Increased Risks of Needing Long-Term Care Among Older Adults Living With Same-Sex Partners

Abstract

Objectives. We examined whether older individuals living with same-sex partners face greater risks of needing long-term care than their counterparts living with different-sex partners or spouses.

Methods. With data on older couples (at least 1 individual aged 60 years or older) from the 2009 American Community Survey, we estimated logistic regression models of 2 activity limitations that signal a long-term care need: difficulty dressing or bathing and difficulty doing errands alone.

Results. When we controlled for age, race/ethnicity, and education, older women who lived with female partners were statistically significantly more likely than those who lived with male partners or spouses to have difficulty dressing or bathing. Older men who lived with male partners were statistically significantly more likely than those who lived with female spouses or partners to need assistance with errands.

Conclusions. Older individuals living with same-sex partners face greater risks of needing long-term care than those living with different-sex partners or spouses, but the role of relationship status differs by gender. These findings suggest more broadly that older gay men and lesbians may face greater risks of needing long-term care than their heterosexual counterparts.

In light of population aging^1,2 and the substantial number of individuals identifying as gay, lesbian, or bisexual,³ understanding the relationship between long-term care needs and sexual orientation has become increasingly important. Several factors may contribute to health limitations among sexual minorities: (1) a history of stigma, prejudice, and discrimination^4–9; (2) discriminatory public policies^10–18; and (3) sexual and reproductive histories.^19–22

First, experiences of discrimination, anticipation of rejection, concealment or disclosure of a stigmatized identity, and internalized negative views of oneself contribute to minority stress, the chronic stress that accompanies a stigmatized social status.^4–6,8 A growing body of literature provides evidence that minority stressors adversely affect the mental and physical health of lesbians and gay men.^4,6,8,9 Minority stressors are linked to higher psychological distress,⁶ depressive symptoms,⁸ and serious physical health problems.⁹ Second, public policies that limit access to health care, employment, housing, public benefits, and legal marriage on the basis of sexual orientation may also have negative health consequences for sexual minorities.¹³ For example, several studies provide evidence of health benefits associated with marriage,^23,24 and a recent study showed that legal marriage to a same-sex partner significantly enhances mental health.¹⁸ Negative health impacts have been found for sexual minorities living in states with public policies that fail to protect them from employment discrimination or hate crimes.¹³ Third, men who have sex with men face greater risks of HIV than do men who have sex with women,²⁰ and differences in reproductive histories combined with minority stressors and discriminatory public policies may cause lesbians to suffer from higher rates of breast cancer than do heterosexual women.^21,22 The cumulative health consequences of these 3 factors may increase the risks of disability among sexual minorities and, in particular, may lead to greater risks of needing long-term care in old age.

Although a growing body of literature provides evidence of mental^15,25–29 and physical^7,30–33 health differences by sexual orientation, collectively this literature offers scant evidence concerning activity limitations among older adults. One recent study begins to fill this gap. Using data from the California Health Interview Survey on individuals aged 50 to 70 years, Wallace et al.³⁴ examined whether a variety of chronic health conditions—including physical disability—varied by sexual identity. The study found that lesbian, gay, and bisexual older adults experienced higher levels of chronic health conditions, including psychological distress and physical disabilities, than did their older heterosexual counterparts.

Our study contributes to this literature by examining whether older individuals in same-sex relationships are more likely than their counterparts in different-sex relationships to experience activity limitations that specifically signal a need for long-term care.³⁵ Long-term care is generally defined as

a range of services and supports [a person] may need to meet health or personal needs over a long period of time. Most long-term care is not medical care, but rather assistance with the basic personal tasks of everyday life.³⁶

We hypothesized that older gay men and lesbians would face greater risks of needing long-term care than their heterosexual counterparts. Without access to federally recognized marriage, we expected gay men and lesbians to be particularly disadvantaged relative to married heterosexuals. With data from the Public Use Microdata Sample of the 2009 American Community Survey (ACS), we focused on 2 activity limitations: (1) self-care difficulty (difficulty dressing or bathing) and (2) difficulty living independently (difficulty doing errands alone). We used logistic regression to compare men living with male partners to men living with female partners or spouses; likewise, we compared women living with female partners to women living with male partners or spouses. Our approach differs from that of Wallace et al.³⁴ in 3 key respects: (1) the use of nationally representative rather than state-level data, (2) the focus on measures that signal a need for long-term care rather than a broad range of chronic health conditions, and (3) the use of a behavioral measure of sexual orientation rather than sexual identity. Thus, our study both broadens the findings of the California study to a national sample and focuses on the implications for older gays and lesbians who have disabilities that could result in a need for a specific type of health care service—long-term care.

