Occupational exposures among hairdressers and the occurrence of hormone-related conditions

Abstract

Objective

To investigate the association between hairdresser exposures and hormone-related conditions.

Methods

Using data from 50 800 eligible Sister Study participants (enrolled 2003–2009, aged 35–74 years), we estimated ORs and 95% CIs for associations between ever working as a hairdresser (n=1803) and prevalent fibroids, endometriosis, hysterectomy and oophorectomy. We estimated HRs and 95% CI for incident fibroids, endometriosis, breast, uterine and ovarian cancers among ever hairdressers versus never hairdressers. We also examined associations of hormone-related diseases and professional use of products such as bleach, perms, chemical straighteners, permanent hair colour, hairspray, barbicide, formaldehyde and alcohol, comparing data from 985 long-term hairdressers who worked ≥2 years to non-long-term hairdressers (never workers and those working <2 years).

Results

Ever-hairdressers were more likely than never-hairdressers to have had a prebaseline hysterectomy (OR=1.23: 95% CI 1.11 to 1.36). Hysterectomies were more common among long-term hairdressers with more frequent applications of perms, chemical straighteners and permanent hair colour compared with less frequent applicators or never hairdressers. Ever-hairdressers had higher rates of incident endometriosis (477 cases, HR=1.61: 95% CI 1.08 to 2.38) compared with never-hairdressers, but there were no notable associations between working as a hairdresser and fibroids (1805 cases, HR=1.04: 95% CI 0.80 to 1.34), breast cancer (4628 cases, HR=0.98: 95% CI 0.83 to 1.16), ovarian cancer (300 cases, HR=1.33: 95% CI 0.77 to 2.29) or uterine cancer (447 cases, HR=1.04: 95% CI 0.60 to 1.77). In race-stratified analyses, Black hairdressers were more likely to be diagnosed with fibroids than Black never-hairdressers (201 cases, HR=1.56: 95% CI 0.93 to 2.62).

Conclusions

Hairdresser occupation was associated with increased odds of hysterectomy and increased rates of incident endometriosis and possibly fibroids among Black women.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Hairdressers are routinely exposed to a range of endocrine disrupting chemicals and potential carcinogens through occupational use of perms, chemical straighteners and dyes. But it is unclear whether these exposures affect their risk of hormone-related diseases.

WHAT THIS STUDY ADDS

• Findings from this study show that hairdressers have increased odds of hysterectomy and higher rates of incident endometriosis compared with non-hairdressers, and that Black hairdressers have an elevated risk of developing fibroids compared with Black non-hairdressers.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study may encourage biomonitoring and toxicological assessments of chemical exposure in hairdressing workplaces to identify opportunities to mitigate risk. This study underscores the importance of training programmes for hairdressers to increase awareness of chemical exposure and safety practices in salons.

Introduction

The Bureau of Labor Statistics projected an 8% increase in hairdresser employment rates from 2022 to 2032, outpacing average job growth rates of 2.8%.¹ With its increasing popularity, worker safety is crucial. Hairdressers are exposed to chemicals with potential long-term health impacts, many not yet fully investigated.2,4

Hairdressers use hair straighteners and dyes, hairsprays, bleaches and disinfecting products such as alcohol, barbicide and formaldehyde. Many of these contain compounds identified as irritants, endocrine disruptors or carcinogens.^{4 5} For example, barbicide contains isopropyl alcohol and dimethyl benzyl ammonium chloride,⁶ and some hair smoothing/straightening products and disinfecting agents contain formaldehyde,^{7 8} a known carcinogen.⁹ Aromatic amines, such as benzidine, o-toluidine and o-dianisidine, historically present in some hair dyes, have been linked to bladder cancer.9,11 The International Agency for Research on Cancer classified hairdressing and barbering occupational exposures as probably carcinogenic, with the strongest evidence for bladder cancer.¹²

Hairdresser exposure to endocrine disrupting chemicals (EDCs),⁴ specifically, may alter oestrogen and progesterone levels and influence hormone-related conditions.¹³ Chemicals of concern include phenols such as bisphenol A, found in some hair dyes, which may promote tumour growth.¹⁴ Volatile organic compounds emitted by products used in salons are also known to have adverse hormonal and respiratory effects.¹⁵ Personal use of hairdressing products is linked to uterine fibroids,¹⁶ endometriosis¹⁷ and cancers of the uterus,¹⁸ ovary and breast.¹⁹ Gynaecological surgery may indicate undiagnosed disease or relevant underlying conditions, including fibroids, which are the leading reason for hysterectomy.²⁰

We investigated the association between working as a hairdresser (or not) and gynaecological conditions and surgeries. Previous studies of this topic have had inconsistent results,^{2 18 21} and our investigation provides new evidence to inform occupational safety measures for hairdressers.

Methods

Study design

The Sister Study includes 50 884 women aged 35–74 years who had a sister with breast cancer but were cancer free at enrolment (2003–2009). Participants resided in the USA, including Puerto Rico, and provided informed consent. Sociodemographic, occupational, lifestyle and reproductive histories were collected via computer-assisted telephone interviews at enrolment, with annual updates on health status and extensive follow-up questionnaire completed every 3 years.²² Data from the Sister Study release 11.1 (follow-up through 9/30/2021) were used. After excluding participants who withdrew from the study (n=5), and those diagnosed with breast cancer before they completed enrolment or with uncertain breast cancer diagnosis (n=79), our analytic sample size was 50 800.

