[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
Purchase Options:
[Skip to Content Landing]
Figure.  Subgroup Analyses for the Association Between Ever Use of Powder in the Genital Area and Risk of Ovarian Cancer, Pooled Hazard Ratios (HRs) and 95% CIsa
Subgroup Analyses for the Association Between Ever Use of Powder in the Genital Area and Risk of Ovarian Cancer, Pooled Hazard Ratios (HRs) and 95% CIsa

aAdjusted for study, race/ethnicity (white, African American, other), education (<high school, some college, ≥college graduate), body mass index (BMI [calculated as weight in kilograms divided by height in meters squared], restricted cubic spline), parity (0, 1, 2, ≥3 births), ever use of oral contraceptives, tubal ligation (yes or no), hysterectomy (yes or no), menopausal status (premenopausal or postmenopausal), ever hormone therapy use. When estimating HRs within a strata of a variable, that variable was not included in the adjustment set.

bNumbers include only participants with complete covariate information.

cEffect estimate based on menopausal status updated throughout follow-up. Of the 2168 cases, 165 were diagnosed while the participant was premenopausal and 2003 occurred after menopause.

dAmong women who were postmenopausal at baseline.

eCalculated as weight in kilograms divided by height in meters squared.

Table 1.  Description of Participating Cohortsa
Description of Participating Cohortsa
Table 2.  Study-Specific and Pooled Risk Differences, Hazard Ratios, and 95% CIs for the Association Between Ever Use of Powder in the Genital Area and Risk of Ovarian Cancer
Study-Specific and Pooled Risk Differences, Hazard Ratios, and 95% CIs for the Association Between Ever Use of Powder in the Genital Area and Risk of Ovarian Cancer
Table 3.  Study-Specific and Pooled Risk Differences, Hazard Ratios, and 95% CIs for the Association Between Duration and Frequency of Powder Use in the Genital Area and Risk of Ovarian Cancer
Study-Specific and Pooled Risk Differences, Hazard Ratios, and 95% CIs for the Association Between Duration and Frequency of Powder Use in the Genital Area and Risk of Ovarian Cancer
Table 4.  Pooled Hazard Ratios and 95% CIs Among Medically Confirmed Cases Overall and by Tumor Invasiveness, Location, and Histotype
Pooled Hazard Ratios and 95% CIs Among Medically Confirmed Cases Overall and by Tumor Invasiveness, Location, and Histotype
1.
World Health Organization International Agency for Research on Cancer.  IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Carbon Black, Titanium Dioxide, and Talc; vol 93; 2010.http://publications.iarc.fr/111 Accessed December 10, 2019.
2.
International Agency for Research on Cancer.  IARC Monographs: Arsenic, Metals, Fibres, and Dusts; vol 100C; 2012.http://publications.iarc.fr/120 Accessed December 10, 2019.
3.
Fiume  MM, Boyer  I, Bergfeld  WF,  et al.  Safety assessment of talc as used in cosmetics.  Int J Toxicol. 2015;34(1)(suppl):66S-129S. doi:10.1177/1091581815586797PubMedGoogle ScholarCrossref
4.
Terry  KL, Karageorgi  S, Shvetsov  YB,  et al; Australian Cancer Study (Ovarian Cancer); Australian Ovarian Cancer Study Group; Ovarian Cancer Association Consortium.  Genital powder use and risk of ovarian cancer: a pooled analysis of 8525 cases and 9859 controls.  Cancer Prev Res (Phila). 2013;6(8):811-821. doi:10.1158/1940-6207.CAPR-13-0037PubMedGoogle ScholarCrossref
5.
Penninkilampi  R, Eslick  GD.  Perineal talc use and ovarian cancer: a systematic review and meta-analysis.  Epidemiology. 2018;29(1):41-49. doi:10.1097/EDE.0000000000000745PubMedGoogle ScholarCrossref
6.
Schildkraut  JM, Abbott  SE, Alberg  AJ,  et al.  Association between body powder use and ovarian cancer: the African American Cancer Epidemiology Study (AACES).  Cancer Epidemiol Biomarkers Prev. 2016;25(10):1411-1417. doi:10.1158/1055-9965.EPI-15-1281PubMedGoogle ScholarCrossref
7.
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-16-0476PubMedGoogle ScholarCrossref
8.
Hsu  T.  Johnson & Johnson told to pay $4.7 billion in baby powder lawsuit. New York Times. July 12, 2018.https://www.nytimes.com/2018/07/12/business/johnson-johnson-talcum-powder.html. Accessed December 10, 2019.
9.
McGinley  L.  Does talcum powder cause ovarian cancer? Washington Post. August 23, 2017.https://www.washingtonpost.com/news/to-your-health/wp/2017/08/23/does-talcum-powder-cause-ovarian-cancer-experts-are-divided/?utm_term=.9342d83c278f. Accessed December 10, 2019.
10.
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:10.1093/jnci/92.3.249PubMedGoogle ScholarCrossref
11.
Houghton  SC, Reeves  KW, Hankinson  SE,  et al.  Perineal powder use and risk of ovarian cancer.  J Natl Cancer Inst. 2014;106(9):dju208. doi:10.1093/jnci/dju208PubMedGoogle Scholar
12.
Gonzalez  NL, O’Brien  KM, D’Aloisio  AA, Sandler  DP, Weinberg  CR.  Douching, talc use, and risk of ovarian cancer.  Epidemiology. 2016;27(6):797-802. doi:10.1097/EDE.0000000000000528PubMedGoogle ScholarCrossref
13.
National Cancer Institute Surveillance, Epidemiology, and End Results Program.  Cancer stat facts: ovarian cancer.https://seer.cancer.gov/statfacts/html/ovary.html. Accessed December 10, 2019.
14.
Bao  Y, Bertoia  ML, Lenart  EB,  et al.  Origin, methods, and evolution of the three nurses’ health studies.  Am J Public Health. 2016;106(9):1573-1581. doi:10.2105/AJPH.2016.303338PubMedGoogle ScholarCrossref
15.
