Objectives. We examined national and state-specific disparities in health insurance coverage, specifically employer-sponsored insurance (ESI) coverage, for adults in same-sex relationships.

Methods. We used data from the American Community Survey to identify adults (aged 25–64 years) in same-sex relationships (n = 31 947), married opposite-sex relationships (n = 3 060 711), and unmarried opposite-sex relationships (n = 259 147). We estimated multinomial logistic regression models and state-specific relative differences in ESI coverage with predictive margins.

Results. Men and women in same-sex relationships were less likely to have ESI than were their married counterparts in opposite-sex relationships. We found ESI disparities among adults in same-sex relationships in every region, but we found the largest ESI gaps for men in the South and for women in the Midwest. ESI disparities were narrower in states that had extended legal same-sex marriage, civil unions, and broad domestic partnerships.

Conclusions. Men and women in same-sex relationships experience disparities in health insurance coverage across the country, but residing in a state that recognizes legal same-sex marriage, civil unions, or broad domestic partnerships may improve access to ESI for same-sex spouses and domestic partners.

There are approximately 646 000 same-sex couples in the United States according to the 2010 decennial census.1 Same-sex couples reside in every state, but each state has its own laws and regulations regarding the legal status of same-sex marriage. At the time of this writing, 16 states and the District of Columbia had recognized legal marriages for same-sex couples; an additional 3 states had extended civil unions or comprehensive domestic partnerships to same-sex couples; and the remaining states had banned same-sex marriage altogether through legislative action or amendments to their state constitutions.2 Differences in same-sex marriage laws can affect access to health insurance for same-sex couples or members of the lesbian, gay, bisexual, and transgender (LGBT) population. When states adopt same-sex marriage or civil unions that extend spousal rights and protections to same-sex couples, fully insured private employers regulated by state insurance laws are often required to treat married same-sex couples as married opposite-sex couples.

The Employee Retirement Income Security Act of 1974 limits the reach of state insurance regulation. Although states maintain jurisdiction over fully insured health plans, employers that self-insure—or assume the risk of health claims out of their own assets—are regulated under the federal Employee Retirement Income Security Act, as health benefits are treated not as insurance but as an employee benefit similar to employer-provided pension plans.3,4 In 2010, more than half of all workers (57.5%) with employer-sponsored insurance (ESI) were covered by self-insured plans.5 Because so many workers are covered by self-insured plans, state-level marriage policies can have a limited effect. Buchmueller and Carpenter, using data from the 2001–2007 California Health Interview Surveys, found that insurance mandates that extended health care benefits to same-sex spouses in California had no statistically significant effect on dependent coverage for gay and bisexual men and only a small positive effect on lesbian and bisexual women.6

The federal Defense of Marriage Act, passed in 1996, created additional barriers for LGBT workers interested in adding their spouses to their ESI plan, even when states acknowledged the legality of same-sex marriage. Section 3 of the Defense of Marriage Act (ruled unconstitutional by the US Supreme Court in 2013) defined marriage as “a legal union between one man and one woman as husband and wife” for federal purposes.7 The federal government does not tax employer contributions to an opposite-sex spouse’s health benefits, but under the Defense of Marriage Act, a same-sex partner’s health benefits were taxed as if the employer contribution was taxable income. LGBT employees paid, on average, $1069 in additional federal income taxes when they added their same-sex spouses to employer health plans.8 These barriers to ESI may have led LGBT persons to enroll in public programs or forgo health insurance and access to affordable health care.

Data on the LGBT population have historically been limited to convenience and nonprobability samples of gay men and lesbians through health care providers and researchers focusing their research on LGBT health.9 Although federal surveys do not ascertain sexual orientation, data have been edited to identify same-sex couples and households. Three previous studies have used intrahousehold information from federal population surveys to compare the health insurance coverage of individuals in same-sex relationships with that of those in opposite-sex relationships.

Heck et al. used the National Health Interview Survey to compare health insurance coverage and access to medical care of adults in same-sex relationships with that of married adults in opposite-sex relationships.10 They used multivariate logistic regression models for men and women and found women in same-sex relationships significantly less likely to have health insurance, to have seen a medical provider in the previous 12 months, and to have a usual source of care. Health insurance coverage, unmet medical needs, and having a usual source of care were not statistically different between men in same-sex relationships and married men in opposite-sex relationships. The authors believed the HIV epidemic motivated gay men to maintain a regular provider. Compared with the other studies using federal surveys, the National Health Interview Survey accommodates the smallest sample size (316 men and 298 women in same-sex relationships)—even after pooling data across a wide time frame (1997–2003).

Ash and Badgett took advantage of larger samples in the Current Population Survey.11 Designed to measure labor force participation and unemployment, the Annual Social and Economic Supplement to the Current Population Survey requires respondents to report health insurance coverage during the previous 16 months for each person in the household. Pooled data between 1996 and 2003 still produced relatively small sample sizes (486 men and 478 women in same-sex relationships), but their study found that both men and women in same-sex couples were 2 to 3 times more likely to be uninsured than were married individuals in opposite-sex relationships.

Buchmueller and Carpenter used a national sample of adults aged between 25 and 64 years in the Behavioral Risk Factor Surveillance System to compare health insurance and utilization of health services of same-sex couples with those of opposite-sex couples (both married and unmarried).12 Again, both men and women in same-sex relationships were significantly less likely to be insured. Married people in opposite-sex relationships had the highest rates and odds of insurance coverage, followed by men and women in same-sex relationships, and then by unmarried men and women in opposite-sex relationships. Although it provides the largest sample to date (2384 men and 2881 women in same-sex relationships), their study pooled data across a wide period (2000–2007) of decline in health insurance coverage, especially for people with ESI.13

These 3 studies were restricted to national-level estimates and surveys with limited sample sizes. Our research builds on the previous work but extends the analysis to all states. To our knowledge, only 1 other study has estimated health insurance disparities for same-sex couples in a single state using the California Health Interview Study.14 Because of the variation in state policies and attitudes toward same-sex couples,15,16 we expected geographic patterns in health insurance. We took advantage of relatively large samples in the American Community Survey (ACS) to compare state-specific health insurance disparities, particularly in ESI coverage. Following recent studies examining the potential for same-sex marriage to improve the health of the LGBT population,17–20 we sought to add early evidence on the relationship between legal same-sex marriage and health insurance coverage.

