Originally published by The Sociological Quarterly in 2019
Although the overall trend in nursing home use has declined, this has not been the case for racial and ethnic minorities. This article elucidates the role of stressors and neighborhood context in nursing home admission. The analysis fits several regression models to predict the effects of stressors, discrimination, and neighborhood context on Blacks’, Hispanics’, and Whites’ odds of nursing home admission between 2010 and 2014 in the Health and Retirement Study. Results revealed that Blacks and Whites had similar odds of nursing home admission with one exception—adding SES to the model increased Blacks’ odds. Hispanics had significantly lower odds than Whites across all models and adding SES and financial strain increased the difference—as did, to a much smaller extent, everyday discrimination and neighborhood disorder. Although financial strain, everyday discrimination, and neighborhood disorder had no effect on Black–White and little influence on Hispanic–White differences, they were found to increase the odds of admission generally. As racial and ethnic minorities’ compared to Whites’ rates of nursing home admission increase in the coming years, it will become even more important to understand why.
Black and Hispanic persons have historically entered nursing homes at a lower rate than Whites, despite having a greater need due to higher rates of disability and poorer health (Akamigbo and Wolinsky 2007; Centers for Medicare and Medicaid Services 2015). However, the last decade has seen a decrease in the percentage of Whites in nursing homes and an increase in the percentage of Blacks and Hispanics. For example, the percentage of Whites in nursing homes declined from a little over 83 percent in 2007 to almost 75 percent in 2017. At the same time, the percentage of Blacks in nursing homes increased from 10 percent to 15 percent and the percentage of Hispanics increased slightly from 5 percent to 5.6 percent (Centers for Medicare and Medicaid 2008, 2017).
The increase in the Black and Hispanic relative to White nursing home population is surprising given cultural differences in expectations regarding elder care and past experiences with health care discrimination. The general sentiment regarding nursing homes is that they are places that are to be avoided, but perhaps even more so among Blacks and Hispanics, who strongly believe that elder care is the family’s responsibility (Dilworth-Anderson et al. 2005; Kirby and Lau 2010). In Hispanic culture, familism encourages family members to put the needs of the family first and to care for the young and old at home (Ruiz and Edward Ransfor 2012). Black caregivers express a greater desire for informal (e.g., respite) rather than for formal (e.g., institutional care) assistance (Desin et al. 2016). Additionally, in the health care system in general, minorities report experiencing discrimination and might even have a lingering distrust of health care providers due to past abuse and medical malpractice (American Medical Association 2016; Feagin and Bennefield 2014). Furthermore, racial/ethnic minorities are overrepresented in low-quality nursing homes that are characterized by financial instability, low staffing ratios, and serious deficiencies—reinforcing the notion that nursing homes are to be avoided at all cost (Mor et al. 2004; Shippee et al. 2016; Smith et al. 2007).
Increases in life expectancy, health status, and cumulative disadvantage could affect Blacks’ and Hispanics’ increasing nursing home admission. Between 2000 and 2014, Blacks gained 3.6 years in life expectancy compared to 2.6 years for Hispanics and 1.4 years for Whites (NCHS Data Brief 2016). Both Blacks and Hispanics are more likely to be living with physical disabilities in older age (Centers for Medicare and Medicaid Services 2015). Compared to Whites, Blacks are two times more likely and Hispanics are one and one-half times more likely than to be living with Alzheimer’s and other dementias (Alzheimer’s Association 2016). Longer life combined with a lifetime of exposure to poor living and working conditions, lower educational attainment, less access to health care, and other stressors are likely result in poorer health at older ages and a greater risk of nursing home admission.
Policy aimed at reducing nursing home utilization could account for some of the decline of Whites in the nursing home. A 1999 Supreme Court decision requires states to shift or rebalance the locus of Long-Term Care (LTC) from institutional settings to less restrictive and less costly community care. A decline in nursing home utilization followed (see Mathematica 2016 for an overview of states’ progress), however, rebalancing efforts have not been even across racial/ethnic groups. Blacks and Hispanics are less likely to live in places with community based LTC options. Furthermore, the high cost of Assisted Living and a lack of Medicaid beds make community options out of reach for many Blacks and Hispanics (Center for Medicare Advocacy 2016; Feng et al. 2011).
The high cost of nursing home care relative to alternatives and the projected population growth of Blacks and Hispanics make understanding the underlying factors that lead to nursing home admission important from a public health perspective.
