NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Research Council (US) Committee on National Statistics; National Research Council (US) Committee on Population. Improving the Measurement of Late-Life Disability in Population Surveys: Beyond ADLs and IADLs, Summary of a Workshop. Washington (DC): National Academies Press (US); 2009.

Cover of Improving the Measurement of Late-Life Disability in Population Surveys

Improving the Measurement of Late-Life Disability in Population Surveys: Beyond ADLs and IADLs, Summary of a Workshop.

Show details

6Research and Development Toward Improved Measures of Late-Life Disability

In the last session of the workshop, participants highlighted their perceptions of some of the research topics and priorities that emerged from the presentations and discussions, regarding important and potentially fruitful areas for research and development, given the goal of improving the measurement of late-life disability in population surveys. The session began with a panel of three workshop presenters who led off the discussion, sharing their initial thoughts on areas for research and development, followed by a general discussion and comments from all participants.


Thomas Gill identified two areas of research opportunities applying methodologies discussed during the workshop. The first area was the application of time-use data to study disability, but in terms of recovery from disability. One cannot have a full and complete understanding of disability unless one understands the recovery process, he said. It is increasingly clear that disability is a highly dynamic process. There are opportunities that have not been fully realized in terms of an expanded array of functional outcome measures with regard to recovery. That is where time-use data could provide some new and unique insights, potentially for shorter periods of time than might otherwise be possible in the context of the traditional disability surveys, because the recovery process typically occurs over a relatively short period of time. Time-use data could prove quite valuable in providing a more complete understanding of how and to what extent individuals who become disabled, say by a hip fracture or other injury or any hospitalization, regain the full array of their predisability function.

The second area is the use of vignettes, a promising method that calls for further research and experimentation. The focus to date has been on cross-national comparisons. Can vignettes be used within the United States to better understand the differences in disability among different ethnic or racial groups or between genders? For some of the disability patterns and disparities described in the workshop, Gill noted that the African American population has a disability profile that mirrors the white population, but for 10 years younger. Could some of that difference be understood through the use of vignettes in identifying the different pathways to disability and distinct populations in the United States?

Arthur van Soest highlighted four areas. First he asked: What is the definition of late life? This workshop is about late-life disability. One can distinguish between the older old (say, 70 years and older) and the younger old (say, ages 50–70). There are differences in the issues concerning these two groups. For the older group, such things as home adjustments, devices, informal and formal care, and neighborhood characteristics seem to be most relevant. For the younger group, the relationship between work and disability seems crucial, including working conditions, employer attitudes and accommodations, job characteristics, stressful and physically demanding work, and stress due to unemployment.

Van Soest noted that most presentations during the workshop focused on the older group, but the younger group also deserves attention for two reasons. First, a substantial percentage in this group already suffers from disabilities; second, health and disability, as well as employment characteristics, at younger ages, are predictive of disabilities at an older age. So the issues may be different for the two groups, but both are worth studying. One of the implications is that longitudinal data are needed to follow people over a long period of time. He noted that the Health and Retirement Study is now reaching the stage at which one can start to study these long-term effects with good longitudinal data.

Second, van Soest noted that health is well known to be related to socioeconomic status, and the same is true for disability. Whichever method is used to measure disability, the data will show a negative association between prevalence of disability and socioeconomic status—education level and income or wealth. Determining the causal mechanisms that may lead to this association is an important research question.

He said that the workshop pointed to several potential mechanisms that may explain a causal effect of socioeconomic status on disability prevalence: working conditions (physically demanding work, stress), differences in health behavior (e.g., obesity is more prevalent among less educated people), as well as the affordability of health care or home modifications and devices. This is also an issue for which international comparisons will be useful, since the extent to which home modifications and special devices are covered by insurance or subsidized for low-income groups varies across countries.

The third area van Soest identified is disability and quality of life. The idea of studying time use in relation to disability is very interesting, particularly in how people cope with their disabilities and how disability affects the quality of their lives—not only the effects on time use, but also on social participation, a person’s economic situation, and, ultimately, well-being and happiness.

The fourth area concerns survey methodology. Many presentations identified existing problems, although often no one knows how to solve these problems. Van Soest noted four issues:

  1. Which questions can we ask and how can we ask them? For example, seemingly simple questions apparently lead to inconsistent answers from husbands and wives.
  2. Some things can be determined better in different ways. Using biomarkers is a promising development that is now feasible. In addition, using administrative data—on income, wealth, pension entitlements, and perhaps also on health care use, as some European countries do—will help to improve data quality. The new approaches are not suggested to replace survey data, but to supplement them.
  3. In terms of sample design and sampling frame, sometimes a representative sample of the population is not needed, but often it is. In that case, the selection problems seem salient: people with a disability will be less likely to participate and more likely to drop out. Proxy interviews offer a partial solution. Although there may be problems with proxies, that does not mean we should not use them. More research on the quality of proxy interviews is warranted.
  4. Internet interviews are a promising and cost-effective way for the future but obviously suffer from selection problems even more than other approaches. The use of proxies in Internet interviews seems something that is worthwhile to test.