METHODS

To examine whether the risk of needing long-term care varied by sexual orientation, we used data from the 2009 ACS, a nationally representative sample of 1 917 748 households with a response rate of 98.0%.³⁷ The ACS includes questions pertaining to each household member’s demographic characteristics and health status. In addition, the survey inquired about each individual’s relationship to the head of the household. With information on the relationship between each household member and the householder as well as the biological sex of each household member, the ACS enables researchers to identify whether the householder was married, living with a different-sex partner, living with a same-sex partner, or not living with a partner. The publicly available ACS files do not enable researchers to distinguish same-sex couples who report their status as married from those who identify themselves as unmarried partners; thus, we use the terms “married” and “spouse” to distinguish heterosexual marriages from heterosexual cohabitations.

We restricted the sample to households that included an older couple (at least 1 individual aged 60 years or older). We used age 60 years as the measure of when a person is “old” because that is the age recognized by the US Congress in the Older Americans Act as the point at which people are likely to need services to support independence and long-term care needs.^38,39 Because of our interest in older adults, we excluded spouses or partners younger than 40 years. We also restricted the sample to individuals who identified their race as White, Black/African American, or Asian, or their ethnicity as Hispanic because of the limited number of individuals in same-sex relationships with other racial/ethnic identities in the ACS. Because of differences in health and mortality between older men and women,^40–42 we stratified our sample by gender. After we applied our sample selection criteria, our sample consisted of 216 852 married men, 6599 men living with female partners, 1348 men living with male partners, 216 386 married women, 6498 women living with male partners, and 1755 women living with female partners.

Measures

Our analysis focused on the behavioral dimension of sexual orientation, in part because the ACS does not provide information about sexual identity or sexual attraction.³⁷ In particular, we compared individuals living with same-sex partners to their counterparts living with or married to different-sex partners or spouses.

We measured risks of needing long-term care in terms of 2 disabilities, self-care difficulty (whether the individual has “difficulty dressing or bathing”) and difficulty living independently (whether “because of a physical, mental, or emotional condition,” the individual has “difficulty doing errands alone such as visiting a doctor’s office or shopping”). We chose these 2 disabilities, because, by definition, a personal care provider may be necessary to complete these tasks. A hallmark of the need for long-term care is the inability to perform personal care tasks or other daily activities without assistance.³⁶ In the ACS, both of these measures are responses to dichotomous yes-or-no questions. It is worth noting that the ACS does not provide information about whether or to what extent these needs for assistance are being met. The ACS provides dichotomous measures of several other disabilities, namely “deafness or difficulty hearing,” “blindness or difficulty seeing,” “difficulty concentrating, remembering, or making decisions,” and “difficulty walking or climbing stairs,” but these disabilities do not necessarily signal a need for long-term care. Thus, as a first step toward understanding the relationship between sexual orientation and the risk of needing long-term care, we focused on 2 disabilities that signal a need for long-term care: self-care difficulty and difficulty living independently.³⁵

Previous studies provide evidence of the health benefits associated with marriage.^23,24 To avoid confounding the effects of living with a different- versus same-sex partner with the effects of federally recognized marriage, our models distinguish between marriage and cohabitation among different-sex couples. We used responses to the question “How is this person related to Person 1?” to identify the spouse or partner of the head of household. Possible responses to this question include “husband or wife,” “unmarried partner,” “biological son or daughter,” and a variety of other relationships. In the case of different-sex couples, determining marital status was straightforward. In the case of same-sex couples, however, the US Census Bureau recoded all individuals identified as “husband or wife” of the head of household as “unmarried partners.” Thus, the treatment of same-sex couples in the publicly available ACS files prevented us from examining whether marriage had a protective effect for same-sex couples, but our approach enabled us to compare individuals living with same-sex partners to their counterparts living with different-sex partners and to those married to different-sex partners. For ease of exposition, we use the term “relationship status” to distinguish among different-sex marriage, different-sex cohabitation, and same-sex cohabitation.