Decades of theory and supporting research beginning with Du Bois’ (1899) seminal work explains health differences between Whites and racial/ethnic minorities as stemming from factors beyond biological, individual, and socioeconomic (SES) explanations. DuBois (1899), for example, in his analysis of health differences between Blacks and Whites in Philadelphia’s Seventh Ward, acknowledged the role of SES in health outcomes but viewed the conditions under which the two groups lived as having a distinct effect. Blacks faced higher poverty and worse working conditions but also lived in neighborhoods with higher crime rates, poor housing conditions, and faced discrimination and prejudice—conditions that even middle class Blacks could not escape. Recent research also calls into question SES as the driving force behind racial/ethnic health differences. Jay, Brown, and Hale (2017) used several measures of Socioeconomic Position (SEP) in their analysis of physical health differences between Blacks and Whites and found that while SEP contributes to some of the observed differences, it was not the primary factor—race was. They conclude that being Black—and race-based stressors and community conditions that Blacks experience—plays a significant role in physical health differences and that SEP is more important for health among Whites.
The Cumulative Advantage/Disadvantage (CAD) theory is a framework that can be used to understand how group differences in health come about over the life course. CAD can be defined as “the systemic tendency for interindividual divergence in a given characteristic (e.g., money, health, or status) with the passage of time” (Dannefer 2003, p. S327). With regard to health and aging, CAD focuses on cohort differences in exposures to health hazards early in life and how they widen health gaps in later life. Some health hazards are onetime events while others can be chronic in nature, and they are not attributable to factors such as chance or selective mortality but rather to events and circumstances that affect groups of individuals unevenly (Dannefer 2003). With regard to race/ethnicity and health among older people, CAD would predict, for example, that the effects of institutional discrimination, experiences of racism and interpersonal discrimination, lower SES, and poor neighborhood conditions would create gaps in health that would be evident in older age.
A large body of research describes how discrimination shapes access to care, health care utilization, quality of care, and subsequent health outcomes. Discrimination affects exposures to health hazards, the prevalence of disease and disability, trust and relationships with health care practitioners, and adherence to treatment (see Hall et al. 2015 for a review). Institutional discrimination is produced by policies that disproportionately and often negatively impact racial and ethnic minorities (Feagin 2006). The rebalancing of LTC from nursing homes to the community and its unintended effect on racial differences in nursing home admission described previously is one such example. Residential segregation is another example of institutional discrimination that has had far-reaching and negative impacts on Black Americans’ lives in particular. Due to racial segregation in housing Blacks compared to other racial/ethnic groups—regardless of class—have disproportionately less access to social goods and are exposed to neighborhood disorder or negative social conditions, which include unequal access to health care, employment, nutritious food, safe places to exercise, exposure to trauma, exposure to violence, and stress (Kramer and Hogue 2009; Massey and Denton 1993; Williams and Collins 2001). Regardless of income, Blacks consume less health care than Whites, indicating a lack of health care providers in the communities in which Blacks live (Charron-Chenier, Fink, and Keister 2017). For Hispanics, the effects of segregation are less clear, with some findings that show that living in Hispanic enclaves exposes individuals to poor living conditions, fewer and more dangerous jobs, and less access to and lower quality health care. However, living in Hispanic enclaves also provides protective factors like greater social networks and healthy behaviors (Bacon, Riosmena, and Rogers 2017; Xie and Gough 2011).
Another type of discrimination, interpersonal discrimination, involves race-based insults, slights, stereotypes, and microaggressions that take place in social interactions. Experiences of institutional and interpersonal discrimination can be thought of as “stressors.” According to the stress process model, when an individual experiences a stressful event, the body produces a fight or flight response, which in turn takes a toll on the body (Dohrenwend and Dohrenwend 1981; Juster, McEwen, and Lupien 2010; Pearlin et al. 1981). Blacks and Hispanics, compared to Whites, are more likely to report being exposed to traumatic events (especially violence) and more likely to report post-traumatic stress disorder as a result (Roberts et al. 2011). Other research has found that racial discrimination acts as a stressor and produces negative mental health outcomes (Carter 2007; Carter and Forsyth 2010). CAD posits that these stressors can accumulate over time, taking a further toll on physical and mental health (Dannefer 2003).
