Environmental Tobacco Smoke and Risk of Late-Diagnosis Incident Fibroids in the Study of Women’s Health Across the Nation

Abstract

Objective

To assess the longitudinal relationship of environmental tobacco smoke (ETS) exposure during midlife, and its interaction with active smoking, to the risk of late-diagnosis incident uterine fibroids during the menopausal transition.

Design

A 13-year prospective cohort study was conducted in the Study of Women’s Health Across the Nation (SWAN). At each near-annual study visit, ETS exposure, smoking, and fibroid occurrence were self-reported via questionnaire, and blood draws were collected.

Setting

A community-based, multi-racial/ethnic cohort from seven cities in the United States.

Patients

The participants were 2575 women aged 42–52 years at baseline, undergoing the menopausal transition.

Interventions

The exposure was time-varying ETS exposure, and its interaction with active smoking.

Main Outcome Measures

Discrete-time proportional odds models were used to estimate the conditional odds ratios (ORs) and 95% Confidence Intervals (CI) of incident fibroids, adjusted for menopausal status, race/ethnicity, site, age, education, estradiol levels, sex hormone use, body mass index, timing of blood draw, age at menarche, alcohol use, and smoking status and pack-years.

Results

Women who were exposed to ETS (≥1 person-hour/week) had 1.28 (95% CI: 1.03, 1.60) times the adjusted odds of incident fibroids in the ensuing year compared the unexposed. The odds were elevated in never smokers (adjusted OR=1.34 (95% CI: 1.06,1.70, p_interaction=0.11)) and former smokers (adjusted OR=2.57 (95% CI:1.05,7.23, p_interaction=0.11)).

Conclusions

ETS exposure in midlife was associated with an increased risk of late-diagnosis incident fibroids in women undergoing the menopausal transition.

Introduction

Environmental tobacco smoke (ETS; secondhand smoke; passive smoke) is a public health burden that has received considerable regulatory scrutiny over the past few decades (1). Although the prevalence of ETS exposure has declined since the 1990’s due to regulatory bans on indoor smoking, the estimated annual economonic costs attributed to smoking and ETS exposure continues to rise and approached $300 billion in the United States in 2014 (2). Most population-based studies have focused on respiratory outcomes; however, ETS has also been associated with adverse reproductive health outcomes in women including low birthweight, preterm birth, and inconsistently with decreased fertility (3–5).

Uterine fibroids are benign tumors of the myometrium that typically emerge during pre- and peri-menopause and are detected in approximately 70–80% of women (6). Symptoms include pelvic pain, abnormal uterine bleeding, decreased fertility, and complications in pregnancy (6). The strongest established risk factors include being African-American, obese, nulliparous, premenopausal, and a user of hormone replacement therapy (7–12). Conversely, active cigarette smoking has been associated with decreased risk of fibroids in some studies (13–15); however, another study found positive associations with diffuse fibroids (16). Preventive strategies are desirable because therapeutic options are limited, and hysterectomy remains the first-line treatment (6). Myomectomy of fibroids is another surgical option for women who want to become pregnant; however, recurrence is still possible.

Late-diagnosis fibroids are an under-investigated health burden in peri- and post-menopausal women, as most cases are diagnosed earlier in the lifecourse. The contributions of ETS to fibroid development have yet to be firmly established, despite possible associations with active smoking. To advance the understanding of the interrelationship between ETS, active smoking, and risk of late-diagnosis fibroids, we leveraged data from the Study of Women’s Health Across the Nation (SWAN), a multi-racial/ethnic cohort of midlife women undergoing the menopausal transition (MT). The objective of this study was to assess the longitudinal relationship between ETS exposure in midlife, its interaction with active smoking, and risk of late-diagnosis incident uterine fibroids. We hypothesized that ETS exposure would be associated with an increased risk of developing uterine fibroids, which would be further elevated in former and current smokers.

