high school or GED, some college, college graduate, missing), marital status (never married, divorced/ widowed/ separated, married / living as married, missing), weight relative to peers in teen years (lighter, same, heavier, missing), and geographic region (Northeast, Midwest, South, West, missing), breast cancer (prebasline/never/incident), ovarian cancer (prebasline/never/incident), uterine cancer (prebasline/never/incident), and cumulative hazard values for time to any of the cancer events of interest.
If report unexposed at baseline but exposed at follow-up and ages contradictory, assume 80% truly exposed; if report exposed at baseline but never exposed at follow-up assume 90% of those truly exposed
Assume 25% of non-frequent, short-term users with ovarian cancer misreport their exposure and that 10% of non-cases who report no use are actually short-term and infrequent users
Analyses jointly considering patency and genital talc use (Supplementary Table 4), relative to never use, showed a potentially stronger association with ovarian cancer among women who used while patent (HR=1.55, 95% CI: 1.14–2.09). Co-adjustment did not notably alter estimates (Supplementary Table 5).
Discussion
Using newly collected data on intimate care product use in a large cohort of US women, we found evidence supporting a positive association between ever genital talc use and incident ovarian cancer. Frequent douching and douching ages 20–39 were also associated with higher rates of ovarian cancer, but neither genital talc use nor douching was consistently associated with breast or uterine cancer. We did not observe clear differences in HRs by subtypes.
Associations between genital talc use and ovarian cancer remained positive, though attenuated, in most quantitative bias analyses addressing missing data biases and potential differential reporting of genital talc use by ovarian cancer status. In an example scenario correcting for mis-reporting in both cases and non-cases, genital talc use was associated with an approximately 40% higher rate of ovarian cancer, compared to never use, with consistently increasing dose-response patterns for both frequency and duration of use.
These results do not establish causality and do not implicate any specific cancer-inducing agent. Those reporting talc use could be recalling products that contained talc, cornstarch, or a mixture, and women may have used different products at different times. Some talc may have been contaminated with asbestos,5 or other potentially harmful chemicals such as phthalates or parabens.1,3 Chronic irritation of the ovaries or fallopian tubes from talc or talc-like products could also potentially contribute to carcinogenesis.
Our findings of a positive association between genital talc use and ovarian cancer are consistent with prior studies. Pooled- or meta-analyses of case-control studies have produced odds ratios of 1.2–1.4.33–37 The HR from a pooled analysis of prospective cohort studies21 also indicated a positive, albeit small association (HR=1.08), and as previously noted, this effect estimate is likely biased toward the null due to non-differential misclassification of exposure. This possibility is well-illustrated by the Sister Study, where we previously reported 27% ever use of genital talc,21 but here observe 40% ever use across a wider age range.
Results from the present analysis suggest ages 20–39 may be a window of susceptibility, which is consistent with prior studies that considered ages of use.38–40 These cover years where hormone levels are high and many women are reproducing. Increased sexual activity during this time period may also correspond to frequent intimate care product use. Additionally, this window occurs before most hysterectomies and tubal ligations are performed, meaning that most women had an intact physical path between application site and the ovaries and fallopian tubes.
Our findings that neither genital talc use nor douching was strongly associated with uterine cancer were consistent with prior literature;24,41,42 we are unaware of any relevant breast cancer studies. We hypothesized that the endocrine disruptors in douche could impact carcinogenesis for any hormone-related cancer, but only the uterus and ovaries could experience adverse effects caused by direct physical contact. Further, unlike the ovaries and fallopian tube epithelium, uterine epithelium (endometrium) sheds and regenerates frequently via menstruation. This process may flush out the tissue and mitigate talc-induced damage to uterine tissue.
