Health Disparities Between Older Lesbian, Gay, and Bisexual Adults and Heterosexual Adults in the Public Mental Health System

In recent years, recognition has grown of the health disparities among sexual minority groups, namely, lesbian, gay, and bisexual (LGB) adults (1). More important, little data exist to assess whether these documented health disparities persist or diminish as this LGB population ages. Fredriksen-Goldsen and Kim (2) estimated the population of LGB older adults (ages ≥50 years) in the United States to be 2.4 million, a number expected to double to >5 million by the year 2030. Research investigating health disparities among LGB older adults in the mental health system has been lacking (3).

Fredriksen-Goldsen et al. (4) reported that older LGB individuals were at higher risk for poor mental health, psychological distress, suicidal ideation, and psychiatric disorders (e.g., depression and anxiety). Depression levels also appear to be elevated among older lesbian women and gay men (5, 6). In terms of substance abuse risk factors, which often co-occur with mental health issues, LGB individuals have higher rates of tobacco, alcohol, and drug use than heterosexual adults (7–10). Research also suggests that older LGB adults may be less likely to seek health services than adults in the general population. Some studies suggest that older LGB individuals do not believe that they will receive respectful care in old age and may delay seeking care for fear of discrimination, stigma, and victimization (which they have experienced across their life course) (11). Fredriksen-Goldsen et al. (12) reported that older LGB adults are less likely to have health insurance and are more likely to face higher financial barriers to health care compared with heterosexual counterparts.

A recent study by Frimpong et al. (13) focused on health outcomes and acute services among LGB adults compared with a matched sample of heterosexual adults served in the mental health system. In the present study, we focused on a subset of this population, primarily older LGB adults (ages ≥50 years). We examined differences between sexual minority older women (lesbian women and bisexual women) and men (gay men and bisexual men) and matched heterosexual adults to investigate disparities in principal psychiatric diagnoses, disabilities and disorders, chronic health conditions, and acute health care access within the New York State mental health system.

Our study included the following major research questions. Compared with heterosexual individuals, what is the health (general medical health, mental health, and substance use) profile of older LGB adults? Relative to heterosexual adults, are there differences in the main outcomes of psychiatric diagnoses; behavioral health disabilities and disorders, including mental health and substance use disorders; chronic medical conditions (CMCs); and acute service utilization (inpatient stays and emergency department [ED] visits)? On the basis of our research questions, we predicted that compared with heterosexual individuals, older LGB individuals would have more severe health profiles, including differences in substance use patterns.

Methods

Study Setting

In this retrospective matched-control study, we used data from the 2015 New York State Office of Mental Health (NYSOMH) Patient Characteristics Survey (PCS) and Medicaid. The PCS is a 1-week, cross-sectional survey of all patients served by the public mental health system in New York State. The PCS is used to collect demographic, clinical, and socioeconomic data from >4,000 licensed or funded NYSOMH programs that serve approximately 180,000 patients. The PCS is not self-reported by patients. It requires providers to collect data from electronic health record (EHR) systems. The survey sample included individuals between ages 50 and 85 years at the time of the survey week (October 19–25, 2015) whose EHR indicated their sexual orientation as gay, lesbian, or bisexual (N=1,659) or as heterosexual (N=1,659). Individuals with unknown sexual orientation were excluded (N=143). The study population was limited to individuals with Medicaid eligibility from October 2014 to October 2015.

Measures

Except for acute service use, all measures were based on PCS data. The PCS was used to identify the study groups, individual characteristics, and factors that are known to influence health disparities (9, 14).

Sociodemographic Characteristics

The sociodemographic characteristics included for matching were sex at birth, sexual orientation, age, race, Hispanic ethnicity, and county of residence. Individual demographic characteristics included four age groups (ages 50–54, 55–59, 60–64, and ≥65 years) as well as race and Hispanic ethnicity (categorized as White non-Hispanic, Black non-Hispanic, and Hispanic), education level, high school graduate (yes or no), living with spouse (yes or no), cohabitation (lives with others or lives alone), homelessness (yes or no), region of residence (New York City or rest of state), employment (yes or no), parental status (yes or no), preferred language English (yes or no), criminal justice involvement (yes or no), and veteran status (yes or no). Two cash assistance benefits were also established: public cash assistance (yes or no) and Supplemental Security Income (SSI) or Social Security Disability Insurance (SSDI) (yes or no).