We examined data from the 2008–2010 ACS 3-year public use microdata sample. The ACS is a general household survey conducted by the US Census Bureau, and it is designed to provide states and communities with reliable and timely demographic, social, economic, and housing information. Replacing the decennial census long-form questionnaire in 2005, the ACS has an annual sample size of about 3 million housing units and a monthly sample size of about 250 000 households. The large samples available in the ACS make it a powerful source for studying relatively small subpopulations, such as same-sex couples, at the state level.21

Like most federal surveys, the ACS did not ascertain sexual orientation. Instead, same-sex couples were identified on the basis of intrahousehold relationships and were assumed to be lesbian, gay, or bisexual persons. Identification strategies cannot ascertain transgender status because of the binary male–female categories of gender identify included in the survey. Lesbian, gay, and bisexual persons were identified when the primary respondent identified another person of the same sex as a husband, wife, or unmarried partner. The Census Bureau edited same-sex spouses as unmarried partners using the husband or wife response categories in the public use files regardless of their legal marriage status.22 Meanwhile, the instruction guide accompanying the survey defines an unmarried partner as a “domestic partner” or “person who shares a close and personal relationship with the reference person.”23(p4) A survey of 602 individuals in same-sex relationships indicated that same-sex couples used both responses depending on the nature and legal status of their relationship when asked identical relationship questions in the 2010 decennial census.24

A question regarding health insurance coverage was added to the ACS in 2008 and requires the respondent to report current health insurance coverage for all members of the household. We used hierarchical assignment to designate a single source of health insurance coverage for each individual, although respondents were able to report multiple sources of coverage. If multiple sources of coverage were reported for a respondent, we assigned the primary source of insurance in the following order:

    Medicare;

    ESI, Tricare, or other military health care, or Veterans Affairs (including for those who have ever enrolled in or used Veterans Affairs health care);

    Medicaid, Medical Assistance, or any kind of government assistance plan for those with low incomes or a disability; and

    Insurance purchased directly from an insurance company.

We classified individuals reporting no source of coverage or only Indian Health Services as uninsured.25 Insurance information was available for both primary respondents and their partners, so the unit of analysis was the individual.

We first examined ESI disparities at the national level for same-sex couples in comparison with married and unmarried opposite-sex couples. We used the following multinomial logistic regression model on the entire sample to control for factors associated with health insurance coverage:

where insurance was 1 of the 4 primary sources of insurance coverage (ESI, directly purchased insurance, Medicaid, and Medicare; uninsured was the reference category) and relationship indexed the type of relationship (same-sex or unmarried opposite-sex; married opposite-sex was the reference category). X was the vector of control variables that included age group (25–34, 35–44, and 45–54 years; 55–64 years was the reference group), race and ethnicity (Hispanic, non-Hispanic Black, non-Hispanic Asian, and non-Hispanic other and multiple races; non-Hispanic White was the reference group), educational attainment (less than high school, high school, and some college; college degree was the reference group), couple’s combined income relative to the federal poverty guidelines ([FPG] ≤ 100%, 101%–200%, 201%–300%, and 301%–400%; > 400% was the reference group), employment status (part-time employment, unemployed, and not in labor force; full-time employment was the reference group), industry of employment (agriculture, mining, construction, manufacturing, wholesale trade, retail trade, transportation, utilities, information, finance, professional, education and health, arts, other, unknown, or not in labor force during previous 5 years; government and military was the reference group), region (Midwest, South, and West; Northeast was the reference group), citizenship (naturalized and noncitizen; citizen was the reference group), the presence of a biological, adopted, or stepchild younger than 18 years in the household, and survey year. Consistent with previous work,12,14 our sample was restricted to adults aged between 25 and 64 years to account for the completion of educational attainment and Medicare coverage starting at age 65 years. We estimated our models separately for men and women, first for the entire sample and then restricted to employed adults to isolate any bias in estimating public insurance enrollment attributable to disability or unemployment. We have reported the relative risk ratios (RRRs) for primary source of coverage using married opposite-sex couples as the reference group.

Our second objective was to estimate state-specific disparities in ESI coverage and to identify how they differ across regions and state marriage policies. We computed unadjusted, or observed, relative differences (RDs) in mean ESI coverage between adults in same-sex relationships and married adults in opposite-sex relationships and tested for statistical significance using a 2-tailed test. We calculated adjusted state-level estimates using predictive margins, or recycled probabilities, from logistic regression models on ESI coverage.26 This procedure allows one to compare mean rates and adjusted RDs in ESI coverage across states while controlling for the noted variables likely to influence insurance coverage. We used RDs in this analysis because they are more intuitive than are odds ratios and are frequently utilized in the health disparities literature, such as the National Healthcare Disparities Report.27

Finally, we estimated unadjusted and adjusted RDs between same-sex couples and married opposite-sex couples on the basis of the legal status of same-sex marriage, civil unions, and broad domestic partnerships available in each respondent’s state according to data from the National Conference of State Legislatures.2 We assigned each respondent 1 of 3 categories: (1) same-sex marriage was present in the respondent’s state during the entire survey year: California (2009, 2010), Connecticut (2009, 2010), Iowa (2010), Massachusetts (2008, 2009, 2010), New Hampshire (2010), and Vermont (2010); (2) civil unions or domestic partnerships extending broad spousal rights were present in the respondent’s state during the entire survey year: California (2008), Connecticut (2008), New Hampshire (2008, 2009), New Jersey (2008, 2009, 2010), Oregon (2009, 2010), Washington (2010), and Vermont (2008, 2009); and (3) neither same-sex marriage, nor civil unions, nor broad domestic partnerships were present in the respondent’s state during the entire survey year (remaining state–year combinations). This last category included states with limited domestic partnerships and registries that only provided some spousal rights to same-sex couples or explicitly protected private employers from having to extend health benefits to domestic partners (Maine, Nevada, Wisconsin, and District of Columbia).2

Because the date when the survey was completed was not available in the ACS public use microdata sample, we classified respondents on the basis of the marriage policy present in their state the entire year. If a state implemented same-sex marriage early in the year or midyear, we did not consider respondents in that state to reside in a policy environment with same-sex marriage until the following year.

Our final sample size included 15 529 men and 16 418 women in same-sex relationships, making this, to our knowledge, the largest analysis of insurance coverage among same-sex couples and the first to compare health insurance disparities across all states. We conducted all regression models and coverage estimates using Stata, version 12,28 with survey weights. We calculated SEs using Taylor linearized series. We measured adjusted estimates on the basis of predictive margins using Stata’s MARGINS command.