Exposure assessment

Participants who reported ever working as a hairdresser, barber or assistant in a barber shop or beauty salon for ≥1 month (N=1803) were classified as ever-hairdressers; long-term hairdressers (ie, those with ≥2 years of work) completed a job module about frequency and duration of professional application of hair dye/colouring, bleach, perms, chemical straighteners, hairspray and agents (barbicide, formaldehyde and alcohol) used to disinfect tools. Short-term hairdressers (<2 years) were not asked these questions. For some analyses, participants with no or short-term hairdresser work are grouped and referred to as non-long-term hairdressers.

Outcome assessment

Prevalent outcomes included hysterectomy or oophorectomy (unilateral, bilateral or partial), uterine fibroids and endometriosis at enrolment. Incident outcomes were breast cancer, fibroids, endometriosis, uterine cancer and ovarian cancer during follow-up. Self-reports were used. Incident cancers were confirmed via pathology reports and other medical records, where possible, with high agreement with self-reports.²³

Covariates

Covariates included age at enrolment (continuous or in age groups, depending on the model), highest level of attained education (≤high school, some college/associate degree, college graduate, postgraduate degree), household income (poor (US$<20 000), low (US$20 000–US$<50 000), middle (US$50 000–US$<100 000) and well-off (>US$100 000)), physical activity (metabolic equivalent hours/week, modelled continuously), body mass index (in kg/m²; underweight (<18.5), healthy weight (18.5 to <25), overweight (25 to <30), obese (≥30), marital status (never married, ever married/living as married, divorced/widowed/separated), cigarette smoking status (never, past, current), alcohol use (never, past, current non-regular, current regular), age at menarche (continuous), hormonal birth control use (none, <2 years, 2–10 years, >10 years), parity (nulliparous, one, two, three or more births), attained menopause (yes, no), race and ethnicity, and a parous (yes, no) by age at most recent childbirth (<25 years, 25–34 years, ≥35 years) interaction term. Recency of childbirth was included because it can impact maternal EDC levels²⁴ and risk of hormone-related conditions.²⁵ Race and ethnicity were self-reported, with participants asked to choose one or more of the following races: American Indian or Alaskan Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, and/or White; and if they considered themselves to be Hispanic or Latina. Based on their responses, they were categorised as non-Hispanic White (NHW), Black (including Hispanic Black), Hispanic White and others. For this study, we conducted stratified analysis within the two largest groups (Black and NHW). All covariates had low missingness (<1% for all covariates), so we carried out complete case analyses, excluding participants with missing data (see table footnotes).

Statistical analysis

We compared baseline characteristics of ever versus never-hairdressers using frequencies, means and SD. We provided descriptive statistics for product use, including a tetrachoric correlation²⁶ matrix describing coexposures to hair products among long-term hairdressers. Unless otherwise stated, analyses were conducted using SAS V.9.4 (SAS Institute).

As all long-term hairdressers worked in a salon, we accounted for passive exposure^{4 27} among non-users by assigning a small constant value (0.01 times/month for 1 year, or 0.11 total uses). We calculated the ‘cumulative usage’ for each hairdresser as the frequency of applications per month multiplied by total duration of reported use in months (total years used products multiplied×11 months assuming 1 month of vacation or personal leave). We categorised hairdressers as ≤median or >median total use (number of uses over the course of their hairdressing career) for each product (hair dye/colouring, bleach, perms, chemical straighteners and hairspray). To date, there is no standardised approach for estimating exposure to each hair product used²⁸; our exposure estimates were based on available information.

There is compelling evidence of harmful chemical composition and health impacts associated with perms, chemical straighteners and permanent hair colour use.^{3 5 29 30} Therefore, we created a coexposure variable which combines the effects of perms, chemical straightener and permanent hair colour.

Prevalence analysis

Using logistic regression, we estimated adjusted ORs and 95% CIs comparing baseline fibroids, endometriosis, hysterectomy and oophorectomy without hysterectomy among ever versus never-hairdressers, adjusting for potential confounders (see table footnotes) selected based on literature review and clinical knowledge. We also calculated ORs comparing ≤median and >median total use of bleach, perms, chemical straightener, permanent hair colour and hairspray among long-term hairdressers, relative to a common referent of non-long-term hairdressers (defined above). We further adjusted for coexposures in models examining hairspray and bleach (coexposure adjusted models).

Incidence analysis

For each outcome of interest, we derived the eligible sample separately. We initially excluded participants diagnosed with the disease of interest prior to enrolment. We next excluded participants missing relevant organs, including those with prebaseline hysterectomies from the analyses of fibroids and uterine cancer or those with prior bilateral oophorectomies from the analyses of ovarian cancer. Lastly, we excluded postmenopausal women from analyses of incident fibroids or endometriosis, as these are considered premenopausal conditions. Using age as a time scale, participants were followed from the time they completed enrolment until the event of interest, with censoring at the end of follow-up, loss-to-follow-up or death. Additional censoring occurred as follows: age at hysterectomy (fibroids, uterine cancer), age at bilateral oophorectomy (ovarian cancer) and age at menopause (fibroids, endometriosis). For endometriosis, we did not exclude participants with hysterectomy because endometriosis can be diagnosed posthysterectomy.

Using Cox proportional hazards regression models, we first assessed covariate-adjusted HRs and 95% CIs comparing incidence rates for fibroids, endometriosis, or breast, uterine or ovarian cancer among ever-hairdressers relative to never-hairdressers. We conducted proportionality assumption tests for each exposure-outcome combination using a Wald test of the time-by-exposure interaction term. We then compared long-term hairdressers to non-long-term hairdressers (ie, never-hairdressers and shorter-term hairdressers). Further, we examined the association of ≤median and >median total use of each product with each hormone-related disease, with and without adjusting for coexposures when assessing hairspray or bleach. We assessed the association of tool cleaning exposures for long-term versus non-long-term hairdressers.