Our exposure assessment included a mix of retrospective and prospective information, integrating some of the strengths and limitations of each data type. Because we only considered incident cancers, self-reported intimate care product use at enrollment was not influenced by recall bias. However, the Sister Study’s overall scope was wide,25 and the initial intimate care product questions were limited to two specific periods: ages 10–13 and the last year, and did not capture lifetime exposure or use during the most likely exposure period of ages 20–39.8
The follow-up questionnaire included a more comprehensive assessment, but did so after some individuals had been diagnosed with cancer, allowing for the possibility of recall bias.22,23 Further, those who died from their disease could not have completed the follow-up questionnaire, allowing for bias due to differential missingness. This was particularly problematic for ovarian cancer, which has a low survival rate.43
Because the Sister Study is a volunteer cohort of women who have a sister with breast cancer, participants have higher levels of attained education, are more likely to identify as non-Hispanic White, and have a higher average risk of both breast and ovarian cancer, compared to the general US population. Given that patterns of intimate care product use differ by some of these factors,2,8 our findings may not generalize to all US women or to international populations. Another important limitation was our sample size. Although the Sister Study is one of the largest studies to collect data on intimate care product use, we lacked statistical power for investigating rare subtypes or differences across subgroups.
Our quantitative bias analyses are a major strength, as they provide a comprehensive illustration of the possible impacts of missing data and recall bias under a variety of scenarios. Though prospectively collected data is preferable for future investigations, our findings demonstrate how retrospective studies can evaluate the possible impact of recall bias. Detailed data on related covariates informed our complex imputations and limited the possibility of residual confounding. However, unmeasured confounding could still be present.
Overall, our findings support the hypothesis that there is a positive association between genital talc use and ovarian cancer incidence, though they do not pinpoint a specific cause or mechanism, and there is still uncertainty as to how much recall bias and missing data could upwardly bias effect estimates. If the underlying biological mechanisms and causal agents can be confirmed, interventions and policies designed to limit exposure to the harmful components of intimate care products have the potential to reduce ovarian cancer incidence.
Supplementary Material
Context Summary.
Key objective:
Are history of genital talc use and douching associated with breast, ovarian, or uterine cancer after correcting for likely biases?
Knowledge generated:
Genital talc use was positively associated with ovarian cancer for a range of plausible bias-correction scenarios, with higher rates seen for frequent and long-term users. Douching frequently or during young adulthood were also positively associated with ovarian cancer, but neither douching nor talc were associated with breast or uterine cancer.
Funding:
This work was support by the Intramural Research Program of the National Institute of Environmental Health Sciences, National Institutes of Health (Z1AES044005 to DPS).
Footnotes
Conflicts of Interest: None of the authors has a conflict of interest to declare.
References
- 1.Gao CJ, Kannan K. Phthalates, bisphenols, parabens, and triclocarban in feminine hygiene products from the United States and their implications for human exposure. Environment International. 2020;136:105465. doi: 10.1016/j.envint.2020.105465 [DOI] [PubMed] [Google Scholar]
- 2.Johnson PI, Favela K, Jarin J, et al. Chemicals of concern in personal care products used by women of color in three communities of California. J Expo Sci Environ Epidemiol. 2022;32(6):864–876. doi: 10.1038/s41370-022-00485-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lin N, Ding N, Meza-Wilson E, et al. Volatile organic compounds in feminine hygiene products sold in the US market: A survey of products and health risks. Environment International. 2020;144:105740. doi: 10.1016/j.envint.2020.105740 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ding N, Batterman S, Park SK. Exposure to volatile organic compounds and use of feminine hygeine products among reproductive-aged women in the United States. Journal of Women’s Health2. Published online 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Food and Drug Administration. FDA Advises Consumers to Stop Using Certain Cosmetic Products. https://www.fda.gov/cosmetics/cosmetics-recalls-alerts/fda-advises-consumers-stop-using-certain-cosmetic-products
