Abstract
Purpose:
Sexual minority women (SMW) may have a different distribution of breast cancer risk factors than their heterosexual peers. Epidemiologic studies of breast cancer in SMW have been limited, and many use only proxy variables to identify SMW in data sets, introducing selection bias. We utilized National Health Interview Survey (NHIS) data to compare breast cancer risk factors, screening, and prevalence between SMW and non-SMW.
Methods:
We identified female respondents to the NHIS from 2013 to 2017, selected women ≥40 years old and stratified by sexual orientation. We compared demographics and health maintenance variables and prevalence of breast cancer diagnosis between groups and performed a multivariable analysis of breast cancer risk.
Results:
Of 58,378 women ≥40 years old, 1162 (2.0%) were identified as SMW. SMW were younger and more likely to use tobacco and alcohol, be younger at menarche, and be nulliparous. SMW also reported less preventive care, and despite reporting equivalent rates of mammography, were more likely to obtain mammograms due to an identifiable problem and not simply for screening purposes. Prevalence of breast cancer was similar between SMW and non-SMW (4.7% vs. 5.0%, p = 0.67), and SMW status was not associated with breast cancer diagnosis on univariate and multivariable logistic regression (p = 0.14 and p = 0.07, respectively).
Conclusion:
Despite finding no difference in breast cancer prevalence between SMW and non-SMW, there was evidence for differences in the utilization of breast care. Further studies of breast cancer incidence, characteristics (including subtype and stage), treatment, and survival for SMW are needed.
Introduction
The Institute of Medicine report concluded that sexual minority women (SMW; women whose sexual identity, orientation, or practices differ from the majority of the surrounding society) experience unique and significant disparities in areas of health care from preventive services to diagnosis and treatment of disease. 1 Studies have found higher rates of substance use, 2 a poorer self-reported assessment of overall health, 2 and a higher prevalence of obesity among SMW3,4 compared with their heterosexual counterparts. In a meta-analysis containing both national and regional surveys, Cochran et al. reported that lesbian/bisexual women were insured at lower rates when compared with standardized estimates for U.S. women from national probability samples, 5 a factor that impacts many aspects of preventive care and disease treatment. In addition to differences in socioeconomic factors, there are likely differences in social determinants of health that underlie these health care disparities such as prejudice, social stigma, and discrimination. 6
Disparities also exist for SMW along the continuum of cancer care. There have recently been many attempts to assess differences in risk, screening, and incidence of breast cancer among SMW as breast cancer is the most common cancer affecting U.S. women.7,8 When compared with non-SMW women, many researchers have reported that SMW have a higher prevalence of independent breast cancer risk factors such as obesity,3,4 nulliparity,9,10 substance abuse, 11 and more prior breast biopsies. 12 Other studies have reported mixed results regarding the rates of breast cancer screening among SMW, with several describing a decrease in the likelihood to obtain such screening13–15 and others showing similarities with heterosexual women. 16 A comprehensive literature review demonstrates that the available studies found both increased and decreased prevalence of breast cancer among SMW. 17 In addition to finding mixed results within and between the articles that they reviewed, the researchers concluded that the majority of the studies were small and of poor quality, limiting their ability to make any strong conclusions about SMW and breast cancer.
Although it is clear that SMW experience health care disparities, it is difficult to investigate these disparities and their causes because many electronic health records and national databases neither capture nor report data related to sexual orientation, 18 and in those that have the ability to track these data, many are missing. As a result, many of the studies examining these health care disparities are based on data from a single institution or a clinic dedicated to the care of SMW, 19 or use proxy variables for identifying SMW,20,21 all of which limit the reliability of the conclusions and generalizability of the studies. This has led to large gaps in the literature addressing disparities in cancer care for SMW. 22
For breast cancer, specifically, multiple strategies have been employed in an attempt to assess these disparities. One study compared breast cancer risk and prevalence between SMW and their heterosexual biological sisters, 12 the results of which are likely not generalizable given the heritable nature of risk factors that ultimately lead to a breast cancer diagnosis. 23 Some groups have attempted to integrate geographic data, such as SMW population density, with adjusted breast cancer incidence.24,25 These studies demonstrate that breast cancer is diagnosed at higher rates in areas with a denser SMW population, but these methods lack the ability to detect the reason for the reported differences.
