4

Population Disparities

OVERVIEW

This session of the workshop addressed a range of topics and questions, including the following:

• How do the timing, types, and severity of disability and disabling morbidity differ across minoritized populations? How prevalent are early-onset disabilities for different groups?
• By what pathways do structural racism/sexism and disadvantage contribute to these differences in disability and in ableism?
• How are inequalities in disability and mortality changing across education levels? How do these inequalities contribute to differential employment opportunities and outcomes and to the occurrence of other types of disability?
• What are the unique challenges to immigrant populations? To what extent are immigrant workers likely to be working in vulnerable occupations (such as care work) and at risk of injury? What access do immigrants of different statuses have to disability programs?
• Concerning maternal morbidity as a source of midlife health disadvantage, how are earlier life events linked to midlife?

The session was moderated by workshop steering committee member Jennifer Karas Montez (Syracuse University). Taylor Hargrove (University of North Carolina) presented her research on the consequences of intersecting social statuses on trends in health and wellbeing across adulthood. Rupa Valdez (University of Virginia) discussed structural ableism and its relationship to the broader topics of population disparities and the impact of intersectionality on individual experiences. Marc A. Garcia (Syracuse University) discussed disability risk among midlife Latinos in the United States with a focus on how sociodemographic factors—including nativity status, citizenship, duration in the United States, age of migration, country of origin, and race—intersect with gender to shape distinct trajectories of impairment.

CONSEQUENCES OF INTERSECTING SOCIAL STATUSES ON TRENDS IN HEALTH AND WELLBEING ACROSS ADULTHOOD

Taylor Hargrove (University of North Carolina) discussed how racism and other systems of inequality become embodied and shape population health in the United States. She first addressed intersectionality—the joint consequences of race/ethnicity, skin color, gender, and socioeconomic status on different health outcomes across the life course. Hargrove pointed to BMI as one example, with Black women having the highest levels of BMI across ages 13–31, while White men and White women had the lowest.

If one adds in parental wealth, then Hispanic men from relatively high-wealth families tended to have some of the worst BMI outcomes, while White men from very wealthy backgrounds had the lowest levels of BMI. Hargrove further explained that the disparities revealed by the analyses can be substantial. For example, 32-year-old dark-skinned women had the same physiological profile (in terms of BMI) as lighter-skinned women who were 23 years older.

Hargrove shared other research looking at the intersection of race and education, where White respondents and light-skinned Black respondents reported better health after achieving higher levels of education in adulthood (generally meaning a college degree or more), whereas dark-skinned Black adults reported worse health after achieving higher levels of education. The relationship between education, race, and cardiometabolic risk also varied by county. In only three out of the 12 county contexts that Hargrove’s team considered did Black young adults with high levels of education experience similar reductions in cardiometabolic risk as their White counterparts; these three counties were characterized by high opportunities for upward mobility, higher per-capita spending on education, and high per-capita spending on health and hospitals. In the other nine county contexts, Black young adults with high levels of education experienced either smaller reductions or no reduction in cardiometabolic risk relative to their White counterparts.

Hargrove reported on additional research that she conducted on school racial contexts and mental health, using five waves of data from Add Health and growth curve models. Those who had the lowest levels of depressive symptoms were White respondents who, as adolescents, went to schools with a low proportion of Black students; those with the highest levels of depressive symptoms, especially by their mid- to early twenties, were Black respondents who went to low-proportion Black schools. Those who also tended to have lower depressive symptoms by adulthood were Black respondents who attended high-proportion Black schools, and a little higher than that group were White respondents who attended high-proportion Black schools. The mean number of depressive symptoms tended to vary over time, declining somewhat from age 12 to the mid-twenties, and then rising again to age 42.

Looking at skin color at schools with higher proportions of Black students, Hargrove found different trajectories depending on the lightness or darkness of their skin. At age 42, those with very dark skin and Whites had the most depressive symptoms. Hargrove also examined the patterns based on whether schools had high or low proportions of White students. These comparisons also showed variations based on whether the students were White or Black, whether they attended high- or low-proportion White schools, and the darkness of the students’ skin color. From these findings, Hargrove drew several implications:

• Social groups are not monoliths, and it is important to look at how the various characteristics intersect.
• Disparities start early in life and are visible by age 12.
• Health consequences associated with socioeconomic attainment exist for marginalized groups. Thus, there could be potential benefit in providing funding or support for these groups, especially as they navigate higher socioeconomic contexts like college environments, to help reduce some of the stress and strain that are associated with upward attainment.