Finally, racial/ethnic differences in health status and mortality could result in differences in nursing home admission. Living longer and having higher rates of disability puts individuals at risk for nursing home admission. Blacks and Hispanics have higher rates of disability in older age than Whites (Centers for Medicare and Medicaid Services 2015) which is consistent with CAD theory. Yet, Hispanics’ life expectancy is similar to Whites’, and Hispanics have lower rates of chronic disease despite having lower SES on average—a phenomena referred to as the Hispanic Paradox. Possible explanations for this paradox have included the Salmon Hypothesis (health related immigration) and protection from disease and mortality afforded by strong social networks and cultural practices. Recent research provides evidence that the Hispanic Paradox is likely not driven by health-related immigration to and from the United States (see Bacon, Riosmena, and Rogers 2017; Ruiz et al. 2016; Ruiz, Steffen, and Smith 2013). Hispanics’ higher rates of disability and a life expectancy that is similar to Whites’ would make nursing home admission more likely, but lower rates of chronic disease makes admission less likely. Blacks have a lower life expectancy than either group, making nursing home admission less likely; however, Blacks who enter the nursing home are younger and living with higher levels of disability and cognitive impairments (Grabowski and Thomas 2009).
This study contributes to the body of literature knowledge about how CAD affects health and health care utilization and how experiences of discrimination, stressors, and neighborhood context contribute to racial and ethnic differences among older people.
Neighborhood Disorder & Unequal Access to Health Care
Most people prefer to remain in their homes and communities, aging in place, rather than uprooting to a nursing home (AARP 2011). The ability to age in place is, in part, dependent upon the characteristics of the neighborhoods in which older people live. As people age in place, the neighborhood becomes the most salient aspect in their lives, and neighborhood conditions have a long-term effect on physical and mental health, and the ability to access resources to remain in the community (Cagney, Browning, and Wen 2005; Glass and Balfour 2003). Neighborhoods that don’t support aging in place are characterized by a lack of transportation, less access to health and social services, social isolation, a lack of affordable housing, and a lack neighborhood safety (Smith 2009; WHO 2017).
Some studies indicate that there is a link between neighborhood context, health, and nursing home access and utilization among older adults. Nursing home closures, for example, are more likely to occur in poor, minority neighborhoods (see Feng et al. 2011). Black and Hispanic communities have fewer and lower quality nursing homes than White communities (Fennell et al. 2010). Furthermore, majority Black and Hispanic nursing homes suffer from severe staffing deficiencies (Lowenstein 2014). One study (Buys et al. 2013) found that among older adults, those who were living with physical impairments and living in disadvantaged neighborhoods were more likely to be admitted to a nursing home, although race was not a significant predictor. Kirby and Lau (2010) examined the racial makeup of neighborhoods and its effect on nursing home admission. They found that the proportion of Blacks in a community did not have an effect on the use of informal versus formal care. By contrast, Hispanics living in block groups that had a Hispanic population of 25 percent or more were more likely to receive informal care—a finding partially attributed to lower English proficiency, immigration status, and citizen status.
Financial Strain, Health, and Health Care Utilization
Blacks and Hispanics are more likely than Whites to experience financial strain in older age, affecting health and access to health care. Older Blacks and Hispanics are more likely than older Whites to report fair or poor health and some studies have found that Blacks’ and Hispanics’ lower self-reported health status is largely attributable to lower SES and in particular, wealth (e.g., Pollack, Page, and LaMontagne 2013). The historical exclusion of Blacks from wealth-generating activities such as self-employment, home ownership, and rising property values—measured by the asset accumulation of the previous generation and intergenerational transfers—has left Blacks with few assets in retirement (Oliver and Shapiro 2006). More recently, the economic recession and declining home values have negatively impacted Blacks and Hispanics more than Whites (Kochhar, Fry, and Taylor 2011). In fact, older non-Hispanic Whites are the least likely group to be living in poverty and have more financial resources in older age (Administration on Aging 2018). While 7 percent of Whites age 65 and older live in poverty, the percentage of older Blacks and Hispanics living in poverty is much higher (18 percent and 20 percent, respectively; Cubanski et al. 2018). With regard to wealth, the median wealth of White families in 2011 was $111,146, compared to $7,113 for Black families and $8,368 for Hispanic households (Traub et al. 2016).
The inability to privately pay for care has been shown to disproportionately concentrate Blacks and Hispanics in low-quality nursing homes (Fennell et al. 2010; Rahman and Foster 2015; Smith et al. 2007; White, Haas, and Williams 2012). Older Blacks and Hispanics are more likely to rely on Medicaid than Whites due to poorer health, a greater need for LTC, and lower incomes (Meyer and Frasier 2013). Over 60 percent of people living in nursing homes rely on Medicaid as the primary source of payment (KFF.Org 2018). Higher quality facilities tend to have a larger proportion of private pay residents while lower quality facilities tend to have a higher proportion of residents paying with Medicaid (Kochhar, Fry, and Taylor 2011). Furthermore, Whites are more likely to have community options like Assisted Living available to them due to the ability to private pay (Centers for Medicare and Medicaid, 2017).
