Health Indicators for Older Sexual Minorities: National Health Interview Survey, 2013–2014

Introduction

In 2014, the Centers for Disease Control and Prevention (CDC) National Center for Health Statistics (NCHS) released the first report from the National Health Interview Survey (NHIS) on sexual minority (SM) data for the U.S. population aged 18 years and older.¹ This publication was the result of several years of effort following the recommendations of the Institute of Medicine to ask a sexual orientation question on a nationally representative survey for all adults.² However, this first analysis from NCHS did not examine older adults as a separate subgroup of the SM population. With this in mind, we undertook this analysis to provide a supplement on older adults using the first 2 years of NHIS SM data.

The United States is experiencing a demographic shift, with the aging population making up a larger segment of the total population and utilizing more health services due to increased life expectancy. The U.S. Census Bureau estimates that by the year 2050, the population aged 65 years and older will reach 83.7 million, nearly double what it was estimated at in 2012.³ Sixty-five years is the eligibility age for Medicare coverage and, thus, it is used by the Census, other federal agencies, and some researchers as a cutoff point to define older adults.

Most research on sexual minorities, including from the NHIS SM data, has focused on adults aged 18 and older (without specific analysis of those aged 65 and older), examined same-sex couples aged 65 years and older instead of using self-reported sexual orientation, or focused on results from state-level surveys.^1,4–8 Prior research on the older SM population has focused on several different age brackets. A common grouping is age 50 years and older, possibly due to American Association of Retired Persons (AARP's) age eligibility cutoff for when some Americans might start to retire.⁶ The age 50 cutoff has also been utilized in some National Institute on Aging surveys and research, possibly to track trends of an aging population more effectively.⁹ Using a lower age cutoff, as well as combining multiple years of data, has improved statistical reliability due to larger sample sizes being used for analysis. With few data sources available for examining the health of older sexual minorities, significant gaps remain.¹⁰

Prior research on older sexual minorities indicates that there are differences by sex for some health outcomes and other health indicators.⁸ However, there are some negative trends for sexual minorities regardless of sex. We aimed at discerning whether those negative trends persisted, subsided, or were exacerbated in older age groups.⁷

Methods

Data overview

The NHIS is an annual survey conducted by the CDC NCHS to monitor the health of the U.S. population.¹¹ Starting in 2013, the NHIS has collected information on the sexual orientation of the respondents, allowing researchers to investigate potential differences in health indicators between heterosexual and SM (defined as lesbian, gay, and bisexual respondents) groups. The NHIS employs a complex sampling strategy designed to produce estimates that are representative of the U.S. population; details of the design, sampling strategy, and weighting can be found elsewhere.¹² NHIS collects race and ethnicity data following guidance from the U.S. Census Bureau.¹²

The number of respondents and response rates for the NHIS vary from year to year. For this analysis, we used the adult sample component of the survey. For this component, one adult is randomly selected from each household to complete survey questions on health-related behaviors, healthcare service utilization, healthcare access, and disability-related items, among other health-related questions. In 2013, the conditional response rate for the adult sample component was 81.7%, corresponding to 34,557 adults, and in 2014 these values were 80.5% and 36,697, respectively.^11,13,14

We then limited our analysis to respondents aged 65 years and older, leaving 7732 respondents (22.4% of the adult sample component) in 2013 and 8644 (23.6%) in 2014. We choose to examine this age group because 65 is the age at which most individuals become eligible for Medicare.

Sexual majority and minority populations

We categorized respondents into three groups: sexual majority (hereafter: heterosexual), SM, and unknown. Respondents were asked about their sexual orientation by using the question “which of the following best represents how you think of yourself.”¹⁵ Response options were: “gay” (for male respondents); “lesbian or gay” (for female respondents); “straight, that is, not gay” (male respondents); “straight, that is, not lesbian or gay” (female respondents); “bisexual”; “something else”; “I don't know the answer”; or refused. Respondents who provided a response of “gay,” “lesbian or gay,” or “bisexual” were combined into a single SM group to increase sample size for analysis. Respondents who provided a response of “straight, that is, not gay” or “straight, that is, not lesbian or gay” comprised the heterosexual group. All other responses were combined into an unknown category and were not included in this analysis (n = 576, 3.6%).