Men and women in same-sex relationships exhibited characteristics that inform predictions of their access to insurance, especially ESI (Table 1). Both men and women in same-sex relationships reported equal or higher levels of income and education, whereas men and women in unmarried opposite-sex relationships reported the lowest levels of income, education, and employment. Forty-eight percent of men in same-sex relationships had a college degree compared with 34% of married men and 18% of unmarried men in opposite-sex relationships. Men in same-sex relationships and married men in opposite-sex relationships reported the highest levels of full-time employment (71% and 77%, respectively). Men in same-sex relationships had the highest income levels of any group. Seventy percent of men in same-sex relationships were members of a couple that earned more than 400% of the FPG, which were $43 320 for an individual and $88 200 for a family of 4 in 2010. Unmarried men in opposite-sex relationships tended to be younger, more racially and ethnically diverse, and more likely to have less than a high school education.

Table

TABLE 1— Sample Descriptive Statistics by Relationship Type: American Community Survey, 2008–2010

TABLE 1— Sample Descriptive Statistics by Relationship Type: American Community Survey, 2008–2010

Men
Women
Same-Sex Relationship, Weighted Mean, %Opposite-Sex Unmarried, Weighted Mean, %Opposite-Sex Married, Weighted Mean, %Same-Sex Relationship, Weighted Mean, %Opposite-Sex Unmarried, Weighted Mean, %Opposite-Sex Married, Weighted Mean, %
Age, y
 25–3419.743.017.721.744.620.3
 35–4432.026.527.028.325.827.1
 45–5431.620.029.931.920.329.4
 55–6416.810.425.418.29.323.2
Race/ethnicity
 Non-Hispanic White76.762.670.377.064.570.8
 Non-Hispanic Black5.413.97.77.411.27.1
 Non-Hispanic Asian3.61.95.72.22.96.6
 Non-Hispanic other/multiple races2.22.91.82.73.11.9
 Hispanic12.018.614.510.618.313.6
Education
 < high school5.817.811.85.714.79.8
 High school diploma or GED16.334.125.916.928.525.1
 Some college or vocational school30.529.928.830.334.131.4
 College degree47.518.233.647.122.733.6
Couple’s combined income relative to FPG
 < 1004.313.08.06.312.78.4
 100–2007.420.013.89.819.513.5
 201–3009.317.515.011.417.014.8
 301–4008.913.914.412.913.714.3
 >40070.235.648.959.637.149.1
Employment
 Full time70.668.477.267.055.147.1
 Part time10.810.97.414.217.119.6
 Unemployed5.09.94.84.57.33.9
 Not in labor force13.610.810.714.420.429.5
Children < 18 y in household12.440.350.224.639.648.3
Region
 Northeast20.719.617.921.719.818.0
 Midwest17.523.022.419.022.622.4
 South33.032.736.631.732.836.6
 West28.924.723.227.624.923.1
Citizenship
 Citizen88.585.280.592.885.080.7
 Naturalized5.73.78.83.94.59.0
 Noncitizen5.811.210.73.310.410.2
Industry
 Public administration/military4.53.96.76.53.63.9
 Agriculture0.61.81.90.50.70.6
 Mining0.30.80.90.20.10.1
 Construction3.716.912.32.71.51.4
 Manufacturing6.913.915.87.67.46.1
 Wholesale trade2.33.63.91.81.91.7
 Retail trade10.69.58.28.811.68.6
 Transportation3.75.86.23.02.21.8
 Utilities0.50.91.50.50.30.4
 Information3.92.32.32.72.01.6
 Finance9.44.35.86.66.97.0
 Professional14.011.011.111.09.88.1
 Education/health18.96.910.030.424.930.9
 Arts8.68.64.66.511.25.4
 Other5.14.33.94.34.64.4
 Unknown or not in labor force past 5 y7.25.65.17.211.318.0
Primary source of health insurance
 Employer70.951.274.572.150.474.1
 Direct purchase8.35.16.16.75.36.9
 Medicaid3.55.93.34.413.14.0
 Medicare3.52.52.73.22.92.3
 Uninsured13.835.313.513.628.412.7
Sample size15 529133 3471 491 38416 418125 8001 569 327

Note. FPG = federal poverty guidelines; GED = general equivalency diploma. Weighted means are for adults aged 25–64 years.

Women in same-sex relationships also reported higher levels of education, income, and full-time employment. Like their male counterparts, women in same-sex relationships had high incomes: 60% of women in same-sex relationships were part of a couple earning more than 400% of the FPG, compared with 49% of married women and 37% of unmarried women in opposite-sex relationships. Almost 70% of women in same-sex relationships were employed full time, a much higher figure than the 47% of women in married opposite-sex relationships.

Similar to previous studies,12,14 we found that adults in married opposite-sex relationships reported the highest levels of having a child in the household, whereas men and women in same-sex relationships reported the fewest. Although men and women in same-sex relationships were more educated, were more likely to work full time, and had higher incomes, they were covered by an employer health plan less frequently than were their married counterparts but more often than their unmarried counterparts in opposite-sex relationships.

Adjusted RRRs for insurance coverage among adults (aged 25–64 years) in same-sex and unmarried opposite-sex relationships are presented in Table 2; married adults in opposite-sex relationships are the reference group. After controlling for demographic and socioeconomic characteristics, we found that men in same-sex relationships were less likely than were men in married opposite-sex relationships to have health insurance through an employer (RRR = 0.50; 95% confidence interval [CI] = 0.46, 0.54), directly purchased from an insurance company (RRR = 0.69; 95% CI = 0.62, 0.76), or through Medicare (RRR = 0.80; 95% CI = 0.70, 0.92) and more likely to have insurance through Medicaid or a government assistance program (RRR = 1.32; 95% CI = 1.14, 1.54). Unmarried men in opposite-sex relationships were far less likely than were men in same-sex relationships and married opposite-sex relationships to have insurance through an employer (RRR = 0.30; 95% CI = 0.30, 0.31), directly purchased from an insurance company (RRR = 0.42; 95% CI = 0.40, 0.43), Medicaid (RR = 0.57; 95% CI = 0.55, 0.59), and Medicare (RRR = 0.44; 95% CI = 0.42, 0.47). When the sample was restricted to employed men, coverage patterns—except for Medicare—were similar in magnitude and direction. Notably, even employed men in same-sex relationships were significantly more likely to maintain coverage through Medicaid (RRR = 1.31; 95% CI = 1.03, 1.66).