Hargrove closed by discussing her current work on structural racism and Alzheimer’s disease–related dementias (ADRDs). Currently about one in eight adults have been diagnosed with ADRD, with estimates suggesting that there will be more than a 40 percent increase in ADRD diagnoses over the period 2020 to 2050. There also are stark racial inequalities in ADRDs, with Black older adults being 1.5 to 2.0 times more likely to be diagnosed with an ADRD than their White counterparts. Hargrove reported that available evidence suggests that structural racism may be an important driver of these inequalities. To address this topic, Hargrove is seeking to (a) build a data repository of repeated multilevel measures of structural racism, and then create domain-specific composite measures of structural racism within the Add Health cohort; (b) examine the impact of education-related structural racism during adolescence on biological risk factors of ADRD in early adulthood; and (c) assess the extent to which interconnected domains of structural racism across the early life course impact ADRD biological risk factors in early midlife.

STRUCTURAL ABLEISM: ITS RELATIONSHIP TO THE BROADER TOPIC OF POPULATION DISPARITIES

Rupa Valdez (University of Virginia) contributed next to the discussion of population health disparities by addressing structural ableism and the impact of intersectionality. She began by defining structural ableism as a complex system of hierarchical and discriminatory processes, policies, and institutions that privilege and prefer able-bodied people and fail to represent or meaningfully include disabled persons’ voices. Valdez explained this environment as grounded in a network of ableist beliefs and practices that maintain and reproduce unequal outcomes for disabled people and favor able-bodied people.

Valdez described structural ableism as an essential concept for understanding why disabled people experience significant disparities across multiple domains of life, and why it is important to collectively dispel the myth that the disparities stem solely from the health conditions themselves

that disabled people face. She pointed out that structural ableism is a relatively new concept that has not yet received widespread attention, and she contrasted the trajectory of the concept with that of structural racism. Queries for structural racism in PubMed produced fewer than 10 entries per year prior to 2017, but 300 or more entries per year in 2021 through 2024. Similar searches for structural ableism produced annual counts of only three to four entries in 2022 through 2024 (and none earlier).

Drawing from a number of sources,¹ Valdez summarized some of the disparities in social determinants of health:

• Nearly two in five imprisoned individuals have at least one disability.
• Thirteen percent of adults with disabilities have received less than a high-school level education, compared to 9.5 percent of adults without disabilities.
• Only 17.8 percent of adults with disabilities are employed, compared to 63.6 percent of adults without disabilities.
• Of people with disabilities, 34 percent have inadequate transportation to health care, compared to 16 percent of adults without disabilities.
• Of working-age individuals with disabilities, 22.3 percent reported cost as a barrier to home internet access compared to 18 percent of non-disabled working-age adults.

Valdez also described barriers to health care access and disparities in health outcomes:

• One in four adults with disabilities do not have a usual health care provider.
• One in four adults with disabilities have an unmet health care need because of cost in the past year.
• One in five adults with disabilities did not have a routine checkup in the past year.
• Adults with disabilities are more likely to be obese, smoke, have heart disease, or have diabetes.
• Women with intellectual and developmental disabilities experience disparities in maternal health outcomes and in contraception provision and have been historically subjected to forced sterilization.

___________________

¹ Bureau of Justice Statistics (2021); Krahn et al. (2015); U.S. Department of Labor (2022).

Valdez listed the following educational and attitudinal barriers:

• Only 41 percent of U.S. physicians reported that they were “very confident” about their ability to provide the same quality of care to people with disabilities.
• Only 57 percent of physicians strongly agreed that they welcome people with disabilities into their practices.
• Among U.S. phyisicians, 35.8 percent reported knowing little or nothing about their legal responsibilities under the Americans with Disabilities Act (ADA).
• Among U.S. phyisicians, 82.4 percent reported the belief that people with significant disability have worse quality of life than non-disabled people.

Regarding physical barriers, Valdez found that greater than 75 percent of individuals with disabilities report experiencing conditions that impede them from using health care and wellness services. Among physicians seeing patients with significant mobility limitations, only around 40 percent always or usually used accessible exam tables or chairs. Thirty-six percent of clinic restrooms do not meet ADA requirements (33% for primary care).

As was stated in discussions of other populations, the disability community is not monolithic and, in Valdez’s view, it is important to think about how data are grouped or disaggregated. Sometimes researchers and policy makers group people together by diagnosis, but people with the same diagnosis can have very different functional capabilities and limitations. People who have been disabled since birth may have different trajectories from those who acquired a disability across the life course. People who have a visible disability may have different experiences as compared with those whose disabilities are not apparent.

Valdez next moved to the topic of intersectionality, noting that talking about structural ableism often involves talking about structural racism, because many of the same people are experiencing both. Valdez explained that disability is overrepresented in certain communities of color. Three in ten individuals who identify as Native and one in four Black Americans are disabled, compared to one in five White Americans. Valdez explained similar dynamics that have been observed with respect to incarceration, financial difficulties, and poverty rates. She also shared geographic interplays; rural areas have a higher rate of disability, but rural areas have fewer personal care aides for people with self-care disability.