Covariates

We controlled for demographic characteristics that may influence the risk of needing long-term care. Because of evidence that the prevalence of activity limitations varies by age and race/ethnicity,⁴³ we controlled for age (measured in years) and race/ethnicity (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian, or Hispanic). As a proxy for economic resources, education may reflect access to health care; in addition, education may influence early life experiences, behavioral risk factors, environmental stress, preventive care, and medical treatment.⁴⁴ Thus, through various mechanisms, education may influence activities and instrumental activities of daily life. In accordance, our analysis controlled for educational attainment (less than a high-school degree, a high-school degree, or at least a college degree).

To estimate the full effect of living with a same-sex partner relative to a different-sex partner or spouse on the odds of experiencing difficulty dressing or bathing or doing errands, our models do not include potentially intervening (endogenous) variables such as psychological distress²⁷ or access to health insurance.⁴⁵ Sexual minorities are more likely than their heterosexual counterparts to face barriers to health insurance, potentially limiting their access to health care.⁴⁵ Delayed access to health care can result in untreated health conditions, thus leading to an increased risk of disability in later life. If structural barriers to health insurance contribute to increased risks of disability among individuals in same-sex relationships, then a model that excludes health insurance coverage (and other intervening variables) will enable us to estimate the total effect of relationship status (including the indirect effect of health insurance coverage) on the risk of experiencing a particular disability. Similar logic applies in the case of psychological distress. Thus, like Conron et al., our models did not control for variables that “may be on the causal pathway between sexual orientation and health.”^7(p1959)

We used SAS Enterprise Guide 4.3 (SAS Institute, Cary, NC) to estimate logistic regression models of self-care difficulty (difficulty dressing or bathing) and difficulty living independently (difficulty doing errands alone). As discussed earlier, our sample consisted of individuals who were married or living with a partner and typically the spouse or partner was also included in our sample. As a consequence, our data were clustered at the level of the household. In the case of clustered data, coefficient estimates are inefficient and standard errors may be underestimated.⁴⁶ Although we estimated separate models by gender, clustering posed a potential problem in the case of same-sex couples. Thus, we used the method of generalized estimating equations to obtain efficient estimates and empirical (robust) standard error estimates for our logistic regression models.

RESULTS

Table 1 displays the demographic characteristics of our sample. The average age varied significantly by relationship status for both men (F = 577.1; P < .001) and women (F = 577.5; P < .001). On average, married men were older than both unmarried men living with female partners and men living with male partners, whereas women living with female partners were older than both married women and unmarried women living with a male partner. Race/ethnicity was not independent of relationship status for men (χ² = 372.2; P < .001) or for women (χ² = 266.8; P < .001). For example, among both men and women, the proportions of Black and Hispanic individuals were highest among those living with but not married to a different-sex partner. Likewise, educational attainment was not independent of relationship status for men (χ² = 318.0; P < .001) or for women (χ² = 272.1; P < .001). For example, the proportion of individuals with a college degree was highest among those living with a same-sex partner.

TABLE 1—

Characteristics of Older Adults Living With a Partner or Spouse by Relationship Status: American Community Survey, 2009

	Men				Women
Characteristic	Married, %, Mean (SD), or No.	Female Partner, %, Mean (SD), or No.	Male Partner, %, Mean (SD), or No.	χ2 or F	Married, %, Mean (SD), or No.	Male Partner, %, Mean (SD), or No.	Female Partner, %, Mean (SD), or No.	χ2 or F
Disability
Difficulty dressing or bathing	4.3	3.7	3.8	6.3*	3.9	4.0	6.4	29.9***
Difficulty doing errands alone	7.3	6.5	9.3	14.2***	7.4	7.8	9.9	16.6***
Educational attainment				318.0***				272.1***
High school degree	53.5	57.9	46.1		63.8	63.9	49.7
College degree	31.1	22.5	41.1		23.3	19.4	34.7
Race/ethnicity				372.2***				266.8***
Black	5.4	9.7	5.1		5.2	8.9	6.4
Asian	2.8	1.2	2.7		3.4	2.2	2.5
Hispanic	5.4	7.9	5.2		5.7	7.9	5.4
Age, y	69.3 (8.1)	66.2 (8.5)	65.7 (10.5)	577.1***	66.0 (8.6)	62.3 (9.3)	67.0 (10.3)	577.5***
Sample size	216 852	6599	1348		216 386	6498	1755

*P < .05; ***P < .001 (statistically significant difference by relationship status).