Table

TABLE 2— Multinomial Logistic Regression Analysis of Health Insurance Coverage by Relationship Type and Sex: American Community Survey, 2008–2010

TABLE 2— Multinomial Logistic Regression Analysis of Health Insurance Coverage by Relationship Type and Sex: American Community Survey, 2008–2010

Adjusted RRR (95% CI)
Employer vs UninsuredDirect Purchase vs UninsuredMedicaid vs UninsuredMedicare vs Uninsured
Men
All
 In a same-sex relationship0.50** (0.46, 0.54)0.69** (0.62, 0.76)1.32** (1.14, 1.54)0.80** (0.70, 0.92)
 In an unmarried, opposite-sex relationship0.30** (0.30, 0.31)0.42** (0.40, 0.43)0.57** (0.55, 0.59)0.44** (0.42, 0.47)
 In a married, opposite-sex relationship (Ref)1.001.001.001.00
Employed
 In a same-sex relationship0.58** (0.53, 0.63)0.75** (0.66, 0.84)1.31* (1.03, 1.66)1.04 (0.79, 1.37)
 In an unmarried, opposite-sex relationship0.34** (0.33, 0.34)0.44** (0.42, 0.46)0.49** (0.47, 0.52)0.48** (0.43, 0.54)
 In a married, opposite-sex relationship (Ref)1.001.001.001.00
Women
All
 In a same-sex relationship0.45** (0.42, 0.48)0.59** (0.53, 0.66)1.19** (1.05, 1.34)1.48** (1.31, 1.67)
 In an unmarried, opposite-sex relationship0.25** (0.25, 0.26)0.40** (0.39, 0.42)1.49** (1.45, 1.54)1.02 (0.97, 1.07)
 In a married, opposite-sex relationship (Ref)1.001.001.001.00
Employed
 In a same-sex relationship0.49** (0.45, 0.54)0.64** (0.57, 0.72)0.90 (0.75, 1.07)1.09 (0.84, 1.42)
 In an unmarried, opposite-sex relationship0.33** (0.32, 0.34)0.46** (0.44, 0.48)1.27** (1.22, 1.33)0.68** (0.61, 0.77)
 In a married, opposite-sex relationship (Ref)1.001.001.001.00

Note. CI = confidence interval; RRR = relative risk ratio. All models control for age group, income group, educational attainment, region, employment status, citizenship status, industry group, the presence of a related child younger than 18 years in the household, and survey year.

*P < .05; **P < .01.

It is important to note that Medicaid includes Medical Assistance and any kind of government assistance plan for those with low incomes or a disability. Some men in same-sex relationships—who may be HIV-positive—may consider the comprehensive primary care received through the Ryan White HIV/AIDS Program when selecting this category.

Women in same-sex relationships were also less likely to have insurance through an employer (RRR = 0.45; 95% CI = 0.42, 0.48) and directly purchased from an insurance company (RRR = 0.59; 95% CI = 0.53, 0.66), but they were more likely to have insurance through Medicaid (RRR = 1.19; 95% CI = 1.05, 1.34) and Medicare (RRR = 1.48; 95% CI = 1.31, 1.67). Like their unmarried male counterparts, unmarried women in opposite-sex relationships were far less likely to have insurance through an employer (RRR = 0.25; 95% CI = 0.25, 0.26) and directly purchased from an insurer (RRR = 0.40; 95% CI = 0.39, 0.42). Both women in same-sex relationships and unmarried women in opposite-sex relationships were more likely to have coverage through Medicaid. When the sample was restricted to employed women, statistical significance for Medicaid coverage diminished for women in same-sex relationships (RRR = 0.90; 95% CI = 0.75, 1.07) but not for unmarried women in opposite-sex relationships (RRR = 1.27; 95% CI = 1.22, 1.33).

State-Specific Disparities in Employer-Sponsored Insurance

The state-specific RDs, shown in Table 3, demonstrated substantial variation in ESI disparities among nonelderly adults in same-sex relationships. Unadjusted ESI coverage rates indicated that men and women in same-sex relationships were less likely to have insurance through an employer nationally, but the observed insurance gaps ranged from modest disparities in the Midwest (RD = −6.6; 95% CI = −8.9, −4.3 for men; RD = −9.3; 95% CI = −11.6, −7.1 for women) to marginal advantages in the Western United States (RD = 0.1; 95% CI = −1.6, 1.7 for men; RD = 3.1; 95% CI = 1.5, 4.8 for women). In California, mean ESI coverage rates for men and women in same-sex relationships were, respectively, 3.3% and 8.9% higher than were those for their counterparts in married opposite-sex relationships. Meanwhile, mean ESI coverage rates for men in Vermont, Hawaii, New Mexico, South Carolina, and Wisconsin and for women in Michigan and Alabama were 15 percentage points lower than ESI coverage for married adults in opposite-sex relationships.

Table

TABLE 3— Relative Differences in Employer-Sponsored Insurance Coverage Between Same-Sex Couples and Married Opposite-Sex Couples: American Community Survey, 2008–2010

TABLE 3— Relative Differences in Employer-Sponsored Insurance Coverage Between Same-Sex Couples and Married Opposite-Sex Couples: American Community Survey, 2008–2010