Sandler  DP, Hodgson  ME, Deming-Halverson  SL,  et al; Sister Study Research Team.  The Sister Study: baseline methods and participant characteristics.  Environ Health Perspect. 2017;125(12):127003. doi:10.1289/EHP1923PubMedGoogle Scholar
16.
Anderson  GL, Cummings  SR, Freedman  LS,  et al; The Women’s Health Initiative Study Group.  Design of the Women’s Health Initiative clinical trial and observational study.  Control Clin Trials. 1998;19(1):61-109. doi:10.1016/S0197-2456(97)00078-0PubMedGoogle ScholarCrossref
17.
Greenland  S, Pearl  J, Robins  JM.  Causal diagrams for epidemiologic research.  Epidemiology. 1999;10(1):37-48. doi:10.1097/00001648-199901000-00008PubMedGoogle ScholarCrossref
18.
DerSimonian  R, Laird  N.  Meta-analysis in clinical trials.  Control Clin Trials. 1986;7(3):177-188. doi:10.1016/0197-2456(86)90046-2PubMedGoogle ScholarCrossref
19.
Peres  LC, Cushing-Haugen  KL, Kobel  M,  et al Invasive epithelial ovarian cancer survival by histotype and disease stage.  J Natl Cancer Inst. 2019;111(1):60-68. doi:10.1093/jnci/djy071PubMedGoogle ScholarCrossref
20.
Xue  X, Kim  MY, Gaudet  MM,  et al.  A comparison of the polytomous logistic regression and joint cox proportional hazards models for evaluating multiple disease subtypes in prospective cohort studies.  Cancer Epidemiol Biomarkers Prev. 2013;22(2):275-285. doi:10.1158/1055-9965.EPI-12-1050PubMedGoogle ScholarCrossref
21.
Breslow  NE.  Discussion of professor Cox’s paper.  J R Stat Soc Ser A (Statistics Soc). 1972;34:216. doi:10.1111/j.2517-6161.1972.tb00900.xGoogle Scholar
22.
Rabin  RC, Hsu  T.  Johnson & Johnson feared baby powder’s possible asbestos link for years. New York Times. December 14, 2018.https://www.nytimes.com/2018/12/14/business/baby-powder-asbestos-johnson-johnson.html. Accessed December 10, 2019.
23.
Zazenski  R, Ashton  WH, Briggs  D,  et al.  Talc: occurrence, characterization, and consumer applications.  Regul Toxicol Pharmacol. 1995;21(2):218-229. doi:10.1006/rtph.1995.1032PubMedGoogle ScholarCrossref
24.
Erickson  BK, Conner  MG, Landen  CN  Jr.  The role of the fallopian tube in the origin of ovarian cancer.  Am J Obstet Gynecol. 2013;209(5):409-414. doi:10.1016/j.ajog.2013.04.019PubMedGoogle ScholarCrossref
25.
Ness  RB, Cottreau  C.  Possible role of ovarian epithelial inflammation in ovarian cancer.  J Natl Cancer Inst. 1999;91(17):1459-1467. doi:10.1093/jnci/91.17.1459PubMedGoogle ScholarCrossref
26.
Henderson  WJ, Joslin  CA, Turnbull  AC, Griffiths  K.  Talc and carcinoma of the ovary and cervix.  J Obstet Gynaecol Br Commonw. 1971;78(3):266-272. doi:10.1111/j.1471-0528.1971.tb00267.xPubMedGoogle ScholarCrossref
27.
Rasmussen  CB, Jensen  A, Albieri  V, Andersen  KK, Kjaer  SK.  Is pelvic inflammatory disease a risk factor for ovarian cancer?  Cancer Epidemiol Biomarkers Prev. 2017;26(1):104-109. doi:10.1158/1055-9965.EPI-16-0459PubMedGoogle ScholarCrossref
28.
Trabert  B, Waterboer  T, Idahl  A,  et al.  Antibodies against chlamydia trachomatis and ovarian cancer risk in two independent populations.  J Natl Cancer Inst. 2019;111(2):129-136. doi:10.1093/jnci/djy084PubMedGoogle ScholarCrossref
29.
Trabert  B, Poole  EM, White  E,  et al.  Analgesic use and ovarian cancer risk: an analysis in the Ovarian Cancer Cohort Consortium.   J Natl Cancer Inst.2019;111(2):137-145. doi:10.1093/jnci/djy100PubMedGoogle ScholarCrossref
30.
Hsu  T.  Risk on all sides as 4800 women sue over Johnson’s baby powder and cancer. New York Times. September 28, 2017. https://www.nytimes.com/2017/09/28/business/johnson-and-johnson-baby-talcum-powder-lawsuits.html. Published Accessed December 10, 2019.
31.
Wentzensen  N, Wacholder  S.  Talc use and ovarian cancer: epidemiology between a rock and a hard place.  J Natl Cancer Inst. 2014;106(9):9-10. doi:10.1093/jnci/dju260PubMedGoogle ScholarCrossref
Original Investigation
January 7, 2020

Association of Powder Use in the Genital Area With Risk of Ovarian Cancer

Author Affiliations
  • 1Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina
  • 2Department of Cancer Epidemiology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Florida
  • 3Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, Massachusetts
  • 4Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington
  • 5Department of Epidemiology, School of Public Health, University of Washington, Seattle
  • 6Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, Washington
  • 7Biostatistics and Computational Biology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina
  • 8Metabolic Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland
  • 9Department of Obstetrics and Gynecology, University of Florida College of Medicine-Jacksonville
  • 10Social and Scientific Systems, Inc, Durham, North Carolina
  • 11Clinical Genetics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, Maryland
JAMA. 2020;323(1):49-59. doi:10.1001/jama.2019.20079
Key Points

Question  Is use of powder in the genital area associated with the risk of developing ovarian cancer?

Findings  In this analysis that pooled data from 4 cohorts with a total of 252 745 women, the hazard ratio for the association between self-reported ever use vs never use of powder in the genital area and incident ovarian cancer was 1.08 (95% CI, 0.99-1.17).

Meaning  Among women from 4 prospective cohorts, there was not a statistically significant association between use of powder in the genital area and ovarian cancer, but the study may have been underpowered to identify a small increase in risk.

