The relationship between education and health: reducing disparities through a contextual approach

Abstract

Adults with higher educational attainment live healthier and longer lives compared to their less educated peers. The disparities are large and widening. We posit that understanding the educational and macro-level contexts in which this association occurs is key to reducing health disparities and improving population health. In this paper, we briefly review and critically assess the current state of research on the relationship between education and health in the United States. We then outline three directions for further research: We extend the conceptualization of education beyond attainment and demonstrate the centrality of the schooling process to health; We highlight the dual role of education a driver of opportunity but also a reproducer of inequality; We explain the central role of specific historical socio-political contexts in which the education-health association is embedded. This research agenda can inform policies and effective interventions to reduce health disparities and improve health of all Americans.

URGENT NEED FOR NEW DIRECTIONS IN EDUCATION-HEALTH RESEARCH

Americans have worse health than people in other high-income countries, and have been falling further behind in recent decades (137). This is partially due to the large health inequalities and poor health of adults with low education (84). Understanding the health benefits of education is thus integral to reducing health disparities and improving the well-being of 21^st century populations. Despite tremendous prior research, critical questions about the education-health relationship remain unanswered, in part because education and health are intertwined over the lifespans within and across generations and are inextricably embedded in the broader social context.

We posit that to effectively inform future educational and heath policy, we need to capture education ‘in action’ as it generates and constrains opportunity during the early lifespans of today’s cohorts. First, we need to expand our operationalization of education beyond attainment to consider the long-term educational process that precedes the attainment and its effect on health. Second, we need to re-conceptualize education as not only a vehicle for social success, valuable resources, and good health, but also as an institution that reproduces inequality across generations. And third, we argue that investigators need to bring historical, social and policy contexts into the heart of analyses: how does the education-health association vary across place and time, and how do political forces influence that variation?

During the past several generations, education has become the principal pathway to financial security, stable employment, and social success (8). At the same time, American youth have experienced increasingly unequal educational opportunities that depend on the schools they attend, the neighborhoods they live in, the color of their skin, and the financial resources of their family. The decline in manufacturing and rise of globalization have eroded the middle class, while the increasing returns to higher education magnified the economic gaps among working adults and families (107). In addition to these dramatic structural changes, policies that protected the welfare of vulnerable groups have been gradually eroded or dismantled (129). Together, these changes triggered a precipitous growth of economic and social inequalities in the American society (17; 106).

Unsurprisingly, health disparities grew hand in hand with the socio-economic inequalities. Although the average health of the US population improved over the past decades (67; 85), the gains largely went to the most educated groups. Inequalities in health (53; 77; 99) and mortality (86; 115) increased steadily, to a point where we now see an unprecedented pattern: health and longevity are deteriorating among those with less education (92; 99; 121; 143). With the current focus of the media, policymakers, and the public on the worrisome health patterns among less-educated Americans (28; 29), as well as the growing recognition of the importance of education for health (84), research on the health returns to education is at a critical juncture. A comprehensive research program is needed to understand how education and health are related, in order to identify effective points of intervention to improve population health and reduce disparities.

The article is organized in two parts. First, we review the current state of research on the relationship between education and health. In broad strokes, we summarize the theoretical and empirical foundations of the education-health relationship and critically assess the literature on the mechanisms and causal influence of education on health. In the second part, we highlight gaps in extant research and propose new directions for innovative research that will fill these gaps. The enormous breadth of the literature on education and health necessarily limits the scope of the review in terms of place and time; we focus on the United States and on findings generated during the rapid expansion of the education-health research in the past 10–15 years. The terms “education” and “schooling” are used interchangeably. Unless we state otherwise, both refer to attained education, whether measured in completed years or credentials. For references, we include prior review articles where available, seminal papers, and recent studies as the best starting points for further reading.

