https://orcid.org/0000-0002-5454-1705
Abstract
To examine whether racial disparities are narrowing or widening with historical time among U.S. middle-aged and older adults, and test the extent to which educational attainment moderates racial disparities over historical time.
Multilevel models were applied to longitudinal data on middle-aged (ages 40–65) and older adults (ages 66 and older) from the Health and Retirement Study. Historical change was indexed as cohort or birth year. The outcomes of focus were depressive symptoms, episodic memory, health conditions, functional limitations, and self-rated health.
Results revealed a differential pattern of racial disparities in historical change between midlife and old age. Across midlife and old age, on average, Blacks and Hispanics reported poorer levels of mental and physical health, compared with Whites. In midlife, racial disparities narrowed with historical time; later-born cohorts of Hispanics but not Whites reported fewer depressive symptoms than their earlier-born peers. Likewise, historical improvements in health were stronger among Hispanics and Blacks than Whites. Conversely, in old age, later-born cohorts across race consistently showed historical improvements in each of the outcomes examined. Regarding educational attainment, we observed little consistent evidence that health-promoting effects of educational attainment differ across race and cohort. Examining questions about heterogeneity, results revealed that in midlife and old age there was greater heterogeneity between race across each of the outcomes.
Our discussion elaborates on reasons behind the documented racial differences in historical changes among U.S. middle-aged and older adults, and how the protective role of education is changing over time.
The course of development is shaped by the sociocultural and historical contexts in which persons live (Bronfenbrenner, 1986; Elder, 1975; Nesselroade & Baltes, 1979; Wahl & Gerstorf, 2018). The last decade has seen an influx of studies examining the nature of historical (cohort) change across midlife and old age. Research from Case and Deaton (2015) brought to light increasing rates of deaths of despair in middle-aged adults, particularly among non-Hispanic White men with a high school education. Following the Great Recession, U.S. middle-aged adults report poorer psychological well-being and exhibit more daily stress and poorer physical health than same-aged peers in the 1990s (Almeida et al., 2020; Kirsch et al., 2019). Findings of historical declines in midlife health and well-being contrast with what has been typically found with older adults. A recent review by Gerstorf et al. (2020) concluded that later-born cohorts of older adults in countries like the United States are exhibiting better performance in cognition and report better physical health and well-being, compared with earlier-born cohorts. A notable limitation of existing research on historical changes in the health, well-being, and cognition of middle-aged and older adults is the lack of consideration of race. Population-level estimates have shown that not only is the United States a graying society, but it is becoming more racially and ethnically diverse, which further signifies the importance of examining whether existing racial disparities and inequities have been increasing or decreasing over historical time among middle-aged and older adults (Colby & Ortman, 2015; Olshansky, 2016).
Our objective is to put these trends in larger perspective by directly comparing the nature of historical changes of mental and physical health in midlife and old age across Whites, Blacks, and Hispanics in the United States. The extent of observed trends of historical declines in the health and well-being of middle-aged adults is yet to be explored, but improvements in older adults are experienced similarly across race. We also consider the role of years of education as a resource that could lessen historical declines or promote improvements across cohorts.
Historical Change in Mental and Physical Health Across Midlife and Old Age
Studying the impact of historical or cohort change takes into consideration societal-level events, normative changes, and other processes (e.g., transforming education systems) that transpire over decades and can have significant implications for individual development (Baltes & Carstensen, 1996). A prime example is research showing that being born before or after the Great Depression had differential implications for development later in life (Elder, 1975). The most common approach to studying historical change has been to examine similarities and differences between same-aged people who were born in different historical times. From reviewing research on old age (65 years and older), for example, Gerstorf et al. (2020) inferred that older adults are faring much better than their same-aged peers born earlier across indices of mental and physical health, as well as psychosocial indices and cognition. Similarly, research has observed cohort differences in late-life adjustment to widowhood; Perrig-Schiello et al. (2016) found that widowed individuals in 2011 reported fewer social and financial difficulties than their counterparts in 1979. At the same time, widowhood-related changes in depressive symptoms and psychological difficulties were consistent across cohorts.