Abstract

Importance  The relationship between use of powder in the genital area and ovarian cancer is not established. Positive associations reported in case-control studies have not been confirmed in cohort studies.

Objective  To estimate the association between use of powder in the genital area and ovarian cancer using prospective observational data.

Design, Setting, and Participants  Data were pooled from 4 large, US-based cohorts: Nurses’ Health Study (enrollment 1976; follow-up 1982-2016; n = 81 869), Nurses’ Health Study II (enrollment 1989; follow-up 2013-2017; n = 61 261), Sister Study (enrollment 2003-2009; follow-up 2003-2017; n = 40 647), and Women’s Health Initiative Observational Study (enrollment 1993-1998; follow-up 1993-2017; n = 73 267).

Exposures  Ever, long-term (≥20 years), and frequent (≥1/week) use of powder in the genital area.

Main Outcomes and Measures  The primary analysis examined the association between ever use of powder in the genital area and self-reported incident ovarian cancer. Covariate-adjusted hazard ratios (HRs) and 95% CIs were estimated using Cox proportional hazards models.

Results  The pooled sample included 252 745 women (median age at baseline, 57 years) with 38% self-reporting use of powder in the genital area. Ten percent reported long-term use, and 22% reported frequent use. During a median of 11.2 years of follow-up (3.8 million person-years at risk), 2168 women developed ovarian cancer (58 cases/100 000 person-years). Ovarian cancer incidence was 61 cases/100 000 person-years among ever users and 55 cases/100 000 person-years among never users (estimated risk difference at age 70 years, 0.09% [95% CI, −0.02% to 0.19%]; estimated HR, 1.08 [95% CI, 0.99 to 1.17]). The estimated HR for frequent vs never use was 1.09 (95% CI, 0.97 to 1.23) and for long-term vs never use, the HR was 1.01 (95% CI, 0.82 to 1.25). Subgroup analyses were conducted for 10 variables; the tests for heterogeneity were not statistically significant for any of these comparisons. While the estimated HR for the association between ever use of powder in the genital area and ovarian cancer risk among women with a patent reproductive tract was 1.13 (95% CI, 1.01 to 1.26), the P value for interaction comparing women with vs without patent reproductive tracts was .15.

Conclusions and Relevance  In this analysis of pooled data from women in 4 US cohorts, there was not a statistically significant association between use of powder in the genital area and incident ovarian cancer. However, the study may have been underpowered to identify a small increase in risk.

Introduction

Quiz Ref IDSome women apply powder to their genitals, either through direct application or on underwear, sanitary napkins, diaphragms or tampons. Most powder products include some mineral talc.1 Talc was first investigated as a carcinogen based on its relationship to asbestos, which has known carcinogenic effects2 and may be mined in the same locations. However, all US-based manufacturers of cosmetic talc agreed to ban asbestos in 1976,3 and the International Agency for Research on Cancer has since concluded there is only “possible” evidence that perineal use of talc-based body powder may be carcinogenic.1

Quiz Ref IDThis classification was largely based on evidence from observational studies. Case-control studies have reported positive associations between ever use of powder in the genital area and ovarian cancer, with an estimated odds ratio of 1.24 in a pooled analysis4 and 1.31 in a meta-analysis.5 However, these findings may be affected by recall bias,6,7 and a recent surge in talc-related lawsuits and media coverage8,9 has increased this possibility. Thus, it is crucial to evaluate the talc–ovarian cancer association using prospective data.

To date, 3 large cohort studies have assessed the association between use of powder in the genital area and ovarian cancer risk, with inconsistent results.10-12 However, ovarian cancer is a rare disease (1.3% lifetime risk in the United States),13 and individual cohort studies are not sufficiently powered to detect modest associations, particularly if restricted to susceptible subgroups, such as women with patent reproductive tracts (ie, having an intact uterus and no tubal ligation).

To better examine the association between use of powder in the genital area and risk of ovarian cancer, 4 large US cohorts that collected the necessary information were identified: the Nurses’ Health Study (NHS), Nurses’ Health Study II (NHSII), Sister Study (SIS), and Women’s Health Initiative Observational Study (WHI-OS). While associations between genital use of powder and ovarian cancer risk have been reported for 3 of these (NHS, WHI-OS, and SIS),10-12 the pooled results reported here incorporate updated data, including additional cases and longer follow-up.

Methods
Study Sample

The study designs of these 4 US-based cohorts have been described in detail elsewhere.14-16 Briefly, the NHS (n = 121 700) enrolled registered nurses living in the United States in 1976, and the NHSII (n = 116 429) did the same in 1989. The study protocols were approved by the institutional review boards of the Brigham and Women’s Hospital, the Harvard T.H. Chan School of Public Health, and those of participating registries, as required. All participants provided written, informed consent. Although the initial questionnaires did not ask about genital use of powder, participants were queried about powder use on the 1982 NHS and 2013 NHSII questionnaires. We only included follow-up time after the questionnaire about use of powder in the genital area was administered and will refer to the questionnaire that assessed powder use as baseline to maintain consistent language across all 4 studies.

Quiz Ref IDGenital use of powder was assessed at enrollment for SIS between 2003 and 2009 (n = 50 884) and for WHI-OS between 1993 and 1998 (n = 93 676). Women were eligible for SIS if they had a sister previously diagnosed with breast cancer but had no personal diagnosis of breast cancer at enrollment. Eligible participants in WHI-OS were postmenopausal women who resided near one of 40 clinical centers. Both studies were approved by the relevant institutional review boards and all participants provided written, informed consent.