As some data suggest a higher prevalence of risk factors associated with breast cancer in SMW, the use of a national dataset for this purpose may more accurately depict the differences in breast cancer risk between SMW and their heterosexual counterparts. We hypothesized that there is a difference in the reported rates of breast cancer risk factors, screening, and prevalence between SMW and non-SMW women in a large, representative national sample.
Methods
Data source
The National Health Interview Survey (NHIS) is a cross-sectional, population-based survey of civilian, noninstitutionalized United States adults ≥18 years old conducted by the National Center for Health Statistics since 1957. One randomly selected adult per household is selected to complete a face-to-face, computer-assisted interview. Each participant completes several questionnaire modules focusing on health care services, behaviors, and health status. In addition to the core questionnaires, a Cancer Control Supplement (CCS) is administered to adult respondents every 5 years to collect data related to family history, cancer screening, and cancer risk assessment. Given that the NHIS is deidentified and publicly available, this study was exempt from Main Line Health System Institutional Review Board review.
Self-reported sexual orientation data have been collected by the NHIS since 2013. 26 Specifically, respondents are asked “How do you think of yourself?” with answer options being: “Lesbian or gay”; “Straight, that is not lesbian or gay”; “Bisexual; “Something else”; “I don't know the answer”; and “Refused or not ascertained.”
Participants
Using NHIS data sets from 2013 to 2017, including the 2015 CCS, we identified female respondents ≥40 years old with a reported sexual orientation. We elected to use these data sets, because 2013 was the first year that sexual orientation data were collected, and 2017 was the most recent data set available. We chose the age cut-off of 40 years, as this is the age at which the American Cancer Society recommends that women of average risk be given the opportunity to initiate breast cancer screening. 27 We stratified the data set by sexual orientation by grouping women who identified as “Lesbian or gay,” “Bisexual,” or “Something else” as SMW. Those who identified as “Something else” were included as SMW, as they do not identify with the sexual majority. 28 We excluded women for whom the answer was not ascertained. Women who identified as “Straight, that is not lesbian or gay” are referred to as non-SMW.
Variables
We analyzed differences in demographic and health maintenance variables between SMW and non-SMW within the dataset. Demographic, biologic, and social variables included were age, race, geographical region, marital and employment status, household income, health insurance, medical comorbidities, body mass index (BMI), and the use of tobacco and alcohol. We also analyzed variables that are used in well-known breast cancer risk models (e.g., Gail Model, 29 Tyrer-Cuzick Model 30 ) such as a personal history of breast biopsy, age at menarche and first live birth, current use of birth control and hormone replacement therapy, and family history of breast cancer to detect differences in the distribution of these risk factors between the groups. The groups were also compared based on their use of and access to preventive care and on breast-specific variables such as time since most recent clinical breast examination and mammogram, and the reasons for pursuing mammogram. Finally, the prevalence of breast cancer diagnosis and reported age at the time of breast cancer diagnosis were analyzed.
Data analyses
Data were analyzed by using SPSS Statistics for Mac, Version 24.0 (IBM Corp., Armonk, NY). Chi-square tests were used to analyze differences in categorical variables, and t-tests and one-way analysis of variance for continuous variables. Mirrored analyses were performed in each instance, comparing SMW with non-SMW as well as women within the SMW group, comparing those who identified as lesbian, bisexual, or something else. Univariate and multivariable binary logistic regressions were utilized to analyze the impact of reported risk factors on the development of breast cancer. Analyses were performed and reported by using weighted values, and p-values were taken to be significant if they were <0.05.