Materials and Methods

Study Population

The characteristics of SWAN have been described elsewhere (17). Briefly, SWAN is a multiracial/ethnic prospective cohort study of women undergoing the menopausal transition. SWAN enrolled 3302 eligible women from seven sites across the United States. At baseline, participants were aged 42–52 years, pre- or early peri-menopausal, reported a recent menstrual period, had an intact uterus and at least one ovary, were not pregnant or lactating, and did not recently use exogenous hormones. Since 1996, participants have been evaluated at near-annual in-person visits, using self-administered questionnaires and interviews to collect demographic, medical, reproductive, and anthropometric information. The institutional review boards of all participating institutions approved the SWAN core study. All participating women provided written informed consent.

The current analyses used a longitudinal design with near-annual repeated measures of ETS exposure, covariates, and fibroid occurrence throughout 13 years of follow-up. Women who reported any cancer diagnosis prior to baseline were excluded. Subsequently, there were 3217 women with ETS exposure data at baseline. The survival analyses further excluded women who reported having fibroids prior to baseline. The resulting analytic sample size was 2575 women. During the follow-up, women who reported having a hysterectomy, incident fibroids, died, or were permanently lost to follow-up since the last visit were censored.

Outcome Assessment: Uterine Fibroids

From visits 0 to 3, occurrence of fibroids was assessed by asking, “Since your last study visit, has a doctor, nurse practitioner or other health care provider told you that you had fibroids, benign growths of the uterus or womb, or treated you for them”. From visits 4 to 13, occurrence was assessed by asking, “Since your last study visit, have you had fibroids (benign growths in the uterus or womb)?” Medical record abstractions for a subset of 99 women who underwent hysterectomy were analyzed using binomial proportions tests for sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), against verified gross and histological clinical findings.

Exposure Assessment: Environmental Tobacco Smoke

Time-varying ETS exposure was assessed at baseline, and visits 3, 7, 8, and 9 using seven questions adapted from a validated survey (18). Participants were asked the number of active smokers at home, work, and social settings, days of ETS exposure in the past week, and daily hours of ETS exposure, at home, work, and other social settings. Based on a previous study, cumulative ETS exposure in the past seven days (person-hours/week) was calculated as follows: (No. of smokers at home)*(Hours/day of ETS exposure at home)*(No. of days of ETS exposure in past 7 days at home) + (No. of smokers at work)*(Hours/day of ETS exposure at work)*(No. of days of ETS exposure in past 7 days at work) + (Hours/day of ETS exposure at other settings)*(No. of days of ETS exposure in past 7 days at other settings) (18). Although legislation has prohibited smoking at work and certain public settings in some states in the 1990’s, these bans would likely generate geographic variability in ETS exposure, which was analytically desirable.

Serum Estradiol Measurements

Total 17β-estradiol (E₂) and Sex Hormone Binding Globulin (SHBG) were measured in near-annual intervals using an Automated Chemiluminescence System (ACS)-180 analyzer (Bayer Diagnostics Corporation, Tarrytown, NY) as previously described (19). During pre- and early peri-menopause, timing of the blood draw was “in window” if the blood was drawn between days 2–5 of a menstrual cycle, and “out of window” otherwise. Level of bioavailable E₂ was presented as a unitless index calculated as the ratio of total E₂ (nM) to SHBG (nM): 100 × (total E₂ (pg/mL) × [0.003671 nM per 1 pg/mL E₂]) / SHBG (nM) (20). E₂ levels were measured because of their established role in fibroid development (6). Inclusion of E₂ as a mediator in the models would isolate the direct effect of ETS, independent of an E₂ related pathway.

Potential Confounders

Covariates assessed at baseline included: age, race/ethnicity (non-Hispanic white, African-American, Hispanic, Japanese, Chinese), study site (Detroit, MI, Boston, MA, Chicago, IL, Oakland, CA, Los Angeles, CA, Newark, NJ, Pittsburgh, PA), educational attainment (no formal schooling, middle school, high school, associates/bachelors, and post-graduate), pack-years of smoking, age at first birth, parity, and age at menarche. Time-varying covariates assessed at each visit included: any exogenous sex hormone use, menopausal status (pre-menopausal, early peri-menopausal, late peri-menopausal, natural post-menopausal, surgical post-menopausal, and unknown status due to hormone use) (17, 21), measured body mass index (kg/m², BMI), active smoking status (never, former, current), timing of blood draw (“in vs. out of window”), and alcohol use (<1 serving/month, 1 serving/month – 2 servings/week, ≥2 servings/week).