Exposure Assessment

The cohorts differed in how they asked participants about use of powder in the genital area (eAppendix in the Supplement). NHS participants were asked whether they “ever commonly used talcum, baby powder or deodorizing powder” on their “perineal (private) area” (no, <1/week, 1-6 times/week, daily) or on sanitary napkins (yes/no). The NHSII questionnaire asked women to report use only if it occurred at least weekly in the “genital/rectal area or on sanitary napkins, tampons, or underwear” and if so, for how long (<1 year, 1-<10 years, 10-<20 years, 20-<30 years, 30+ years). In SIS, the question specifically focused on use of talcum powder and application to “a sanitary napkin, underwear, diaphragm, or cervical cap, or directly to the vaginal area” in the last year or at the ages of 10 to 13 years. Participants were queried about their frequency of use in the year prior to enrollment (never, <1/mo, 1-3 times/mo, 1-5 times/week, >5 times/week), as well as use during the ages 10-13 (did not use, sometimes, frequently). Women in WHI-OS were asked if they had ever used powder on their “private parts (genital areas)” (yes/no) and for how long they had used it (<1 year, 1-4 years, 5-9 years, 10-19 years, 20+ years), with similar questions for powder use on diaphragms or sanitary pads.

To harmonize across the 4 studies, we defined women as ever vs never users of powder on genital areas. For SIS, ever use included use in the last year or at ages 10 to 13 years. We were also able to examine long-term use, which we defined as use of powder on genitals for at least 20 years (NHSII and WHI-OS) or use at ages 10 to 13 years and also in the last year (SIS). Frequent users were those who reported use of powder in the genital area at least once per week (NHS, NHSII), at least once per week in the last year, or “frequently” during ages of 10 to 13 years (SIS).

Outcome Assessment

For NHS and NHSII, follow-up questionnaires were distributed every 2 years, at which point participants were asked to report recent cancer diagnoses. Those reporting incident cancers were asked to grant access to their medical records, which were reviewed for confirmation of the diagnosis and disease details. Additional cases were identified from among deceased participants via National Death Index searches. The protocol for SIS was similar, except follow-up questionnaires were collected annually and most participants provided pathology reports rather than complete medical records. Participants in WHI-OS were also asked to self-report cancers on annual questionnaires, but only medically confirmed cases were counted. All 4 studies categorized tumors originating in the ovary, peritoneum, and fallopian tubes as ovarian cancers.

For NHS, NHSII, and SIS, delays in the confirmation process and incomplete retrieval of medical records meant that not all self-reported cases could be medically confirmed. We ran sensitivity analyses limited to medically confirmed cases but included all self-reported diagnoses in our main analyses. Subtype analyses were limited to medically confirmed cases.

Covariates

All 4 studies had substantial covariate data, which we harmonized into a common set of potential confounders or effect modifiers. The following data were included: age at baseline (continuous), race (white, black, other), education (≤high school, some college, completed college), body mass index (BMI [calculated as weight in kilograms divided by height in meters squared], restricted cubic spline), parity (nulliparous, 1 birth, 2 births, ≥3 births), smoking status (never, former, current), oral contraceptive use (ever/never), hormone therapy use (ever/never), tubal ligation status (yes/no), hysterectomy status (yes/no), and menopausal status (premenopausal/postmenopausal). Race was self-reported by the participant, based on provided categories. It was considered to be an important confounder because both ovarian cancer rates13 and genital powder use vary by race/ethnicity. Only baseline levels of these covariates were considered as confounders, though we did consider postbaseline changes in menopausal status when assessing effect modification.

Statistical Analyses

We used Cox proportional hazards models with age as the primary time scale to estimate hazard ratios (HRs) and 95% CIs measuring the association between genital use of powder and incident ovarian cancer, adjusting for potential confounders. We selected potential confounders using a directed acyclic graph framework,17 considering covariates that were possibly related to use of powder in the genital area and also ovarian cancer risk.

We excluded women who had ovarian cancer or a bilateral oophorectomy prior to baseline, or who were missing information on powder use or age at ovarian cancer diagnosis. For regression analyses, we additionally excluded women with missing data for 1 or more covariates. Women underwent follow-up from age at baseline until ovarian cancer diagnosis, with censoring at bilateral oophorectomy, end of follow-up, or death from causes other than ovarian cancer. An exception was made for WHI-OS because postbaseline oophorectomy data were not collected. Participants in SIS and WHI-OS who were no longer actively responding to follow-up requests were censored at age of last contact, although their follow-up continued via linkage to the National Death Index.

To better control for differences across studies, we allowed the baseline hazard function to vary across cohorts by implementing study-stratified Cox models. We tested for study heterogeneity by conducting likelihood ratio tests comparing models with and without study × powder interaction terms. For the primary analysis of ever vs never powder use and ovarian cancer risk, we additionally calculated the effect estimate and the P value for heterogeneity from a random-effects meta-analysis.18 Proportional hazards assumptions were tested via likelihood ratio tests of powder × time interaction terms.

Because patency is required for there to be a direct physical pathway between the powder application area and the ovaries, we hypothesized a priori that women with patent reproductive tracts would be more susceptible to the effects of powder use in the genital area on ovarian cancer. We therefore conducted analyses restricted to this subgroup. When estimating the effects of duration of powder use on ovarian cancer risk, we compared long-term (≥20 years) and non–long-term users with never users. Similarly, we compared frequent users (≥1/week) and nonfrequent users with never users. We conducted trend tests using the ordinal forms of these variables.

We also conducted exploratory analyses to examine whether the association between powder use in the genital area and ovarian cancer varied by subgroup. These categorizations were selected based on the existing literature or hypotheses about potential biological mechanisms and included age, race/ethnicity, menopausal hormone therapy use, BMI, and parity. We also considered time-varying menopausal status and follow-up time as effect modifiers and more formally compared subgroups defined by hysterectomy, tubal ligation and patency status. We evaluated heterogeneity across strata of each potential effect modifier by conducting likelihood ratio tests of the interaction between that factor and powder use in the genital area.

For analyses limited to medically confirmed cases of ovarian cancer, we censored unconfirmed cases at their self-reported age of diagnosis. For type-specific analyses, the medically confirmed cases were further divided by invasiveness status (invasive vs borderline), tumor location (epithelial ovarian, peritoneal, or fallopian tube), or histotype (serous, endometroid, mucinous, clear-cell, or other). For an alternative histotype analysis, we defined high-grade serous as grades 2 to 4 serous or grades 3 to 4 endometroid tumors.19 We estimated the HRs for each set of subtypes using joint Cox proportional hazards models,20 utilizing likelihood ratio tests to compare model fit for models that did and did not allow the main-effect estimates to differ by subtype. These test results are reported as P values for heterogeneity.