Results
Demographics and comorbidities
A total of 58,378 women ≥40 years old with a reported sexual orientation were identified in the NHIS data set from 2013 to 2017, 1162 (2.0%) of whom were identified as SMW. Comparisons of demographic variables are shown in Table 1; although the SMW and non-SMW groups differed on most variables, there were fewer demographic differences between the SMW subgroups. SMW were younger and a larger proportion lived in the Northeast, were never married, and were employed compared with non-SMW (all p < 0.001). Among women who reported being unemployed, SMW more often indicated health issues as the reason, whereas non-SMW more frequently indicated being retired or taking care of family. There were no differences in reported household income or health insurance between the groups.
Demographics, Medical Comorbidities, and Social History of Sexual Minority Women and Non Sexual Minority Women in the Full National Health Interview Survey 2013–2017 Cohort
Statistical comparisons were performed by using chi-square, Student's t test, and ANOVA between aSMW and non-SMW and between bSMW subgroups (“Lesbian or gay,” “Bisexual,” and “Something else”).
ANOVA, analysis of variance; SD, standard deviation; SMW, sexual minority women.
Related to medical comorbidities, SMW had similar reported rates of hyperlipidemia, diabetes, coronary artery disease, cerebral vascular accident, and chronic obstructive pulmonary disease, but lower reported rates of hypertension (39.0% vs. 46.3%, p < 0.001) compared with non-SMW. SMW had a higher mean BMI (28.8 vs. 27.8, p < 0.001), and a higher proportion of women qualifying as obese (34.5% vs. 27.8%, p < 0.001) compared with non-SMW. SMW also had higher rates of tobacco and alcohol use than non-SMW, with more than half reporting being current or former smokers (53.6% vs. 38.7%, p < 0.001) and the majority reporting being current or former users of alcohol (85.3% vs. 74.0%, p < 0.001).
Personal and family history
Table 2 summarizes reported rates of personal and family history factors that are important to breast cancer. There was no difference in the rate of previous breast biopsy between groups. SMW were, however, twice as likely to have early menarche (≤11 years of age) compared with non-SMW (8.6% vs. 4.0%, p = 0.003). In addition, SMW women were much more likely to report being nulliparous (55.6% vs. 15.9%), and of those who had given birth, SMW were less likely to report being 30 years of age or older at the time of their first live birth (7.8% vs. 10.2%, p < 0.001). Although there were no differences in the rates of birth control use, SMW had lower reported rates of hormone replacement therapy use than non-SMW (9.5% vs. 14.9%, p = 0.03). Reported rates of first-degree relatives with breast cancer were similar between groups. SMW were more likely than non-SMW to report breast cancer diagnoses in second-degree relatives (26.3% vs. 21.9%, p = 0.04). There were no differences in the reported rates of breast cancer in first- and second-degree relatives when the SMW subgroups were compared.
Reported Personal and Family History of Sexual Minority Women and Non Sexual Minority Women in the 2015 National Health Interview Survey Cancer Control Supplement Cohort
Statistical comparisons were performed by using chi-square, Student's t test, and ANOVA between aSMW and non-SMW and between bSMW subgroups (“Lesbian or gay,” “Bisexual,” and “Something else”).
Preventive care and breast cancer screening
Comparisons of the rates of reported preventive care and breast cancer screening (clinical breast examination and mammogram) are summarized in Table 3. SMW were slightly more likely than non-SMW to report not receiving preventive care (4.9% vs. 3.8%, p = 0.001), and SMW who did receive preventive care were more likely to report receiving it in “some other place” than a clinic or doctor's office (2.7% vs. 1.3%, p = 0.004) compared with non-SMW. SMW were also more likely to report having trouble finding a doctor or provider in the year before survey administration (5.4% vs. 3.0%, p < 0.001). In addition, when asked about the last time they were seen by a health professional, SMW were more likely to report having not been seen for a year or more (9.5% vs. 7.8%, p = 0.03).
Comparisons of Reported Rates of Preventive Care, Clinical Breast Examination, and Mammography Between Sexual Minority Women and Non Sexual Minority Women in the Full National Health Interview Survey 2013–2017 Cohort and the 2015 National Health Interview Survey Cancer Control Supplement Cohort
Statistical comparisons were performed by using chi-square, Student's t test, and ANOVA between aSMW and non-SMW and between bSMW subgroups (“Lesbian or gay,” “Bisexual,” and “Something else”). Data are from the 2015 NHIS Cancer Control Supplement cohort or from the full NHIS 2013–2017 cohortc.