- 6.IARC Working Groups. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Arsenic, Metals, Fibres and Dusts. Vol 100C.; 2012:219–309. https://monographs.iarc.fr/iarc-monographs-on-the-evaluation-of-carcinogenic-risks-to-humans-19/ [PMC free article] [PubMed] [Google Scholar]
- 7.Ferranti M “From birth control to that fresh feeling”: A historical perspective on feminine hygiene in medicine and media. Women and Health. 2009;49(8):592–607. doi: 10.1080/03630240903496069 [DOI] [PubMed] [Google Scholar]
- 8.O’Brien KM, Ogunsina K, Wentzensen N, Sandler DP. Douching and genital talc use: Patterns of use and reliability of self-reported exposure. Epidemiology. 2023;Publish Ahead of Print. doi: 10.1097/EDE.0000000000001589 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Branch F, Woodruff TJ, Mitro SD, Zota AR. Vaginal douching and racial/ethnic disparities in phthalates exposures among reproductive-aged women: National Health and Nutrition Examination Survey 2001–2004. Environ Health. Published online 2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Martino JL, Vermund SH. Vaginal Douching: Evidence for Risks or Benefits to Women’s Health. Epidemiol Rev. 2002;24(2):109–124. doi: 10.1038/mp.2011.182.doi [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.O’Brien KM, Weinberg CR, D’Aloisio AA, Moore KR, Sandler DP. The association between douching, genital talc use, and the risk of prevalent and incident cervical cancer. Scientific Reports. 2021;11(1):1–10. doi: 10.1038/s41598-021-94447-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gonzalez N, O’Brien KM, D’Aloisio AA, Sandler DP, Weinberg CR. Douching, talc use, and risk of ovarian cancer. Epidemiology. 2017;28(2):797–802. doi: 10.1097/EDE.0000000000000528 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gabriel IM, Vitonis AF, Welch WR, Titus L, Cramer DW. Douching, Talc Use, and Risk for Ovarian Cancer and Conditions Related to Genital Tract Inflammation. Cancer Epidemiology, Biomarkers & Prevention. 2019;28(11):1835–1844. doi: 10.1158/1055-9965.EPI-19-0375 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wentzensen N, O’Brien KM. Talc, body powder, and ovarian cancer: A summary of the epidemiologic evidence. Gynecologic Oncology. Published online 2021. doi: 10.1016/j.ygyno.2021.07.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cramer DW, Welch WR, Scully RE, Wojciechowski CA. Ovarian Cancer and Talc: A Case-Control Study. Cancer. 1982;50:372–376. [DOI] [PubMed] [Google Scholar]
- 16.Harlow BL, Cramer DW, Bell DA, Welch WR. Perineal exposure to talc and ovarian cancer risk. Obstet Gynecol. 1992;80(1):19–26. [PubMed] [Google Scholar]
- 17.Hartge P, Hoover RN, Lesher LP, McGowan L. Talc and ovarian cancer. JAMA. 1983;250(14):1844. [PubMed] [Google Scholar]
- 18.Whittemore AS, Wu ML, Paffenbarger RS, et al. Personal and environmental characteristics related to epithelial ovarian cancer. Am J Epidemiol. 1988;128(6):1228–1240. [DOI] [PubMed] [Google Scholar]
- 19.Gertig DM, Hunter DJ, Cramer DW, et al. Prospective Study of Talc Use and Ovarian Cancer. J Natl Cancer Inst. 2000;92(3):249–252. [DOI] [PubMed] [Google Scholar]
- 20.Houghton SC, Reeves KW, Hankinson SE, et al. Perineal powder use and risk of ovarian cancer. J Natl Cancer Inst. 2014;106(9). doi: 10.1093/jnci/dju208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.O’Brien KM, Tworoger SS, Harris HR, et al. Association of Powder Use in the Genital Area with Risk of Ovarian Cancer. JAMA - Journal of the American Medical Association. 2020;323(1):49–59. doi: 10.1001/jama.2019.20079 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Schildkraut JM, Abbott SE, Alberg AJ, et al. Association between Body Powder Use and Ovarian Cancer: The African American Cancer Epidemiology Study (AACES). Cancer Epidemiology Biomarkers & Prevention. 2016;25(10):1411–1417. doi: 10.1158/1055-9965.epi-15-1281 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Trabert B Body power and ovarian cancer risk - what is the role of recall bias? Cancer Epidemiol Biomarkers Prev. 2016;25(10):1369–1370. doi: 10.1158/1055-9965.EPI-16-0476 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.O’Brien KM, D’Aloisio AA, Shi M, Murphy JD, Sandler DP, R WC. Perineal talc use, douching and the risk of uterine cancer. Epidemiology (Cambridge, Mass)2. 2019;published. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Sandler DP, Hodgson ME, Deming-Halverson SL, et al. The Sister Study: Baseline methods and participant characteristics. Environ Health Perspect. 2017;125(12):127003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.American Cancer Society: Ovarian Cancer Stages. Published online April 18, 2018. Accessed May 19, 2023. https://www.cancer.org/cancer/types/ovarian-cancer/detection-diagnosis-staging/staging.html