David Weir spoke to the issues of definition and measurement. He said he was sure most of the participants had been struck by the box diagram presentations on what disability is. The concept of disability is complex because it is relational. The key elements are health, the physical and mental capacities or abilities of an individual; the environment, much of which is completely beyond the individual’s control; and then, in the middle, the behaviors and psychological adaptations through which people translate their abilities into activities and rewards they are able to achieve.

What should surveys do in light of this conceptualization? The nationally representative population surveys, first and foremost, have to be very clear about the measurement of health and about the measurement of abilities and capacities. Weir noted that the description of the Washington Group’s effort was great. That kind of rigor in developing self-report questions will allow conduct of comparable studies across population groups.

Clearly, there is a place for performance measures, which are, by their nature, attempts to measure capacities and abilities. The idea is to abstract as much as possible from the environment so that a person with one set of capacities might be disabled in one environment but not in another. However, information about a person’s health is also important, in part because if one is focused on late-life disability, one will want to know about aging and how that translates into life experience. Weir said that an operating definition of aging is a decline of physical and mental abilities and capacities, and no one has found a solution to that. Researchers want to know about what is being modified—whether medical intervention or people’s behavior or something is changing the pace at which people’s capacities decline over time. The only way to have confidence in those types of findings is to have solid comparable measures of those capacities and abilities in surveys. So that should be an important priority.

Given the nature of declining function with age, any survey is going to need to rely on proxy reporting to capture some of the people who have a disability. If people who cannot respond for themselves are not represented in a survey, it will miss what is going on. So there is a need for measures that are interchangeable between proxies and self-reports. Weir noted that one area that was not discussed much during the workshop is whether the existing health measures cover all of the domains of health that are relevant to disability in the sense of limitations in activity. Turning next to the other end of the spectrum, the environment, surveys of people can address what is going on in the environment, and one can also survey environments. That is actually being done; for example, looking at how transportation systems function, at whether people have access to places of work, and at shopping malls.

Finally, work on adaptation is the research frontier. Surveys have a role to play in the sense that people can be queried about whether they are using assistive devices and how they are adapting. They can also be asked how they feel about their health and how they feel about the activities they are able, or unable, to engage in. These types of inquiries have to be viewed as more of a research activity for which nationally representative population surveys are not necessarily required because they would be for monitoring trends and evaluating health policies.


During the general discussion, participants identified several areas for further research and development that may lead to improved or enhanced measurement of late-life disability. Among the several comments and questions, six subject areas surfaced as the focus of most of the discussion: (1) clinical classification of disability, (2) measuring levels and changes in the levels, (3) vignettes, (4) inclusion of the working-age population, (5) the purpose of surveys, and (6) measures of participation and the environment.

Clinical Classification of Disability

Bob Wallace (University of Iowa) said that all the areas for further research identified so far are important, but he did not think that disabling illnesses of older people are being characterized very well. A good deal of surgery, as well as many functional disorders that affect the disability, such as musculoskeletal disability, are not being talked about or measured. A commodity called disability, even if it is perfectly measured, really provides little information. For example, neuropathy is different from constipation, which is different from old fractures and old football injuries. Better clinical classification of disability is needed of the kinds of dysfunctions and disabilities of older people. Only with such classification will it be possible to move to the next level, which is prevention. He observed that there are ways to prevent various kinds of disabilities, and they should be assessed in population surveys.

Wallace also noted that late-life disability should be distinguished from disabilities that people have had most of their lives—these people are different, and their trajectories are different. It was noted that because of sustaining medical care, for example, there will be more older people with conditions such as Down syndrome and cystic fibrosis than before.

Jack Guralnik observed that although significant advances have been made moving forward from the medical model of disability to assessing function and disability in the whole older person, to understand trends in disability and to consider interventions to prevent disability, researchers will need to start looking more at the diseases underlying disability. These types of data would be very valuable for determining intervention strategies to prevent disability. He noted that this topic was not discussed at the workshop. Some surveys have included a question on the main cause of disability, but no one has a good sense of the validity of the responses. There have been statistical analyses looking at the association of disease and disability, but it is hard to know if it shows just a cross-sectional association or is causal. Even longitudinal studies have some problems. There is a tremendous amount of comorbidity in the older population and it is hard to sort out the main cause of disability.

Difference Between Measuring Levels and Measuring a Change in That Level

John Haaga (National Institute on Aging) observed that information on change in the level of functioning is more important and a better prognostic indicator of potential disability for an individual than just information on levels. This information is useful in clinical work, and it can be measured even one time in a survey. For example, there is no right amount of time to sleep, and the number of hours someone sleeps is not a good measure of anything other than sleep, but a recent change in the number of hours of sleep is a warning sign. Similarly, gait speed and other items that are good measures of disability become even better measures or better prognostic indicators if there has been a recent change in the item (in either direction).