Valdez concluded her remarks with the guidance that addressing disparities of the types described above will require transdisciplinary efforts at multiple levels in partnership with the disability community.

VARIABILITY IN DISABILITY RISK AMONG MIDLIFE WORKING-AGE LATINOS IN THE UNITED STATES

Marc A. Garcia (Syracuse University) led off the session with the observation that the Latino population in the United States has increased tremendously, from 14.6 million in 1980 to 62.1 million in 2020. It is a diverse population—in terms of background, racial composition, culture, and language—originating from more than 30 countries. Concerning Latinos in the United States, there is a concept known as the “healthy immigrant effect”; that is, Latino immigrants tend to be healthier on average than populations in both their native country of origin and the United States. However, Garcia also shared that with longer duration as residents in the United States, Latino immigrants are more likely to adopt negative health behaviors such as drinking, smoking, and poor diet, and their health patterns become more similar to those of the general U.S. population.

A person’s age at the time of migration has been shown to be a more useful indicator of health than duration in the United States, as it captures both the type of migration (family- vs. labor-based) and the degree of selectivity among Latino foreign-born populations, and it also captures exposure both to their native country environment and the environment in the United States. Garcia discussed prior research that shows those who immigrate between ages 18 and 34 are the most likely to have health advantages, whereas those who immigrate at younger ages are more likely to resemble their U.S.-born counterparts (Gubernskaya, 2014). Notably, not all health impacts of acculturation are negative. For example, with increased duration in the United States, immigrants—particularly those who migrate early in life—have more opportunities to integrate educationally and occupationally into mainstream institutions, which can result in lower disability risk.

To document trends among the U.S. Latino population, Garcia presented data collected by the Census Bureau’s American Communities Survey covering the period 2008 to 2019. In the relevant sample, 41 percent were U.S.-born and the majority of Latinos (59.1%) were of Mexican origin. The focus of Garcia’s analysis was on four outcomes: ambulatory difficulty (serious difficulty walking or climbing stairs), cognitive difficulty (difficulties learning, remembering, concentrating, or making decisions because of a physical, mental, or emotional condition), self-care difficulty (difficulties bathing or dressing), and independent living difficulty (difficulties doing errands alone such as visiting a doctor’s office because of a physical, mental, or emotional problem).

Using a logistic-regression model stratified by gender, nativity status, citizenship, duration in the United States, age of migration, country of origin, and race, Garcia found that, as the population aged (from 45 to

65 years old), Latinos’ risk of disability increased roughly threefold overall. The following findings were observed for specific subgroups:

• Consistent with the healthy immigrant effect, foreign-born Latinos, regardless of gender, exhibited a lower probability of ambulatory and cognitive difficulties than U.S.-born Latinos (Figure 4-1).
• Among the foreign-born, there was little difference in disability risk between those who were naturalized citizens and noncitizens.
• There was little variation in disability risk among midlife Latinos based on their duration in the United States, except that those who had been in the United States for more than 21 years had a higher predicted probability of disability than their foreign-born counterparts.

• Latinos who migrated to the United States prior to the age of 18 had disability patterns similar to U.S.-born Latinos, reflective of their exposure to U.S. society.
• Puerto Ricans and “All Other Hispanics,” regardless of gender, exhibited a higher predicted probability of ambulatory and cognitive difficulties compared to other country-of-origin groups; whereas South Americans, Central Americans, and to a lesser extent Cubans (ambulatory only) had a lower disability risk.
• Indigenous Latinos and Black Latinos had a greater disability risk than White Latinos and Asian Latinos.

Garcia pointed to these findings to emphasize that Latinos are not a monolithic group. The research by Garcia and colleagues shows that differing cultural, socioeconomic, migratory, and behavioral experiences have lasting effects that contribute to varied patterns of disability risk among Latino subgroups. These findings imply that researchers should use a gendered intersectional life-course approach that considers diversity along multiple dimensions such as race/ethnicity, nativity, and country of origin when assessing disability risk among Latinos. Additionally, Garcia stressed that state policies and state contexts matter, stating that Latino subgroups (i.e., foreign-born/U.S.-born and men/women) residing in states with more liberal private labor policies exhibit a lower probability of any difficulty compared to states with more conservative labor policies.

DISCUSSION

A consistent theme throughout the session, summarized by Montez in her comment as moderator, was that the term “disabled” is not a monolith, because it applies to many conditions and a wide range of people. Zajacova asked about the degree of nonlinearity in the data on disability patterns. She cited Zoya Gubernskaya’s (2014) work on the “immigrant health paradox”—the observed pattern whereby immigrants perform better by a number of health indicators than do native-born peers with similar demographic and socioeconomic characteristics. Zajacova noted that the paradox diminishes as immigrants remain longer in the host country and suggested that, in part due to this effect, younger immigrants were healthier on average than older immigrants relative to their age-peer groups.