Sensitivity and supplemental analyses

Because women often undergo simultaneous hysterectomy and oophorectomy, we explored different combinations of these conditions in sensitivity analyses. We conducted race/ethnicity (Black and NHW) stratified analyses due to known differences in hair product use patterns and the presence of potentially more harmful ingredients in products marketed specifically to Black women.³¹ Here, we conducted a Wald test of a hairdresser by race/ethnicity interaction term for the incident fibroids and endometriosis analyses to directly assess HR heterogeneity. We also examined the association of total professional product use and incident hormone-related diseases with adjustment for personal use. To determine the impact of combining short-term hairdressers with never hairdressers in analyses of specific hair products, we conducted additional analyses examining the odds of gynaecological conditions among short-term hairdressers, long-term hairdressers who did not use the product of interest and long-term hairdressers who did use the products, relative to never-hairdressers. For perms, chemical straighteners and permanent hair colour we conducted additional analyses for each product, adjusting for the other two products.

We used quantile g-computation (qgcomp) for additional exploratory analysis to evaluate the effects of joint exposure to hair product mixtures on hormone-related diseases.³² Non-long-term hairdressers made up the lowest exposure group, with long-term hairdressers separated into tertiles by amount of hair processing chemical exposure. Outcome-specific weights for each component of the mixture were calculated using a Cox proportional hazards model (qgcomp³² package, R V.4.3). The resulting HRs can be interpreted as the change in the hazard per one-unit simultaneous categorical increase in each component of the hair product mixture.

Results

A total of 1803 study participants reported ever working as a hairdresser, leaving 48 997 never-hairdressers (table 1). Compared with never-hairdressers, ever-hairdressers had lower education (12% vs 27% college degree; 8% vs 25% postgraduate degree), and were less likely to report being well off financially (20% vs 34%). Ever-hairdressers were also less likely to be never smokers (41% vs 57%). Both groups were predominately NHW, but ever-hairdressers were more likely to identify as non-Hispanic Black (13% vs 9%) or ‘other’ race/ethnicity (12% vs 8%) than never-hairdressers.

Table 1. Baseline characteristics of ever-hairdressers and never-hairdressers in the Sister Study, 2003–2021.

Characteristics	All* N=50 800 (100%)	Ever-hairdressers† N=1803 (3%)	Never-hairdressers N=48 997 (97%)
Age at menarche (years, SD)	12.6±1.5	12.6±1.6	12.6±1.5
Physical activity MET hours/week‡	14.3±17.7	11.9±16.2	14.3±17.7
Age groups (years)
35–44	6571 (13)	281 (16)	6290 (13)
45–54	17 500 (34)	607 (34)	16 893 (34)
55–64	17 876 (35)	719 (40)	17 157 (35)
≥65	8853 (17)	196 (11)	8657 (18)
Body mass index (Kg/M2)
Underweight (<18.5)	591 (1)	10 (0)	581 (1)
Healthy weight (18.5 to <25)	19 007 (37)	544 (30)	18 463 (38)
Overweight (25 to <30)	16 035 (32)	615 (34)	15 420 (31)
Obese (≥30)	15 150 (30)	633 (35)	14 517 (30)
Race/Ethnicity
Non-Hispanic White	42 486 (84)	1358 (75)	41 128 (84)
Non-Hispanic Black	4459 (9)	235 (13)	4224 (9)
Others	3840 (8)	209 (12)	3631 (8)
Marital status
Never married	2755 (5)	71 (4)	2684 (5)
Ever married/living as married	37 940 (75)	1307 (73)	36 633 (75)
Divorced/widowed/separated	10 091 (20)	423 (23)	9668 (20)
Highest level of education at enrolment
High school or less	7790 (15)	498 (28)	7292 (15)
Some college/associate degree	17 162 (34)	939 (52)	16 223 (33)
College graduate	13 690 (27)	224 (12)	13 466 (27)
Post graduate degree	12 146 (24)	141 (8)	12 005 (25)
Household income at enrolment
Poor (<US$20 000)	2391 (5)	169 (9)	2222 (5)
Low (US$20 000–<US$50 000)	10 707 (21)	530 (29)	10 177 (21)
Middle (US$50 000–<US$100 000)	20 725 (41)	738 (41)	19 987 (41)
Well off (>US$100 000)	16 977 (33)	366 (20)	16 611 (34)
Cigarette smoke status at enrolment
Never	28 511 (56)	744 (41)	27 767 (57)
Past smoker	18 108 (36)	784 (44)	17 324 (35)
Current smoker	4165 (8)	273 (15)	3892 (8)
Alcohol use at enrolment
Never	1945 (4)	63 (4)	1882 (4)
Past	7715 (15)	320 (18)	7395 (15)
Current non-regular	4177 (8)	158 (9)	4019 (8)
Current regular	36 950 (73)	1260 (70)	35 690 (73)
Hormonal birth control use
None	7468 (15)	213 (12)	7255 (15)
<2 years	7736 (15)	312 (17)	7424 (15)
2–10 years	23 676 (47)	863 (48)	22 813 (47)
>10 years	11 628 (23)	405 (23)	11 223 (23)
Parity
Nulliparous	9194 (18)	225 (12)	8969 (18)
One	7333 (14)	280 (16)	7053 (14)
Two	18 658 (37)	706 (39)	17 952 (37)
Three or more	15 580 (31)	590 (33)	14 990 (31)
Age at most recent birth
No birth	9194 (18)	225 (13)	8969 (18)
<25 years	8044 (16)	474 (26)	7570 (15)
25–34 years	26 171 (52)	922 (51)	25 249 (52)
≥35 years	7338 (14)	179 (10)	7159 (15)
Menopausal
No	17 014 (33)	610 (34)	16 404 (33)
Yes	33 786 (67)	1193 (66)	32 593 (67)

Percentages may not add up to 100 due to rounding up.