THE ASSOCIATION BETWEEN EDUCATION AND HEALTH

Conceptual toolbox for examining the association

Researchers have generally drawn from three broad theoretical perspectives to hypothesize the relationship between education and health. Much of the education-health research over the past two decades has been grounded in the Fundamental Cause Theory (75). The FCT posits that social factors such as education are ‘fundamental’ causes of health and disease because they determine access to a multitude of material and non-material resources such as income, safe neighborhoods, or healthier lifestyles, all of which protect or enhance health. The multiplicity of pathways means that even as some mechanisms change or become less important, other mechanisms will continue to channel the fundamental dis/advantages into differential health (48). The Human Capital Theory (HCT), borrowed from econometrics, conceptualizes education as an investment that yields returns via increased productivity (12). Education improves individuals’ knowledge, skills, reasoning, effectiveness, and a broad range of other abilities, which can be utilized to produce health (93). The third approach, the Signaling or Credentialing perspective (34; 125) has been used to explain the observed large discontinuities in health at 12 or 16 years of schooling, typically associated with the receipt of a high school and college degrees, respectively. This perspective views earned credentials as a potent signal about one’s skills and abilities, and emphasizes the economic and social returns to such signals. Thus all three perspectives postulate a causal relationship between education and health and identify numerous mechanisms through which education influences health. The HCT specifies the mechanisms as embodied skills and abilities, FCT emphasizes the dynamism and flexibility of mechanisms, and credentialism identifies social responses to educational attainment. All three theoretical approaches, however, operationalize the complex process of schooling solely in terms of attainment and thus do not focus on differences in educational quality, type, or other institutional factors that might independently influence health. They also focus on individual-level factors: individual attainment, attainment effects, and mechanisms, and leave out the social context in which the education and health processes are embedded.

Observed associations between education and health

Empirically, hundreds of studies have documented “the gradient” whereby more schooling is linked with better health and longer life. A seminal 1973 book by Kitagawa and Hauser powerfully described large differences in mortality by education in the United States (71), a finding that has since been corroborated in numerous studies (31; 42; 46; 109; 124). In the following decades, nearly all health outcomes were also found strongly patterned by education. Less educated adults report worse general health (94; 141), more chronic conditions (68; 108), and more functional limitations and disability (118; 119; 130; 143). Objective measures of health, such as biological risk levels, are similarly correlated with educational attainment (35; 90; 140), showing that the gradient is not a function of differential reporting or knowledge.

The gradient is evident in men and women (139) and among all race/ethnic groups (36). However, meaningful group differences exist (60; 62; 91). In particular, education appears to have stronger health effects for women than men (111) and stronger effects for non-Hispanic whites than minority adults (134; 135) even if the differences are modest for some health outcomes (36). The observed variations may reflect systematic social differences in the educational process such as quality of schooling, content, or institutional type, as well as different returns to educational attainment in the labor market across population groups (26). At the same time, the groups share a common macro-level social context, which may underlie the gradient observed for all.

To illustrate the gradient, we analyzed 2002–2016 waves of the National Health Interview Survey (NHIS) data from adults aged 25–64. Figure 1 shows the levels of three health outcomes across educational attainment levels in six major demographic groups predicted at age 45. Three observations are noteworthy. First, the gradient is evident for all outcomes and in all race/ethnic/gender groups. Self-rated health exemplifies the staggering magnitude of the inequalities: White men and women without a high school diploma have about 57% chance of reporting fair or poor health, compared to just 9% for college graduates. Second, there are major group differences as well, both in the predicted levels of health problems, as well as in the education effects. The latter are not necessarily visible in the figures but the education effects are stronger for women and weaker for non-white adults as prior studies showed (table with regression model results underlying the prior statement is available from the authors). Third, an intriguing exception pertains to adults with “some college,” whose health is similar to high school graduates’ in health outcomes other than general health, despite their investment in and exposure to postsecondary education. We discuss this anomaly below.

Figure 1.

Predicted Probability of Health Problems

Source: 2002–2016 NHIS Survey, Adults Age 25–64

Pathways through which education impacts health

What explains the improved health and longevity of more educated adults? The most prominent mediating mechanisms can be grouped into four categories: economic, health-behavioral, social-psychological, and access to health care. Education leads to better, more stable jobs that pay higher income and allow families to accumulate wealth that can be used to improve health (93). The economic factors are an important link between schooling and health, estimated to account for about 30% of the correlation (36). Health behaviors are undoubtedly an important proximal determinant of health but they only explain a part of the effect of schooling on health: adults with less education are more likely to smoke, have an unhealthy diet, and lack exercise (37; 73; 105; 117). Social-psychological pathways include successful long-term marriages and other sources of social support to help cope with stressors and daily hassles (128; 131). Interestingly, access to health care, while important to individual and population health overall, has a modest role in explaining health inequalities by education (61; 112; 133), highlighting the need to look upstream beyond the health care system toward social factors that underlie social disparities in health. Beyond these four groups of mechanisms that have received the most attention by investigators, many others have been examined, such as stress, cognitive and noncognitive skills, or environmental exposures (11; 43). Several excellent reviews further discuss mechanisms (2; 36; 66; 70; 93).