Health limitations also varied by relationship status. As shown in Table 1, neither self-care difficulty (χ² = 6.3; P < .05 for men; χ² = 29.9; P < .001 for women) nor difficulty living independently (χ² = 14.2; P < .001 for men; χ² = 16.6; P < .001 for women) was independent of relationship status. Among women in our sample, difficulty dressing or bathing was more prevalent among those in same-sex relationships than among those in different-sex relationships. Among men in our sample, those who were married were the most likely to report self-care difficulty. Among men and women in our sample, those in same-sex relationships were the most likely to report difficulty living independently.

Tables 2 and 3 present the estimated odds ratios associated with our logistic regression models. For older women but not for older men, the risk of experiencing difficulty dressing or bathing varied significantly by relationship status. Women living with women were significantly more likely than those living with or married to men to experience self-care difficulty, after we controlled for age, race/ethnicity, and educational attainment. The estimated odds of self-care difficulty were 1.50 times greater among women living with female partners than among married women (adjusted odds ratio [AOR] = 1.50; 95% confidence interval [CI] = 1.21, 1.85). The estimated odds of experiencing self-care difficulty were 1.30 times greater among women living with female partners than among women living with male partners (AOR = 1.500/1.158 = 1.30; 95% CI = 1.01, 1.66). For older men and women, the risk of needing assistance with errands varied significantly by relationship status. Men living with male partners were significantly more likely than married men (AOR = 1.65; 95% CI = 1.34, 2.02) and men living with female partners (AOR = 1.647/1.021 = 1.61; 95% CI = 1.28, 2.03) to need assistance with errands. Women living with but not married to male partners were significantly more likely than married women to need assistance with errands (AOR = 1.26; 95% CI = 1.14, 1.38), but the differences between women living with women and those living with or married to men were not statistically significant.

TABLE 2—

Logistic Regression Model of Difficulty Dressing or Bathing: American Community Survey, 2009

	Men (n = 224 799)		Women (n = 224 639)
Characteristic	AOR (95% CI)	SE	AOR (95% CI)	SE
Intercept	0.001*** (0.000, 0.001)	0.105	0.001*** (0.001, 0.001)	0.1061
Relationship status
Married (Ref)	1.000		1.000
Living with same-sex partner	1.060 (0.788, 1.425)	0.151	1.500*** (1.215, 1.852)	0.108
Living with different-sex partner	0.956 (0.838, 1.091)	0.067	1.158* (1.017, 1.317)	0.066
Age, y	1.069*** (1.067, 1.072)	0.001	1.063*** (1.060, 1.066)	0.001
Race and ethnicity
White (Ref)	1.000		1.000
Black	1.676*** (1.553, 1.808)	0.039	1.812*** (1.673, 1.963)	0.041
Asian	0.970 (0.844, 1.115)	0.071	0.758*** (0.654, 0.878)	0.075
Hispanic	1.310*** (1.207, 1.423)	0.042	1.180*** (1.081, 1.287)	0.044
Educational attainment
< high school degree (Ref)	1.000		1.000
High school degree	0.638*** (0.606, 0.670)	0.026	0.555*** (0.526, 0.586)	0.028
College degree	0.371*** (0.348, 0.396)	0.033	0.323*** (0.299, 0.350)	0.040
Within-cluster correlation	0.075 (0.022, 0.128)	0.027	0.099** (0.052, 0.145)	0.024

Note. AOR = adjusted odds ratio; CI = confidence interval. Quasi likelihood under the independence model criterion: men = 74 754.831; women = 70 125.574. The SEs are associated with the parameter estimates and not with the AORs.

*P < .05; **P < .01; ***P < .001; values represent statistically significant difference in odds of difficulty dressing or bathing from reference category.

