7

Synthesis and Takeaways—Research Gaps, Data and Measurement Needs, and Policy Directions

The final session of the workshop was created to provide participants with an opportunity to offer a synthesis of the presentations that took place over the two-day event and to encourage discussion of actions that could be taken to advance research and policy priorities. The session opened with members of the planning committee offering their comments and was followed by open discussion of key takeaways among workshop participants and attendees. Given the format, the discussion was not organized systematically by topic. This summary re-orders the discussion on a topic-by-topic basis, providing the names of speakers at the end of each comment.

DEFINITIONS AND MEASUREMENT

Defining, conceptualizing, and measuring disability, along with data needs—topics covered in Session 1—were themes carried forward throughout the workshop. Session 1 presenters Swenor, Landes, and Forber-Pratt emphasized that, since definitions and measurement undergird almost all aspects of research, improvements on these fronts are needed. At the same time, participants recognized that there are challenges in determining the best way to move forward, especially on data infrastructure (Landes).

There is a need to look systematically to ensure that, to the extent practical, everyone is using the same definition of disability or, at least, it is clear what different definitions are being used and for what research and policy purposes (Montez). Training in the use of definitions may be an important part of that work (Montez and Agree). Often researchers are asked why

they have used a particular measure, and the answer is typically that this is what is available but not necessarily what would be optimal (Landes).

Agree supported the view expressed by others that work on definitions is needed but also recognized that a solid groundwork has already been created by researchers. Much of this work has been focused on older populations, while children and youth are a distinct group requiring additional attention (Agree). The foundational activities of daily living (ADLs) measurement approach, first developed by Katz and colleagues (1963), was designed to assess needs for care in community facilities such as hospitals, rehabilitation centers, nursing homes, and home care programs. ADL measures were adapted for use in survey research and for aging research in particular, but they were conceptualized from within a clinical approach and context. Many users of ADL are not aware of this origin (Agree). Saad Nagi (1965) introduced a broader definition than that underlying ADLs, capturing the notion that it is the gap between an individual’s capacity and abilities, on the one hand, and the demands of their social and physical environment, on the other, that is the most important measure of disability (Agree).

The National Health and Aging Trends Study framework, which gathers information on a nationally representative sample of Medicare beneficiaries ages 65 and older, was built from these foundations, and therefore incorporates multiple dimensions that affect individuals’ capacities; these include health conditions, impairments, body functions and structures, and capacity; and, for mobility, upper body, lower body, sensory, cognitive functions, and modified by environment and by accommodations that people use (Agree).

Rather than trying to develop a unified view of a comprehensive way of understanding disability, several Day 2 presenters—Hargrove, Valdez, and Garcia, in particular—argued for a more intersectional concept of disability cast in the context in which it is being examined. Researchers working on intersectional concepts of disability recognize that a condition may be disabling or enabling for people with different inherent capacities (Agree). Within this perspective, room exists for varying definitions of disability, and perhaps what is most needed is a set of definitions that are understood in terms of how they relate to one another (Maestas). That said, while multiple, often overlapping, measures are needed, and which depend on the context, there is value—particularly for use in general population surveys—in working toward a unifying measure in order to obtain accurate measures of population prevalence (Landes). Calls for consistency across data sources were tempered among participants by the recognition that so much heterogeneity exists in the large disabled population (67 million) and the multitude of purposes for which data are used. Employment surveys, health surveys (National Health Interview Survey [NHIS]—20 functional

measures), and income and education surveys all have specific requirements where a set of identical questions may be far from optimal.

Beyond the medical model, it is also important to develop alternatives such as those looking at self-perceptions and those using a social justice framework (Beltrán-Sánchez). A key point in setting definitions and conceptualizing people’s well-being is that disabling situations are determined not just by the functional impairment of individuals but also by the accessibility features driven by technology, the physical built environments, and other factors. In the workplace, the same condition may be disabling for a person employed in one job type and not in another. The result is that people with the same physical condition may or may not identify as disabled, and society and policy play major roles in this interaction (Clarke, Landes). Current measurement approaches focus much more on the individual medical aspect of this interaction and less on the social model and context; the National Institute on Aging is seeking to address this imbalance in its support of researchers (Jain).

Even for more intersectional measurement approaches, there is value in thinking about cardiometabolic risks and other biological mechanisms as a potential downstream or upstream set of indicators that would eventually lead to functional mobility or perhaps disability (Beltrán-Sánchez). Within the traditional approaches commonly in use, the developmental origins of health and disease and how our physiology and health may be changing should be incorporated to provide a broader sense of functional limitations and disability (Beltrán-Sánchez).

The presentation on pain also highlighted an aspect of disability that has not received much attention but is crucial in teasing out some of the disparities that are present but sometimes missed (Landes). This ties into the idea of measuring self-rated disability in a way that parallels the use of self-rated health scales in surveys. While there are limitations to self-rated measures, they have the potential to capture otherwise hidden disabilities, like pain. These conditions are not cognitive—rather they are physical—but they are disabling in a way that affects people’s ability to participate in day-to-day life (Zajacova).

There can be a tendency to focus on the social model and the understanding that disparities exist with the social structure, within an ableism mindset. Yet, as Hughes and Paterson (1997) argued, the social model abandons the disabled body to the medical community. By making that bifurcation between disability and impairment, it allows impairment in the disabled body to stay within the hands of medicine and rehabilitation. The research community and public more broadly tend to overemphasize the role of functional limitations in capturing social differences. While functional limitations are a real piece of the overall picture, how disabled people are engaging in the built environment is also importantly deter-

mined by ableism, by access to resources, and by social and economic status (Landes).