In a sensitivity analysis, we attempted to isolate participants who were possibly exposed to asbestos-contaminated talc by limiting analysis to women in WHI-OS and NHS, most of whom were born before 1945. In the age-adjusted and fully adjusted models, we additionally estimated cumulative risk of ovarian cancer by age 70 years and assessed differences in absolute risk among ever vs never users of powder in the genital area using the Breslow method.21

Statistical tests were 2-sided, and a P value less than .05 was considered statistically significant. Because of the potential for type I error due to multiple comparisons, findings from subgroup and sensitivity analyses should be interpreted as exploratory. All analyses were conducted in SAS 9.4.

Results

After initial exclusions, we had data from 257 044 women, including 2213 who developed incident ovarian cancer (Table 1). Use of powder in the genital area was common overall (39%) but varied by cohort with 53% of participants reporting ever use in WHI-OS, 41% in NHS, 27% in SIS, and 26% in NHSII. Long-term use was reported by 16% in WHI-OS and by 6% in both SIS and NHSII; frequent use was reported by 27% in NHS, 26% in NHSII, and 7% in SIS.

After further excluding women with missing covariates (<3% of all participants), 2168 participants with ovarian cancer (1884 medically confirmed) and 250 577 without ovarian cancer remained. Most NHS and WHI-OS participants were born between 1915 and 1944 and most NHSII and SIS participants were born in 1945 or later (eTable 1 in the Supplement), and there appeared to be a generational trend in use of powder in the genital area, with older cohorts more likely to report use. Overall, this was a highly educated group (most completed college) and most participants were white (84%-98% of each cohort). Compared with never users, ever users of powder in the genital area were more likely to be black (6% vs 3%; eTable 2 in the Supplement), to be obese (26% vs 19%), or to have had a hysterectomy (22% vs 18%), and less likely to have used oral contraceptives (57% vs 64%).

A total of 2168 women developed ovarian cancer (58 cases per 100 000 person-years; Table 2). Consistent with mean age at enrollment, incidence was highest in WHI-OS (63 cases per 100 000 person-years) and lowest in NHSII (34 cases per 100 000 person-years). In the pooled sample, estimated crude cumulative incidence of ovarian cancer at age 70 years was 1.3%, with higher risk among participants in NHS (1.3%) and SIS (1.4%) than in NHSII (0.7%) or WHI-OS (0.9%).

Considering all 4 cohorts, the estimated incidence of ovarian cancer was 61 per 100 000 person-years among ever users and 55 among never users. The estimated adjusted cumulative risk of ovarian cancer by age 70 years among unexposed participants was 1.16%, with an estimated covariate-adjusted risk difference of 0.09% (95% CI, −0.02% to 0.19%) comparing with those who were exposed. The HR for the association between ever powder use and incident ovarian cancer was 1.08 (95% CI, 0.99 to 1.17; Table 2). There was no evidence of heterogeneity across cohorts (P value for heterogeneity = .81) and no evidence of a proportional hazards assumption violation (P > .99). The estimated HR from the random-effects model was 1.07 (95% CI, 0.98 to 1.17; P value for heterogeneity = .71).

When restricted to women with patent reproductive tracts at baseline, the HR was 1.13 (95% CI, 1.01 to 1.26) and the estimated covariate-adjusted risk difference was 0.15% (95% CI, 0.01% to 0.29%). Among women without patent reproductive tracts, the estimated HR was 0.99 (95% CI, 0.86 to 1.15) and the P value for heterogeneity comparing the result for women with patency vs without was .15 (Figure). The remaining stratified analyses are also presented in the Figure and in eTable 3 in the Supplement.

The covariate-adjusted risk difference for long-term (≥20 years) vs never use was 0.01% (95% CI, −0.21% to 0.24%), and the HR was 1.01 (95% CI, 0.82 to 1.25; P value for trend = .57; Table 3). The covariate-adjusted risk difference for frequent use (≥1/week) vs none was 0.10% (95% CI, −0.05% to 0.25%), and the HR was 1.09 (95% CI, 0.97 to 1.23; dose-response P value for trend =.20). The covariate-adjusted risk difference for the association between frequent powder use and ovarian cancer among women with patent reproductive tracts was 0.22% (95% CI, 0.02% to 0.42%), and the HR was 1.19 (95% CI, 1.03 to 1.37; P value for trend = .03).

When the outcome was limited to medically confirmed cases, the HR was attenuated (Table 4; HR, 1.05 [95% CI, 0.96 to 1.16] for ever use vs never use). There were no notable differences in estimates by invasive status, tumor location, or histotype. This was also true for analyses limited to women with patent reproductive tracts (eTable 4 in the Supplement). When limited to the older cohorts (NHS and WHI-OS), the estimated pooled HR was 1.09 (95% CI, 0.99 to 1.19) for ever use vs never use. The estimated HR from the young cohorts (NHSII and SIS) was 0.97 (95% CI, 0.75 to 1.26).

Discussion

Quiz Ref IDIn this pooled analysis of 4 large US cohorts, there was no statistically significant association between self-reported use of powder in the genital area and risk of ovarian cancer. There were no clear dose-response trends for duration and frequency of powder use in the genital area in relation to ovarian cancer risk. Although the study was underpowered to detect small changes in risk, this is, to our knowledge, the largest study of this topic to date, and it is believed that no other large prospective cohorts have collected data on powder exposure in the genital area.

One of the primary drivers of research on genital use of talc-based products and ovarian cancer has been the potential link between talc and asbestos, which can occur together in nature. In an analysis limited to the older cohorts in which women may have started using powder before the asbestos ban of 1976, the estimated effect remained consistent, with no association observed in the younger cohorts. However, it was recently suggested that some products may have contained asbestos after 1976, meaning that there may not be a clearly defined time period in which talc-based products did or did not contain asbestos.22 Further, although most cosmetic powder products include some quantity of mineral talc,1 the percentage varies widely,23 and exposure to asbestos (through talc) would also depend on the type of product used and the method of application (eg, underwear vs diaphragm).