Principal Psychiatric Diagnosis

Serious mental illness and psychiatric diagnosis were categorized as present or not. Serious mental illness referred to adults having been given one or more diagnoses of mental disorders combined with significant functional impairment. Primary psychiatric disorders included schizophrenia spectrum and other psychotic disorders, depression, bipolar disorder, and anxiety.

Health and Substance Use Disorders

Disability and disorder variables were based on a significant disability as indicated in the EHR system and were characterized as present or not and included intellectual disability, developmental disability, autism spectrum disability, and physical disability (hearing, vision, or mobility). Indicators of substance use disorders included alcohol-related disorder (yes or no), drug- or substance use–related disorder (yes or no), and tobacco use (yes or no).

CMCs

CMC variables were based on a significant disability as indicated in the EHR system and were defined as having at least one of the following general health conditions: hyperlipidemia, high blood pressure, diabetes, obesity, heart attack, stroke, pulmonary condition, Alzheimer’s disease, kidney disease, liver disease, endocrine condition, neurological condition, traumatic brain injury, joint and connective tissue condition, cancer, and other chronic general medical conditions. In addition to at least one CMC condition, these medical conditions were each coded individually as present or not.

Acute Service Use: Inpatient Stays and ED Visits

We assessed acute services by using Medicaid claims and encounters for ED visits or inpatient stays in the 12 months preceding the PCS week. Mental health inpatient stays comprised psychiatric-related stays at licensed general, psychiatric specialty, or state psychiatric hospitals. Inpatients stays related to substance use comprised detoxification services, rehabilitation services, and other substance use–related stays. Nonbehavioral inpatient stays were inpatient stays for general health conditions.

Mental health ED visits comprised psychiatric-related visits at general hospitals and comprehensive psychiatric emergency programs, and substance use ED visits were substance use–related visits. Nonbehavioral ED visits were neither mental health nor substance use–related visits. We categorized the inpatient stays and ED services by using a combination of rate codes, diagnosis-related group codes, specialty codes, procedure codes, revenue codes, mental health or substance use diagnosis, and hospital provider type. The primary outcomes included principal psychiatric conditions, disabilities and disorders, and CMCs. Secondary outcomes included length of inpatient stays and frequency of ED visits.

Statistical Analysis

A matched case-control approach was used to control for potential confounding and to form groups sufficiently balanced to provide efficient statistical analysis. We performed the matching of LGB individuals by using the PROC SURVEYSELECT procedure from SAS, version 9.4, as outlined elsewhere (15). LGB individuals were matched 1:1 to heterosexual adults on exact values of assigned sex at birth, age, race, Hispanic ethnicity, and county of residence. In this procedure, we selected matched LGB individuals and heterosexual adults through a simple random sampling approach without replacement from independent samples stratified on the matched criteria. Although race was used as a variable in the matching approach, we understand that the concept of race is multidimensional and that matching on this variable does not account for all the cultural dimensions represented by race.

To examine the differences in health conditions and acute service use between LGB individuals and heterosexual adults, we calculated crude and adjusted ratios (risk and rate) using generalized estimating equation (GEE) models (accounting for matching pairs), with Poisson or negative binomial distribution, and an independent working correlation structure (16, 17). For binary outcome (health condition present or not), a GEE-modified Poisson regression model was used (18). For count outcome (inpatient stays or ED visits), a GEE negative binomial regression model was used.

The models were fitted for each LGB group with adjustment for employment status, education, cohabitation, parental status, criminal justice status, language, income (SSI or SSDI), and homelessness. These variables were maintained in the adjusted models regardless of statistical significance. Risk and rate ratios with 95% confidence intervals are presented. The study protocol was approved by the Nathan S. Kline Institute Institutional Review Board with a full waiver of informed consent. All statistical analyses were performed with SAS, version 9.4.

Results

Population Characteristics

In 2015, PCS collected information on 179,096 individuals. Overall, 9,683 (5.4%) from the EHRs indicated their sexual orientation as gay, lesbian, bisexual, or other. Of 9,205 individuals reporting a sexual orientation, 2,579 were gay men (28.0% of male-assigned sex at birth), 1,062 were bisexual men (11.5%), 2,469 were lesbian women (26.8% of female-assigned sex at birth), 3,095 were bisexual women (33.6%), and 478 (5.2%) identified as other sexual orientation. The matched cohort included 1,659 LGB adults and 1,659 heterosexual adults. Detailed descriptive characteristics of the matched-control sample are presented in Table 1. Compared with heterosexual older women, sexual minority (lesbian and bisexual) older women were less likely to have graduated from high school, more likely to be living alone, more likely to have paid employment, and less likely to have children. Compared with heterosexual older men, sexual minority (gay and bisexual) older men were less likely to have graduated from high school and were more likely to be living alone, to be unemployed, to prefer speaking English, and to have no children. They were also more likely to be receiving public assistance, less likely to be involved with the criminal justice system, and less likely to have veteran status than their heterosexual counterparts.