Missing: physical activity n=46, age at menarche n=46, BMI n=17, race/ethnicity n=15, marital status n=14, education n=12, smoke status n=16, alcohol use n=13, any hormonal birth control use n=292, parity n=35, age at most recent birth n=53.

^*Excludes those who withdrew (n=5), and those who were diagnosed with breast cancer or had an uncertain diagnosis before completing enrolment (n=79).

^†Includes all who responded yes to ever worked as a hairdresser, barber or assistant in a barber’s shop or beauty salon.

^‡Physical activity total METs hour per week.

BMI, body mass index; MET, metabolic equivalent.

Of the 1803 ever-hairdressers, 985 (55%) worked ≥2 years (online supplemental Table S1). Most reported using hairspray (89%), perms (84%), permanent hair colour (84%) and bleach (75%). The least reported agent was chemical straightener use (38%), but 66% of Black hairdressers reported use. The most reported tool cleaning agent was barbicide (88%). After assigning low-level exposure values to non-users, the median total use was highest for hairspray (9240 total applications), followed by permanent hair colour (1192), perms (954), bleach (220) and chemical straightener (0.11). Applications of hairspray, permanent hair colour, bleach and perms were highly correlated (Tetrachoric R range=0.85–0.92; online supplemental Figure S1).

Compared with never-hairdressers, ever-hairdressers had increased odds of hysterectomy OR=1.23 (95% CI 1.11 to 1.36) with or without oophorectomy (figure 1a) and, slightly increased odds of fibroids OR=1.06 (95% CI 0.95 to 1.17), but not endometriosis OR=0.95 (95% CI 0.83 to 1.09) or oophorectomy alone OR=0.88 (95% CI 0.65 to 1.17).

Figure 1. ORs (95% CIs) for prevalent gynaecological conditions (a) and HRs (95% CIs) for incident hormone-related diseases (b) among ever versus never hairdressers. BMI, body mass index.

Compared with never-hairdressers, ever-hairdressers showed a higher incidence of endometriosis (477 cases, HR=1.61, 95% CI 1.08 to 2.38) (figure 1), but no evident associations were observed for breast cancer (4628 cases, HR=0.98: 95% CI 0.83 to 1.16), fibroids (1805 cases, HR=1.04, 95% CI 0.80 to 1.34) or uterine cancer (447 cases, HR=1.04: 95% CI 0.60 to 1.77). The association with ovarian cancer (300 cases, HR=1.33, 95% CI 0.77 to 2.29) was elevated, but statistically imprecise.

When we examined product use, we observed only a few strong associations, with little to no evidence of a monotonic dose-response across categories of increasing exposure (table 2 and online supplemental Table S11). One exception was the association between applications of bleach and prevalent fibroids after adjusting for coexposure to other hair products (OR=1.72, 95% CI 0.96 to 3.11 for <median exposure, OR=2.09, 95% CI 0.97 to 4.05 for >median exposure, relative to non-long-term hairdressers; p-for-trend=0.15).

Table 2. Covariate-adjusted ORs (95% CIs) for associations between cumulative occupational exposure to hair products and prevalent gynaecological conditions among long-term hairdressers^* compared with non-long-term hairdressers^* in the Sister Study.

Exposure	Fibroid OR (95% CI)	Endometriosis OR (95% CI)	Hysterectomy OR (95% CI)	Oophorectomy only OR (95% CI)
Applied bleach
Non-long-term hairdresser	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.04 (0.89 to 1.22)	0.93 (0.76 to 1.14)	1.31 (1.12 to 1.53)	0.68 (0.41 to 1.12)
Main model
≤Median total use	1.05 (0.86 to 1.27)	0.85 (0.66 to 1.10)	1.14 (0.94 to 1.37)	1.17 (0.73 to 1.88)
>Median total use	1.08 (0.88 to 1.31)	0.92 (0.71 to 1.19)	1.21 (1.00 to 1.47)	0.47 (0.22 to 0.99)
P trend	0.40	0.26	0.02	0.13
Coexposure adjusted
≤Median total use	1.72 (0.96 to 3.11)	0.70 (0.32 to 1.52)	0.81 (0.47 to 1.51)	1.56 (0.29 to 8.37)
>Median total use	2.09 (0.97 to 4.05)	0.71 (0.26 to 1.94)	0.81 (0.38 to 1.73)	0.69 (0.07 to 6.83)
P trend	0.15	0.87	0.73	0.21
Applied perms
Non-long-term hairdresser	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.03 (0.88 to 1.19)	0.89 (0.73 to 1.08)	1.20 (1.04 to 1.39)	0.68 (0.42 to 1.09)
≤Median total use	1.08 (0.89 to 1.31)	0.89 (0.69 to 1.14)	1.05 (0.86 to 1.27)	1.11 (0.68 to 1.81)
>Median total use	1.04 (0.86 to 1.27)	0.88 (0.68 to 1.14)	1.31 (1.08 to 1.59)	0.53 (0.26 to 1.07)
P trend	0.48	0.20	0.01	0.16
Applied chemical straightener
Non-long-term hairdresser	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	0.98 (0.78 to 1.22)	0.91 (0.69 to 1.21)	1.24 (1.00 to 1.54)	1.01 (0.56 to 1.80)
≤Median total use	1.13 (0.95 to 1.35)	0.85 (0.67 to 1.08)	1.13 (0.95 to 1.35)	0.69 (0.39 to 1.19)
>Median total use	0.95 (0.76 to 1.19)	0.94 (0.71 to 1.25)	1.24 (1.00 to 1.54)	1.05 (0.59 to 1.88)
P trend	0.71	0.28	0.02	0.57
Applied permanent hair colour
Non-long-term hairdresser	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.03 (0.88 to 1.19)	0.91 (0.75 to 1.10)	1.20 (1.04 to 1.39)	0.76 (0.49 to 1.19)
≤Median total use	1.18 (0.97 to 1.43)	0.80 (0.61 to 1.04)	1.15 (0.95 to 1.40)	1.26 (0.79 to 2.01)
>Median total use	0.95 (0.77 to 1.16)	0.97 (0.76 to 1.25)	1.20 (0.99 to 1.45)	0.39 (0.17 to 0.88)
P trend	0.79	0.35	0.02	0.10
Used hairspray on hair
Non-long-term hairdresser	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.05 (0.90 to 1.21)	0.91 (0.75 to 1.10)	1.20 (1.04 to 1.39)	0.72 (0.46 to 1.12)
Main model
≤Median total use	1.20 (0.99 to 1.45)	0.82 (0.64 to 1.06)	1.11 (0.92 to 1.34)	1.34 (0.86 to 2.09)
>Median total use	0.92 (0.75 to 1.13)	0.95 (0.74 to 1.23)	1.25 (1.02 to 1.52)	0.27 (0.10 to 0.73)
P trend	0.87	0.31	0.02	0.07
Coexposure adjusted
≤Median total use	1.24 (0.66 to 2.35)	0.76 (0.33 to 1.74)	0.87 (0.47 to 1.64)	0.65 (0.11 to 3.70)
>Median total use	0.97 (0.40 to 2.34)	0.85 (0.27 to 2.66)	0.88 (0.37 to 2.11)	0.10 (0.01 to 1.24)
P trend	0.25	0.73	0.98	0.00