Research examining cohort differences in midlife has painted a different picture as to what has been found in older adults. Case and Deaton (2020) observed that current cohorts of middle-aged adults are reporting more pain and perceive their health to be worse than middle-aged adults in earlier historical times. Numerous studies have found that middle-aged adults following the Great Recession report poorer daily well-being and overall mental and physical health (Almeida et al., 2020; Kirsch et al., 2019). When compared with their same-aged peers in Europe and Asia, cohorts of middle-aged adults in the United States exhibit higher rates of disability and chronic illness (Avendano et al., 2015; Lee et al., 2018). More recently, Infurna et al. (2021) took a cross-national approach to examine whether such trends of historical decrements in midlife functioning are specific to the United States or similarly occurring in other nations. They observed that later-born cohorts of U.S. middle-aged adults are exhibiting historical declines in mental health and cognition and only slight gains in physical health, whereas later-born cohorts in Germany, South Korea, and Mexico are showing historical gains consistently in mental and physical health.
Racial Disparities
A substantial amount of research has explored racial disparities in adulthood and old age across a diverse set of outcomes (Williams, 2012), but not whether such disparities are increasing or decreasing over historical time. Generally speaking, most of this literature finds poorer outcomes among Black and Hispanic older adults that can be attributed to social disadvantages (Weden et al., 2017). When focusing on health, Blacks face rates of diabetes, hypertension, and obesity that are 25%, 49%, and 59% higher than those found in Whites, respectively. Hispanics experience rates of diabetes and obesity that are 25% and 50% higher than their White counterparts (National Center for Health Statistics (US), 2016). Blacks have also shown to have lower life expectancies, worse physical functioning, higher rates of physical and mental illnesses, and steeper declines in health and cognitive functioning over time (Asnaani & Hall-Clark, 2017; McClendon, Bogdan, et al., 2019; McClendon, Jackson, et al., 2019; Thorpe et al., 2016; Zahodne, Sol, & Kraal, 2019). A more complex picture emerges for Hispanics that has been coined the Hispanic Paradox. The association between health and socioeconomic status (SES) is mixed for Hispanics; foreign-born Hispanics have better outcomes on aspects of health than White Americans, independent of SES (Markides & Eschbach, 2005; Ruiz et al., 2013; Weden et al., 2017). Despite these differences across race, many questions still remain regarding the social, physical, and economic conditions that lead to or result from these inequalities.
Reasons for Differences in Historical Change Across Race
It stands to reason that the noted racial disparities could have changed over the past decades because the noted historical shifts may have affected some groups more than others. To begin with, Infurna et al. (2020) have argued that middle-aged adults are contending with many more challenges in the realm of intergenerational dynamics (caregiving for aging parents and raising children) and are suffering more from financial vulnerabilities than their age peers several decades ago. For example, there are more multigenerational households today and marriage and first children being born are occurring later in life than in previous cohorts. Importantly, such considerations differ by race. To illustrate, Black and Hispanic caregivers are more likely to be coresidents with their families, spend more time providing care, and use fewer formal services of care than their White peers (Johnson & Wiener, 2006). Similarly, Blacks and Hispanics are more likely to be custodial grandparents (Hayslip et al., 2017), have less access to school activities and sports, and rely more on families than their White peers (Vincent & Maxwell, 2016). Likewise, employment patterns often dictate financial stability among racial minoritized groups. It is long known that Blacks and Hispanics throughout their work life have less access to pension or retirement plans, which in turn increases the risk for retirement income insecurity (Gassoumis, Wilber, and Torres-Gil, 2009).
Historical Dynamics Across Education in Midlife and Old Age
Individual differences in educational attainment are known to be associated with health inequalities and longevity (Oakes & Rossi, 2003). We conceptualize educational attainment as one form of SES to study insights into broader social dynamics (see Stephens et al., 2014). People in low SES strata, on average, are faced with higher risks for poor mental and physical health (Adler et al., 1994). In the United States, these well-documented SES disparities are often closely intertwined with racial and ethnic differences (Crimmins & Saito, 2001; Guralnik et al., 1993; Olshansky, 2016). For example, studies have shown that a large part of the racial divide is still driven by SES disparities despite seeing reductions in rates of mortality and morbidity over the years (Do et al., 2012). Studies have also shown that racial inequities in health increased in magnitude as educational attainment increases (Bell et al., 2020). Because mental and physical health are strong predictors of mortality and morbidity, fully explicating racial differences in health is integral to understanding these race inequities (Beck et al., 2014; Landrine & Corral, 2014).