Men
Women
Unadjusted RD, % (95% CI)Adjusted RD, % (95% CI)Unadjusted RD, % (95% CI)Adjusted RD, % (95% CI)
Northeast−3.2** (–5.1, −1.3)−6.3** (−8.4, −4.2)−1.6 (−3.5, 0.3)−7.8** (−9.6, −5.9)
 Connecticut−0.1 (−7.7, 7.4)−1.2 (−9.6, 7.3)−5.2 (−12.6, 2.2)−7.0 (−15.3, 1.3)
 Maine−6.4 (−19, 6.3)−8.5 (−19.2, 2.1)−7.4 (−18.3, 3.6)−12.1* (−21.9, −2.3)
 Massachusetts−4.4a (−8.8, 0.0)−6.8** (−11.5, −2.2)−2.5 (−5.8, 0.8)−8.4** (−11.8, −5)
 New Hampshire−10.3 (−23.0, 2.5)−5.7 (−13.2, 1.7)3.2 (−3.5, 9.9)1.1 (−7.3, 9.4)
 New Jersey−2.8 (−7.2, 1.7)−5.6* (−9.7, −1.5)−2.0 (−7.5, 3.4)−8.7** (−13.2, −4.1)
 New York0.4 (−2.6, 3.3)−6.0** (−8.9, −3.1)1.1 (−2.2, 4.4)−7.4** (−10.4, −4.4)
 Pennsylvania−5.1a (−10.1, −0.1)−6.7* (−11.8, −1.6)−3.7 (−9.2, 1.8)−6.9** (−11.4, −2.3)
 Rhode Island−11.5 (−25.2, 2.2)−12.2* (−21.7, −2.6)−3.5 (−13.6, 6.6)−9.7a (−18.8, −0.6)
 Vermont−18.8* (−35.4, −2.2)−12.0a (−24, −0.1)−6.8 (−19.9, 6.4)−11.4a (−23.1, 0.3)
Midwest–6.6** (−8.9, −4.3)−6.7** (–8.7, −4.7)–9.3** (−11.6, −7.1)–11.7** (−13.5, −9.9)
 Illinois−3.1 (−7.1, 0.9)−6.1** (−9.6, −2.6)−10.4** (−15.8, −5.1)−12.6** (−16.3, −9)
 Indiana−9.1* (−16.4, −1.7)−8.0* (−14.5, −1.6)−10.5** (−16.9, −4.1)−12.0** (−16.3, −7.6)
 Iowa−8.9 (−23.7, 5.8)−4.6 (−11.9, 2.7)−2.4 (−13.4, 8.5)−10.9* (−20.1, −1.8)
 Kansas−9.4 (−27.0, 8.2)−10.5a (−21.1, 0.1)1.2 (−6.4, 8.8)−4.9 (−11.2, 1.3)
 Michigan−1.7 (−7.1, 3.8)−4.5a (–9.3, 0.3)−16.0** (−22.9, −9.1)−14.3** (−19.1, −9.5)
 Minnesota−9.6* (−17.4, −1.8)−9.9** (−15.6, −4.2)−2.5 (−9.7, 4.7)−7.3* (−13.0, −1.7)
 Missouri−7.0 (−15.2, 1.2)−7.3* (−13.7, −0.9)−13.2** (−21.1, −5.2)−14.5** (−19.8, −9.3)
 Nebraska−17.6b (−43.5, 8.2)2.3b (−11.0, 15.6)−11.4 (−24.4, 1.7)−10.0 (−23.7, 3.8)
 North Dakota−4.8b (−36.7, 27.1)−0.6b (−28.0, 26.8)−11.8b (−36.0, 12.4)−12.5b (−27.4, 2.4)
 Ohio−5.7* (−10.4, −1.0)−4.5a (–9.0, −0.1)−8.6** (−12.9, −4.3)−12.1** (−16.1, −8.2)
 South Dakota−0.4b (−30.8, 30.1)0.8b (−18.9, 20.6)17.0*,b (5.5, 28.4)20.3 **,b (11.4, 29.2)
 Wisconsin−15.6** (−24.7, −6.5)−12.0** (−19, −4.9)−7.6a (−14.8, −0.4)−12.6** (−18.3, −6.8)
South−4.5** (-6.0, −3.0)−7.0** (−8.5, −5.5)−2.1* (−3.6, −0.5)−8.1** (−9.5, −6.7)
 Alabama−10.3 (−22.4, 1.8)−6.1 (−15.4, 3.2)−15.4** (−25.5, −5.2)−11.7** (−18.2, −5.2)
 Arkansas6.1 (−4.3, 16.5)6.1 (−3.5, 15.7)−6.4 (−18.7, 5.9)−2.7 (−13.2, 7.8)
 Delaware−10.8 (−22.3, 0.8)−10.9a (−21.7, −0.1)4.0 (−6.3, 14.3)2.5 (−11.5, 16.4)
 District of Columbia−1.7 (−6.8, 3.4)−8.2* (−13.1, −3.3)−1.7 (−10.4, 7.1)−10.4* (−18.1, −2.8)
 Florida−4.7** (−7.9, −1.5)−7.0** (−9.6, −4.4)0.3 (−3.0, 3.6)−9.3** (−12.1, −6.4)
 Georgia−4.3a (−8.7, 0.0)−6.8** (−10.8, −2.9)−3.7 (−8.5, 1.2)−9.1** (−13.4, −4.8)
 Kentucky−7.0 (−15.1, 1.2)−7.3 (−15.5, 0.8)−3.0 (−11.2, 5.1)−2.3 (−9.9, 5.3)
 Louisiana−12.0* (−20.9, −3.1)−10.1** (−16.0, −4.1)−9.2a (–18.6, 0.3)−10.2** (−16.8, −3.6)
 Maryland−8.7* (−15.9, −1.6)−8.8** (−14.7, −3)2.8 (−1.4, 6.9)−3.2 (−7.5, 1.2)
 Mississippi−13.8a (−26.6, −1.0)−11.4* (−21.3, −1.4)−5.6 (−18.1, 7.0)−9.7 (−21.6, 2.1)
 North Carolina−1.4 (−6.9, 4.1)−2.1 (−6.3, 2.0)−2.8 (−7.9, 2.3)−7.0** (−11.2, −2.8)
 Oklahoma−8.2 (−19.9, 3.5)−12.9** (−21.5, −4.2)−0.9 (−9.8, 8.0)−6.3a (−12.7, 0.1)
 South Carolina−15.8** (−24.6, −7.0)−10.6** (−17.5, −3.7)−7.6 (−16.8, 1.6)−5.8 (−13.0, 1.4)
 Tennessee−5.2 (−11.4, 0.9)−8.7** (−14.0, −3.3)−9.3* (−16.1, −2.5)−10.0** (−14.7, −5.3)
 Texas1.2 (−2.2, 4.7)−6.9** (−9.8, −4.1)1.7 (−1.6, 5.0)−9.9** (−12.7, −7.1)
 Virginia−8.7* (−15.5, −1.9)−6.8** (−11.3, −2.3)−1.6 (−6.6, 3.4)−5.4* (−10.0, −0.7)
 West Virginia−12.0 (−30.0, 6.1)−19.9** (−33.0, −6.8)−2.5 (−18.9, 14.0)−7.9 (−17.9, 2.0)
West0.1 (−1.6, 1.7)−6.7** (−8.5, −5.0)3.1** (1.5, 4.8)−7.6** (−9.3, −6.0)
 Alaska6.5b (−6.4, 19.5)−2.2b (−17.4, 12.9)0.5 (−12.4, 13.4)−11.0 (−23.6, 1.6)
 Arizona−6.1* (−12.3, 0.1)−10.2** (−14.9, −5.4)−1.2 (−7.3, 4.9)−11.3** (−16.1, −6.5)
 California3.3** (1.2, 5.4)−6.8** (−8.8, −4.8)8.9** (6.8, 11.0)−5.0** (−7.3, −2.8)
 Colorado0.2 (−6.3, 6.8)−4.7 (−10.7, 1.2)−3.6 (−9.5, 2.3)−11.9** (−16.8, −7.1)
 Hawaii−16.9* (−32.5, −1.3)−23.1** (−39.5, −6.7)−12.2 (−26.9, 2.6)−12.2a (−24.9, 0.5)
 Idaho10.6b (−6.7, 27.8)3.4b (−12.9, 19.8)0.7 (−10, 11.4)−1.8 (−9.9, 6.2)
 Montana−4.7b (–37.0, 27.5)−0.7b (−27.6, 26.2)10.0 (−1.0, 20.9)0.1 (−13.4, 13.6)
 Nevada−8.5a (−16.6, −0.4)−12.0** (−17.7, −6.2)−5.7 (−17.4, 6.1)−11.2* (−19.9, −2.5)
 New Mexico−16.1* (−27.4, −4.8)−14.3** (−23.8, −4.8)−6.7 (−16.0, 2.7)−13.5** (−20.3, −6.7)
 Oregon−0.9 (−7.3, 5.6)−4.5 (−9.9, 0.9)0.6 (−5.5, 6.7)−4.1 (−9.3, 1.1)
 Utah3.9 (−4.9, 12.6)−2.9 (−10.9, 5.1)−2.7 (−14.4, 9.1)−13.5** (−22.5, −4.6)
 Washington−0.2 (−5.9, 5.5)−2.1 (−7.1, 2.9)−2.0 (−6.9, 3.0)−8.9** (−13.2, −4.7)
 Wyoming−56.0**,b (−78.5, −33.5)−30.8**,b (−45.3, −16.2)−2.2b (−25.7, 21.2)−12.6b (−30.1, 4.9)
United States–3.5** (−4.4, −2.6)–6.7** (−7.9, −5.5)–2.0** (−2.9, −1.1)–8.5** (−9.5, −7.6)