Amelia Karraker (National Institute on Aging) followed up by asking about the “salmon bias”—the concept that people in poor or declining health are more likely to emigrate back to their countries of origin (also noting that health should not be conflated with disability). Garcia responded that research shows that some salmon bias has been happening, but not to the extent that it alters the conclusion about age-health relationships. Indeed,

she noted that the effect is balanced by older people migrating to the United States through family reunification. Beltrán-Sánchez commented that the extent of the salmon bias depends on the cohort being examined. Older cohorts experienced migration during a period when more circular migration was occurring, while more recent cohorts migrate to the United States and stay in place in the United States. Garcia added that another nuance is which countries of origin are involved; not all immigrants are able to return to their country of origin.

Beltrán-Sánchez asked Garcia about comparability when analyzing data on Puerto Ricans, given that Puerto Ricans are U.S. citizens and thus face a unique set of circumstances relative to other immigrant populations. Garcia agreed that having U.S. citizenship makes Puerto Rican immigrants very different, noting that another important difference is that Puerto Ricans also differ by being more phenotypically dark-skinned than other Latino populations.

Agree asked the presenters to comment on the new Office of Management and Budget standards on measuring race and ethnicity in federal statistics. Garcia responded that, while statistical agencies have attempted to collect data identifying more racial categories, what is ultimately happening is that respondents are being categorized based on the fill-in boxes into “some other race.” This is problematic for data accuracy since Latinos responding to surveys typically are not identifying as some other race. Valdez commented that there is a need to look at data with more granularity, and Hargrove added that there is a need for more data on structural contexts.

Landes asked workshop participants how to develop measures of structural ableism. Valdez responded that structural ableism is not yet conceptually well defined, so her and her colleagues’ research is seeking first to understand the dimensions of structural ableism. Once this is established, a multimethod approach might be applied to look at both historical texts and policy documents to examine how the experiences of disabled persons are shaped. She added that a qualitative component—focusing on health outcomes and how different systems shape them—is also needed to understand the different domains of society that impact and shape the experiences of disabled persons. The second aim of Valdez’s research is to develop measures at an individual level of how experiences of discrimination and bias shape different aspects of their lives and the outcomes they have experienced. For community-level measures they are looking at markers of structural ableism for well-specified groups, such as the percentage of people in a given geographic region denied a mortgage who are disabled versus the percentage who are non-disabled, or the percentage of crosswalks in a metro region that have curb cuts versus those that do not. They are also looking at the intersection of structural racism and structural ableism, although this work is still exploratory in nature.

Valdez asked what the next step is to come up with an explanation for the large disparities across intersections. Hargrove responded that data availability is very important, including the need to take some of the more nuanced measures developed in regional surveys and incorporate them more widely. Researchers also need to learn more about how overall contexts shape the intersectional patterns. Garcia mentioned how the uneven distribution of key populations across the United States makes it difficult to make comparisons across groups and states. For example, 95 percent of the Latino population resides in 25 states, and 70 percent of the Cuban population resides in Florida and New Jersey.

Ellen Meara (Harvard Chan School of Public Health) asked about the use of alternative data sources, noting that rich information is available through social media data consisting of millions of observations. Hargrove responded by listing some of the alternative sources she has used: the Eviction Lab to compare areas based on eviction rates; the Home Mortgage Disclosure Act to measure residential redlining; and criminal legal data on incarceration rates in a county or state, the policing environment of an area, and the nature of encounters with the police. Valdez supported the idea of exploring alternative data sources but raised the issue that there are often substantial gaps for key subpopulations. In particular, for large studies like the University of Michigan’s Health and Retirement Study and the National Health and Aging Trends Study, some subpopulations, such as disabled young adults, are not well represented. Ne’eman noted that political science makes great use of nonprobability online panels, combined with statistical adjustments to make the samples more representative. He added that probability-based surveys are becoming worse, largely due to declining respondent participation, at the same time that nonprobability-based surveys are becoming better. With declining telephone (and other) response rates, even very reputable polling firms and research projects are increasingly coming under pressure to explore nonprobability methods with statistical adjustment.

Karraker asked for additional discussion about how to improve measures of disability going forward. Ne’eman responded that there is an urgent need for a body of methodological work to better ensure that particular disability definitions correspond with particular public programs or research purposes. The ACS-6 and Washington Group questions provide very little indication about the diagnosis or the disability identity profile of who those questions identify. Ne’eman suggested that more work is therefore needed on which disability definitions may be more appropriate for, say, Social Security—a long-term care, ADA 504 context. Landes added that more qualitative work is needed; specifically, input is needed from people with disabilities to better understand how they describe their own disabilities, something that has not been done well.

This page intentionally left blank.