Richard Suzman (National Institute on Aging) observed that vignette methodology remains a promising experimental technique that is worth much more experimentation and that much of the success is going to come from writing good vignettes. Perhaps presenting them in different ways and ensuring that the respondents have understood the vignette when they give answers to the questions will enhance comparability of self-reports across important population groups.

Including the Working-Age Population

Andrew Houtenville (New Editions Consulting) stressed the need to include younger ages in population-based surveys. Researchers working with disabilities among the working-age population would benefit from help in terms of defining disability either in population-based surveys or for programs such as Social Security Disability Insurance and Supplemental Security Income benefits. With the increase in the age for receiving Social Security benefits and other economic forces, work is going to be an increasing part of older life.

The work done with vignettes focuses on work limitations. A participant noted that, unfortunately, questions on work limitation have been eliminated from two basic population surveys, the American Community Survey (ACS) and the Current Population Survey (CPS) monthly questions, possibly because of a limitation of space, but also because of distrust of that methodology.

Barbara Altman explained that ACS and CPS have not eliminated questions on work, but only on work limitations. The reasoning was that to get work limitations, one can cross-tabulate people with functioning limitations with whether or not they work. ACS questions are going to be used in CPS; they are also going to be used in the National Crime Victimization Survey for the Department of Justice and in the National Health Interview Survey of the National Center for Health Statistics.

She agreed that the age focus needs to be expanded if any kind of preventive intervention is considered. She also identified some other areas that should be considered for addition or expansion. From a population survey perspective, the function and assessment aspects of the questions need to be expanded in order to expand one’s understanding of the population that is at risk. Organized activities need to be added to population surveys on a regular basis; organized activities that are important to people’s participation in multiple ways (such as transportation) give a clue about the kinds of barriers that a person might face. Finally, there is need for longitudinal data on populations.

Richard Suzman pointed out that cross-sectional population surveys are different from the collection of longitudinal data and one would not necessarily use the same questions. Also, different surveys have different purposes, and one should not impose a cross-sectional official survey framework on a research question.

Purpose of a Survey

Jay Magaziner observed that we need to ask why we are asking these measurement questions. What is it that we really want to answer? If the focus is issues that relate to underlying diseases and prevention of disability, then we need to develop questions and studies around that. If the purpose is having an item that can be used nationally to represent disability for policy purposes, that may be a different kind of question and a different set of items to make the responses more interpretable and useful. So, when talking about disability, it is important to think about the purpose of the inquiry, whether it is an individual research study or a global population survey.

Measures of Participation and the Environment

Alan Jette (Boston University) noted that if a key issue is the extent to which loss of function and diseases of late life affect the lives of individuals in society, then it is important to attempt to measure participation, at least in some dimensions. This is particularly important if one wants to assess the effectiveness of interventions for older populations at the societal level. The ability of older people to maintain engagement in important social roles in their lives is an extremely critical element, and Jette said that he sees it as part of the disablement process. There is a role for it in population surveys, and it can be measured. There is some evidence now that with techniques such as the item response theory and computer-adaptive testing (CAT), one can begin to get at some elements of important social roles in late life in a way that might be feasible to build into population surveys.

A participant commented that a barrier to obtaining better measures of participation and the environment in population surveys has been the burden of including such a large number of questions in a survey questionnaire. Given these new methodologies, is there now an opportunity to make these measurements more relevant, given the new psychometric techniques? The work could really benefit from short forms or CAT.

Jette responded that it is important to have improved measures of the environment, about which even less is known than in the measurement of participation. Measurements of the environment, if incorporated into population surveys, will help provide a better handle on the question: If the environment is changed, will it have a beneficial impact on the lives of older individuals? However, measures of the environment are not yet ready for incorporation in population surveys, but that is an important area for research.

Vicki Freedman noted that the environment does not mean only physical challenge, that is, it is not just the steps of the grocery store, if there are any, or how wide the aisles are, because technology has changed the meaning of the environment. For example, people can pick up the telephone to order prescriptions and have them delivered by the drugstore, which has changed the whole notion of environment.

It was pointed out that it is important to conceptualize the environment, not just as the physical environment, but also as the social environment and the public environment.


The workshop brought together a large group of participants from many different disciplines who presented and discussed the various approaches to measurement of disability, some tested and some still being tested. In this final session participants identified some of the major research areas and priorities that are emerging that they would like to see advanced by NIA and other organizations in terms of research and development, given the goal of improving the measurement of late-life disability in population surveys beyond ADLs and IADLs.

Copyright © 2009, National Academy of Sciences.
Bookshelf ID: NBK28482
PubReader format: click here to try


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.2M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...