Health outcomes parameters

We investigated the 12 health indicators available in the weighted 2013–2014 NHIS data^13,14 comparing the heterosexual population with the SM population. The selected outcomes are similar to those used in the National Health Statistics Report and include health status, health-related behaviors, healthcare service utilization, and healthcare access.¹ The functional limitations, visual impairment, and hearing impairment were added to our analysis, in divergence from the NCHS report on sexual orientation data, as a proxy for disability-related measures and they are often included specifically for older adult populations. Measures of health insurance status were omitted because this population, aged 65 years and older, was assumed to be covered by Medicare and disability measures were added because of their relevance to the health and well-being of older populations. All 12 health indicators derived from the NHIS are described in Table 1.

Statistical analysis

To test the statistical significance of the difference between the SM and heterosexual groups, we used a Pearson's chi-square test adjusted for the complex sample design of the NHIS. The estimates, confidence intervals, and statistical tests were computed from weighted 2013 and 2014 NHIS survey years. Estimates were weighted according to guidelines published by the NHIS.¹² Sample weights from each year were divided by two, given that 2 years of data were pooled.

Because there are relatively few respondents who identify as an SM, the tests we performed have limited power, and the detectable differences are correspondingly large. For certain outcomes with very small samples sizes or low response rates, we determined that there were insufficient data to draw a conclusion. This determination was made based on the size of the standard error relative to the point estimate and by using rules recommended by CDC for reliability of estimates.¹⁶ Outcomes where the standard error was more than 30% of the point estimate for the SM group were not included in tests of statistical significance. These outcomes were hearing impairment, psychological distress, and failing to obtain needed medical care. We used Stata version 13 (StataCorp LLC, College Station, TX) for all data processing and statistical tests. We assumed that each of the statistical tests was independent and considered P-values less than 0.05 to be statistically significant. This research was exempt from Institutional Review Board review because it uses aggregate de-identified public use data.

Results

In the combined 2013–2014 NHIS, there were 15,800 respondents aged 65 years and older, of whom 147 respondents (0.90%) identified as an SM. These included 68 (0.88%) of the 2013 sample (7732 respondents aged 65 years and older) and 79 (0.91%) of the 2014 sample (8644 respondents aged 65 years and older). The majority of the respondents in the SM group identified as gay or lesbian (n = 111), and the remaining 36 identified as bisexual. Table 2 shows the demographic characteristics of the SM and heterosexual groups. On average, the SM respondents were younger (P < 0.001) and less likely to be female (P = 0.011) than their heterosexual respondent counterparts. Self-reported race and ethnicity distributions were similar for heterosexual and SM respondents. Compared with heterosexual respondents, SM respondents were less likely to be married and had lower total household income (P < 0.001 for both).

Health outcomes

Table 3 shows the population percentages for the selected health-related indicators for both SM and heterosexual groups. For most indicators, no statistically significant differences were detected. However, for four of the indicators—self-rated excellent/very good health, influenza vaccination, HIV testing, and binge drinking behavior—significant differences were observed.

SM respondents were more likely than heterosexual respondents to report that they are in excellent or very good health (62.7% and 45.9%, respectively, P = 0.003), have received an influenza vaccine during the past year (79.7% and 68.8%, respectively, P = 0.023), and have ever been tested for HIV (51.3% and 15.8%, respectively, P < 0.001). Having five or more alcoholic drinks in 1 day at least once in the past year, an indicator of binge drinking, was also more likely among SM respondents compared with heterosexual respondents (14.5% and 6.7%, respectively, P = 0.003).

For functional limitations, the difference was relatively large between the heterosexual and SM groups (64.0% and 53.4%, respectively), although this difference did not reach statistical significance. Similar proportions of SM and heterosexual older adults were obese, reported being a current cigarette smoker, having a usual source of medical care, and having a visual impairment. The remaining indicators—psychological distress, failing to obtain needed medical care, and hearing impairment—were not tested for statistical significance due to a relative standard error greater than 30% of the estimate, which is not considered to produce reliable inferences.