Existing research suggests that educational attainment is a key component in understanding development in midlife and old age in the context of historical change. Previous research shows that historical declines in midlife are exacerbated for individuals who attained fewer years of education (Kirsch et al., 2019) and even attaining more years of education may not be as beneficial for later-born cohorts when compared with earlier-born cohorts (Infurna et al., 2021). Health disparities are well-documented in the United States and in part reflect racial and ethnic differences as well as educational differences directly linked to behavior (Adler & Ostrove, 1999; Crimmins et al., 1989; Crimmins & Saito, 2001). One caveat to the issue of health disparities and educational attainment is that there is some controversy as to whether educational attainment conveys similar benefits across race. Glymour and Manly (2008) suggest that educational attainment connotates different meanings across race, such that the use of quality of education and literacy differs between Whites and Blacks (Sisco et al., 2015).
The Present Study
Our overarching goal is to examine whether the historical declines in midlife and historical improvements in older adults’ health, well-being, and cognition are similarly experienced across Whites, Blacks, and Hispanics in the United States. Using historical change as a lens constitutes a tool that helps us identify how and why gaps between population segments currently exist, and whether these have been narrowing or widening over the past decades. To address our research questions, we utilize longitudinal panel data from the Health and Retirement Study. Given structural inequalities across macro-level processes and daily experiences of discrimination that accumulate over time (Barnes et al., 2012; Ferraro et al., 2009), it is possible that historically increased challenges in midlife are particularly pronounced among racially minoritized groups. On the other hand, because of differential patterns of family formation across Blacks and Hispanics, compared with Whites, this could have created a context in which family lives differ, resulting in less pronounced differences in historical change (Carr & Utz, 2020). We also seek to empirically test whether these historical trends are generalizable across or differ by levels of educational attainment. We expect that, generally speaking, those individuals who have more years of education to exhibit better overall mental and physical health given that educational attainment has generally led to the availability of better jobs, more job security, access to health care and higher pay (Case & Deaton, 2020). Focusing on educational attainment in the context of race, we expect that obtaining more education operates as a protective resource that buffers racial inequalities in mental and physical health.
Method
Participants and Procedures
We used data from the Health and Retirement Study (HRS), which is a nationally representative sample of households in the United States of adults aged 50 years and older and their spouse (spouses younger than age 50 were included as participants). Participants provide biennial reports on a wide range of measures including economic, sociological, psychological, mental, and physical health information. The HRS collects data through in-person and telephone interviews and every 6 years recruits a new cohort of participants to refresh the sample. The HRS has made efforts to oversample racially minoritized groups in recently included cohorts. Descriptions of participants, procedures, and data accessibility for the HRS are reported in previous publications (McArdle et al., 2007). We use biennial data obtained between 1992 and 2018 on functional limitations ([instrumental] activities of daily living, [I]ADL; Rodgers & Miller, 1997), self-rated health (single-item), health conditions, depressive symptoms (Center for Epidemiologic Studies—Depression scale (CES-D); Radloff, 1977), and episodic memory (immediate and delayed recall).
Because of our focus on examining racial differences in cohort effects in midlife and old age, we have two samples for our analyses. One sample includes persons who provided observations while they were in midlife during the course of the study (ages 40–65) and the other sample includes persons who provided observations while they were in old age during the course of the study (ages 66 and older). Our total sample for our analyses pertaining to midlife included 28,892 participants who were aged 40–65 during the course of the study. The birth years represented in the data ranged from 1930 to 1969 (M = 1948, SD = 10.62); 55% were women; attained, on average, 12.78 years of education (SD = 3.08); and 65% of our sample was White, 21% Black, and 14% Hispanic. For old age, we included 23,955 participants who were aged 66–103 during the course of the study. If the participants were younger than age 65, then they were a part of the midlife sample. However, once they crossed the age of 65, they became a part of the old age sample, meaning that an individual could take part in both groups. Participants who turned from midlife into old age during the course of study were part of both samples. The observations in which they were younger than age 66 were used in the midlife analyses and observations in which they were age 66 and older were used in the old age analyses. The birth years represented in the data ranged from 1900 to 1952 (M = 1938, SD = 8.06); 54% were women; attained, on average, 12.33 years of education (SD = 3.31); and 73% were White, 17% were Black, and 10% were Hispanic.