Causal interpretation of the education-health association

A burgeoning number of studies used innovative approaches such as natural experiments and twin design to test whether and how education causally affects health. These analyses are essential because recommendations for educational policies, programs, and interventions seeking to improve population health hinge on the causal impact of schooling on health outcomes. Overall, this literature shows that attainment, measured mostly in completed years of schooling, has a causal impact on health across numerous (though not all) contexts and outcomes.

Natural experiments take advantage of external changes that affect attainment but are unrelated to health, such as compulsory education reforms that raise the minimum years of schooling within a given population. A seminal 2005 study focused on increases in compulsory education between 1915 and 1939 across US states and found that a year of schooling reduced mortality by 3.6% (78). A re-analysis of the data indicated that taking into account state-level mortality trends rendered the mortality effects null but it also identified a significant and large causal effect on general health (88). A recent study of a large sample of older Americans reported a similar pattern: a substantial causal effect of education for self-rated health but not for mortality (47). School reform studies outside the US have reported compelling (122) or modest but significant (32) effects of schooling on health, although some studies have found nonsignificant (4), or even negative effects (7) for a range of health outcomes.

Twin design studies compare the health of twins with different levels of education. This design minimizes the influence of family resources and genetic differences in skills and health, especially for monozygotic twins, and thus serves to isolate the effect of schooling. In the US, studies using this design generated robust evidence of a causal effect of education on self-rated health (79), although some research has identified only modest (49) or not significant (3; 55) effects for other physical and mental health outcomes. Studies drawing on the large twin samples outside of the US have similarly found strong causal effects for mortality (80) and health (14; 16; 51) but again some analyses yielded no causal effects on health (13; 83) or health behaviors (14). Beyond our brief overview, readers may wish consult additional comprehensive reviews of the causal studies (40; 45; 89).

The causal studies add valuable evidence that educational attainment impacts adult health and mortality, even considering some limitations to their internal validity (15; 88). To improve population health and reduce health disparities, however, they should be viewed as a starting point to further research. First, the findings do not show how to improve the quality of schooling or its quantity for in the aggregate population, or how to overcome systematic intergenerational and social differences in educational opportunities. Second, their findings do take into account contexts and conditions in which educational attainment might be particularly important for health. In fact, the variability in the findings may be attributable to the stark differences in contexts across the studies, which include countries characterized by different political systems, different population groups, and birth cohorts ranging from the late 19^th to late 20^th centuries that were exposed to education at very different stages of the educational expansion process (9).

TOWARD A SOCIALLY-EMBEDDED UNDERSTANDING OF THE EDUCATION-HEALTH RELATIONSHIP

To date, the extensive research we briefly reviewed above has identified substantial health benefits of educational attainment in most contexts in today’s high-income countries. Still, many important questions remain unanswered. We outline three critical directions to gain a deeper understanding of the education-health relationship with particular relevance for policy development. All three directions shift the education-health paradigm to consider how education and health are embedded in life course and social contexts.

First, nearly universally, the education-health literature conceptualizes and operationalizes education in terms of attainment, as years of schooling or completed credentials. However, attainment is only the endpoint, although undoubtedly important, of an extended and extensive process of formal schooling, where institutional quality, type, content, peers, teachers, and many other individual, institutional, and interpersonal factors shape lifecourse trajectories of schooling and health. Understanding the role of the schooling process in health outcome is relevant for policy because it can show whether interventions should be aimed at increasing attainment, or whether it is more important to increase quality, change content, or otherwise improve the educational process at earlier stages for maximum health returns. Second, most studies have implicitly or explicitly treated educational attainment as an exogenous starting point, a driver of opportunities in adulthood. However, education also functions to reproduce inequality across generations. The explicit recognition of the dual function of education is critical to developing education policies that would avoid unintended consequence of increasing inequalities. And third, the review above indicates substantial variation in the education-health association across different historical and social contexts. Education and health are inextricably embedded in these contexts and analyses should therefore include them as fundamental influences on the education-health association. Research on contextual variation has the potential to identify contextual characteristics and even specific policies that exacerbate or reduce educational disparities in health.