There were no differences between SMW and non-SMW in reported rates of and time since last clinical breast examination. Similarly, there were no differences between the two groups in reported rates of ever having a mammogram and time since last mammogram. Only approximately half of all women reported having a mammogram within the last year (54.1% in the entire cohort and 55.1% in the 2015 CCS cohort), with 13% of women having not undergone a mammogram in more than 5 years. Although they obtained mammograms at similar rates, there were differences between groups in the reasons for undergoing their most recent mammogram. SMW were more likely to report pursuing mammography because of a problem and not for a routine screening (10.4% vs. 4.8%, p = 0.001) compared with non-SMW and more likely to report that a doctor recommended a mammogram (69.0% vs. 57.7%, p = 0.001). After mammography, SMW were also more likely to report having been told they had dense breasts on mammogram (29.2% vs. 20.4%, p = 0.003), and require additional tests such as repeated imaging or biopsy as a result of their mammogram (14.4% vs. 8.8%, p = 0.005) compared with non-SMW.
Breast cancer diagnosis
Prevalence of breast cancer diagnosis was similar between the SMW and non-SMW groups (4.7% vs. 5.0%, p = 0.67; Table 4), although SMW who reported having been diagnosed with breast cancer were diagnosed at a younger age than non-SMW (51.5 years vs. 56.5 years, p = 0.008). There were no statistically significant differences in breast cancer diagnosis or the age of breast cancer diagnosis between SMW subgroups.
Reported Prevalence of Breast Cancer Diagnosis and Age at Breast Cancer Diagnosis in the Full National Health Interview Survey 2013–2017 Cohort
Statistical comparisons were performed by using chi-square, Student's t test, and ANOVA between aSMW and non-SMW and between bSMW subgroups (“Lesbian or gay,” “Bisexual,” and “Something else”).
On univariate analysis, White race, older age, early menarche, first live birth at age 30 or older, use of hormone replacement therapy, dense breasts, previous benign breast biopsy, and first-degree relative with breast cancer were all associated with reported breast cancer diagnosis (Table 5, all p < 0.05). BMI and reported use of alcohol and tobacco had no statistically significant association with breast cancer diagnosis. On multivariable analysis, older age, dense breasts, previous benign breast biopsy, first-degree relative with breast cancer, and the use of hormone replacement therapy remained significantly associated with reported breast cancer diagnosis (p < 0.05). SMW status was not significantly associated with reported breast cancer diagnosis on univariate analysis (p = 0.14), nor when included in the multivariable regression model (p = 0.07).
Univariate and Multivariable Analyses of the Odds of Developing Breast Cancer Based on Reported Breast Cancer Risk Factors
Unless specified, all reference categories are the converse of the reported category (e.g., nondense breasts is the reference category for dense breasts). Alcohol and tobacco use categories combine current and former use.
Continuous variable.
CI, confidence interval; OR, odds ratio.
Discussion
In our analysis of the NHIS, we corroborated the findings of previous studies in identifying a higher prevalence of independent breast cancer risk factors among SMW such as obesity, early menarche, nulliparity, and substance abuse.5,19,31,32 In the NHIS cohort, we found no difference between SMW and non-SMW in reported rates of obtaining mammograms. SMW were, however, less likely to report undergoing simple screening mammography and more likely to report “a problem” (such as feeling a lump or breast distortion) and a doctor's recommendation to undergo mammography. Although Bazzi et al. report that bisexual women undergo screening mammography at a lower rate than lesbian and heterosexual women, 33 there was no difference within the SMW group in this cohort. Similar to analyses of other data sets,8,17 we did not find differences in the prevalence of breast cancer diagnosis when comparing SMW and non-SMW women and identifying as a SMW was not a risk factor for developing breast cancer in univariate and multivariable analyses.