- 27.van den Brandt PA, Spiegelman D, Yaun SSSS, et al. Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol. 2000;152(6):514–527. [DOI] [PubMed] [Google Scholar]
- 28.Schoemaker MJ, Nichols HB, Wright LB, et al. Association of Body Mass Index and Age With Subsequent Breast Cancer Risk in Premenopausal Women. JAMA Oncology. Published online 2018:1–10. doi: 10.1001/jamaoncol.2018.1771 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Peres LC, Cushing-Haugen KL, Kobel M, et al. Invasive epithelial ovarian cancer survival by histotype and disease stage. Journal of the National Cancer Institute. 2019;111(1). doi: 10.1093/jnci/djy071 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Clarke MA, Devesa SS, Harvey SV, Wentzensen N. Hysterectomy-corrected uterine corpus cancer incidence trends and differences in relative survival reveal racial disparities and rising rates of nonendometroid cancers. J Clin Oncol. 2019;37(22):1895–1908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Rubin Donald. Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons, Inc; 1987. [Google Scholar]
- 32.White IR, Royston P. Imputing missing covariate values for the Cox model: Imputing missing covariate values for the Cox model. Statist Med. 2009;28(15):1982–1998. doi: 10.1002/sim.3618 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Penninkilampi R, Eslick GD. Perineal Talc Use and Ovarian Cancer. Epidemiology. 2018;29(1):41–49. doi: 10.1097/EDE.0000000000000745 [DOI] [PubMed] [Google Scholar]
- 34.Taher MK, Farrhat N, Karyakina NA, et al. Critical review of the association between perineal use of talc powder and risk of ovarian cancer. Reproductive Toxicology. Published online 2019. doi: 10.1016/j.ejbt.2019.07.002 [DOI] [PubMed] [Google Scholar]
- 35.Terry KL, Karageorgi S, Shvetsov YB, et al. Genital powder use and risk of ovarian cancer: A pooled analysis of 8,525 cases and 9,859 controls. Cancer Prevention Research. 2013;6(8):811–821. doi: 10.1158/1940-6207.CAPR-13-0037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Berge W, Mundt K, Luu H, Boffetta P. Genital use of talc and risk of ovarian cancer: a meta-analysis. European Journal of Cancer Prevention. 2017;00(00):0-0. doi: 10.1097/CEJ.0000000000000340 [DOI] [PubMed] [Google Scholar]
- 37.Davis CP, Bandera EV, Bethea TN, et al. Genital Powder Use and Risk of Epithelial Ovarian Cancer in the Ovarian Cancer in Women of African Ancestry Consortium. Cancer Epidemiology, Biomarkers & Prevention. 2021;30(9):1660–1668. doi: 10.1158/1055-9965.EPI-21-0162 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Mills PK, Riordan DG, Cress RD, Young HA. Perineal talc exposure and epithelial ovarian cancer risk in the central valley of California. International Journal of Cancer. 2004;112(3):458–464. doi: 10.1002/ijc.20434 [DOI] [PubMed] [Google Scholar]
- 39.Rosenblatt KA, Weiss NS, Cushing-Haugen KL, Wicklund KG, Rossing MA. Genital powder exposure and the risk of epithelial ovarian cancer. Cancer Causes and Control. 2011;22(5):737–742. doi: 10.1007/s10552-011-9746-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Cramer DW, Vitonis AF, Terry KL, Welch WR, Titus LJ. The association between talc use and ovarian cancer a retrospective case-control study in two us states. Epidemiology. 2016;27(3):334–346. doi: 10.1097/eDe.0000000000000434 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.O’Brien KM, Tworoger SS, Harris HR, et al. Genital powder use and risk of uterine cancer: A pooled analysis of prospective studies. International Journal of Cancer. 2021;(November 2020):1–10. doi: 10.1002/ijc.33470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Neill AS, Nagle CM, Spurdle AB, Webb PM. Use of talcum powder and endometrial cancer risk. Cancer Causes and Control. 2012;23(3):513–519. doi: 10.1007/s10552-011-9894-5 [DOI] [PubMed] [Google Scholar]
- 43.Cancer Fact Sheets: Ovarian Cancer. National Cancer Institute. Surveillance, Epidemiology, and End Results Program; 2023. https://seer.cancer.gov/statfacts/html/ovary.html [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.