Models adjusted for age group at baseline, BMI, age at menarche, physical activity, race/ethnicity, education, income, hormonal birth control use, parity, alcohol and smoking.

Coexposure adjusted includes additional adjustment for perms, chemical straightener and permanent hair colour.

Hysterectomy includes women who had ever had a hysterectomy or a hysterectomy and oophorectomy (67%), oophorectomy includes oophorectomy only.

Missing/don’t know/refused: oophorectomy n=77, hysterectomy n=1, endometriosis=0, fibroids=0.

Cumulative use over the duration of their career (frequency of use multiplied by total duration of use).

Median total use bleach 220, perms 953.5, chemical straightener 0.11, permanent hair colour 1191.9, hairspray 9240 applications over the total duration of their career.

Long-term hairdressers who reported no/unknown/missing use were assigned a very small constant value of 0.11 for total use of hair products to account for passive exposure that occurs in a salon.

Less than 40% of the hairdressers reported use of chemical hair straighteners.

^*Long-term hairdressers include only study participants who have worked as hairdressers, barbers or assistants in a barber’s shop or beauty salon for 2 or more years, while non-long-term hairdressers include never-hairdressers and hairdressers with less than 2 years of experience (short-term hairdressers).

BMI, body mass index.

History of hysterectomy was positively associated with all 5 exposures of interest in models unadjusted for coexposures. The respective ORs associated with >median total use versus non-use and p-for-trends were as follows: OR=1.21 (95% CI 1.00 to 1.47) and p-trend=0.02 for bleach, OR=1.31 (95% CI 1.08 to 1.59) and p trend=0.01 for perms, OR=1.24 (95% CI 1.00 to 1.54) and p trend=0.02 for chemical straightener, OR=1.20 (95% CI 0.99 to 1.45) and p trend=0.02 for permanent hair colour, and OR=1.25 (95% CI 1.02 to 1.52) and p trend=0.02 for hairspray. However, the positive associations for bleach and hairspray were no longer present after adjusting for coexposure to perms, chemical straighteners and permanent hair dye.

No statistically significant associations were observed for use of bleach, perms, chemical straighteners, permanent hair colour or hairspray and incident breast cancer, fibroids, endometriosis, uterine cancer or ovarian cancer (table 3 and online supplemental Table S12).

Table 3. Covariate-adjusted HRs (95% CIs) for associations between cumulative occupational exposure to hair products and incidence of hormone-related diseases among long-term hairdressers^* relative to non-long-term hairdressers^* in the Sister Study.