Analyses

The distribution and normality of continuous variables were assessed using histograms, QQ plots, and Shapiro-Wilks tests. Between those exposed and unexposed to ETS, differences in normally distributed variables were assessed using a two-sample Student’s t-test, while non-normally distributed variables were assessed using Wilcoxon Rank Sum tests. Differences in categorical variables were evaluated using Chi-square tests.

Because data were collected in discrete time-intervals in SWAN, discrete-time proportional odds survival models were used to estimate the conditional odds ratio (OR) and 95% Confidence Intervals (CI) of late-diagnosis incident fibroids, as described by Singer et al. (1993) (22). The follow-up period was from visit 0 to 13. ETS exposure was time-varying and was defined as cumulative exposure in the past seven days before each visit. Time-varying ETS exposure was operationalized as exposed (≥1 person-hour/week) versus unexposed (0 person-hours/week) based on previous analyses in SWAN (23). Both unadjusted and multivariable-adjusted models were performed. In the adjusted models, the timescale was stratified by potential confounders including a race/ethnicity/study site combination variable and menopausal status (post- and late peri-menopause versus pre- and early peri-menopause), fitting separate baseline intercepts for each category. Additionally, the following covariates were parametrically adjusted based on causal criteria (24), previous literature, and a statistical association with fibroids of p<0.15: time-varying bioavailable E₂ (≥0.5, <0.5); time-varying exogenous sex hormone use (user vs. nonuser); age at baseline (years); time-varying BMI (<25, 25–30, ≥30 kg/m²); time-varying active smoking status (never, former, current); timing of blood draw (“in window”, “out of window”); time-varying alcohol use; educational attainment; age at menarche; and continuous pack-years of smoking up to baseline. Parity at baseline, age at first birth, and health insurance coverage at baseline were considered, but did not meet the criteria and were not included. Time-varying ETS exposure and covariates, except bioavailable E₂, were lagged by one visit. Interaction terms between ETS exposure and active smoking status were included to assess effect modification. Separate analyses were also restricted to non-Hispanic White and African-American women.

No blood draws occurred at visit 11. Women from Newark, NJ were censored after visit 6 due to administrative issues. Missing time-varying exposure and covariate data was imputed using last observation carried forward. All analyses were performed using SAS v9.4 (SAS Institute Inc., Cary, NC, USA). P-values <0.05 for main effects and global p_interactions<0.20 were considered statistically significant.

Results

Baseline Characteristics of the SWAN Study Population

At baseline, the median age was 46 years (Table 1). There were 642 cases of self-reported diagnosis of fibroids at baseline. Non-Hispanic White and African-American women were more likely to be exposed to ETS compared to Hispanic, Chinese and Japanese women. Compared to women unexposed to ETS, those exposed to ETS had a lower proportion of those who were premenopausal, a greater proportion who were early peri-menopausal. Furthermore, exposed women were more likely to be heavy drinkers, current smokers, and have higher median BMIs. Those exposed to ETS who were current or former active smokers had greater pack-years of smoking compared to unexposed women. Among the 809 former smokers at baseline, 17.0% had quit in the previous five years. Among the 556 current active smokers at baseline, 35.8% quit at least once during the follow-up.

Table 1.

Baseline Characteristics of the SWAN Study Population by Exposure to Environmental Tobacco Smoke