Note. CI = confidence interval; RD = relative difference. Adjusted models control for age group, income group, educational attainment, region, employment status, citizenship status, industry group, the presence of a related child younger than 18 years in the household, and survey year.

aRD approached significance at P < .1

bFewer than 30 adults in our sample of same-sex couples.

*P < .05; **P < .01.

Adjusted RDs in ESI allow a comparison of state-specific disparities as if each state shared the same demographic and socioeconomic characteristics, yet many of the ESI disparities persevered or were increased. We found the largest adjusted RDs among men in same-sex relationships in the South (RD = −7.0; 95% CI = −8.5, −5.5) and among women in same-sex relationships in the Midwest (RD = −11.7; 95% CI = −13.5, −9.9). Adjusted RDs in ESI coverage also varied across the states. The adjusted ESI coverage gaps between men in same-sex relationships and married opposite-sex relationships were greater than 10% and statistically significant at the 95% CI level in 12 states. No states reported statistically significant advantages for men in same-sex relationships. Women in same-sex relationships also faced ESI disparities in every region. Moreover, adjusted differences in ESI were wider for women in same-sex relationships than they were for men in 30 states. The adjusted ESI coverage gaps between women in same-sex relationships and women in married opposite-sex relationships exceeded 10% in 16 states.

Same-Sex Marriage, Civil Unions, and Employer-Sponsored Insurance

Table 4 presents unadjusted and adjusted RDs between same-sex couples and married opposite-sex couples on the basis of the legal status of same-sex marriage or civil unions. Observed (unadjusted) ESI disparities were negligible for men (RD = 0.0; 95% CI = −2.23, 2.23) but were positive and favorable for women in same-sex relationships (RD = 6.19; 95% CI = 4.23, 8.15) in states that recognized legal same-sex marriage. Unadjusted differences in ESI were not statistically significant in states that offered civil unions or broad domestic partnerships. Mean ESI coverage, meanwhile, was approximately 4% lower for men and women in same-sex relationships living in states without broad marriage provisions or with same-sex marriage bans.

Table

TABLE 4— Relative Differences in Employer-Sponsored Insurance Coverage Between Same-Sex Couples and Married Opposite-Sex Couples by State Marriage Policy: American Community Survey, 2008–2010

TABLE 4— Relative Differences in Employer-Sponsored Insurance Coverage Between Same-Sex Couples and Married Opposite-Sex Couples by State Marriage Policy: American Community Survey, 2008–2010

Men
Women
Unadjusted RD, % (95% CI)Adjusted RD, % (95% CI)Unadjusted RD, % (95% CI)Adjusted RD, % (95% CI)
Same-sex marriage0.00 (−2.23, 2.23)−6.30** (−8.84, −3.84)6.19** (4.23, 8.15)−5.76** (−8.14, −3.38)
Civil unions or broad domestic partnerships0.29 (−2.12, 2.70)−5.20** (−7.86, −2.58)2.35 (−0.16, 4.86)−6.05** (−8.81, −3.28)
Same-sex marriage bans or no provisions−4.66** (−5.74, −3.58)−7.00** (−8.28, −5.66)−4.07** (−5.13, −3.01)−9.43** (−10.51, −8.35)

Note. CI = confidence interval; RD = relative difference. Adjusted models control for age group, income group, educational attainment, employment status, citizenship status, industry group, the presence of a related child younger than 18 years in the household, region, and survey year.

**P < .01.

After adjusting for economic and demographic factors, differences in ESI coverage between same-sex couples and married opposite-sex couples grew wider, but the differences were smaller in states that offered legal same-sex marriage or civil unions than in states without these provisions. Men in same-sex relationships experienced the narrowest gaps in ESI in states with civil unions or broad domestic partnerships (RD = −5.20; 95% CI = −7.86, −2.58), whereas women in same-sex relationships experienced the narrowest gaps in states with legal same-sex marriage (RD = −5.76; 95% CI = −8.14, −3.38). The largest disparities in ESI were still found in states with same-sex marriage bans or no comprehensive marriage provisions for both men (RD = −7.00; 95% CI = −8.28, −5.66) and women (RD = −9.43; 95% CI = −10.51, −8.35).