Discrimination, Health, and Health Care
Discrimination influences health. Researchers have found, in a national survey of adults, that those with disadvantaged statuses (and especially those with more than one disadvantaged status) were more likely than non-disadvantaged persons to report mental and physical health problems as well as more functional limitations and that multiple forms of discrimination, at least partially, explained the relationship (Grollman 2014). In a sample of Black older adults, Mouzon et al. (2017) found that both race-based and non-race-based experiences of everyday discrimination were related to worse mental health. Some research on discrimination and older adults shows that Whites are more susceptible to its effects on health. Barnes and colleagues (2008) found that perceived discrimination was a stronger predictor of mortality for Whites compared to Blacks. Similarly, another study (Ayalon and Gum 2011) found that experiences of everyday discrimination were negatively related to mental health more so for Whites than Blacks. The researchers posited that because Blacks have had a lifetime of experience with discrimination, older Blacks have developed coping mechanisms and social support, whereas older Whites, who may be experiencing discrimination for the first time, have not developed such buffers.
Discrimination influences the way that different racial and ethnic groups interact with the health care system. Whites, for example, are less likely than Blacks or Hispanics to have had a race-based negative experience with the health care system. Age, preferred language, and cultural differences could make these experiences worse. Older Blacks, for example, may have a cultural mistrust of the system attributable to remembering past medical abuses and legal, racial segregation of health care facilities (Smith 1998). Hispanics whose preferred or first language is Spanish may have difficulty communicating with health care workers and understanding/navigating the health care system (White, Haas, and Williams 2012). In a recent survey, 16 percent of Hispanics reported feeling that nursing homes could not provide for cultural needs, including preferred spoken and written language (AP-NORC 2017). There is also a concern about the lack of racial, ethnic, and cultural diversity among health care providers. Medical professionals have been found to associate negative words with Blacks and Hispanics, indicating racial bias (Bean et al. 2013; Blair et al. 2013; Sabin, Rivara, and Greenwald 2008). With regard to end of life care, both Blacks and Hispanics have been found to use hospice less than Whites and are less likely to have advanced care directives such as a living will—findings that are often attributed to cultural and religious differences (although one study found in a NJ sample that differences were largely attributable to formal and legal obstacles; Carr 2011).
In sum, experiences of discrimination negatively affect health and possibly accumulate. Therefore, those who experience discrimination are expected to have worse health and higher health care utilization. However, for racial and ethnic minorities, past experiences of discrimination in health care could lead to lower health care utilization.
Health care utilization and, in particular, nursing home use varies by race and ethnicity. Blacks and Hispanics are less likely than Whites to use nursing home care despite having a greater need. Therefore, Blacks’ and Hispanics’ lower use is likely due to factors above and beyond need. The goal of the current study is to answer two research questions: Are Blacks and Hispanics more or less likely than Whites to enter the nursing home? Second, controlling for other factors, how do eventful and chronic stressors, experiences of discrimination, and neighborhood conditions affect racial/ethnic differences in nursing home admission? The following hypotheses follow from prior research and will be examined in the present study: H1:
Although Blacks and Hispanics will have a greater need due to health than Whites, they will be less likely to enter a nursing home.H2:
Eventful and chronic stressors will increase the odds of Blacks and Hispanics entering a nursing home because these groups experience more stressors than Whites, which in turn affects health.H3:
Although Blacks and Hispanics may experience discrimination in the health care system more often than Whites, thus making it less likely to seek care, this study expects that experiences of discrimination will increase the odds of Blacks and Hispanics entering a nursing home because discrimination has been linked to poorer health.H4:
Blacks and Hispanics are more likely to live in neighborhood contexts that experience disorder and a lack of nursing facilities, making nursing home admission less likely.
Design and Methods
Data and Sample
Data come from the Health and Retirement Study (HRS), a nationally representative, longitudinal survey of adults aged 50 and over. The HRS is sponsored by the National Institute on Aging and is conducted by the University of Michigan. The core survey has been administered every two years since it began in 1992 (Health and Retirement Study, 2014). This study also uses the Psychosocial and Lifestyle Questionnaire, which has been administered alongside the HRS and during off years. This study uses HRS data because it contains in-depth health, financial, and psychosocial data pertaining to older adults. Additionally, the survey oversamples Blacks and Hispanics, and the Psychosocial and Lifestyle Questionnaire contains questions relating to eventful and chronic stressors, discrimination, and neighborhood conditions (Health and Retirement Study, 2008).