Chemical exposure	Breast cancer	Fibroid	Endometriosis	Uterine cancer	Ovarian cancer
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
Applied bleach
Non-long-term hairdresser	Ref	Ref	Ref	Ref	Ref
Main model
Hairdresser ‘yes’ use	1.05 (0.82 to 1.34)	0.75 (0.46 to 1.21)	0.98 (0.43 to 2.22)	1.24 (0.59 to 2.62)	1.07 (0.44 to 2.60)
≤Median total use	0.96 (0.71 to 1.31)	1.10 (0.71 to 1.72)	1.33 (0.62 to 2.85)	1.29 (0.54 to 3.13)	- -
>Median total use	1.09 (0.81 to 1.47)	0.63 (0.31 to 1.25)	- -	- -	- -
P trend	0.71	0.36	- -	- -	- -
Coexposure adjusted
Hairdresser ‘yes’ use	0.72 (0.28 to 1.86)	0.90 (0.14 to 5.84)	3.59 (0.24 to 54.9)	- -	0.66 (0.03 to 13.9)
≤Median total use	0.71 (0.30 to 1.70)	1.17 (0.25 to 5.57)	--	2.11 (0.42 to 10.7)	- -
>Median total use	0.73 (0.22 to 2.37)	0.68 (0.09 to 5.33)	- -	- -	- -
P trend	0.93	0.31	- -	- -	- -
Applied perms
Non-long-term hairdresser	Ref	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.05 (0.83 to 1.32) 0.94 (0.69 to 1.29)	0.86 (0.56 to 1.31)	1.10 (0.54 to 2.24)	1.06 (0.50 to 2.24)	1.11 (0.49 to 2.52)
≤Median total use		1.05 (0.66 to 1.65)	1.48 (0.72 to 3.05)	1.27 (0.53 to 3.07)
>Median total use	1.11 (0.82 to 1.49)	0.71 (0.37 to 1.36)	- -	- -	1.62 (0.66 to 3.97)
P trend	0.67	0.44	- -	- -	- -
Applied chemical straighteners
Non-long-term hairdresser	Ref	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	0.97 (0.68 to 1.39) 1.03 (0.78 to 1.35)	1.04 (0.61 to 1.79)	0.55 (0.14 to 2.25)	0.66 (0.16 to 2.64)	1.26 (0.40 to 3.95)
≤Median total use		0.80 (0.48 to 1.34)	1.71 (0.87 to 3.37)	1.28 (0.57 to 2.88)	- -
>Median total use	1.02 (0.71 to 1.45)	1.08 (0.64 to 1.85)	- -	- -	- -
P trend	0.86	0.86	- -	- -	- -
Applied permanent hair colour
Non-long-term hairdresser	Ref	Ref	Ref	Ref	Ref
Hairdresser ‘yes’ use	1.03 (0.81 to 1.30)	0.95 (0.64 to 1.41)	1.05 (0.51 to 2.15)	1.22 (0.61 to 2.47)	0.94 (0.38 to 2.27)
≤Median total use	0.88 (0.64 to 1.22)	1.19 (0.76 to 1.87)	1.39 (0.65 to 2.97)	- -	1.67 (0.68 to 4.06)
>Median total use	1.17 (0.88 to 1.56)	0.60 (0.31 to 1.16)	--	- -	- -
P trend	0.54	0.30	--	- -	- -
Used hairspray on hair
Non-long-term hairdresser	Ref	Ref	Ref	Ref	Ref
Main model
Hairdresser ‘yes’ use	1.06 (0.84 to 1.32)	0.80 (0.52 to 1.23)	0.92 (0.43 to 1.96)	1.13 (0.56 to 2.29)	1.07 (0.47 to 2.42)
≤Median total use	1.04 (0.77 to 1.39)	1.05 (0.65 to 1.69)	1.15 (0.50 to 2.61)	1.46 (0.65 to 3.27)	1.62 (0.66 to 3.95)
>Median total use	1.01 (0.73 to 1.38)	0.75 (0.41 to 1.37)	1.41 (0.57 to 3.47)	- -	- -
P trend	0.89	0.46	0.43	- -	. - -
Coexposure adjusted
Hairdresser ‘yes’ use	0.80 (0.34 to 1.88)	0.96 (0.16 to 5.88)	2.27 (0.18 to 28.3)	4.07 (0.49 to 33.6)	--
≤Median total use	0.55 (0.21 to 1.44)	1.43 (0.28 to 7.19)	- -	1.14 (0.12 to 10.76)	0.31 (0.02 to 6.35)
>Median total use	0.41 (0.10 to 1.66)	1.15 (0.13 to 10.02)	- -	- -	- -
P trend	0.31	0.72	- -	- -	- -

Models adjusted for age at baseline, BMI, age at menarche, physical activity, race/ethnicity, education, income, hormonal birth control use, parity*age of last childbirth, alcohol, parity and smoking.

Co-exposure adjusted includes additional adjustment for perms, chemical straightener and permanent hair colour.

NA=HR could not be estimated, - - unstable estimates with n<5 are not reported.

Use caution interpreting some HRs with wide CIs.

Median total use bleach 220, perms 953.5, chemical straightener 0.11, permanent hair colour 1191.9, hairspray 9240 applications over the total duration of their career.

Long-term hairdressers who reported no/unknown/missing use were assigned a very small constant value of 0.11 for total use of hair products to account for passive exposure that occurs in a salon.

Less than 40% of the hairdressers reported use of chemical hair straighteners.

^*Long-term hairdressers include only study participants who have worked as hairdressers, barbers or assistants in a barber’s shop or beauty salon for 2 or more years, while non-long-term hairdressers include never hairdressers and hairdressers with less than 2 years’ experience (short-term hairdressers).

BMI, body mass index.

Overall, we did not observe any notable associations for breast cancer and use of barbicide (HR=1.02, 95% CI 0.80 to 1.29), formaldehyde (HR=0.99: 95% CI 0.66 to 1.48) or alcohol (HR=1.05, 95% CI 0.68 to 1.63), compared with non-long-term hairdressers (table 4). Race/ethnicity-specific results were similarly null. Results for other hormone-related diseases are not presented due to sample size limitations.

Table 4. HRs for the association of occupational exposure to cleaning agents and incident breast cancer among long-term hairdressers^* relative to non-long-term hairdressers in the Sister Study.

Chemical exposure	All hairdressers		Black hairdressers		NHW hairdressers
	E/n	HR (95% CI)	E/n	HR (95% CI)	E/n	HR (95% CI)
Barbicide
Non-hairdresser	4542/49815	Ref	363/4334	Ref	3893/41724	Ref
Hairdresser non-user	9/111	0.94 (0.48 to 1.84)	- -/22	- -	8/75	1.20 (0.58 to 2.48)
Hairdresser users	72/827	1.03 (0.81 to 1.30)	7/98	0.91 (0.43 to 1.92)	58/648	1.05 (0.81 to 1.36)
Formaldehyde
Non-hairdresser	4542/49815	Ref	363/4334	Ref	3893/41724	Ref
Hairdresser non-user	57/668	1.04 (0.79 to 1.35)	7/90	0.96 (0.45 to 2.02)	45/509	1.05 (0.78 to 1.41)
Hairdresser users	25/272	0.99 (0.66 to 1.48)	- -/31	- -	21/217	1.06 (0.69 to 1.63)
Alcohol
Non-hairdresser	4542/49815	Ref	363/4334	Ref	3893/41724	Ref
Hairdresser non-user	61/712	1.00 (0.78 to 1.29)	5/64	0.96 (0.40 to 2.30)	51/588	1.00 (0.76 to 1.32)
Hairdresser users	21/233	1.05 (0.68 to 1.63)	- -/57	- -	14/142	1.14 (0.68 to 1.93)

E/n indicates total number of events per sample size.