TABLE 3—

Logistic Regression Model of Difficulty Doing Errands Alone: American Community Survey, 2009

	Men (n = 224 799)		Women (n = 224 639)
Characteristic	AOR (95% CI)	SE	AOR (95% CI)	SE
Intercept	0.000*** (0.000, 0.000)	0.084	0.001*** (0.001, 0.001)	0.080
Relationship status
Married (Ref)	1.000		1.000
Living with same-sex partner	1.647*** (1.342, 2.020)	0.104	1.192 (0.995, 1.427)	0.092
Living with different-sex partner	1.021 (0.920, 1.132)	0.053	1.258*** (1.143, 1.384)	0.049
Age, y	1.084*** (1.081, 1.086)	0.001	1.072*** (1.070, 1.075)	0.001
Race and ethnicity
White (Ref)	1.000		1.000
Black	1.552*** (1.457, 1.653)	0.032	1.611*** (1.511, 1.718)	0.033
Asian	1.226*** (1.109, 1.354)	0.051	1.022 (0.928, 1.124)	0.049
Hispanic	1.224*** (1.145, 1.309)	0.034	1.164*** (1.090, 1.243)	0.034
Educational attainment
< high school degree (Ref)	1.000		1.000
High school degree	0.577*** (0.555, 0.601)	0.020	0.521*** (0.501, 0.543)	0.021
College degree	0.345*** (0.328, 0.362)	0.026	0.312*** (0.295, 0.331)	0.029
Within-cluster correlation	0.055 (0.002, 0.108)	0.027	0.211*** (0.166, 0.255)	0.024

Note. AOR = adjusted odds ratio; CI = confidence interval. Quasi likelihood under the independence model criterion: men = 107 604.024; women = 110 161.041. The SEs are associated with the parameter estimates and not with the AORs.

*P < .05; **P < .01; ***P < .001; values represent statistically significant difference in odds of difficulty doing errands alone from reference category.

Both the likelihood that an older individual had difficulty dressing or bathing and the likelihood that he or she needed assistance with errands varied by age, race/ethnicity, and educational attainment. As expected, for both men and women, the risk of developing either disability increased significantly with age. Relative to their White counterparts, Black and Hispanic individuals faced significantly greater risks of developing either disability. Relative to their White counterparts, Asian women were significantly less likely to experience difficulty dressing or bathing, whereas Asian men were significantly more likely to need assistance with errands. Finally, individuals with greater educational attainment were less likely to experience either disability.

DISCUSSION

We examined whether older individuals in same-sex relationships faced greater risks of needing long-term care than their counterparts in different-sex relationships. We measured risks of needing long-term care in terms of self-care difficulty (difficulty dressing or bathing) and difficulty living independently (difficulty doing errands alone).

When we controlled for age, race/ethnicity, and educational attainment, women living with female partners were statistically significantly more likely than both married women and women living with male partners to need help bathing or dressing, whereas men living with male partners were statistically significantly more likely than both married men and men living with female partners to need help with errands. Thus, both men and women living with same-sex partners faced significantly greater risks of needing long-term care than their counterparts in different-sex relationships, but the specific findings differ by gender.

The reason for this gender difference remains a question for future work. It may stem from differences in underlying health conditions that contribute to self-care difficulty or difficulty living independently. For example, the prevalence of obesity, a condition associated with higher risks of experiencing difficulty dressing or bathing,⁴⁷ is higher among older lesbians and bisexual women than among older heterosexual women.⁴⁸ The prevalence of HIV infection, a condition associated with physical challenges such as doing errands (and also bathing),⁴⁹ is higher among older gay and bisexual men than among older heterosexual men.⁴⁸

Our article contributes to a growing body of literature documenting health differences by sexual orientation.^{4–9,19,33,45,50} Our analysis of activity limitations experienced by older individuals fills an important gap in this literature, especially in light of population aging. Moreover, our use of a large, nationally representative sample enabled us to draw inferences about the underlying population of older adults in the United States living with a spouse or partner.

By indicating that older men living with men are more likely to need help with errands and that older women living with women are more likely to need help dressing or bathing than their counterparts living with or married to different-sex partners, our findings suggest more broadly that older gay men and lesbians may face greater risks of needing long-term care than their heterosexual counterparts. Whether older gay men and lesbians in general or those living with same-sex partners in particular face greater risks of needing long-term care, our findings have important policy implications.