By irritating epithelial ovarian tissue or fallopian tubes24 directly, powder could induce an inflammatory response even in the absence of asbestos. This could set off a cascade of increased oxidative stress levels, DNA damage, and cell division, all of which could contribute to carcinogenesis.25 In this analysis, there was a possible positive association among women with patent reproductive tracts (no history of hysterectomy or tubal ligation), although because the association was not significantly different from that observed in women with nonpatent reproductive tracts, this finding should be considered only exploratory and hypothesis generating. This observation lends support to the hypothesis that powder with or without asbestos could irritate and inflame the reproductive tract, as patency is required for there to be a direct physical path between the genitals and the fallopian tubes or ovaries.26Quiz Ref ID The positive relationships between pelvic inflammatory disease and ovarian cancer27 and chlamydia infection and ovarian cancer28 also support an inflammation-mediated mechanism, as does the inverse association between regular aspirin use and ovarian cancer.29

One of the main concerns about previous case-control studies on this topic is the possibility for recall bias, which would result if case participants were more likely to report using powder than control participants. As highlighted by Trabert,7 the African American Cancer Epidemiology Study6 found evidence supporting this phenomenon. Based on the timing of the first major talc lawsuits,30 Schildkraut et al6 stratified their results by year of interview (earlier than 2014 vs 2014 or later), observing that among women interviewed earlier, ever use of powder in the genital area was less-strongly associated with ovarian cancer (odds ratio [OR], 1.19 [95% CI, 0.87 to 1.63]) than among women interviewed later (OR, 2.91 [95% CI, 1.70 to 4.97]). This difference was driven by an increase in the reported prevalence of powder use among case participants (36.5% vs 51.5% of women interviewed early vs later), while self-reported use in the control partcipants remained stable (34.0% vs 34.4%). However, most of the case-control studies that have examined this association recruited well before 2014, and a large pooled analysis published in 2013 reported an OR of 1.24 (95% CI, 1.15 to 1.33).4 For the current analysis, recall bias was avoided by excluding those with preexisting ovarian cancer.

The strengths of this study were large sample size and long follow-up time. The main analysis included 2168 ovarian cancer cases that developed over 3.8 million person-years. This far exceeds a previous meta-analysis of the published NHS, SIS, and WHI-OS results (890 cases over 182 000 person-years).5 However, power to investigate links to peritoneal or fallopian tube cancers or histotypes other than serous was still low. Improvements in the classification of tumor types may contribute new insights, especially for fallopian tube cancers, which may be the true point of origin for most serous ovarian cancers.24 This and other subtype-specific associations should be better examined in the future.

Limitations

This study has several limitations. First, the included cohorts varied widely in how they assessed exposure, particularly the duration and frequency of powder use. There was no evidence of between-study heterogeneity for either the pooled or meta-analysis models of ever use vs never use, but because the 2 largest studies were missing information on duration (NHS) and frequency (WHI-OS) of powder use, the dose-response analyses are underpowered compared with the main results and thus difficult to interpret. Second, use of powder in the genital area could not be assessed as a time-varying factor, as none of the 4 studies collected data on use after baseline.

Third, specific exposure windows could not be examined, nor could type of powder used or patency status at time of powder use. As previously noted, information on powder exposure is typically more limited in cohort studies compared with case-control studies, particularly with respect to dose and duration of use.31 Therefore, ongoing or future cohort studies should collect detailed information on these topics.

Fourth, as with all observational studies, residual confounding is possible. All 4 included studies recorded detailed information on many potential confounders, which were harmonized across cohorts and adjusted for in multivariable models. However, residual confounding may still be present if the harmonized covariates did not adequately capture the relationship or if any key confounders were missing.

Fifth, the study may have limited generalizability. All 4 cohorts included predominately white, well-educated women, approximately half of whom had a BMI of less than 25, which could raise concerns about generalizability, especially since these factors may be related to powder use. However, these studies have high retention rates and accurate self-reported data, increasing internal validity.

Sixth, confounding by indication is another potential limitation, and it would occur if women with other underlying conditions that were associated with ovarian cancer were also more likely to use powder in the genital area. It is also possible that if powder use is associated with increased risk of other gynecologic conditions (eg, fibroids, ovarian cysts), it can affect whether women receive oophorectomies, hysterectomies, or tubal ligations and alter their risk of developing ovarian cancer. Seventh, because tests to confirm patency were not performed, it is possible that not all women categorized as having a patent reproductive tract in this analysis had truly patent tubes.

Conclusions

In this analysis of pooled data from women in 4 US cohorts, there was not a statistically significant association between self-reported use of powder in the genital area and incident ovarian cancer. However, the study may have been underpowered to identify a small increase in risk.

Back to top
Article Information

Corresponding Author: Katie M. O’Brien, Epidemiology Branch, National Institute of Environmental Health Sciences, Research Triangle Park, NC, 27709 (obrienkm2@niehs.nih.gov).

Accepted for Publication: November 16, 2019.

Author Contributions: Dr O’Brien had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Joint senior authors: Drs Sandler and Wentzensen.

Concept and design: O'Brien, Tworoger, Sandler, Wentzensen.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: O'Brien, Weinberg, Trabert, Wentzensen.

Critical revision of the manuscript for important intellectual content: O'Brien, Tworoger, Harris, Anderson, Trabert, Kaunitz, D'Aloisio, Sandler, Wentzensen.

Statistical analysis: O'Brien, Tworoger, Harris, Trabert.

Obtained funding: Sandler.

Administrative, technical, or material support: O'Brien, Tworoger, Harris, Kaunitz, D'Aloisio.

Supervision: Sandler, Wentzensen.

Conflict of Interest Disclosures: Dr Tworoger reported receipt of grants from the US Department of Defense Ovarian Cancer Research Program (OCRP) and the National Institutes of Health (NIH) both during the conduct of the study and outside the submitted work. Dr Anderson reported receipt of grants from the National Heart Lung Blood Institute (NHLBI) during the conduct of the study. Dr Kaunitz reported provision of consultancy services to the University of Florida, which receives research funding from companies involved with products related to contraception and treatment of menopausal symptoms; personal fees for consultancy services from Pfizer (injectable contraception), AMAG (treatment of genital atrophy), Mithra (contraceptive and menopausal hormone products), and Merck (implantable and vaginal ring contraception), but no companies involved with sales of powder; royalties from UpToDate; and funding for clinical trials through the University of Florida from Medicines 360 (intrauterine devices), Allergan (treatment of uterine fibroids), Myovant (treatment of uterine fibroids), and Endoceutics (treatment of genital atrophy). No other disclosures were reported.