Table 2 presents data on the prevalence of psychiatric diagnoses, health and substance use disorders, and CMCs. Compared with heterosexual older women, sexual minority women had higher rates of bipolar disorder, obesity, liver diseases, cancer, alcohol and substance use disorders, and tobacco use. They had lower rates of intellectual disability, schizophrenia, and anxiety disorders than their heterosexual counterparts. Compared with heterosexual older men, sexual minority men had lower rates of serious mental illness, schizophrenia, alcohol and substance use disorders, and tobacco use. They had higher rates of bipolar disorder, depression, anxiety, diabetes, other cardiac issues, and at least one CMC compared with their heterosexual counterparts.

Health Indicators

Table 3 presents data on the risk ratios of primary psychiatric diagnoses, health and substance use disorders, and CMCs, controlled for demographic characteristics. All models were adjusted for employment status, education, cohabitation, parental status, criminal justice status, language, income (SSI or SSDI), and homelessness.

Gay and bisexual older men were significantly more likely than heterosexual older men to have a primary diagnosis of depression (adjusted risk ratio [ARR]=1.34, p<0.05) or anxiety (ARR=1.70, p<0.05) and less likely to have diagnosis of schizophrenia (ARR=0.66, p<0.05) or serious mental illness (ARR=0.93, p<0.05). After adjusting for demographic characteristics, we found that the sexual minority older women were significantly more likely than heterosexual older women to have a primary diagnosis of bipolar disorder (ARR=1.68, p<0.05) and less likely to have a primary diagnosis of schizophrenia (ARR=0.76, p<0.05) or anxiety (ARR=0.56, p<0.05).

Prevalence of health and substance use disorders indicated that gay and bisexual older men reported no differences from their heterosexual counterparts. After adjusting for demographic characteristics, we observed that the sexual minority older women reported significantly higher rates of hearing and visual disabilities (ARR=1.63, p<0.05), alcohol use disorders (ARR=1.73, p<0.05), substance use disorders (ARR=2.10, p<0.05), and tobacco use (ARR=1.39, p<0.05) than did heterosexual women.

Gay and bisexual older men were significantly more likely than heterosexual older men to have experienced a cardiometabolic disorder (ARR=1.13, p<0.05) and diabetes (ARR=1.28, p<0.05). The sexual minority older women differed from heterosexual women only in the prevalence of liver disease (ARR=3.17, p<0.05).

Utilization Measures

Table 4 presents data on rate ratios and on the prevalence of inpatient stays and ED visits, controlled for demographic characteristics. All models were adjusted for employment status, education, cohabitation, parental status, criminal justice status, language, income (SSI or SSDI), and homelessness.

Gay and bisexual older men were significantly less likely than heterosexual older men to have experienced inpatient stays that were related to substance use disorder (ARR=0.45, p<0.05). No statistically significant differences were found in ED visits. The sexual minority older women differed from heterosexual women in both measures related to substance use: inpatient stays related to substance use disorder (ARR=1.84, p<0.05) and frequency of ED visits related to substance use disorder (ARR=2.02, p<0.05).

Discussion

This study builds on the broader literature of health disparities among sexual minority populations and extends the findings to the older (≥50 years old) LGB population within a mental health system. Using a larger sample, Frimpong et al. (13) found that LGB individuals appear to have more chronic general medical conditions and disabilities and tend to use fewer inpatient and ED services than do heterosexual individuals. They also found that gay men and lesbian women were significantly less likely than their heterosexual counterparts to have used a mental health–related inpatient service or ED during the previous 12 months. Bisexual men and women were more likely than heterosexual individuals to have shorter substance use–related inpatient stays and to have fewer substance use–related ED visits. The sample of sexual minority older adults in that study was not large enough to allow investigation of health disparities among subgroups. Gay and bisexual men were combined, likewise, lesbian women and bisexual women were combined, for comparison with heterosexual individuals.