First, as discussed earlier, health limitations among sexual minorities may stem in part from public policies burdening access to health care, employment, housing, public benefits, and legal marriage. Thus, policies that prohibit discrimination in these realms may reduce health disparities by sexual orientation.^10–18 This recommendation is supported by 2 recent studies that provide evidence of the positive health impacts of marriage equality and antidiscrimination statutes. A study tracking rates of medical and mental health care visits and costs in the years before and after same-sex marriage was legalized in Massachusetts found that male sexual minorities had a statistically significant decrease in medical and mental health care visits and costs following the legalization of same-sex marriage.¹⁸ Another study found that lesbians and gay men living in states that had no protections against hate crimes and employment discrimination had significantly greater psychiatric disorders than those who lived in states with protective public policies.¹³

Second, policies that target the special needs of older disabled sexual minorities may improve the quality of their care. As a result of societal stigma, prejudice, and discrimination, older gays and lesbians can be isolated, and they are more likely than their heterosexual counterparts to be estranged from their families of origin.⁵¹ With the notable exception of organizations such as Services and Advocacy for Gay, Lesbian, Bisexual, and Transgender Elders (SAGE),⁵² aging network service providers have largely ignored lesbian, gay, bisexual, and transgender (LGBT) elders in terms of both outreach to and provision of specialized services to this population.⁵³ Yet, our study suggests that older sexual minorities face greater risks of needing long-term care than their heterosexual counterparts. Thus, long-term care policies could be designed to better serve this population. Better training of long-term care service providers on the existence and special needs of LGBT older adults as well as best practices and policies for increased inclusion and safety of this population may improve the quality of care received by older sexual minorities. Recently, the federal Administration on Aging began to address this public policy issue by funding the Technical Assistance Resource Center: Promoting Appropriate Long Term Care Supports for LGBT Elders.⁵⁴ This center offers mainstream service providers in the aging network as well as LGBT service organizations training related to advance planning for the specialized long-term care needs of older sexual minorities.

The third policy implication concerns the high costs of long-term care. Medicaid is one of the major funding sources for nursing home care and formal home health care. Current policies protect the incomes, savings, and homes of different-sex married couples when one spouse needs nursing home or formal home health care,^55,56 but these protections, for the most part,⁵⁷ are unavailable to same-sex married couples. Thus, older partnered gays and lesbians who need long-term care can face significant financial costs. In light of our findings that these individuals are at greater risk than their heterosexual counterparts of needing long-term care, equality of access to Medicaid financing of long-term care services would help alleviate the financial burden of long-term care for older same-sex couples.

Several data limitations of our study are worth noting. In the absence of a large, nationally representative longitudinal data set, we relied on cross-sectional data, namely the 2009 ACS. The ACS does not provide information on an individual’s sexual orientation identity or attraction but only on the individual’s sex and that of his or her partner or spouse. Thus, our analysis focused exclusively on the behavioral dimension of sexual orientation with the caveat that we only observe current relationship status. Because we cannot observe an individual’s relationship history, we cannot distinguish between exclusive lifetime same-sex (or different-sex) behavior and a history of relationships with both men and women; we do not observe the duration of the current relationship; and we cannot infer anything about the sexual orientation of unmarried individuals who are not living with a partner. Moreover, we cannot distinguish between same-sex couples who identify themselves as spouses from those who identify themselves as unmarried partners. Nevertheless, as a large nationally representative sample with information on each household member’s sex and relationship to the head of the household, the ACS presents a rare opportunity to examine differences in long-term needs between older individuals in same-sex relationships and their counterparts in different-sex relationships. Although earlier waves of the ACS inflated the number of same-sex couples, the US Census Bureau implemented a redesigned survey and improved data editing techniques in 2008.⁵⁸

As better data sources become available, we will investigate health differences among older adults by sexual identity and sexual attraction in addition to the behavioral measure used in this study to examine the robustness of our findings across different dimensions of sexual orientation. For example, we will examine whether individuals who identify as gay, lesbian, or bisexual face greater risks of needing long-term care than do those who identify as heterosexual, regardless of current relationship status. In addition, we will examine whether state-level policies concerning employment discrimination and relationship recognition influence the health of sexual minorities as they age. Collectively, these studies will inform policy debates concerning issues such as health care and same-sex marriage.