Variable		ETS Exposure 0 person hours/week n=1446		ETS Exposure ≥ 1 person hour/week n=1771
Self-Reported Previous Medical Diagnosis of Fibroids, n (%)		267	18.5	375	21.2	*
Age, years (median, IQR)		46.3	44.3–48.4	46.0	44.0–48.2
Bioavailable 17β-Estradiol (median, IQR)		0.49	0.30–0.87	0.52	0.32–0.86	*
Race/Ethnicity, n (%)
	Non-Hispanic White	590	40.8	913	51.6	***
	African-American	316	21.9	592	33.4
	Hispanic	178	12.3	106	6.0
	Japanese	171	11.8	105	5.9
	Chinese	191	13.2	55	3.1
Body Mass Index, kg/m2 (median, IQR)		25.5	22.2–30.3	27.7	23.7–33.4	***
Menopausal status, n (%)
	Early Peri-menopause	604	41.8	850	48.0	**
	Pre-menopause	811	56.1	876	49.5
Past sex hormone-use for any reason, n (%)		159	11.0	207	11.7
Smoking status, n (%)
	Current Smokers	122	8.4	421	23.8	***
	Former Smokers	363	25.1	446	25.2
	Never Smokers	955	66.0	886	50.0
Smoking pack-years in active and past smokers, (median, IQR)		7.5	2.7–18.0	12.5	4.0–24.0	***
ETS Exposure, person-hrs/week (median, IQR)		-	-	5	2 – 26
Parity, children (median, IQR)		2	1 – 3	2	1 – 3
Age at First Live Birth, years (median, IQR)		26	21 – 30	23	20 – 28	***
Educational Attainment, n (%)
	No formal education	141	9.8	94	5.3	***
	Middle	215	14.9	351	19.8
	High school	394	27.2	624	35.2
	Associates / Bachelors	337	23.3	309	17.4
	Post-graduate	345	23.9	377	21.3
Alcohol consumption, n (%)
	<1 serving per month, low	779	53.9	825	46.6	***
1 serving/month – 2 servings/week, moderate		390	27.0	527	29.8
	>2 servings per week, high	273	18.9	413	23.3

^*p-values <0.05 were considered significant.

^**p<0.001.

^***p<0.0001.

Discrepancies were due to missing data and rounding.

Medical Confirmation of Self-Reported Fibroid Occurrence in Women who underwent Hysterectomy

When comparing self-reported and verified gross evidence of uterine fibroids in 99 women who underwent hysterectomy, the sensitivity was 62%, the specificity was 57%, the PPV was 95%, and the NPV was 10%. When compared to histological evidence, the sensitivity was 62%, the specificity was 55%, the PPV was 92%, and the NPV was 15%.

Risk of Late-diagnosis Incident Fibroids

A total of 512 women self-reported incident fibroids in 23,069 woman-years of follow-up (Table 2). The estimated unadjusted rate of incident fibroids was 22.2 cases per 1000 woman-years. In the unadjusted analyses, exposure to ETS (≥1 person-hour/week) was significantly associated with risk of incident fibroids in the ensuing year compared to being unexposed (OR=1.29, 95% CI: 1.08, 1.55) (Table 3). Similarly, in the multivariable-adjusted analyses, ETS exposure (≥1 person-hour/week) was associated with a significant 1.28 (95% CI: 1.03, 1.60) times adjusted odds of incident fibroids in the ensuing year compared to being unexposed (Table 4). When examining the interaction between ETS exposure and active smoking, positive associations between ETS exposure and risk of incident fibroids were observed in those who never smoked (aOR: 1.34, 95% CI: 1.06, 1.70, p_interaction=0.11) and in those who formerly smoked (aOR: 2.57, 95% CI: 1.05, 7.23, p_interaction=0.11). However, the number of cases in those who formerly smoked was small.

Table 2.

Incidence Rates of Late-Diagnosis Uterine Fibroids in SWAN

Race/Ethnicity - Study Site		Number of Incident Cases	Number of Person-Years	Incidence Rate per 1000 p.y.*	95% CI Lower	95% CI Upper
Entire Study Population		512	23069	22.2	22.1	22.3
non-Hispanic White
Detroit, MI		27	1570	17.2	17.0	17.4
Boston, MA		47	2123	22.1	21.9	22.3
Chicago, IL		32	1672	19.1	18.9	19.4
Oakland, CA		22	1796	12.2	12.1	12.4
Los Angeles, CA		42	1702	24.7	24.4	24.9
New Jersey, NJ		21	496	42.3	41.8	42.9
Pittsburgh, PA		60	2244	26.7	26.5	27.0
African-American
Detroit, MI		44	1776	24.8	24.5	25.0
Boston, MA		49	1152	42.5	42.2	42.9
Chicago, IL		39	1545	25.2	25.0	25.5
Pittsburgh, PA		30	1016	29.5	29.2	29.9
Hispanic
New Jersey, NJ		31	1153	26.9	26.6	27.2
Japanese
Los Angeles, CA		39	2474	15.8	15.6	15.9
Chinese
Oakland, CA		29	2350	12.3	12.2	12.5

^*Non-standardized incidence rates of fibroids in the study population throughout a 13-year follow-up period.