Interestingly, our findings may demonstrate that SMW wait to seek care until they detect a problem on their own, and forego routine screening due to stigma within the health care arena. We did observe that SMW women reported less engagement with the health care system, with a higher proportion of these women indicating that they do not receive preventive care, have not seen a health professional in a year or more, and have had trouble finding a doctor or provider in the 12 months before the survey. Although we were unable to determine the reasons for these findings, it is worth noting that there were no differences in income or reported rates of health insurance between SMW and non-SMW. This indicates that the reasons for disengagement may be social rather than financial.
Previous studies from the Boston Lesbian Health Project noted an increase in use of primary care and mammography among lesbian women in the 1980s and 1990s, although the rate of increase was lower than expected.34,35 As Baptiste-Roberts et al. reported, sexual minority populations experience health disparities that may be caused, in part, by stigma and prejudice within the health care environment. 6 Implicit bias against sexual minority individuals by health care providers decreases the effectiveness of encounters with patients, and it likely impacts the rates at which these individuals seek preventive care,36,37 which ultimately may impact breast cancer screening rates.
In this analysis, however, we did not find consistent differences between SMW and non-SMW in factors that would demonstrate stigma and health care disparities for SMW. For example, despite reporting more difficulty finding a provider, SMW reported undergoing clinical breast examination and mammography at similar rates and time intervals when compared with non-SMW. These similarities may reflect that, when compared with prior decades, health care stigma is reducing for the majority of SMW.
Strengths and limitations
The strength of using the NHIS data for this analysis is in the large, representative national study population, but there are also significant limitations. First, as with any large database, the data are not annotated, and we were therefore unable to analyze the reasons for our observations at a more granular level. Second, as the NHIS relies on interviews, recall bias is likely to play a factor in either over- or under-reporting certain factors. There is not any reason to believe, however, that recall bias affects SMW any differently than non-SMW.
The issue of nondisclosure likely means that there are women included in the non-SMW group who truly identify as SMW but were not comfortable disclosing their SMW identity during survey administration. In addition, although the NHIS allows identification of SMW through one survey question about sexual orientation, the database does not allow for the identification of transgender individuals who have important and unique risks for breast cancer, including exposure to exogenous hormones and lack of screening due to gender stigmatization.8,33 Next, although the NHIS is a large database, we grouped all SMW for this analysis to improve our statistical power, and we did not investigate differences (such as those between races and ethnicities 38 ) in this heterogenous group. Finally, women who have died from breast cancer or other causes are not able to participate in the NHIS. In addition to precluding an assessment of breast cancer incidence, this may obscure true differences in how breast cancer affects SMW and non-SMW if members of one group are diagnosed more often with late-stage disease, leading to increased mortality and under-representation in survey data.
Conclusion
Despite differences in breast cancer risk factors between SMW and non-SMW, we found no difference in breast cancer prevalence. Although many of these risk factors are biological, others are affected by socioeconomic determinants of health. Taken together, reports in the literature and our findings underscore the need for continued work in addressing health care disparities for sexual minority individuals in general, and for SMW and breast health specifically. Both the American Cancer Society and the American Society of Clinical Oncology have recognized this need and have developed resources aimed at tackling cancer disparities in the LGBT community.39,40 The Healthy People 2020 campaign, sponsored by the Centers for Disease Control and Prevention, also has articulated goals and objectives for addressing health care disparities affecting sexual minority individuals, including in breast cancer screening, detection, and treatment. 41 Further studies of breast cancer incidence, characteristics (including subtype and stage), treatment, and survival for SMW are needed. In the meantime, primary care and specialty providers should pay special attention to creating welcoming health care environments in which disparities among sexual minority individuals can be recognized, barriers to health care access can be broken down, and equitable health care can be delivered.
Disclaimer
Portions of these data were presented at the 36th GLMA Annual Conference on LGBTQ Health, Las Vegas, Nevada, October 10–13, 2018.
Footnotes
Author Disclosure Statement
No competing financial interests exist.
Funding Information
No funding was received for this article.