The sample selection process occurred in several stages. First, individuals who had never been admitted to a nursing home and who were not currently institutionalized, those who were 65 or older (the age at which nursing home admission begins to become more common), and those who were administered either the 2006 or 2008 Psychosocial and Lifestyle Questionnaire were selected, along with their demographics from the HRS Tracker file. This study excludes those who were not able to be identified in the data as being Black, White, or Hispanic (n= 239). Second, individuals from the HRS Tracker file (n = 8,781) were merged with their responses to either the 2006 or 2008 Psychosocial and Lifestyle Questionnaire (half of the sample was administered the survey in 2006, the other half in 2008). SAS’ PROC MI procedure was used to examine patterns of missing data and revealed no clear pattern across groups and a small percentage of cases with missing data (6 percent). Comparing imputing data for these cases (n = 561) to the observed data showed no improvement in model fit or predictive ability. Therefore, these cases were deleted. Third, RAND HRS Data, Version O (RAND Center for the Study of Aging 2016) regarding respondents’ health status, nursing home admission, and SES were merged onto the file. The RAND HRS Data file is an easy to use version of the HRS data. The analyses are based on 6,769 Whites, 926 Blacks, and 525 Hispanics for a total sample of 8,220 individuals.
Nursing home admission (1 = yes, 0 = no) is the outcome variable. Respondents were asked if they entered a nursing home, convalescent home, or other long-term health care facility within the past 2 years. Respondents were tracked forward from either 2006 or 2008 (depending on when they were administered the Psychosocial and Lifestyle Questionnaire) through 2014 or the most recent available HRS survey.
Race/ethnicity—the primary independent variable—is measured as a series of mutually exclusive dummy variables indicating whether a respondent is non-Hispanic Black (1 = yes, 0 = no), Hispanic (1 = yes, 0 = no), or non-Hispanic White (1 = yes, 0 = no). Whether a respondent was foreign born is included (1 = foreign born, 0 = U.S. born).
Several factors contribute to individuals’ increased risk of nursing home admission, regardless of race and ethnicity. These include older age, being unmarried, being female, lower income, recent hospitalization, and having a greater need with Activities of Daily Living (ADLs; see Gaugler et al. 2007 for a review). Key demographic control variables are measured as gender (1 = woman, 0 = man), marital status (1 = married, 0 = other marital status), and age (65 or older; continuous). Whether a respondent receives Medicaid benefits is measured as a dichotomous variable (1 = yes, 0 = no).
ADLs, Instrumental Activities of Daily Living (IADLs), and hospitalization are important indicators of older adults’ health status and nursing home admission (Boyd et al. 2008; Gill et al. 2009; Han, Barnard, and Chapman 2009; Wong and Miller 2008). ADLs were measured as an index (0–5). These items include any difficulty with dressing, bathing, eating, bed transfers, and walking across a room. An index was also created to measure difficulty with performing IADLs (0–5). These items include any difficulty with using the phone, taking and handling money, shopping, and preparing meals. Prior hospitalizations come from a question that asks if the respondent stayed overnight in a hospital during the past two years (1 = yes, 0 = no).
Three variables measure SES: wealth, education, and Medicaid benefits. Wealth is measured as the log transformation of the respondent’s household, non-housing assets minus debt. Prior to the log transformation, respondents who reported negative wealth were assigned a value of zero in order to retain them. Those reporting zero wealth were assigned a constant (+1) prior to the transformation. Education is measured as a continuous variable representing the number of years of education (0–17). Whether a respondent receives Medicaid benefits is measured as a dichotomous variable (1 = yes, 0 = no).