Men and women in same-sex relationships enjoy higher income and education levels. Yet, when we control for education and income, among other factors, we find that they do not enjoy the same levels of ESI. Furthermore, there is a significant relationship between access to ESI and the legality of same-sex marriage. ESI coverage gaps are smaller in states that recognize same-sex marriage—particularly for women. Without the legal status of same-sex marriage and civil unions, LGBT workers face barriers to adding their partners to their health plans. Thus, same-sex marriage remains an imperative health policy issue and part of the public policy goal of expanding access to health care through employer health plans.17

Although 16 states and the District of Columbia have adopted marriage equality laws, 35 states continue to limit the rights of same-sex couples similar to the federal Defense of Marriage Act.2 Yet, many private firms, typically large and self-insured employers, are ahead of state policies and voluntarily extend health benefits to same-sex couples. According to the 2012 Employer Health Benefits Survey sponsored by the Kaiser Family Foundation and the Health Research and Educational Trust, almost half (42%) of large employers with 200 or more employees offer health benefits to same-sex domestic partners. Among all employers surveyed, 31% offered health benefits to same-sex domestic partners in 2012—up from 22% in 2008.29 Further research should investigate the economic and health effects of extending employer benefits to same-sex couples.

Interestingly, couples who are less likely to have ESI are unmarried men and women in opposite-sex relationships. These couples are less likely than are their married counterparts and those in same-sex relationships to have insurance through an employer. In this regard, the lower levels of income and education do translate into lower levels of health insurance coverage, but as the category of people refraining from marriage grows, so too will the rates of the uninsured among working age adults.

Study Limitations

There were several limitations to using data from the ACS for this study. Foremost, researchers and demographers are concerned with data quality when using intrahousehold and relationship information to identify same-sex couples. Misreporting gender among married opposite-sex couples, although uncommon, unintentionally includes heterosexuals as false positives among our same-sex partners.30 The computer-assisted telephone and personal interview versions of the ACS verify the gender of the husband, wife, and unmarried partner if it matches the primary respondent’s gender. After restricting our sample to the respondents using the computer-assisted telephone and personal interview versions of the ACS, we estimated RRRs similar in direction, magnitude, and significance to the results presented in Table 2. Additionally, our identification strategy may be missing some same-sex couples. We knew only each person’s relationship to the primary respondent, so our analyses excluded same-sex couples unrelated to the primary respondent or same-sex partners that were identified as a roommate, relative, or nonrelative.

Some of our results, particularly in Table 4, may be affected by composition and selection bias. State marriage policies may influence how respondents identify their same-sex partners. We found smaller ESI disparities in states with same-sex marriage; however, this may not be explained by the policies themselves but as a response to who reports their same-sex relationship status. Indeed, our sample of same-sex couples living in states with same-sex marriage or civil unions had higher incomes and more education. In addition, selection into a partnership may explain the sensitivity to covariates in our adjusted results, as partnered LGBT adults report elevated socioeconomic measures compared with nonpartnered LGBT persons.31 Very high levels of income and education may explain why same-sex couples remain disadvantaged in our adjusted models because they continue to exhibit lower levels of ESI than their high socioeconomic status predicts. Finally, lesbian women are also more likely to be partnered and report being in state-sanctioned partnerships than are gay men,31,32 which may explain why we found favorable unadjusted RDs and narrower adjusted RDs among women living in states with same-sex marriage (Table 4).

Our study would have benefited from additional information missing in the ACS. For instance, our method of identifying same-sex couples cannot verify the sexual orientation or the transgender identity of our sample, so bisexual and transgender people were missing from our analysis if they were in an opposite-sex relationship. Knowing sexual orientation would have also assisted the analysis of nonpartnered LGBT adults. Furthermore, we do not know the legal status of the same-sex couple’s relationship; we cannot decipher whether the same-sex couple is legally married, in a state-sanctioned civil union or domestic partnership, or unmarried cohabitating partners. The Census Bureau reassigns same-sex couples identified as husband or wife to unmarried partners without providing edit flags in the public use files. Withholding reassignment flags for these edits in the public use files prevents researchers from examining differences between unmarried same-sex couples and married same-sex spouses.

Our study would have also benefited from additional variables related to ESI coverage. The size of firm an individual works for and the individual’s health status contribute to whether an employee is offered health insurance and eventually enrolls in an ESI plan. We attempted to complete the lack of information on employment by including the respondent’s employment industry in our models. Notwithstanding these limitations, we leveraged the large sample of adults in same-sex relationships in the ACS to document disparities in health insurance coverage for same-sex couples across the country, and we did so by combining relatively few years of data. The ACS is the largest survey conducted in the United States—second to the decennial census—and permits demographers at the Census Bureau to regularly and reliably tabulate and describe same-sex households.1

Conclusions

Same-sex couples reside in every state but face various marriage discrimination laws. Although health insurance disparities among same-sex couples have been well documented, we used one of the largest national surveys to estimate differences in health insurance coverage, particularly ESI, across the United States. We found that men in the South and women in the Midwest in same-sex relationships experienced the largest ESI disparities between 2008 and 2010. We also found that ESI disparities were narrower for same-sex couples living in states that had adopted legal same-sex marriage, civil unions, and broad domestic partnerships. This finding contributes to the growing body of evidence that same-sex marriage confers health benefits to sexual minorities, including expansions in employer-sponsored health insurance.

Acknowledgments

This work was funded in part by a grant from the Robert Wood Johnson Foundation to the State Health Access Data Assistance Center, University of Minnesota, School of Public Health, Division of Health Policy & Management (grant 65902).

We thank the conference participants at the 2012 AcademyHealth Gender Interest Group meeting and the 2012 American Public Health Association Annual Meeting for their useful comments.

Human Participant Protection

No protocol approval was needed for this study because the data were obtained from secondary sources.

