Abstract
This cross-sectional study uses Behavioral Risk Factor Surveillance System data to examine the association of gender identity with self-reported lifetime prevalence of skin cancer.
Gender identity refers to one’s personal sense of gender and includes identifying as cisgender (ie, a gender identity that aligns with the sex assigned at birth), transgender (ie, a gender identity that does not align with the sex assigned at birth), and gender nonconforming (GNC) (ie, a gender identity that does not follow others’ ideas about how a person should look or act based on the sex assigned at birth). There has been increasing national focus on examining cancer risks of transgender and gender-nonconforming (TGNC) patients1 and, although prior research has examined skin cancer prevalence among sexual minority populations,2 this study is the first, to our knowledge, to evaluate skin cancer history by gender identity.
Methods
This cross-sectional study analyzed data from annual Behavioral Risk Factor Surveillance System (BRFSS) surveys of adults aged 18 years and older from 2014 through 2018. Self-reported sex and skin cancer history were ascertained as described elsewhere.3 Gender identity was ascertained through BRFSS’s optional sexual orientation and gender identity module, which was introduced in 2014 and has since been administered in 37 US states. Respondents self-identified as cisgender, transgender (male to female or female to male), or GNC. We stratified cisgender respondents by sex and excluded respondents who answered “don’t know” or “not sure” or refused to answer any questions regarding sex, gender identity, or skin cancer history. This study was deemed exempt from review by the Partners Healthcare institutional review board because it used publicly available data. Accordingly, informed consent procedures were not required.
We weighted prevalence estimates to account for BRFSS’s complex survey design. We performed univariate regression analyses to determine associations between gender identity (Table 1) and other covariates (eTable in the Supplement) with skin cancer history. We performed multivariable logistic regression analyses to calculate adjusted odds ratios of skin cancer history by gender identity using R software, version 3.5.1 (R Foundation for Statistical Computing). A P value of .05 was used for statistical significance.
Table 1. Age-Adjusted Lifetime Prevalences, Univariate Odds Ratios, and Adjusted Odds Ratios of Skin Cancer History by Gender Identity.
| Characteristic | Age-Adjusted Lifetime Prevalence, % (95% CI)a | Univariate Odds Ratio (95% CI) | P Value | Adjusted Odds Ratio (95% CI)b | P Value |
|---|---|---|---|---|---|
| Cisgender | |||||
| Men | 6.6 (6.5-6.8) | 1 [Reference] | NA | 1 [Reference] | NA |
| Women | 6.4 (6.2-6.5) | 0.96 (0.93-0.99) | .02 | 0.85 (0.82-0.88) | <.001 |
| Transgender | |||||
| Men | 6.0 (4.1-8.8) | 0.87 (0.51-1.47) | .60 | 1.12 (0.66-1.92) | .67 |
| Women | 5.8 (3.5-9.5) | 0.90 (0.60-1.36) | .63 | 1.19 (0.73-1.93) | .48 |
| Gender-nonconforming individuals | 7.1 (4.1-11.9) | 1.08 (0.61-1.90) | .80 | 2.11 (1.01-4.39) | .046 |
Abbreviation: NA, not applicable.
Age-adjusted prevalence was calculated using direct standardization with cisgender male respondents from the weighted sample as the standard population group.
Adjusted for age, geographic region, race/ethnicity, education level, employment status, insurance status, current alcohol consumption, smoking history, and history of another cancer diagnosis.
Results
The unweighted study sample included 368 197 cisgender men, 492 345 cisgender women, 1214 transgender men, 1675 transgender women, and 766 GNC individuals (Table 2). Age-adjusted lifetime prevalence of skin cancer was 6.6% (95% CI, 6.5%-6.8%) among cisgender men, 6.4% (95% CI, 6.2%-6.5%) among cisgender women, 6.0% (95% CI, 4.1%-8.8%) among transgender men, 5.8% (95% CI, 3.5%-9.5%) among transgender women, and 7.1% (95% CI: 4.1%-11.9%) among GNC individuals (Table 1). Compared with cisgender men, adjusted odds ratios of skin cancer history were significantly lower among cisgender women (0.85 [95% CI, 0.82-0.88]), higher among GNC individuals (2.11 [95% CI, 1.01-4.39]) (Table 1), and not significantly different among transgender men or transgender women.