We illustrate the key conceptual components of future research into the education-health relationship in Figure 2. Important intergenerational and individual socio-demographic factors shape educational opportunities and educational trajectories, which are directly related to and captured in measures of educational attainment. This longitudinal and life course process culminates in educational disparities in adult health and mortality. Importantly, the macro-level context underlies every step of this process, shaping each of the concepts and their relationships.

Figure 2.

Enriching the conceptualization of educational attainment

In most studies of the education-health associations, educational attainment is modeled using years of schooling, typically specified as a continuous covariate, effectively constraining each additional year to have the same impact. A growing body of research has substituted earned credentials for years. Few studies, however, have considered how the impact of additional schooling is likely to differ across the educational attainment spectrum. For example, one additional year of education compared to zero years may be life-changing by imparting basic literacy and numeracy skills. The completion of 14 rather than 13 years (without the completion of associated degree) could be associated with better health through the accumulation of additional knowledge and skills as well, or perhaps could be without health returns, if it is associated with poor grades, stigma linked to dropping out of college, or accumulated debt (63; 76). Examining the functional form of the education-health association can shed light on how and why education is beneficial for health (70). For instance, studies found that mortality gradually declines with years of schooling at low levels of educational attainment, with large discontinuities at high school and college degree attainment (56; 98). Such findings can point to the importance of completing a degree, not just increasing the quantity (years) of education. Examining mortality, however, implicitly focused on cohorts who went to school 50–60 years ago, within very different educational and social contexts. For findings relevant to current education policies, we need to focus on examining more recent birth cohorts.

A particularly provocative and noteworthy aspect of the functional form is the attainment group often identified as “some college:” adults who attended college but did not graduate with a four-year degree. Postsecondary educational experiences are increasingly central to the lives of American adults (27) and college completion has become the minimum requirement for entry into middle class (65; 87). Among high school graduates, over 70% enroll in college (22) but the majority never earn a four-year degree (113). In fact,, the largest education-attainment group among non-elderly US adults comprises the 54 million adults (29% of total) with some college or associate’s degree (113). However, as in Figure 1, this group often defies the standard gradient in health. Several recent studies have found that the health returns to their postsecondary investments are marginal at best (110; 123; 142; 144). This finding should spur new research to understand the outcomes of this large population group, and to glean insights into the health returns to the postsecondary schooling process. For instance, in the absence of earning a degree, is greater exposure to college education in terms of semesters or earned credits associated with better health or not? How do the returns to postsecondary schooling differ across the heterogeneous institutions ranging from selective 4-year to for-profit community colleges? How does accumulated college debt influence both dropout and later health? Can we identify circumstances under which some college education is beneficial for health? Understanding the health outcomes for this attainment group can shed light on the aspects of education that are most important for improving health.

A related point pertains to the reliability and validity of self-reported educational attainment. If a respondent reports 16 completed years of education, for example, are they carefully counting the number of years of enrollment, or is 16 shorthand for “completed college”? And, is 16 years the best indicator of college completion in the current context when the median time to earn a four-year degree exceeds 5 years (30)? And, is longer time in college given a degree beneficial for health or does it signify delayed or disrupted educational pathways linked to weaker health benefits (132)? How should we measure part-time enrollment? As studies begin to adjudicate between the health effects of years versus credentials (74) in the changing landscape of increasingly ‘nontraditional’ pathways through college (132), this measurement work will be necessary for unbiased and meaningful analyses. An in-depth understanding may necessitate primary data collection and qualitative studies. A feasible direction available with existing data such as the National Longitudinal Survey of Youth 1997 (NLSY97) is to assess earned college credits and grades rather than years of education beyond high school.