Women were censored after reported death, deactivation, hysterectomy, and incident fibroids. Excludes the 646 women who self-reported medical diagnosis of fibroids prior to baseline. Person-years (p.y.).

Table 3.

Unadjusted Associations of Environmental Tobacco Smoke Exposure and Active Smoking to Risk of Late-Diagnosis Incident Fibroids in the SWAN Study Population

		a) All Women (n=2575)					b) Restricted to Non-Hispanic White (n=1255)				c) Restricted to African-Americans (n=631)
		Cases	OR	95%CI Lower	95%CI Upper		Cases	OR	95%CI Lower	95%CI Upper	Cases	OR	95%CI Lower	95%CI Upper
Environmental Tobacco Smoke Exposure
≥1 vs. 0 person-hours/week		512	1.29	1.08	1.55	*	251	1.18	0.91	1.52	162	1.12	0.80	1.57
Active Smoking
Former Smoker vs. Never		512	0.65	0.31	1.39		251	0.36	0.09	1.45	162	0.52	0.16	1.65
Current Smoker vs. Never		512	0.82	0.63	1.07		251	0.77	0.52	1.15	162	0.61	0.40	0.93	*

^*p-values < 0.05 were considered statistically significant.

Unadjusted discrete-time proportional odds models were used to obtain the conditional odds ratio (OR) of incident self-reported diagnosis of fibroids. Time-varying ETS exposure was lagged by 1 follow-up year. All analyses were restricted to women without self-reported fibroids and cancer at baseline.

Table 4.

Multivariable-Adjusted Associations Between Exposure to Environmental Tobacco Smoke and Risk of Late-Diagnosis Incident Fibroids in the SWAN Study Population

		a) All Women (n=2575)					b) Restricted to Non-Hispanic White (n=1255)				c) Restricted to African-Americans (n=631)
		Cases	OR	95%CI Lower	95%CI Upper		Cases	OR	95%CI Lower	95%CI Upper	Cases	OR	95%CI Lower	95%CI Upper
Model with Main Effects
ETS Exposure >1 vs. 0 person-hours/week		512	1.28	1.03	1.60	*	251	1.30	0.98	1.73	162	1.05	0.73	1.50
Former Smoker vs. Never			0.82	0.51	1.33			0.84	0.43	1.63		0.68	0.33	1.41
Current Smoker vs. Never			0.78	0.55	1.13			0.82	0.50	1.34		0.60	0.35	1.02
Model with ETS x Smoking Status Interactions
ETS Exposure >1 vs. 0 person-hours/week, in Never Smokers		439	1.34	1.06	1.70	¥	220	1.38	0.99	1.93	132	1.15	0.73	1.82
ETS Exposure >1 vs. 0 person-hours/week, in Former Smokers		7	2.57	1.05	7.23	¥	2	2.36	0.70	10.71	3	1.31	0.29	6.87
ETS Exposure >1 vs. 0 person-hours/week, in Current Smokers		66	0.68	0.38	1.24		29	0.97	0.37	2.83	27	0.63	0.26	1.63

^*p-values < 0.05 were considered statistically significant.

^¥p-interactions < 0.20 were considered statistically significant.

Discrete-time proportional odds models were used to obtain the conditional odds ratio (OR) of incident self-reported diagnosis of fibroids. Time-varying passive smoke exposure and covariates were lagged by 1 follow-up year. The multivariable-adjusted models controlled for time-varying bioavailable estradiol, time-varying sex hormone use for any reason, age at baseline, time-varying body mass index, time-varying smoking status (never, past, current), being within 2–5 days of menstrual cycle at blood draw, time-varying physical activity, time-varying alcohol use, highest educational attainment at baseline, age at menarche, pack years of smoking at baseline. The timescale of the models were further stratified by study site - race/ethnicity and menopausal status (post- and late-peri menopause vs. pre- and early peri-menopause), fitting separate baseline intercepts for each strata. (b) Model was restricted to non-Hispanic White women (c) Model was restricted to African-American women. Analyses were not restricted to Hispanic, Chinese, and Japanese women due to small sample sizes. All analyses were restricted to women without self-reported fibroids and cancer at baseline.