Funding/Support: This work was supported by the Intramural Research Program of NIH, National Institute of Environmental Health Sciences (Z01-ES044005 to Dr Sandler); the Intramural Research Program of the National Cancer Institute; US Department of Defense OCRP (W81XWH-12-1-0561); NIH (UM1 CA186107, P01 CA87969, UM1 CA176726, and R01 CA67262 [Nurses’ Health Study {NHS} and Nurses’ Health Study II {NHSII}]); NHLBI, NIH/US Department of Health and Human Services (HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C [Women’s Health Initiative Observational Study {WHI-OS}]); and the National Institute of Environmental Health Sciences (Sister Study [SIS]). Dr Harris is supported by an NIH grant (K22 CA193860). Statistical analysis services from Westat (Durham, North Carolina) were provided through a support contract from the National Institute of Environmental Health Sciences with Social and Scientific Systems (Durham, North Carolina).

Role of the Funder/Sponsor: None of the sponsors had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Information: Statistical analyses were replicated by Westat, an independent contractor (Durham, North Carolina).

Additional Contributions: We would like to thank the participants and staff of the participating cohorts for their valuable contributions as well as the following state cancer registries for their help in the NHS/NHSII: Alabama, Arizona, Arkansas, California, Colorado, Connecticut, Deleware, Florida, Georgia, Idaho, Illinois, Indiana, Iowa, Kentuckyy, Louisiana, Maine, Maryland, Massachusetts, Michigan, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, Oklahoma, Oregon, Pennsylvania, Rhode Island, South Carolina, Tennessee, Texas, Virginia, Washington, and Wyoming. The authors acknowledge the Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital, as the home of the NHS and the following investigators from the WHI, from the NHLBI, Bethesda, Maryland, Jacques Rossouw, MD, Shari Ludlam, MPH, Joan McGowan, PhD, Leslie Ford, MD, and Nancy Geller, PhD; Fred Hutchinson Cancer Research Center, Seattle, Washington, Garnet Anderson, PhD, Ross Prentice, PhD, Andrea LaCroix, PhD, and Charles Kooperberg, PhD; Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts, JoAnn E. Manson, MD, DrPH (); MedStar Health Research Institute/Howard University, Washington, DC, Barbara V. Howard, PhD; Stanford Prevention Research Center, Stanford, California, Marcia L. Stefanick, PhD; The Ohio State University, Columbus, Rebecca Jackson, MD; University of Arizona, Tucson/Phoenix, Cynthia A. Thomson, PhD, RD; University at Buffalo, Buffalo, NY, Jean Wactawski-Wende, PhD; University of Florida, Gainesville/Jacksonville, Marian Limacher, MD; University of Iowa, Iowa City/Davenport, Jennifer Robinson, MD MPH; University of Pittsburgh, Pittsburgh, Pennsylvania, Lewis Kuller, MD, DrPH; Wake Forest University School of Medicine, Winston-Salem, North Carolina, Sally Shumaker, PhD, and Mark Espeland, PhD; and University of Nevada, Reno, Robert Brunner, PhD. None of aforementioned individuals received compensation for their role in this article.