Discussion

Using the same analytic tools as the NCHS, our results indicate that challenges persist with small sample sizes, which impede understanding of the SM adult population aged 65 years and older.¹ Our analysis indicates that heavy alcohol consumption continues to stand out as a negative health behavior, which is higher in SM older adults compared with heterosexual older adults.^8,17 Similarly, heightened alcohol consumption has been found in sexual minorities across the life course.¹ Although our analysis was unable to test for differences between lesbian and bisexual women, there is evidence to support that when more data are available lesbians would likely have a higher rate of excessive drinking than bisexual or heterosexual women.^8,18 The long-term and complex side effects from heavy or binge alcohol use throughout the life course can lead to increased health issues in older populations, including decreased life expectancy.¹⁹

Interestingly, our analysis did not detect a difference between older sexual minorities and their heterosexual peers in reported current cigarette smoking, obesity, usual sources of medical care, visual impairment, and functional limitations. Psychological distress, failing to obtain needed medical care, and hearing impairment were not tested for statistical significance due to already small and unreliable estimates. Psychological distress is an indicator that often comes up for older adults, especially sexual minorities, as they adjust to limited physical functionality, changing social circumstances, and the stress of adjusting to the aging process.^18,20 Continued investigation as more years of data become available, along with more complex multivariate analysis, should elucidate why our research was unable to detect statistically significant differences.

Older sexual minorities appear to fare better on preventive health measures such as ever having obtained an HIV test and receiving an influenza vaccine in the past year. The results for HIV testing may be influenced by adult sexual minorities in other NHIS analyses: 51% for older sexual minorities compared with 69% for gay/lesbian individuals and 54% for bisexual individuals aged 18–64, respectively.¹ It is interesting that heterosexual older adults have a substantially lower rate of ever having received an HIV test, at ∼15% compared with 42% in the 18- to 64 year-old heterosexual population.¹

Older sexual minorities also report substantially higher rates of excellent or very good health compared with their older heterosexual peers. In younger adult populations, the gap between sexual minorities and heterosexuals in terms of self-rated excellent or very good health is much less pronounced.¹ It is unclear as to whether this response is the result of persistent resilience among older sexual minorities or that older heterosexuals experience a decrease in perceived good health.^18,21

Limitations

In this analysis of the 2013–2014 NHIS data, we combined gay, lesbian, and bisexual older adult respondents into a single group due to the small number of older SM respondents. However, prior literature in the elderly and nonelderly populations suggests that there may be important differences between each group and/or between genders. When we examined gay, lesbian, and bisexual respondents separately, we observed some differences that we would expect based on previous research, but none of these differences were statistically significant and, thus, they were excluded from our results. We did not find significant differences between gay, lesbian, and bisexual respondents, most likely due to the small sample sizes for each of the groups. Other studies that were able to examine differences between by gender or differences between gay, lesbian, and bisexual groups combined more years of data than we had available at the time of our analysis.^6–8 Further, in some cases where there may be differences between gay, lesbian, and bisexual older adult groups, combining groups will conceal differences with the heterosexual group. It is possible that subgroup differences could be observed with more data. As more years of NHIS data become available, we will be able to discern trends more effectively, and move beyond the limitations of bivariate analysis.

In statistical testing, we assumed each comparison was independent and considered P-values less than 0.05 as indicating statistically significant findings. If instead, we do not assume independence, and correct for multiple comparisons, the threshold for statistical significance would decrease to P < 0.006. Using this threshold would imply that the difference for influenza vaccination is statistically insignificant.

In addition, it was challenging to explore differences between other demographic variables for older sexual minorities with only 2 years of data. Future research should include multivariable analysis and explore the influence of socioeconomic and other demographic factors on these health indicators for older sexual minorities.

Conclusion

Our objective of providing an analysis of NHIS sexual orientation data on health indicators for older sexual minorities, to supplement the work published by NCHS, was accomplished. We were able to show that self-reported excellent or very good overall health persists into older ages for sexual minorities and that health behaviors such as HIV testing and receiving influenza vaccines also continue. We also found that alcohol consumption continues to be an issue for older sexual minorities. These findings are important to the field of SM health research because, although limited, they provide a nationally representative supplement to the work of the NCHS that focused on the adult population aged 18–64.¹ Continued analysis is needed as more years of data become available, and the population aged 18–64 years, for which data were reported previously, ages into the Medicare population.

Acknowledgments

The authors thank Josef Clift, DrPH, for his insight and editorial review of this article and Erin Ewald for additional statistical consultation. This work was supported by the Centers for Medicare & Medicaid Services under Contract Number HHSM-500-2011-00002I with NORC at the University of Chicago.

Disclaimers

The views expressed in this article are those of the authors and do not necessarily represent the views of the Centers for Medicare & Medicaid Services or the U.S. Department of Health and Human Services. Portions of this work appear in the CMS OMH data highlight.²²

Author Disclosure Statement

No competing financial interests exist.