When the analyses were restricted to non-Hispanic White women, the association between ETS exposure and late-diagnosis incident fibroids was marginally non-significant (aOR=1.30, 95% CI: 0.98, 1.73) (Table 4). Among never-smoking non-Hispanic White women, the relationship was also marginally non-significant (aOR=1.38, 95% CI: 0.99, 1.93, p_interaction=0.75). No statistically significant associations were observed in African-American women; however, the associations were in the same direction.

Non-significant inverse associations were observed between current and former smoking, and risk of fibroids in both the unadjusted and adjusted analyses. Among African-American women, current smoking was significantly inversely associated with risk of fibroids in the unadjusted analyses (OR=0.61, 95% CI: 0.40, 0.93) (Table 3); however, the association became non-significant after adjusting for covariates (Table 4).

Discussion

We found that midlife women who reported ETS exposure had substantially higher odds of developing late-diagnosis incident fibroids in the ensuing year during the menopausal transition, compared to the unexposed. This finding was observed in never and former smokers, but not in current smokers. Our findings also suggest that the effect of ETS exposure may differ between former and never smokers. Women who were exposed to high levels of ETS and formerly smoked had over twice the risk of developing fibroids compared to unexposed former smokers. However, the number of cases was small in former smokers, which may have led to chance findings. Therefore, caution is recommended when interpreting the findings.

The absence of an association in current smokers may have been due to insufficient statistical power, or the potential effect of ETS may have been masked or subsumed by active smoking. Further, ETS and active smoking may be difficult to disentangle statistically due to high correlation. When restricting the analyses to African-American women, significant associations were not detected, which was unexpected given the higher prevalence and severity of fibroids at younger ages. These findings are most likely due to decreased statistical power; however, it may also be due to selection bias. African-American women who developed fibroids earlier in life and had hysterectomies would not be eligible for entry into this study of late-diagnosis fibroids. The finding may also be due to the presence of undetected fibroids, which may mask the association.

Unlike previous studies, we did not detect significant associations between active smoking status and incident fibroids in this population of midlife women (14–16). We observed a borderline non-significant, inverse association with current smoking in African-American women, which was consistent with some studies. In general, the associations between smoking and risk of fibroids have been inconsistent in the body of literature. A cross-sectional study by Dragomir et al. (2010) found that current smoking was only associated with diffuse uterine fibroids detected with ultrasound screening, but not other types of fibroids. Because diagnoses of uterine fibroids were self-reported in SWAN, we were unable to assess location and subtype of fibroids as in the study by Dragomir et al. (2010). Thus, potentially true associations between current smoking and fibroids and fibroid subtypes may have been obscured by outcome misclassification. In contrast to the study by Dragomir et al. (2010), Templeman et al. (2009) found inverse associations with active smoking (15) in a large prospective cohort study of teachers, which has been postulated to be due to smoking’s anti-estrogenic effects (25). The seemingly protective effect may also be attributed to detection bias, because active smokers might be less likely to seek medical attention. However, this potential caveat would probably bias estimates towards the null, rather than downwards away from the null. Similar to our study, no associations with active smoking were found in the prospective Nurses’ Health Study II, which consisted of mostly pre-menopausal White women (26, 27). Differences between the studies may be attributed to variations in population characteristics, accuracy of exposure and outcome assessment, study design, control of covariates and sample size. Given the inconsistencies with active smoking in the literature, the mechanism of action of ETS on fibroid development remains elusive. However, the differences in chemical composition and toxicity between ETS and directly inhaled mainstream smoke is well-established (1) and could account for differences in findings regarding the relation of ETS versus active smoking to risk of fibroids.