Measures
Depressive symptoms
Eight items from the CES-D scale (Radloff, 1977) were used to assess depressive symptoms. Items asked participants whether they had (1) or had not (0) experienced the following symptoms “much of the time during the past week”: feeling depressed, everything was an effort, restless sleep, was not happy, felt lonely, did not enjoy life, felt sad, and could not get going. The sum across items is taken as an indicator of the number of depressive symptoms an individual experienced frequently (range is 0–8). The shorter scale with different response format (yes or no vs. several response categories ranging from rarely or none of the time to most or all of the time) used in the HRS has demonstrated highly similar construct and external validity as the standard CES-D (see Kohout et al., 1993; Steffick, 2000).
Functional limitations
Functional limitations were measured using a composite sum index of the number of everyday activities participants reported having any difficulty completing, including walking several blocks, climbing one flight of stairs, pushing or pulling large objects, lifting or carrying 10 lb (4.53 kg) of weight, and picking up a dime (range is 0–5). Higher scores represent greater functional limitations or poorer physical functioning (Rodgers & Miller, 1997). Although abbreviated versions of standard activities of daily living (ADL) and instrumental activities of daily living questionnaires (IADL) were used, the HRS’s measures of functional limitations are comparable with the standard scales (see Fonda & Herzog, 2004; Rodgers & Miller, 1997).
Health conditions
Health conditions were assessed with a sum index of the number self-reported physician-diagnosed medical conditions, including high blood pressure, diabetes, cancer or malignant tumor, lung disease, heart condition, stroke, psychiatric problems, and arthritis. A sum score was created and higher scores indicate reporting more health conditions (range is 0–8).
Self-rated health
Self-rated health was assessed using a single item asking participants to rate his or her health on a five-point scale ranging from 1 (excellent) to 5 (poor). The item was reverse-scored, so that higher scores indicate better self-reported health. This item has been widely used and shows strong prediction of key outcomes, including longevity (Idler & Benyamini, 1997).
Episodic memory
Episodic memory was measured using a unit-weighted composite of performances on the immediate and delayed free-recall tests (see Ofstedal, Fisher, & Herzog, 2005). The immediate recall test was typically given during the first interview quarter and asked participants to recall as many nouns as possible from a list of 10 nouns selected from four lists. For the delayed recall test, interviewers asked participants after a period of about 5 minutes again to recall as many nouns as possible out of the original word list. We used the percentage of words correctly remembered from both tests, ranging from 0 to 20, with higher scores representing more words remembered or better memory (range is 0–100). In the 1992 and 1994 assessments, the word lists included 20 words, instead of 10 words. Because the data were skewed and scores were much lower for these assessments, we did not include them in our analyses. Furthermore, our statistical models accounted for practice effects by including occasion-specific parameters (see Ghisletta et al., 2014).
Statistical Analyses
Time-in-study, age, and cohort
Following Gerstorf et al. (2019) and Infurna et al. (2021), we examined intraindividual change as time-in-study, a time-varying variable quantified for each assessment as the number of years since baseline (T1) and centered at the middle of each individual’s repeated-measures time series. Age-related differences (age gradients) were examined as individuals’ chronological age (at their middle assessment) and centered at age 55 for the midlife analyses and 75 for the old age analyses, close to the average age within each of the samples. Cohort-related differences were examined as individuals’ birth year. Birth year was centered such that the earliest-born participants within the study served as the reference.
Data analysis
Intraindividual changes, age-related, and history-related differences were examined using growth curve models (Grimm et al., 2017). Two models were estimated for each outcome. The first model examined whether there were racial differences in cohort differences in levels and rates of change across each outcome and specified as follows:
where person i’s score on the outcome at observation t, outcometi, is modeled as a function of a person-specific intercept, β 0i; a person-specific linear slope coefficient, β 1i; a person-specific quadratic slope coefficient, β 2i; and residual error eti. Individual differences in the person-specific coefficients were modeled as
(i.e., Level 2 model) where γ 00, γ 10 and γ 20 are the sample means or fixed effects from the model and u0i and u1i estimate between-person differences in each parameter and are assumed to be normally distributed, correlated with each other, and uncorrelated with the residual errors, eti. In order to compare differences by race, we created dummy coded variables where Whites would be the comparison group. To evaluate the best fitting model to the data in regards to the random effects, we estimated two nested models (for discussion, see Hertzog et al., 2008). We first estimated models for each outcome with only the random intercept. Second, we estimated a model with the random intercept, in addition to the slope variance and intercept–slope covariance. We then compared the change or difference in the −2LL between the two models and found that the latter model that estimated the random intercept and slope and intercept–slope covariance provided a more parsimonious or better fit to the data.