- - Unstable estimates with n<5 are not reported.

Use caution interpreting some HRs due to wide CIs.

Models adjusted for age at baseline, BMI, age at menarche, physical activity, race/ethnicity, education, income, hormonal birth control use, parity*age of last childbirth, alcohol, parity and smoking.

^*Long-term hairdressers include only study participants who have worked as hairdressers, barbers, or assistants in a barber’s shop or beauty salon for 2 or more years, while non-hairdressers includes hairdressers with less than 2 years’ experience (short-term hairdressers).

BMI, body mass index.

Sensitivity analysis and supplemental findings

The OR for ever working as a hairdresser was slightly higher for hysterectomy without oophorectomy (OR=1.26, 95% CI 1.10 to 1.46) than both oophorectomy and hysterectomy (OR=1.10, 95% CI 0.98 to 1.23) (online supplemental Figure S2), but both estimates were positive.

The previously noted positive association between>median bleach use and hysterectomy, after adjustment for co-exposures, was observed for hysterectomy alone (OR=3.40, 95% CI 1.20 to 9.65, p trend=0.01) (online supplemental Table S2). In contrast, we observed an inverse association for >median bleach use and having had both hysterectomy and oophorectomy (OR=0.36, 95% CI 0.16 to 0.85), p trend=0.05. However, both estimates were imprecise. The ever-hairdresser and prebaseline hysterectomy association was similar among NHW (OR=1.24, 95% CI 1.10 to 1.39) and Black (OR=1.24, 95% CI 0.94 to 1.66) participants (online supplemental Table S3).

Among Black participants, the HRs comparing ever-hairdressers to never-hairdressers were 1.56 (201 cases, 95% CI 0.93 to 2.62) for fibroids and 2.04 (39 cases, 95% CI 0.68 to 6.06) for endometriosis (online supplemental Figure S3). Among NHW participants, the HR comparing ever-hairdressers with never-hairdressers was 0.85 (95% CI 0.61 to 1.18) fibroids and 1.36 (393 cases, 95% CI 0.83 to 2.22) for endometriosis. We found that race/ethnicity modifies the association of hairdressing occupation with fibroid incidence (p interaction=0.04), but not incident endometriosis (p interaction=0.55) (online supplemental Table S4).

Race-stratified HRs for associations between cumulative exposure to hair products and incident hormone-related diseases are provided in online supplemental Tables S5 and S6. We did not observe any clear associations for any product among Black women. Among NHW women, the HR for ≤median total use of permanent hair colour, versus being a non-hairdresser, and incident breast cancer was 0.74 (95% CI 0.49 to 1.11). However, this association was elevated for >median total use of permanent hair colour (HR=1.31, 95% CI 0.97 to 1.76; p trend=0.35).

For chemical straightener use, additional adjustment for coexposure resulted in HRs that were much stronger, yet highly imprecise. For ≤median total use of chemical straightener, versus being a non-hairdresser, the HR for incident fibroids was 2.76 (95% CI 0.74 to 10.4) relative to HR=0.80 without coadjustment, and the HR for >median total use of chemical straightener, versus being a non-hairdresser was HR=4.09 (95% CI 1.10 to 21.8; p trend=0.84) relative to 1.08 without coadjustment. Findings for perm and permanent hair colour with additional coexposure adjustments were similar to findings reported in the main analyses (online supplemental Table S7).

In sensitivity analyses that included adjustments for personal use of products, the HRs for exposure to specific products and incidence of hormone-related diseases remained similar (online supplemental Table S8). The association between fibroids, endometriosis, hysterectomy and oophorectomy with hair product use was similar for short-term hairdressers and long-term hairdressers’ who reported use (online supplemental Table S9). This was particularly evident with hysterectomy.

The quantile g-computation estimates did not provide compelling evidence of an association with the total exposure mixture and any of the outcomes of interest (online supplemental Table S10).

Discussion

We found that hairdressing is positively associated with history of hysterectomy and incident endometriosis. Additionally, we observed a positive association with incident fibroids among Black hairdressers. Longer-term hairdressers with higher exposure to perms, chemical straighteners or permanent hair colour were more likely to have had a hysterectomy than non-long-term hairdressers, with consistent dose-response trends. Higher exposure to permanent hair colour may also be associated with higher rates of incident breast cancer among NHW women, but there was insufficient evidence to support associations with uterine or ovarian cancer.