As indicated in Figure 2, beyond a more in-depth usage of the attainment information, we argue that more effective conceptualization of the education-health relationship as a developmental life course process will lead to important findings. For instance, two studies published in 2016 used the NLSY97 data to model how gradual increases in education predict within-individual changes in health (39; 81). Both research teams found that gradual accumulation of schooling quantity over time was not associated with gradual improvements in health. The investigators interpreted the null findings as an absence of causal effects of education on health, especially once they included important confounders (defined as cognitive and noncognitive skills and social background). Alternatively, perhaps the within-individual models did not register health because education is a long-term, developing trajectory that cannot be reduced to point-in-time changes in exposure. Criticisms about the technical aspects of theses studies notwithstanding (59), we believe that these studies and others like them, which wrestle with the question of how to capture education as a long-term process grounded in the broader social context, and how this process is linked to adult health, are desirable and necessary.

Education as (re)producer of inequality

The predominant theoretical framework for studying education and health focuses on how education increases skills, improves problem-solving, enhances employment prospects, and thus opens access to other resources. In sociology, however, education is viewed not (only) as increasing human capital but as a “sieve more than a ladder” (126), an institution that reproduces inequality across generations (54; 65; 103; 114). The mechanisms of the reproduction of inequality are multifarious, encompassing systematic differences in school resources, quality of instruction, academic opportunities, peer influences, or teacher expectations (54; 114; 132). The dual role of education, both engendering and constraining social opportunities, has been recognized from the discipline’s inception (52) and has remained the dominant perspective in sociology of education (18; 126). Health disparities research, which has largely dismissed the this perspective as “specious” (93), could benefit from pivoting toward this complex sociological paradigm.

As demonstrated in Figure 2, parental SES and other background characteristics are key social determinants that set the stage for one’s educational experiences (20; 120). These characteristics, however, shape not just attainment, but the entire educational and social trajectories that drive and result in particular attainment (21; 69). Their effects range from the differential quality and experiences in daycare or preschool settings (6), K-12 education (24; 136), as well as postsecondary schooling (5; 127). As a result of systematically different experiences of schooling over the early life course stratified by parental SES, children of low educated parents are unlikely to complete higher education: over half of individuals with college degrees by age 24 came from families in the top quartile of family income compared to just 10% in the bottom quartile (23).

Unfortunately, prior research has generally operationalized the differences in educational opportunities as confounders of the education-health association or as “selection bias” to be statistically controlled, or best as a moderating influence (10; 19). Rather than remove the important life course effects from the equation, studies that seek to understand how educational and health differences unfold over the life course, and even across generations could yield greater insight (50; 70). A life course, multigenerational approach can provide important recommendations for interventions seeking to avoid the unintended consequence of increasing disparities. Insofar as socially advantaged individuals are generally better positioned to take advantage of interventions, research findings can be used to ensure that policies and programs result in decreasing, rather than unintentionally widening, educational and health disparities.

Education and health in social context

Finally, perhaps the most important and policy-relevant emerging direction to improving our understanding of the education-health relationship is to view both as inextricably embedded within the broad social context. As we highlight in Figure 2, this context underlies every feature of the development of educational disparities in health. In contrast to the voluminous literature focusing on individual-level schooling and health, there has been a “startling lack of attention to the social/political/economic context” in which the relationships are grounded (33). By context, we mean the structure of a society that varies across time and place, encompassing all major institutions, policy environments, as well as gender, race/ethnicity, age, and socioeconomic stratification. Under what circumstances, conditions, and policies are the associations between education and health stronger or weaker?

Within the United States, the most relevant units of geo-political boundaries generating distinct policy contexts are states, although smaller geographic units are also pertinent (44; 100). Since the 1980s, the federal government has devolved an increasing range of key socioeconomic, political, and health-care decisions to states. This decentralization has resulted in increasing diversity across states in conditions for a healthy life (96; 101). A recent study demonstrates how different environments across US states yield vastly different health returns to education (100). State-level characteristics had little impact on adults with high education, whose disability levels were similarly low regardless of their state of residence. In contrast, disability levels of low-educated adults were not only high but also varied substantially across states: disability was particularly high in states that have invested less in the social welfare of its residents, such as Mississippi, Kentucky, and West Virginia. Highly-educated adults, particularly white adults and men who can convert education into other resources most readily, use personal resources to protect their health like a ‘personal firewall’ (97). Their less-educated peers, meanwhile, are vulnerable without social safety nets. Demonstrating the potential for informing policy in this area, the findings directly identify state policies that influence the extent to which educational attainment matters for health and longevity. These include economic policies including state income tax structures and education expenditures per capita, as well as policies influencing social cohesion in a state, such as income inequality and unemployment rates. Beyond the US, investigators can leverage differences in political systems across countries to assess the impact of different welfare regimes on the education-health associations, as some European researchers began generating (41; 82).