We did not observe significant associations between parity at baseline and risk of late-diagnosis fibroids. Previous investigations in the Black Women’s Health Study (28) and Nurses’ Health Study (26) found increased parity to be protective against fibroids. These discrepancies may be due to age-effects, as SWAN assessed late-diagnosis fibroids during the menopausal transition. However, the relationship with parity is difficult to interpret in this older cohort due to selection bias, in which women with their desired number of children may be more likely to get hysterectomies compared to women desiring pregnancy. Further, parous women have ultrasound examination during pregnancy, which would likely detect fibroids, and make them ineligible for cohort studies of fibroid incidence. However, nulliparous women are less likely to be screened and may enter cohort studies with undetected fibroids.

The most prominent strength of the present investigation was its longitudinal study design, which permitted establishment of temporality between ETS exposure, potential confounders, and incident fibroids across midlife. Furthermore, the repeated assessments of ETS exposure allowed for the long-term trajectory of ETS exposure to be evaluated in relation to the odds of incident fibroids. Additionally, SWAN is a well-powered multi-racial/ethnic cohort; therefore, our findings may be generalizable to midlife women undergoing the menopausal transition across the United States.

This investigation was not without limitations. First, the age range at entry was 45–52 years, beyond the ages at which most women develop uterine fibroids. Indeed, the prevalence of fibroids at baseline was 20%, much lower than the prevalence in the general population. Thus, findings from this study may be generalizable to older women with late-diagnosis fibroids. Further, because fibroids may have been present but undetected for years, the time sequence of ETS in relation to fibroids could not be definitively established. Second, ETS exposure and active smoking were not biochemically confirmed via urinary cotinine, which may have led to misclassification of exposure. Although the survey instrument for ETS exposure was previously validated (29), random misclassification may have occurred if the participants did not accurately recall the number of active smokers and/or hours per week of exposure. However, this would have biased the estimates toward the null, leading to conservative estimates. Third, fibroid occurrence was self-reported and only medically confirmed in a subset of women who underwent hysterectomy. The overall sensitivity in this subset was slightly higher than previous analyses that used ultrasound diagnosis, whereas the specificity was significantly lower (30). However, assuming the low sensitivity was non-differential between those exposed and unexposed to ETS, under-detection would only lead to underestimation of the true effect. Fourth, SWAN did not assess fibroid size, number, or classification. Many uterine fibroids are asymptomatic; therefore, most of the reported cases were likely to be symptomatic. Differential misclassification of the outcome may occur if higher ETS exposure affected fibroid characteristics and altered detection rates (30). Given the similarities between the toxic components of ETS and mainstream smoke, in addition to previously observed associations between smoking and heavy bleeding (31, 32), a relationship may also exist between ETS and heavy bleeding. If women exposed to ETS were more likely to visit a doctor for the complaint of heavy bleeding, they may be more likely to undergo ultrasound in which fibroids may be diagnosed, resulting in differential misclassification of the outcome and bias away from the null. Conversely, under-detecting asymptomatic women with fibroids with higher ETS exposure would likely bias the estimates downward and lead to conservative estimates. Fifth, unmeasured confounders may have potentially introduced bias. However, major potential confounders from previous literature were controlled. Lastly, the relative contribution of ETS exposure on fibroid development in current smokers may be difficult to disentangle due to high correlation between current smoking and ETS exposure, which may have masked potential associations.

Findings from our study suggest that midlife women exposed to ETS during the menopausal transition may have an increased risk of late-diagnosis incident fibroids compared to those unexposed. This risk was observed in never and former smokers, but not current smokers. Future longitudinal studies would benefit from longer follow-up times spanning the adult life course. Additionally, biochemical confirmation of ETS and smoking would improve exposure assessment and accuracy of the findings. Furthermore, regular clinical screening for fibroids throughout the follow-up or medical confirmation of self-reported fibroids would bolster the validity of the findings in longitudinal studies. In closing, ETS is a specter of cigarette smoking that has health ramifications extending beyond smokers themselves. Its seemingly detrimental effects are not limited to the respiratory system, but may also affect the reproductive health of women.

In the longitudinal Study of Women’s Health Across the Nation (SWAN), women exposed to environmental tobacco smoke (ETS) during midlife had a significantly increased risk of late-diagnosis uterine fibroids.