In the second model, we examined the role of education in the person-specific intercepts and linear rates of change, β 0i and β 1i; and tested all interaction terms with the birth year variable. To maintain parsimony, if birthyear2 was not statistically significant, we dropped it and its interactions from the model. Person-level predictors were effect-coded/centered, so that parameters indicated the average trajectory and the extent of differences associated with a particular variable (rather than a particular group). Models were fit using SAS (Proc Mixed; Littell et al., 2006). Incomplete data were accommodated under usual missing at random assumptions (Little & Rubin, 1987).
Results
We first report our findings pertaining to cohort effects across Whites, Blacks, and Hispanics and the moderating role of educational attainment in midlife, followed by these similar questions in old age. Due to space constraints, the tables that report the parameter estimates from our multilevel models can be found in the Supplementary Materials.
Historical Trends of Mental and Physical Health in Midlife: Variations by Race
Results from growth curve models that examined cohort-related differences across race in midlife mental and physical health are shown in Supplementary Table S1. We observed differences in levels of each outcome across Whites, Blacks, and Hispanics. On average, Blacks and Hispanics reported higher levels of depressive symptoms, health conditions, and functional limitations, and lower levels of self-rated health and episodic memory, compared with Whites. Figures 1–4 show our findings of racial differences in cohort effects for depressive symptoms, health conditions, self-rated health, and episodic memory. Later-born cohorts of Whites reported more depressive symptoms than earlier-born cohorts of Whites (see Figure 1A). In contrast, such historical increase was reversed for Blacks and Hispanics among whom, on average, later-born cohorts reported fewer depressive symptoms than earlier-born cohorts (see Figure 1B and C). There were no differences across race in cohort effects for functional limitations; later-born cohorts reported fewer functional limitations, but this trend was weaker for those born in the 1960s. As can be obtained from Figure 2, later-born cohorts of Whites reported fewer health conditions than earlier-born cohorts (Figure 2A), and this historical trend was stronger for Hispanics (see Figure 2B) and for early-midlife Blacks (see Figure 2C). On self-rated health, Whites showed less historical improvements over cohorts (see Figure 3A) than Blacks and Hispanics for whom self-rated health showed steady improvements across cohorts (see Figure 3B and C). On episodic memory, later-born cohorts across race, on average, exhibited historical improvements, and this was strongest for Hispanics (see Figure 4).
Historical change in depressive symptoms in midlife across race. Later-born cohorts of Whites reported higher levels of depressive symptoms (A), whereas for Blacks (B) and Hispanics (C), the trend was reversed.
Historical change in health conditions in midlife across race. Later-born cohorts across race reported fewer health conditions and this was stronger for Blacks (B) and Hispanics (C), compared with Whites (A).
Historical change in self-rated health in midlife across race. Later-born cohorts of Whites reported poorer self-rated health (A), whereas later-born cohorts of Blacks (B) and Hispanics (C) showed steady improvements.
Historical change in episodic memory in midlife across race. Later-born cohorts of Hispanics exhibited stronger gains in episodic memory (C), compared to whites (A) and Blacks (B).
Education
Results from the second set of models that included education as a moderator of cohort differences across race are shown in Supplementary Table S3. Overall, attaining higher levels of education was associated with fewer depressive symptoms, better physical health (i.e., higher self-rated health and fewer health conditions and functional limitations), and better memory. There were instances in which education was differentially associated with cohort effects across race. Later-born cohorts of Blacks who attained more years of education exhibited fewer gains in episodic memory than Whites. Later-born cohorts of Hispanics who attained more years of education reported more health conditions than Whites.
Historical Trends of Mental and Physical Health in Old Age: Variations by Race
Results from growth curve models that examined cohort-related differences across race in older adults’ mental and physical health are shown in Supplementary Table S2. We observed differences in levels of each outcome across Whites, Blacks, and Hispanics. On average, Blacks and Hispanics reported higher levels of depressive symptoms, health conditions, and functional limitations, and lower levels of self-rated health and episodic memory, compared with Whites. For our major research question, there was little evidence of differences across Whites, Blacks, and Hispanics in the size and direction of historical trends. To illustrate, Figure 5 shows that later-born cohorts report fewer depressive symptoms than earlier-born cohorts, and this trend was similar across White (Figure 5A), Black (Figure 5B), and Hispanic (Figure 5C) older adults. As a consequence, racial disparities observed among the earliest-born cohorts (born 1900–1919) continued in direction and size among the latest-born cohorts (born 1940–1952). With three notable exceptions, a highly similar pattern of results was obtained for the other outcome variables. First, we observed that successive cohorts of older adults, on average, reported fewer functional limitations and this waned for those born in the 1930s. Second, for self-rated health, later-born cohorts across race, on average, reported improvements, but these improvements were weaker for Blacks and Hispanics. Third, each successive cohort of older adults across race, on average, exhibited better episodic memory, and this waned for those born in the 1930s.