After controlling for socioeconomic covariates, here we found many significant differences between older LGB individuals and matched heterosexual adults. Consistent with the literature on mental health–related issues (4–6), in the present study older gay and bisexual men had higher rates of both depression and anxiety than heterosexual men. Rates of bipolar disorder were higher among older lesbian women and among older bisexual women than among matched heterosexual women.

The most striking difference was related to drug- and alcohol-related behaviors. Older lesbian and bisexual women had significantly more disabilities and disorders related to alcohol, substance, or tobacco use; in addition, they also had significantly higher rates of liver disease. This pattern was consistent with the services data, indicating significantly longer substance use inpatient stays and more substance use–related ED visits among LGB women than among heterosexual women. This pattern among older LGB women was not consistent with results from Frimpong et al.’s (13) study comparing lesbian and bisexual women with matched heterosexual women. In fact, in that study, lesbian and bisexual women used acute substance use treatment services less frequently; in addition, they had on average shorter substance use inpatient stays and fewer substance use–related ED visits. However, the pattern observed here was consistent with those observed in other studies indicating that older LGB women have higher rates of tobacco, alcohol, and drug use than heterosexual adults (7–10).

Gay and bisexual men had significantly higher rates of diabetes and cardiometabolic disorders compared with matched heterosexual men. This finding is consistent with those of Wallace et al. (19), who found elevated rates of diabetes and hypertension among gay and bisexual men (ages 50–70 years) relative to heterosexual men. For general health, we found that lesbian and bisexual women had higher rates of obesity and liver disease compared with matched heterosexual women; however, only liver disease was statistically significant in the adjusted model. The higher rates of obesity among lesbian and bisexual women we found in the present study are consistent with findings in several studies (20, 21); furthermore, the increased rates of liver disease among lesbian and bisexual women may be related to the higher rates of alcohol use and substance use disorder we found in this population.

Another key finding with the older LGB cohort was a difference in the frequency of both inpatient stays and ED visits for substance use disorder. Lesbian and bisexual women were more likely to have longer inpatient stays and more ED visits related to substance use disorders compared with heterosexual women. This finding is consistent with the higher rates of alcohol and substance use disorder in this population compared with heterosexual individuals. Older gay and bisexual men had shorter lengths of inpatient stays related to substance use disorders, consistent with lower (unadjusted) rates of alcohol and substance use compared with heterosexual men.

Substance use among sexual minority groups has been attributed to stress among members of these groups, namely, the stigma, prejudice, and discrimination uniquely experienced by those in a marginalized social group (22). Such stress may result from interpersonal experiences of rejection or discrimination, internalized stigma, or structural factors. This chronic social stress is associated with elevated risk for behavioral health issues among sexual minority individuals compared with their heterosexual counterparts, including substance use (23), psychological distress, and depression (24–26).

This study had several limitations. First, the PCS is conducted every 2 years and uses a cross-sectional survey methodology. Second, mental health providers outside of the public mental health system, such as private practitioners, were not surveyed. Third, services not covered under Medicaid, such as those paid out of pocket, were not accounted for in this study. Fourth, because EHR data were used, more complex definitions of sexual minority status were limited (e.g., both self-perceived identity and sexual behavior). Finally, the study used data from Medicaid-insured individuals and may not be generalizable to other populations.

Conclusions

This study provides evidence of health disparities among sexual minority older adults within a public mental health system in the United States. Our results suggest that these health disparities persist into older adulthood and that new health concerns emerge with the aging of the sexual minority population. Overall, compared with matched heterosexual men, gay and bisexual men had more chronic general medical conditions (diabetes and cardiometabolic diseases) and mental health problems (depression and anxiety); moreover, they used fewer inpatient services related to substance use disorders. Most concerning were the high rates of tobacco, alcohol, and substance use disorders among the older lesbian and bisexual women in the sample compared with their matched sample of heterosexual women. This finding was consistent with the significant increase in inpatient lengths of stay and frequency of ED visits related to substance use disorders. This finding might be related to the age of the populations studied because it was not consistent with Frimpong et al.’s (13) study of lesbian and bisexual women, in which the authors reported fewer acute substance use service uses, shorter substance use inpatient stays, and fewer substance use–related ED visits. In summary, targeted prevention and intervention programs are needed to effectively engage older LGB adults into treatments for general and mental health illnesses as well as substance use disorders.