Similar to variation across geo-political boundaries, research on variation across time can highlight policies and conditions that mitigate or inflate health disparities. How has the education-health association changed over time? In recent decades, the association has become increasingly strong, with widening disparities in health outcomes across education (53; 77; 86; 116; 143). These increases started in the 1980s (17) at the same time that social inequality began rising with the political embrace of pro-market neoliberal policies (33). Since then, the United States has been increasingly marked by plummeting economic wellbeing (except for the wealthiest Americans), growing economic segregation, emerging mass incarceration, downward social mobility, and despair in many working-class communities (17; 95; 129). Conversely, in the two decades prior (1960s and 70s), social disparities in health were decreasing (1; 72). During those decades, many pro-social policies such as Civil rights legislation, War on Poverty programs, and racial desegregation were improving social inequalities. Macro-level political forces, clearly, can influence not only social but also health inequalities (104). Two facts follow: growing disparities are not inevitable and changes in the education-health relationship may be strongly linked to social policies. While some of the growth in educational inequalities may be attributable to changes in educational composition of the population with increasingly negatively select groups of adults at the lowest levels of schooling, these compositional changes likely play only a minor role in the overall trends (38; 58). Linking education and health to the broader social context brings to the forefront the ways in which we, as individuals and a collective society, produce and maintain health disparities.

Implications for Policy and Practice

Reducing macro-level inequalities in health will require macro-level interventions. Technological progress and educational expansion over the past several decades have not decreased disparities; on the contrary, educational disparities in health and mortality have grown in the US. Moreover, the consistent, durable relationship between education and health and the multitude of mechanisms linking them suggests that programs targeting individual behaviors will have limited impact to counteract disparities. Thus, we argue that future findings from the new research directions proposed here can be used to intervene at the level of social contexts to alter educational trajectories from an early age, with the ultimate goal of reducing health disparities. We note two promising avenues for policy development.

One potential solution may focus on universal federal and state-level investment in the education and well-being of children early in the life course to disrupt the reproduction of social inequalities and change subsequent educational trajectories. Several experimental early-education programs such as the Perry Preschool Project and Carolina Abecedarian Project have demonstrated substantial, lasting, and wide-ranging benefits, including improved adult health (25; 57; 102). These programs provided intensive, exceptionally high-quality, and diverse services to children, and it is these characteristics that appear central to their success (138). Further research on the qualitative and social dimensions of education and their effects on health can inform future model educational programs and interventions across all ages.

Another important issue for both researchers and policymakers pertains to postsecondary enrollment and attrition, and their effects on health. Educational expansion in the college-for-all era has yielded high post-secondary enrollment, but also unacceptable dropout rates with multiple detrimental consequences, including high rates of student debt (64) and stigma (76), which may negatively affect health. Emerging studies found that college dropouts fail to benefit from their postsecondary investments. Next we need to understand under what circumstances college goers do reap health benefits, or how their postsecondary experience can be modified to improve their health.

For both of these avenues, effective implementation will need further research on the specific institutional characteristics and social contexts that shape the schooling effects. However, in designing interventions and policies, we need to be aware of the dual role of education as a drive and reproducer of inequality. Individuals from advantaged backgrounds may be better positioned to take advantage of new educational opportunities, and thus any interventions and programs need to ensure that marginalized populations have equal or greater access in order to avoid the unintended consequence of further intensifying disparities. Finally, researchers and policymakers should engage in a dialogue such that researchers effectively communicate their insights and recommendations to policymakers, and policymakers convey the needs and challenges of their practices to researchers.

CONCLUSION

Education and health are central to individual and population well-being. They are also inextricably embedded in the social context and structure. Future research needs to expand beyond the individual-focused analyses and hypothesize upstream (96), taking a contextual approach to understanding education and health. Such an approach will require interdisciplinary collaborations, innovations in conceptual models, and rich data sources. The three directions for further research on health returns to education we outlined above can help generate findings that will inform effective educational and health policies and interventions to reduce disparities. During this critical time when health differences are widening and less educated Americans are experiencing social and health declines, research and policy has the opportunity to make a difference and improve the health and well-being of our population.