Eventful and Chronic Stressors
Eventful and chronic stressors are comprised from questions in the HRS Leave Behind Psychosocial survey. They are an inventory of the number of affirmative responses to 10 questions about traumatic events (e.g., have you ever been in a major natural disaster, were you the victim of a serious physical attack in your life, were you ever physically abused in your life), 5 questions about stressful events (e.g., have you involuntarily lost a job, have you moved to a worse residence or neighborhood, were you robbed or burglarized), and 1 question about financial strain (how difficult is it to meet monthly payments on your bills). More detailed information regarding the construction of these variables is available here: http://hrsonline.isr.umich.edu/sitedocs/userg/HRS2006LBQscale.pdf
Two measures of discrimination were created. The first is an index of 5 questions regarding experiences of everyday discrimination (e.g., you are treated with less courtesy or respect than others, people act as if they think you are not smart, you are threatened or harassed, Cronbach’s α = 0.80). The second is the sum of affirmative responses to 6 questions measuring experiences of major lifetime discrimination (e.g., have you ever been unfairly dismissed from a job, have you ever been unfairly denied a bank loan, have you ever been unfairly stopped, searched, questioned, physically threatened, or abused by the police). More detailed information regarding the construction of these variables is available here: http://hrsonline.isr.umich.edu/sitedocs/userg/HRS2006LBQscale.pdf
The neighborhood conditions measures are made up of two indexes. The first asks respondents 8 questions about neighborhood disorder in the community in which they live (e.g., people would be afraid to walk alone, vandalism and graffiti are a big problem in this area, and there are many vacant or deserted houses or storefronts in this area, Cronbach’s α = 0.74). The second asks about neighborhood cohesion (e.g., people feel safe walking alone in this area, there is no problem with vandalism and graffiti in this area, and there are no vacant or deserted houses or storefronts in this area, Cronbach’s α = 0.85). More detailed information regarding the construction of these variables is available here: http://hrsonline.isr.umich.edu/sitedocs/userg/HRS2006LBQscale.pdf
Table 1 summarizes the measures used in the current study.
The analysis begins by providing descriptive statistics by race/ethnicity and statistical tests for differences for each variable. Pooled two sample t-tests were used to test significance when the variances were equal and Satterthaite t-tests were used when the variance was unequal. Next, the logistic regression models proceed in several steps. The first model estimates the baseline relationship between race/ethnicity and nursing home admission. Model 2 adds health status variables, ADLs, IADLs and prior hospitalization. Model 3 adds wealth, years of education, and Medicaid benefits. Model 4 adds eventful and chronic stressors (traumatic events, stressful events, and financial strain). Model 5 adds interaction effects between race/ethnicity and financial strain (the only statistically significant interaction between race/ethnicity and key psychosocial variables). Model 6 examines discrimination stressors including perceived daily discrimination and major experiences with discrimination. Model 7 adds neighborhood conditions. Model 8 includes all variables to show the full relationship.
Table 2 provides descriptive statistics by race and ethnicity. The proportion of Blacks (0.16) who enter the nursing home during the study period does not significantly differ from the proportion of Whites (0.17). However, a significantly lower proportion of Hispanics (0.10) compared to Whites (0.17) enter the nursing home. Blacks and Hispanics report needing help with ADLs and IADLs more than Whites. Additional analyses (not shown) revealed that traumatic events and financial strain predicted help with ADLs, IADLs, and hospitalization, while perceived everyday discrimination predicted ADLs and IADLs, and experiences with major discrimination and neighborhood cohesion predicted ADLs. With regard to SES, Blacks and Hispanics have less wealth, fewer years of education, and are more likely to report Medicaid benefits than Whites. Blacks are more likely than Whites to report traumatic events, and Blacks and Hispanics are significantly more likely than Whites to experience stressful events and financial strain. Blacks and Hispanics report higher averages than Whites for the neighborhood cohesion measure. Blacks compared to Whites report higher levels of perceived everyday discrimination, major discrimination, and neighborhood disorder.
Table 3 reports odds ratios from logistic regression analyses of nursing home admission. The results for Model 1 suggest that being Hispanic compared to White is associated with a decrease in the odds of nursing home admission (0.61), which is consistent with Hypothesis 1. Being Black compared to White was not associated with a significant difference in the odds of nursing home admission which is inconsistent with Hypothesis 1. Being foreign-born compared to U.S. born also showed no significant difference. The odds of nursing home admission are 1.29 higher for women than men. Being married was associated with a decrease in the odds of nursing home admission while older age was associated with an increase.
Model 2 fits health status variables to the baseline model. Those who needed help with a greater number of ADLs and IADLs, and those who experienced a hospitalization during the study period had higher odds of nursing home admission. Being Hispanic was associated with decreased odds (0.54) compared to Whites. The chi-square difference between Model 2 and the baseline is significant (p < 0.001).