References

1. Lofquist D, Lugailia T, O’Connell M, Feliz S. Households and Families: 2010. Washington, DC: US Census Bureau; 2012. 2010 Census Briefs C2010BR-14. Google Scholar
2. National Conference of State Legislatures. Defining marriage: Defense of Marriage Acts and same-sex marriage laws. Available at: http://www.ncsl.org/issues-research/human-services/same-sex-marriage-overview.aspx. Accessed January 11, 2013. Google Scholar
3. Badgett MVL. The economic value of marriage for same-sex couples. Drake Law Rev. 2010;102(58):10811116. Google Scholar
4. Employee Retirement Income Security Act, 29 USC §1002 (1974). Google Scholar
5. Crimmel BL. Self-Insured Coverage in Employer-Sponsored Health Insurance for the Private Sector, 2000 and 2010. Rockville, MD: Agency for Healthcare Research and Quality; 2011. Statistical Brief 339. Google Scholar
6. Buchmueller TC, Carpenter CS. The effect of requiring private employers to extend health benefit eligibility to same-sex partners of employees: evidence from California. J Policy Anal Manage. 2012;31(2):388403. CrossrefGoogle Scholar
7. Defense of Marriage Act, Pub. L. 104-199, 1 USC §7 and 28 USC §1738C (1996). Google Scholar
8. Badgett MVL. Unequal taxes on equal benefits: the taxation of domestic partner benefits. 2007. Available at: http://williamsinstitute.law.ucla.edu/wp-content/uploads/Badgett-UnequalTaxesOnEqualBenefits-Dec-2007.pdf. Accessed January 16, 2013. Google Scholar
9. Institute of Medicine. The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better Understanding. Washington, DC: National Academies Press; 2011. Google Scholar
10. Heck JE, Sell RL, Sheinfield Gorin S. Health care access among individuals involved in same-sex relationships. Am J Public Health. 2006;96(6):11111118. LinkGoogle Scholar
11. Ash M, Badgett L. Separate and unequal: the effect of unequal access to employment-based health insurance on gay, lesbian, and bisexual people. Contemp Econ Policy. 2006;24:582599. CrossrefGoogle Scholar
12. Buchmueller T, Carpenter CS. Disparities in health insurance coverage, access, and outcomes for individuals in same-sex versus different-sex relationships, 2000–2007. Am J Public Health. 2010;100(3):489495. LinkGoogle Scholar
13. Holahan J, Cook A. The US economy and changes in health insurance coverage, 2000–2006. Health Aff (Millwood). 2008;27(2):w135w144. Crossref, MedlineGoogle Scholar
14. Ponce NA, Cochran SD, Pizer JC, Mays VM. The effects of unequal access to health insurance for same-sex couples in California. Health Aff (Millwood). 2010;29(8):15391548. Crossref, MedlineGoogle Scholar
15. Lax JR, Phillips JH. Gay rights in the states: public opinion and policy responsiveness. Am Polit Sci Rev. 2009;103(3):367386. CrossrefGoogle Scholar
16. Lupia A, Krupnikov Y, Levine AS, Piston S, Von Hagen-Jamar A. Why state constitutions differ in their treatment of same-sex marriage. J Polit. 2010;72(4):12221235. CrossrefGoogle Scholar
17. Buffie WC. Public health implications of same-sex marriage. Am J Public Health. 2011;101(6):986990. LinkGoogle Scholar
18. Lau H, Strohm CQ. The effects of legally recognizing same-sex unions on health and well-being. Law Inequal. 2011;29:107148. Google Scholar
19. Hatzenbuehler ML, O’Cleirigh C, Grasso C, Mayer K, Safren S, Bradford J. Effect of same-sex marriage laws on health care use and expenditures in sexual minority men: a quasi-natural experiment. Am J Public Health. 2012;102(2):285291. LinkGoogle Scholar
20. Wight RG, LeBlanc AJ, Badgett MVL. Same-sex legal marriage and psychological well-being: findings from the California Health Interview Survey. Am J Public Health. 2013;103(2):339346. LinkGoogle Scholar
21. Lofquist D. Same-Sex Couple Households. Washington, DC: US Census Bureau; 2011. Publication ACSBR/10–03. Google Scholar
22. O’Connell M, Gooding G. Editing unmarried couples in Census Bureau data. Paper presented at the Annual Meeting of the Population Association of America; March 30, 2012; Los Angeles, CA. Google Scholar
23. US Census Bureau. Your Guide for the American Community Survey. Washington, DC. Publication ACS-30(2009)(8–2008). Google Scholar
24. Gates G. Same-sex couples in US Census Bureau data: who gets counted and why. Available at: http://williamsinstitute.law.ucla.edu/wp-content/uploads/Gates-Who-Gets-Counted-Aug-2010.pdf. Accessed January 23, 2013. Google Scholar
25. US Census Bureau. ACS health insurance definitions. Available at: http://www.census.gov/hhes/www/hlthins/methodology/definitions/acs.html. Accessed January 23, 2013. Google Scholar
26. Kleinman LC, Norton EC. What’s the risk? A simple approach for estimating adjusted risk measures from nonlinear models including logistic regression. Health Serv Res. 2009;44(1):288302. Crossref, MedlineGoogle Scholar
27. Agency for Healthcare Research and Quality. National Healthcare Disparities Report 2011. Rockville, MD: US Department of Health and Human Services; 2012. AHQRQ Publication 12–0006. Google Scholar
28. Stata Statistical Software, Version 12.0. [computer program]. College Station, TX: StataCorp LP; 2011. Google Scholar
29. Claxton G, Rae M, Panchel N, et al. Employer health benefits 2012 annual survey. Available at: http://ehbs.kff.org/pdf/2012/8345.pdf. Accessed January 23, 2013. Google Scholar
30. Gates G, Steinberger MD. Same-sex unmarried partner couples in the American Community Survey: the role of misreporting, miscoding and misallocation. Paper presented at the Annual Meeting of the Population Association of America; April 30, 2009; Detroit, MI. Google Scholar
31. Carpenter C, Gates GJ. Gay and lesbian partnership: evidence from California. Demography. 2008;45(3):573590. Crossref, MedlineGoogle Scholar
32. Black D, Gates G, Sanders S, Taylor L. Demographics of the gay and lesbian population in the United States: evidence from available systematic data sources. Demography. 2000;37(2):139154. Crossref, MedlineGoogle Scholar

Related

No related items

TOOLS

SHARE

ARTICLE CITATION

Gilbert Gonzales, MHA, and Lynn A. Blewett, PhDGilbert Gonzales and Lynn A. Blewett are with the Division of Health Policy and Management and the State Health Access Data Assistance Center, School of Public Health, University of Minnesota, Minneapolis. “National and State-Specific Health Insurance Disparities for Adults in Same-Sex Relationships”, American Journal of Public Health 104, no. 2 (February 1, 2014): pp. e95-e104.

https://doi.org/10.2105/AJPH.2013.301577

PMID: 24328616