References
1.
World Health Organization International Agency for Research on Cancer.  IARC Monographs on the Evaluation of Carcinogenic Risks to Humans: Carbon Black, Titanium Dioxide, and Talc; vol 93; 2010.http://publications.iarc.fr/111 Accessed December 10, 2019.
2.
International Agency for Research on Cancer.  IARC Monographs: Arsenic, Metals, Fibres, and Dusts; vol 100C; 2012.http://publications.iarc.fr/120 Accessed December 10, 2019.
3.
Fiume  MM, Boyer  I, Bergfeld  WF,  et al.  Safety assessment of talc as used in cosmetics.  Int J Toxicol. 2015;34(1)(suppl):66S-129S. doi:10.1177/1091581815586797PubMedGoogle ScholarCrossref
4.
Terry  KL, Karageorgi  S, Shvetsov  YB,  et al; Australian Cancer Study (Ovarian Cancer); Australian Ovarian Cancer Study Group; Ovarian Cancer Association Consortium.  Genital powder use and risk of ovarian cancer: a pooled analysis of 8525 cases and 9859 controls.  Cancer Prev Res (Phila). 2013;6(8):811-821. doi:10.1158/1940-6207.CAPR-13-0037PubMedGoogle ScholarCrossref
5.
Penninkilampi  R, Eslick  GD.  Perineal talc use and ovarian cancer: a systematic review and meta-analysis.  Epidemiology. 2018;29(1):41-49. doi:10.1097/EDE.0000000000000745PubMedGoogle ScholarCrossref
6.
Schildkraut  JM, Abbott  SE, Alberg  AJ,  et al.  Association between body powder use and ovarian cancer: the African American Cancer Epidemiology Study (AACES).  Cancer Epidemiol Biomarkers Prev. 2016;25(10):1411-1417. doi:10.1158/1055-9965.EPI-15-1281PubMedGoogle ScholarCrossref
7.
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-16-0476PubMedGoogle ScholarCrossref
8.
Hsu  T.  Johnson & Johnson told to pay $4.7 billion in baby powder lawsuit. New York Times. July 12, 2018.https://www.nytimes.com/2018/07/12/business/johnson-johnson-talcum-powder.html. Accessed December 10, 2019.
9.
McGinley  L.  Does talcum powder cause ovarian cancer? Washington Post. August 23, 2017.https://www.washingtonpost.com/news/to-your-health/wp/2017/08/23/does-talcum-powder-cause-ovarian-cancer-experts-are-divided/?utm_term=.9342d83c278f. Accessed December 10, 2019.
10.
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:10.1093/jnci/92.3.249PubMedGoogle ScholarCrossref
11.
Houghton  SC, Reeves  KW, Hankinson  SE,  et al.  Perineal powder use and risk of ovarian cancer.  J Natl Cancer Inst. 2014;106(9):dju208. doi:10.1093/jnci/dju208PubMedGoogle Scholar
12.
Gonzalez  NL, O’Brien  KM, D’Aloisio  AA, Sandler  DP, Weinberg  CR.  Douching, talc use, and risk of ovarian cancer.  Epidemiology. 2016;27(6):797-802. doi:10.1097/EDE.0000000000000528PubMedGoogle ScholarCrossref
13.
National Cancer Institute Surveillance, Epidemiology, and End Results Program.  Cancer stat facts: ovarian cancer.https://seer.cancer.gov/statfacts/html/ovary.html. Accessed December 10, 2019.
14.
Bao  Y, Bertoia  ML, Lenart  EB,  et al.  Origin, methods, and evolution of the three nurses’ health studies.  Am J Public Health. 2016;106(9):1573-1581. doi:10.2105/AJPH.2016.303338PubMedGoogle ScholarCrossref
15.
Sandler  DP, Hodgson  ME, Deming-Halverson  SL,  et al; Sister Study Research Team.  The Sister Study: baseline methods and participant characteristics.  Environ Health Perspect. 2017;125(12):127003. doi:10.1289/EHP1923PubMedGoogle Scholar
16.
Anderson  GL, Cummings  SR, Freedman  LS,  et al; The Women’s Health Initiative Study Group.  Design of the Women’s Health Initiative clinical trial and observational study.  Control Clin Trials. 1998;19(1):61-109. doi:10.1016/S0197-2456(97)00078-0PubMedGoogle ScholarCrossref
17.
Greenland  S, Pearl  J, Robins  JM.  Causal diagrams for epidemiologic research.  Epidemiology. 1999;10(1):37-48. doi:10.1097/00001648-199901000-00008PubMedGoogle ScholarCrossref
18.
DerSimonian  R, Laird  N.  Meta-analysis in clinical trials.  Control Clin Trials. 1986;7(3):177-188. doi:10.1016/0197-2456(86)90046-2PubMedGoogle ScholarCrossref
19.
Peres  LC, Cushing-Haugen  KL, Kobel  M,  et al Invasive epithelial ovarian cancer survival by histotype and disease stage.  J Natl Cancer Inst. 2019;111(1):60-68. doi:10.1093/jnci/djy071PubMedGoogle ScholarCrossref
20.
Xue  X, Kim  MY, Gaudet  MM,  et al.  A comparison of the polytomous logistic regression and joint cox proportional hazards models for evaluating multiple disease subtypes in prospective cohort studies.  Cancer Epidemiol Biomarkers Prev. 2013;22(2):275-285. doi:10.1158/1055-9965.EPI-12-1050PubMedGoogle ScholarCrossref
21.
Breslow  NE.  Discussion of professor Cox’s paper.  J R Stat Soc Ser A (Statistics Soc). 1972;34:216. doi:10.1111/j.2517-6161.1972.tb00900.xGoogle Scholar
22.
Rabin  RC, Hsu  T.  Johnson & Johnson feared baby powder’s possible asbestos link for years. New York Times. December 14, 2018.https://www.nytimes.com/2018/12/14/business/baby-powder-asbestos-johnson-johnson.html. Accessed December 10, 2019.
23.
Zazenski  R, Ashton  WH, Briggs  D,  et al.  Talc: occurrence, characterization, and consumer applications.  Regul Toxicol Pharmacol. 1995;21(2):218-229. doi:10.1006/rtph.1995.1032PubMedGoogle ScholarCrossref
24.
Erickson  BK, Conner  MG, Landen  CN  Jr.  The role of the fallopian tube in the origin of ovarian cancer.  Am J Obstet Gynecol. 2013;209(5):409-414. doi:10.1016/j.ajog.2013.04.019PubMedGoogle ScholarCrossref
25.
Ness  RB, Cottreau  C.  Possible role of ovarian epithelial inflammation in ovarian cancer.  J Natl Cancer Inst. 1999;91(17):1459-1467. doi:10.1093/jnci/91.17.1459PubMedGoogle ScholarCrossref
26.
Henderson  WJ, Joslin  CA, Turnbull  AC, Griffiths  K.  Talc and carcinoma of the ovary and cervix.  J Obstet Gynaecol Br Commonw. 1971;78(3):266-272. doi:10.1111/j.1471-0528.1971.tb00267.xPubMedGoogle ScholarCrossref
27.
Rasmussen  CB, Jensen  A, Albieri  V, Andersen  KK, Kjaer  SK.  Is pelvic inflammatory disease a risk factor for ovarian cancer?  Cancer Epidemiol Biomarkers Prev. 2017;26(1):104-109. doi:10.1158/1055-9965.EPI-16-0459PubMedGoogle ScholarCrossref
28.
Trabert  B, Waterboer  T, Idahl  A,  et al.  Antibodies against chlamydia trachomatis and ovarian cancer risk in two independent populations.  J Natl Cancer Inst. 2019;111(2):129-136. doi:10.1093/jnci/djy084PubMedGoogle ScholarCrossref
29.
Trabert  B, Poole  EM, White  E,  et al.  Analgesic use and ovarian cancer risk: an analysis in the Ovarian Cancer Cohort Consortium.   J Natl Cancer Inst.2019;111(2):137-145. doi:10.1093/jnci/djy100PubMedGoogle ScholarCrossref
30.
Hsu  T.  Risk on all sides as 4800 women sue over Johnson’s baby powder and cancer. New York Times. September 28, 2017. https://www.nytimes.com/2017/09/28/business/johnson-and-johnson-baby-talcum-powder-lawsuits.html. Published Accessed December 10, 2019.
31.
Wentzensen  N, Wacholder  S.  Talc use and ovarian cancer: epidemiology between a rock and a hard place.  J Natl Cancer Inst. 2014;106(9):9-10. doi:10.1093/jnci/dju260PubMedGoogle ScholarCrossref
×