Historical change in depressive symptoms in old age across race. Racial disparities are observed, such that Whites (A), on average, report fewer depressive symptoms compared with Blacks (B) and Hispanics (C). There were no differences in cohort effects across race.
Education
The results from the second set of models that included education as a moderator of cohort differences across race are shown in Supplementary Table S4. Overall, attaining higher levels of education was associated with fewer depressive symptoms, better physical health (i.e., higher self-rated health and fewer health conditions and functional limitations), and better memory. We did not find evidence to suggest that education was differentially associated with cohort effects across race.
Discussion
The overarching goal of this study was to examine the extent to which recent trends of historical declines in middle-aged adults’ mental and physical health and historical improvements in older adults’ mental and physical health differs across race in the United States. We demonstrated that across midlife and old age, Blacks and Hispanics, on average, reported poorer levels of each outcome, compared with Whites. In midlife, we observed race differences in cohort differences, with later-born cohorts of Hispanics and Blacks showing general improvements compared to Whites. Conversely, in old age, we observed little evidence to suggest for race differences in historical change in mental and physical health. For education, we observed little consistent evidence that health-promoting effects of educational attainment would differ across race and cohort. Our discussion elaborates on the potential underlying reasons for differences across race and historical time, with an emphasis on macro-level structural factors.
Trends of Mental and Physical Health in Midlife: Variations by Race
Our findings build upon the study of cohort effects and racial disparities across adulthood and old age through life span and course perspectives. O’Rand (1996) discusses how examining heterogeneity and inequality within aging cohorts is a main concern to the study of the life course perspective. Dannefer’s (2003) work discusses the importance of incorporating the concept of cumulative disadvantages to better understand societally structured individual differences (see also Ferraro et al., 2009). Notions of studying inequality and diversity in aging falls in line with our approach of examining whether the size and direction of racial disparities have differed over historical time in midlife and old age. Research from Weden et al. (2017) and Thorpe et al. (2016) documented how racial disparities in race across adulthood are evident across mental and physical health. Our findings of racial disparities in both midlife and old age is similar to previous research, showing that Blacks and Hispanics, on average, report poorer physical health than Whites (Asnaani & Hall-Clark, 2017; Weden et al., 2017). However, it is important to note that these associations are not as clear-cut, with evidence also showing that mental health in Blacks is better than in Whites (Jackson et al., 2010).
Our findings complement those of Colen et al. (2018), who observed racial disparities in health among Whites, Hispanics, and Blacks. Here, we observed that in midlife and in some cases old age, Hispanics and Blacks are worse off than White Americans in mental and physical health. Moreover, our findings appear to contradict some aspects of the Hispanic Paradox, for example, in that we find Hispanic Americans in midlife worse off in measures of mental and physical health (Weden et al., 2017). This complements recent research discussing how midlife adults are now dealing with more challenges than ever before (such as more parenting pressures and caregiving for aging relatives; Infurna et al., 2020). With the larger number of multigenerational households, it has become increasingly important to evaluate the individual structure of roles and family structures of Hispanic and Blacks. One study suggests that greater disease burden is associated with declines in specific cognitive abilities in midlife Blacks (Byrd, Thorpe, et al., 2020). On the other hand, research suggests that experiencing financial hardship can directly affect mental health (Byrd, Gonzales, et al., 2020).