DATA SOURCES AND NEEDS

Often research on disability prevalence and trends relies on data from either the Health and Retirement Survey or the American Community Survey (ACS). Researchers are now more commonly looking to use alternative data sources, including those originally developed for other purposes (Montez). One motivation is to achieve greater granularity. Policies and programs are often implemented or enacted on a local or even hyperlocal level, yet geographic information in traditional data sources is often at the state level. An example of the need to use smaller units of geography, identified in the presentation by Maestas, is in research requiring information on how appellate judges are a variable in decisions about program eligibility (Agree).

Another need is to disentangle the relationship between disability and health, including issues such as discrimination (Agree). As of 2019, the NHIS dropped every question previously included about intellectual and developmental disability. As a result, this population goes essentially unmeasured within federal surveys—the data no longer exist (Landes).

Limitations in government survey data led to a discussion of alternative (nonsurvey/non-probability sample) data sources—a research strand that the National Institute on Aging is interested in pursuing. Among the promising alternative datasets for researchers to explore, one example is the National Neighborhood Data Archive (NaNDA), referenced by Philippa Clarke in her presentation and used to link neighborhood characteristics at the ZIP code level. NaNDA can leverage AI to analyze high-resolution aerial imagery to detect sidewalks, curb cuts, crosswalks, and other elements of the built environment. Another example is the National Organization on Disability (NOD)—referenced by Douglas Kruse in his presentation—which collected data on 49 disability employment practices and on eight million workers to help inform their mission to improve opportunities for employees with disabilities. Access to these kinds of alternative data sources should continue to be explored and be drawn upon to supplement the federal survey data typically used by researchers (Jain). With regard to COVID-19, the National Institute on Aging funded the Social, Behavioral, & Economic COVID-19 Coordinating Center, exploring the social, behavioral, and economic effects of the pandemic. This could be an important area of exploration as well (Jain).

In terms of a strategy for advancing data collection, several workshop participants acknowledged the need to inform a very wide range of measurement purposes. Data are needed to generate core demographic profiles and trends (like race, gender, and geolocation) for the population; at the

same time, there is a strong research and policy need for measures of the functioning ability of individuals with very different health and disability profiles. To answer specific questions, flexibility in the data infrastructure is needed in a way that incorporates a variety of data sources (Landes, Swenor, Mullen).

The challenge to the measurement strategy is reflected in the fact that the Interagency Committee on Disability Research¹ references as many as 73 different definitions. Categories are not always well captured (e.g., mental disabilities being dropped altogether on NHIS). The debate over disability measurement in federal surveys (beyond ACS 6 or Washington Group questions) also requires a resolution (Landes). Ideally, more and more precisely worded questions would be included, but this is not easy to achieve in large general-population surveys such as the Current Population Survey, ACS, and the Survey of Income and Program Participation, where space is extremely limited and valuable. Nonetheless, a consensus is needed about how much space would be needed to satisfactorily measure disability for population monitoring purposes (Landes).

IMPACTS OF POLICIES

In considering policies and practices, there is a need to explore the variety of pathways faced by individuals with disabilities that result in different risks depending on the environment. One example is examining how people with disabilities who are unable to engage in physical activity may have a higher risk of diabetes or metabolic disorders, and another is learning how to create enabling environments that will address people’s limitations (Mehta, Valsez, Clarke).

In building an evidence base on this topic, researchers examine which policies successfully reduce disability disparities and which ones exacerbate them. As described by Meara in her presentation, some policies may be protective even when they are not specifically labeled as disability policies (Maestas). Social Security Disability Insurance, the Americans with Disabilities Act, and Medicaid were all found to have large effects, something that is rare in policy research. However, there are many other policies that are relevant and for which research should be performed, including those related to minimum wage, right-to-work laws, the Earned Income Tax Credit, and paid leave (Montez). It is important to look at the details of policy implementation, including duration, sequence, and timing of exposure, all of which can have fairly profound effects (Jain).

An important topic for which data do not yet satisfactorily exist is subminimum wages paid to people with intellectual or developmental dis-

___________________

¹ https://icdr.acl.gov/home#gsc.tab=0

abilities. More effort is needed to tease out data to find out what is going on with these populations, particularly when there are exploitive employment practices at work (Landes).

FUTURE RESEARCH

Currently, researchers often do not examine trends beyond 2020 because of the complications and irregularities introduced by the COVID-19 pandemic and related changes in policies and programs. One goal of the workshop was to nudge participants to undertake or report on more recent trend analyses, although much more is needed (Montez). In particular, more work is needed to understand the recent experiences occurring for subpopulation groups such as Latinos and other people of color who tend to be disadvantaged (Beltrán-Sánchez).

Workshop participants reiterated the potential value in re-energizing research on trends and disparities, even if some of this will necessarily be correlational, associational work. This could be something that funders might be encouraged to pursue. As articulated by Ari Ne’eman during his presentation, changes in the composition of the disabled population are occurring and have to be understood for effective policy. Similarly, work on projections of future trends would be valuable (Zajacova). It was also noted that there is considerably more work being done on measuring trends for the older population; but there is not similar work being done for younger ages (Agree).

Even though we have imperfect measures, the signal is very clear that, since around 2000, we have had deteriorating functional health among working-aged individuals. This marks a clear deviation from the trajectories seen since the 1980s and 1990s (Mehta). Workshop chair Emily Agree concluded that additional research into understanding these changes would have great value.