Model 3 explores the role of SES in nursing home admission. Higher levels of wealth are associated with decreased odds of nursing home admission. Conversely, greater years of education and being a Medicaid recipient increase the odds. In this model, being Black compared to White was significantly associated with decreased odds of nursing home admission (0.76). Comparing the change in direction and significance of the logistic regression coefficients (not shown in table) from Model 1 (0.01) to Model 3 (−0.28*) indicates that SES is a suppressor variable when comparing Blacks to Whites in nursing home admission. Adding SES to the model magnifies (as opposed to reducing or mediating) the effect of being Black compared to White on nursing home admission (see MacKinnon, Krull, and Lockwood 2000 for a review of mediation, confounding, and suppression effects). SES also magnifies the effect of being Hispanic compared to White on nursing home admission (0.45) compared to the baseline model (0.61). The chi-square difference between Model 3 and the baseline is significant (p < 0.001).
Model 4 adds eventful and chronic stressors to the baseline model. Results were largely inconsistent with Hypothesis 2. Neither traumatic events nor stressful events significantly predicted nursing home admission. Financial strain, however, was associated with an increased odds of nursing home admission (1.17). Being Hispanic compared to White was associated with lower odds of being admitted to a nursing home (0.56). Financial strain magnifies the relationship between Hispanic ethnicity and nursing home admission from the baseline model, indicating a suppressor effect not predicted by Hypothesis 2. The chi-square difference between Model 4 and the baseline is significant (p < 0.001).
Model 5 adds interaction terms between race/ethnicity and financial strain (the only statistically significant interaction between race/ethnicity and psychosocial variables). Figure 1 illustrates how the predicted probabilities of nursing home admission varied between Whites and Hispanics across levels of financial strain. Hispanics had lower probabilities of nursing home admission compared to Whites across all levels of financial strain, a trend that increased as levels of financial strain also increased. The chi-square difference between Model 5 and the baseline is significant (p < 0.001).
Figure 1. Predicted probability of nursing home admission by race/ethnicity and financial strain with 95 percent confidence limits.
Model 6 fits discrimination stressors. Consistent with Hypothesis 3, perceived everyday discrimination was associated with an increase in the odds of admission (1.24). Experiences of major discrimination were not significantly associated with admission, a finding that is inconsistent with Hypothesis 3. Hispanics’ odds (0.60) of admission were less than Whites’, but not much less than their odds in the baseline model (0.61), which is inconsistent with Hypothesis 3. The chi-square difference between Model 6 and the baseline is significant (p < 0.001).
Model 7 adds neighborhood conditions. Neighborhood disorder was associated with an increase in the odds of nursing home admission (1.04). Neighborhood cohesion did not significantly affect odds of admission. Hispanics had lower odds (0.59) of admission than Whites, which is consistent with Hypothesis 4, however, their odds were not much lower than in the baseline model. The chi-square difference between Model 7 and the baseline is significant (p < 0.001).
Model 8 fits the full model. Hispanics’ odds (0.60) of nursing home admission were significantly lower than Whites’. Being married was also associated with a significant decrease in the odds of a nursing home admission. Being a woman, older age, needing greater assistance with ADLs and IADLS, hospitalization, higher levels of education, receiving Medicaid benefits, perceiving everyday discrimination, and neighborhood disorder were all significantly related to a higher odds of admission. The interaction effect for Hispanic*Financial Strain was also significant. The chi-square difference between the full model and the baseline model is significant (p< 0.001).
Overall the results show that Blacks and Hispanics are less likely to enter the nursing home compared to Whites, which is consistent with Hypothesis 1. While Hispanics were significantly less likely than Whites to be admitted in all of the models, Blacks were only significantly less likely than Whites when the SES variables were added to the baseline model. SES acts as a suppressor for Blacks and Hispanics, magnifying the relationship between race/ethnicity and nursing home admission. Of all the variables considered in this study, SES had the largest effect on Black-White and Hispanic-White differences in admission when compared to the baseline model.
Financial strain, perceived everyday discrimination, and neighborhood disorder were associated with an increased odds of admission in general, however, they had no effect on Black-White differences, and a very small suppressor effect on Hispanic-White differences—findings that are inconsistent with Hypotheses 2, 3, and somewhat with Hypothesis 4. The final model which contained all study variables, a significant interaction, and controls revealed no significant differences in the odds of admission between Blacks and Whites, and that Hispanics had significantly lower odds of admission compared to Whites.