While studies have considered disease rates among hairdressers or separately investigated the association between individual hair products and hormone-related diseases, we have expanded on the literature by also exploring the association of hairdressers’ use of specific products, both individually and as mixtures, with hormone-related diseases. We additionally considered differences by race, which can be an important predictor of the types of hair products women use.³¹

A meta-analysis by Takkouche et al reported an increased risk of gynaecologic cancers among hairdressers, ranging from 6% for breast cancer to 12% for ovarian cancer.³⁰ Similarly, a record linkage study of five Nordic countries by Pukkala et al demonstrated an excess risk of ovarian cancer among hairdressers.³³ In our sample, working as a hairdresser was associated with an elevated but imprecise rate of ovarian cancer, but not breast cancer. However, permanent hair colour use showed a potential association with breast cancer, particularly among NHW hairdressers. The literature on the association of permanent hair dye and breast cancer is inconclusive.^{19 34} In a prior Sister Study analysis, Eberle et al reported a positive association between personal permanent hair dye use (HR=1.09, 95% CI 1.00 to 1.19) and incident breast cancer, as well as between non-professional application of semipermanent hair dye (HR=1.28, 95% CI 1.05 to 1.56) and breast cancer.¹⁹ Personal hair dye use was not strongly associated with ovarian or uterine cancer in the Sister Study.¹⁸ The Nurses’ Health Study reported increased ovarian cancer incidence with frequent permanent hair dye use.³⁴ Although we did not observe an association between applying chemical straighteners as a hairdresser and hormone-related conditions, we previously reported a positive association between personal use of chemical straighteners and breast, uterine¹⁸ and ovarian cancer,³⁵ as well as uterine fibroids.¹⁶

Hair care products contain over 5000 chemicals, many of which possess mutagenic and endocrine-disrupting properties.³⁶ Hairdressers are regularly exposed to these chemicals through dermal contact, inhalation and ingestion.^{3 4 12 27} Hazardous chemicals include phthalates, parabens, siloxanes,^{37 38} formaldehyde, mineral oil/petrolatum, benzene, toluene and ammonia, all contributing to health risk.²⁷ Phthalates and parabens mimic hormones, disrupting hormonal balance.³⁹ Product use varies by race, with Black women more likely to use products that contain EDCs than NHW women.³¹ The use of heat for styling hair increases release of chemicals into the room air.¹⁵

Hairdressers in our study had higher rates of endometriosis than never-hairdressers, with Black hairdressers also showing higher rates of fibroids. The aetiologies of endometriosis and fibroids are largely unknown, but exposure to EDCs has been identified as a significant risk factor for both.^{21 39} In a previous study, we also found a positive association between personal use of hair straighteners and fibroids.¹⁶ These conditions are leading indications for hysterectomy,⁴⁰ including among women in the Sister Study, where over 50% of those with pre-baseline hysterectomies reported fibroids, and over 30% reported endometriosis. Higher prevalence of hysterectomy in our study was associated with increased applications of perms, chemical straighteners and permanent hair colour use. More research is needed to understand the individual contributions of each hair product. However, race-stratified analyses may not fully capture exposure variation if a stylist’s clientele is racially heterogeneous or differs from her own racial/ethnic background. Long periods of bending and standing in hairdressing may also drive hairdressers to seek more definitive treatments like hysterectomy for fibroids or endometriosis diagnosis.

A major limitation is the small sample sizes in some subgroup analyses. Self-reported data may introduce response bias, though any such bias in prospective incidence analyses is likely non-differential. Binary measures of exposure for tool cleaning agents and absence of measures of exposure for short-term hairdressers may have limited the ability to detect associations, or attenuated findings associated with hair product use.

The cohort’s average enrolment age of 55 years likely excluded some earlier-onset incident cases, which may have distinct aetiologies, particularly for conditions like endometriosis and fibroids, which are much more common in younger women. As all occupational exposures and gynaecological conditions were reported at baseline, we cannot establish temporality for the prevalent analyses. Further, this approach provides an opportunity for recall bias (ie, differential reporting of exposure by those with vs without a condition). Analyses of incident outcomes avoid these concerns but cannot be used to study earlier-onset occurrences. Because participants were aged 35–74 at enrolment, the prevalence analysis allowed us to include diagnoses and conditions such as hysterectomy, fibroids and endometriosis that often occur at younger ages. Hence, we conducted both incidence and prevalence analyses. The highly correlated product use also hindered the ability to isolate the effect of individual chemicals. We also did not consider other occupational groups with potentially high cleaning chemical exposures (eg, healthcare workers, housecleaners). Unmeasured potential confounders such as family history of disease, ventilation in salons and personal protective equipment use may have influenced the results.

The strengths of the Sister Study cohort are its large size and extensive follow-up data on personal, environmental and occupational exposures, and hormone-related diseases. The relatively large number of hairdressers in the study enabled the exploration of several hormone-related diseases and gynaecological conditions. While the cohort is predominantly NHW, the Sister Study also includes a sizeable sample of non-Hispanic Black women, allowing stratification by race to better address differences in types of products used.

Given the observed associations between hairdresser occupational exposures and hormone-related conditions in this study, this field of research could benefit from studies that obtain more objective measures of chemical exposures, especially among younger cohorts and in situations with poor glove use or inadequate ventilation. Some hairdressers also work as cosmetologists and beauticians¹; therefore, studies examining individual or combined exposures to cosmetics, nail products and hair products may be informative. The study also underscores the need for more detailed investigations into what specific hair care products or chemicals within those products, individually or in combination, directly impact the health of hairdressers. Additionally, the timing of the hairdressing occupation relative to benchmarks such as pregnancy and menopause should be considered in future studies.

Conclusions

We observed positive associations between occupational exposures among hairdressers and hormone-related conditions. Hairdressers had higher odds of prebaseline hysterectomy and higher rates of endometriosis compared with non-hairdressers, and Black hairdressers were more likely to develop fibroids. Increased application of permanent hair dye was modestly associated with increased incident breast cancer, mirroring prior findings of positive associations between personal permanent hair dye use and breast cancer. These findings underscore the need to better understand the chemical constituents of hair products and their effects, particularly in occupational settings where exposures may be high. Evidence-based interventions to reduce harmful exposures could benefit hairdressers, especially those of reproductive age who may face prolonged exposure during critical windows of disease development.

Data availability statement

Data are available on reasonable request.