Racial Disparities in Historical Change Differ Across Midlife and Old Age
We observed that in midlife, but not old age, there were cohort differences across race. There could be several plausible explanations for these age-group differences and later-born cohorts of White middle-aged adults doing more poorly in depressive symptoms, health conditions, and self-rated health. Research from Case and Deaton (2015) observed trends of increasing deaths of despair among middle-aged Whites in the United States. The opioid pandemic has particularly affected Whites in the United States with this group surpassing all other racial groups in opioid analgesic consumption and deaths (Hasegawa et al., 2014; Jones & McAninch, 2015; Mendoza et al., 2019). The labor market has been stagnant in recent decades, and labor market changes driven by globalization and some technological change have led to lessened job opportunities and upward mobility (Case & Deaton, 2017; Gaydosh et al., 2019). It has been argued that socioeconomic hardship could potentially be driving increases in deaths of despair (Case & Deaton, 2020). The changes in labor market volatility, stagnant wages, as well as rising student debt can place great strains on mental and physical health. On the other hand, we see a picture of optimism in midlife with racial disparities narrowing over successive birth cohorts in Blacks and Hispanics. This has not always been the case, as depicted by Olshansky et al. (2012), who reported widening gaps in education and life expectancy among racial groups. One potential factor that could be driving these upward health trends could be the availability of resources such as health care access and participation in retirement savings programs among these later-born groups who aim to build financial assets.
Historical Dynamics Across Education in Midlife and Old Age
One of our other consistent findings is that individuals who attained more education were not necessarily more likely to exhibit better functioning across the outcomes included. This finding, with a few exceptions, did not differ across cohorts, suggesting that attaining more years of education may not always provide the buffering effects that coincide with providing the opportunities for attaining higher education. This is contrary to what other studies such as those conducted by Buhler and Nikitin (2020) have previously suggested about educational attainment. They noted that the majority of work on social development assumes that there is a universal age-related process. But they argue instead that while older adults ascribe to social ties and limited life path options, younger cohorts are socialized in radically changing social structures, such as the increasing importance of educational attainment. There are some possibilities as to why this might be. Work by Colen et al. (2018) suggests that exposure to discrimination plays a role in the relationship between unfair treatment and health—especially among higher SES groups. Moreover, work by Hudson et al. (2012) demonstrates that there is a known association between acute and chronic discrimination and depression among Blacks with higher educational attainment. It is quite possible that discrimination even among higher SES groups could still be relevant for a number of reasons. As mentioned before, the ever-changing social nature of later-born cohorts (digitalization, living arrangements) can significantly affect how age-related change can take place and drive differential developmental trajectories in comparison to previous current and future cohorts.
Limitations
We note several limitations with our study. First, we used a U.S.-based sample to address our research questions. We note that our findings pertaining to racial disparities observed and cohort differences in midlife may not necessarily be similar in other nations (see Infurna et al., in press). Likewise, although we were able to examine differences in race, the sample used was predominantly White, and thus this could limit the generalizability of this study. Second, although we had access to a wide-array of pertinent mental and physical health outcomes, in some instances, we only had one indicator of each. For example, we were only able to examine depressive symptoms, and it is unclear whether similar findings would be observed using measures of well-being, such as life satisfaction. Similarly, episodic memory relates to more fluid abilities, and future research should examine the extent to which crystallized abilities are affected by historical change across race. As another limitation of our measures, our focus was at the descriptive level of documenting whether there are racial differences in historical change in mental and physical health in midlife and old age. Future research is warranted that explores whether there are individual-level and macro-level factors that could account for the differences we observed. For example, could macro-level forces pertaining to access to education or employment opportunities as well as health care costs account for the racial disparities observed. In addition, our measure of education focused solely on attainment, and when considering race differences, especially in cognition, it is important to also examine the quality of education (Sisco et al., 2015). Fourth, an important point of consideration is that our findings are based on data obtained up to 2018; we were not able to assess how ongoing societal events in the United States such as Black Lives Matter and the coronavirus disease-2019 pandemic may affect future historical trends. Additional waves of data will position researchers to examine the extent to which such societal events, coupled with society becoming increasingly diverse, affect the historical trends observed here and whether existing disparities are mitigated or magnified for future cohorts of middle-aged and older adults (Infurna, 2021). Lastly, another promising direction for future research is analytic advancements, such as the incorporation of a Bayesian approach that may help accommodate and work around dependencies of Age, Period, and Cohort effects (Fosse, 2019; 2020; Yang et al., 2008).
Conclusion
Our findings portray how historical changes in midlife and old age adults’ mental and physical health unfold differently across race. Future research needs to explore to potential underlying individual-level differences, in addition to whether our findings extend to other nations with difference cultural backgrounds and perspectives. Insights into nation-level differences can help identify factors that can make a difference in promoting mental and physical health for the greater good. Additional areas of inquiry include examining individual-level factors that could operate as buffers that individuals can rely on to deter historical declines in mental and physical health across race.
Funding
The HRS (Health and Retirement Study) is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan.
Conflict of Interest
The authors declare that there is no conflict of interest.