Discussion and Implications
This study contributes to the body of research on health care differences by adding knowledge about how stressors, discrimination, and neighborhood context affect nursing home admission. It also focuses on an older population and racial and ethnic minorities which could provide information about the future needs of this population, a major public health concern. Previous research has largely focused on how a health determinate model or socioeconomic factors affect differences. Findings here show that a health determinate model does not significantly affect the observed Black-White differences in nursing home admission but that it does magnify Hispanic-White differences. Black-White disparities in nursing home admission only appeared after the addition of SES, indicating the importance of including SES in analyses that wish to examine similar differences in admission. The relationship between Hispanic ethnicity and nursing home admission is also better understood when SES is taken into account. These findings are largely consistent with CAD theory and reconfirm the importance of SES in racial differences in health care utilization. This study also extends our understanding by looking at inequalities that stem from social determinates of health and access to care.
Findings revealed that the magnitude of Hispanic-White differences in nursing home admission was masked until stressors, discrimination, and neighborhood context were taken into consideration. Although the effect was small, accounting for stressors, discrimination, and neighborhood context resulted in Hispanics, who already had lower odds than Whites, being even less likely to enter the nursing home despite worse health. In fact, when considering increasing levels of financial strain, Hispanics compared to Whites were even less likely to enter the nursing home. Interestingly, the proportion of Hispanics compared to Whites who reported experiencing discrimination and neighborhood disorder was not significantly different, and yet, not including them in the analytic model appeared to slightly mask the overall size of Hispanic-White differences in nursing home admission. One possible explanation is that Hispanics have a greater preference than Whites for home and community-based care. Hispanics in this study reported high levels of neighborhood cohesion, which could reduce feelings of vulnerability in older age and make living in the community an attractive option. Furthermore, a cohesive neighborhood could buffer against experiences of discrimination, especially in largely Hispanic communities. Hispanic elders might fear institutionalization because nursing homes might not be sensitive to cultural needs. Because the suppression effect of these variables in this study was small, more research is needed before a definitive conclusion can be made about the role of discrimination and neighborhood disorder in Hispanic-White differences in nursing home admission.
Black-White differences in admission followed a pattern that was different than Hispanic-White ones, which highlights the importance of going beyond White-minority only comparisons. Although Blacks reported higher levels of stressors, discrimination, and neighborhood disorder, these factors didn’t seem to significantly affect Black-White differences in admission. Instead, Blacks were only significantly less likely than Whites to enter to the nursing home when SES was taken into account. These findings could indicate a tension between need, the ability to private pay or use Medicaid benefits, and the availability of LTC choices.
This study also contributes important findings above and beyond documenting racial/ethnic differences. Regardless of race/ethnicity, psychosocial and structural variables—stressors, discrimination, and neighborhood conditions—shed light on why some individuals enter the nursing home. Those who reported financial strain, perceived everyday discrimination, and neighborhood disorder had higher odds of nursing home admission. These findings stand in contrast to predictions made using the CAD hypothesis, which would predict that racial and ethnic minorities in particular would be more likely to enter the nursing home due to experiences of multiple forms of discrimination, worse health, and lower SES. Although Blacks and Hispanics were disadvantaged compared to Whites on many of the measures in this study, they were less likely to enter the nursing home. One possible explanation is that Blacks and Hispanics are advantaged by strong social networks and coping mechanisms that buffer against discrimination and disadvantages. This explanation more closely aligns with Ferraro and Shippee’s (2009) Cumulative Inequality Theory, which allows for a less deterministic, nonlinear relationship between disadvantages, advantages, and life course trajectories. Future research should explore how certain racial and ethnic groups develop mechanisms to counteract cumulative disadvantage, if population truncation is giving the appearance of decreasing inequality, how individual agency shapes the life course trajectory after exposure to disadvantage, and how individuals use the resources available to them.
Despite its unique contributions, this study contains several limitations. The first limitation is that Black-White and Hispanic-White comparisons were made without the ability to make other racial and ethnic comparisons due to small sample sizes. The second limitation lies in the inability to analyze home and community based care as an alternative to nursing home admission. The HRS does include some measures of informal caregiving and some home health care information, however, the data are limited. The fourth limitation is the inability to assess the quality of the nursing homes in the study. Other research has assessed nursing home quality by race and has found that minorities are concentrated in lower quality facilities (Center for Medicare Advocacy 2016). Future research should track nursing home risk over time, include a more racially diverse sample, examine differences due to experiences of discrimination by health care providers, and consider other alternative long-term care arrangements that might be available to an individual.
Shifts in the U.S. population and the changing demographics of the nursing home population make research on LTC differences even more important. Understanding why some groups enter the nursing home and not others could help policy makers and the nursing home industry to anticipate future care needs and improve quality of care in nursing homes for individuals of many different races and ethnicities, especially with regard to cultural preferences.