Table 2. Characteristics of Respondents, Stratified by Gender Identity.
| Characteristic | Individuals, % (95% CI) | P Valuea | ||||
|---|---|---|---|---|---|---|
| Cisgender | Transgender | Gender-Nonconforming Individuals | ||||
| Men | Women | Men | Women | |||
| Unweighted, No.b | 368 197 | 492 345 | 1214 | 1675 | 766 | NA |
| Weighted, No. | 76 276 501 | 83 960 441 | 274 578 | 355 057 | 168 428 | NA |
| Age, mean (95% CI), y | 47.3 (47.2-47.4) | 49.2 (49.1-49.3) | 43.0 (40.7-45.3) | 45.9 (44.2-47.6) | 40.2 (37.6-45.3) | NA |
| Region | ||||||
| Northeast | 16.9 (16.7-17.0) | 17.2 (17.0-17.4) | 13.7 (10.5-16.9) | 16.5 (13.4-19.7) | 21.4 (16.1-26.7) | <.001 |
| South | 39.1 (38.8-39.4) | 39.9 (39.6-40.2) | 51.3 (45.1-57.6) | 46.6 (41.5-51.7) | 37.9 (30.5-45.3) | |
| Midwest | 21.2 (21-21.3) | 20.9 (20.8-21.1) | 20.1 (15.7-24.5) | 21.7 (18.3-25) | 17.5 (13.1-21.8) | |
| West | 22.9 (22.6-23.2) | 22.0 (21.7-22.3) | 14.9 (9.3-20.5) | 15.2 (11.2-19.2) | 23.3 (15.6-30.9) | |
| Race/ethnicity | ||||||
| Non-Hispanic white | 64.9 (64.5-65.3) | 64.4 (64.0-64.7) | 59.1 (52.7-65.5) | 55.0 (49.8-60.2) | 56.3 (48.7-63.9) | <.001 |
| Non-Hispanic black | 10.8 (10.6-11.1) | 12.4 (12.1-12.6) | 16.7 (11.6-21.8) | 15.3 (11.4-19.1) | 16.4 (9.4-23.4) | |
| Hispanic | 8.1 (7.9-8.4) | 7.7 (7.4-7.9) | 5.8 (3-8.5) | 11.1 (7.7-14.4) | 8.4 (5.2-11.5) | |
| Otherc | 16.1 (15.8-16.5) | 15.6 (15.3-15.9) | 18.5 (12.9-24) | 18.7 (13.9-23.4) | 18.9 (13-24.8) | |
| Education level | ||||||
| High school or less | 43.4 (43-43.7) | 39.8 (39.5-40.2) | 53.7 (47.5-59.9) | 60.7 (55.9-65.5) | 45.2 (37.8-52.6) | <.001 |
| Some college or more | 56.6 (56.3-57) | 60.2 (59.8-60.5) | 46.3 (40.1-52.5) | 39.3 (34.5-44.1) | 54.8 (47.4-62.2) | |
| Employment status | ||||||
| Unemployed | 35.4 (35.1-35.8) | 50.4 (50.0-50.7) | 46.2 (40.0-52.4) | 48.5 (43.5-53.5) | 58.0 (51.0-65.0) | <.001 |
| Employed | 64.6 (64.2-64.9) | 49.6 (49.3-50) | 53.8 (47.6-60) | 51.5 (46.5-56.5) | 42.0 (35.0-49.0) | |
| Insurance status | ||||||
| No | 12.7 (12.5-13) | 10.1 (9.9-10.4) | 16.2 (11.8-20.6) | 19.0 (14.5-23.6) | 14.6 (9.5-19.7) | <.001 |
| Yes | 87.3 (87-87.5) | 89.9 (89.6-90.1) | 83.8 (79.4-88.2) | 81.0 (76.4-85.5) | 85.4 (80.3-90.5) | |
| Smoking history | ||||||
| Never | 53.2 (52.9-53.6) | 64.3 (64.0-64.6) | 61.3 (55.2-67.4) | 56.6 (51.7-61.5) | 65.1 (58.6-71.7) | <.001 |
| Former | 28.7 (28.4-29.1) | 21.6 (21.3-21.8) | 18.7 (14-23.4) | 24.2 (20.2-28.2) | 17.3 (12.8-21.8) | |
| Current | 18.0 (17.7-18.3) | 14.1 (13.9-14.3) | 20.0 (14.9-25.2) | 19.2 (15.6-22.8) | 17.6 (12.4-22.8) | |
| Alcohol used | ||||||
| None | 41.0 (40.6-41.3) | 53.0 (52.7-53.4) | 60.8 (54.8-66.7) | 57.0 (52.1-61.8) | 48.2 (40.7-55.6) | <.001 |
| Light | 43.2 (42.8-43.6) | 41.4 (41.1-41.8) | 30.9 (25.4-36.4) | 33.3 (28.7-37.9) | 39.2 (31.5-46.9) | |
| Moderate or heavy | 15.9 (15.6-16.1) | 5.6 (5.4-5.7) | 8.3 (5.1-11.6) | 9.8 (7.1-12.4) | 12.6 (8.1-17.1) | |
| Other cancer diagnosis | ||||||
| No | 94.4 (94.2-94.5) | 91.6 (91.4-91.7) | 90.3 (85.4-95.2) | 93.2 (90.9-95.4) | 93.5 (89.8-97.2) | <.001 |
| Yes | 5.6 (5.5-5.8) | 8.4 (8.3-8.6) | 9.7 (4.8-14.6) | 6.8 (4.6-9.1) | 6.5 (2.8-10.2) | |
Abbreviation: NA, not applicable.
Analysis of variance was used to determine the P value for age, and χ2 tests were used to determine the remaining P values.
The unweighted sample excluded those with missing data for the exposure of interest (gender identity) and outcome of interest (skin cancer), which was 100 374 individuals (10.0%). We also excluded those with missing data for any covariate (35 083 individuals [3.9%]).
Other race/ethnicity was defined as all respondents who self-identified as “American Indian or Alaska Native only, non-Hispanic,” “Asian only, non-Hispanic,” “Native Hawaiian or other Pacific Islander only, non-Hispanic,” “other race only, non-Hispanic,” or “multiracial, non-Hispanic.”
Nondrinking was defined as 0 alcoholic beverages per week, light alcohol use was defined as 1 to 7 drinks per week, and moderate or heavy alcohol use was defined as more than 7 alcoholic beverages per week.
Discussion
Compared with cisgender men, GNC individuals but not transgender men or women had a higher self-reported lifetime prevalence of skin cancer. While the reasons for this finding are unclear and require further evaluation, our leading hypothesis would be increased engagement in skin cancer risk behaviors. Cisgender women had significantly lower odds of lifetime prevalence of skin cancer than cisgender men, which is consistent with existing data on gender differences in skin cancer in the United States.4
These results are subject to 4 distinct limitations. First, because BRFSS’s sexual orientation and gender identity module was not administered in every state during each year, the findings may not be generalizable to the entire US population. Second, the study relied on self-reported skin cancer diagnoses and did not include data on risk factors, such as ultraviolet exposure, use of hormone replacement therapy,5 and HIV status. Third, the absolute event rate in TGNC populations was low compared with that of cisgender groups, yielding wide 95% CIs. Finally, although the lifetime prevalence estimates of skin cancer among transgender men and women were not different from that of cisgender men, disproportionately lower engagement in cancer screening behaviors and overall health care utilization6 by TGNC individuals compared with cisgender people may lead to fewer skin cancer diagnoses among TGNC individuals.
Additional studies are required to confirm these findings. Future research may identify avenues for public health intervention.
eTable. Univariate Odds Ratios for Lifetime Diagnosis of Skin Cancer Diagnosis Among Respondents Who Provided Gender Identity.
References
- 1.Burkhalter JE, Margolies L, Sigurdsson HO, et al. . The national LGBT cancer action plan: a white paper of the 2014 National Summit on Cancer in the LGBT Communities. LGBT Health. 2016;3(1):19-31. doi: 10.1089/lgbt.2015.0118 [DOI] [Google Scholar]
- 2.Mansh M, Katz KA, Linos E, Chren M-M, Arron S. Association of skin cancer and indoor tanning in sexual minority men and women. JAMA Dermatol. 2015;151(12):1308-1316. doi: 10.1001/jamadermatol.2015.3126 [DOI] [PubMed] [Google Scholar]
- 3.Singer S, Tkachenko E, Hartman R, Mostaghimi A.. Lifetime prevalence of skin cancer diagnosis among gay, bisexual, and heterosexual individuals in the United States [published online February 5, 2020]. JAMA Dermatology. doi: 10.1001/jamadermatol.2019.4197 [DOI] [Google Scholar]
- 4.Holman DM, Freeman MB, Shoemaker ML. Trends in melanoma incidence among non-Hispanic whites in the United States, 2005 to 2014. JAMA Dermatol. 2018;154(3):361-362. doi: 10.1001/jamadermatol.2017.5541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cervenka I, Mahamat-Saleh Y, Dartois L, Boutron-Ruault M-C, Fournier A, Kvaskoff M. Exogenous hormone use and cutaneous melanoma risk in women: the European prospective investigation into cancer and nutrition. Rev Epidemiol Sante Publique. 2018;66:S253. doi: 10.1016/j.respe.2018.05.051 [DOI] [Google Scholar]
- 6.James SE, Herman JL, Rankin S, Keisling M, Mottet L, Anafi M; National Center for Transgender Equality The report of the 2015 U.S. transgender survey. https://transequality.org/sites/default/files/docs/usts/USTS-Full-Report-Dec17.pdf. Published 2016. Accessed December 19, 2019.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. Univariate Odds Ratios for Lifetime Diagnosis of Skin Cancer Diagnosis Among Respondents Who Provided Gender Identity.