| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
|
|
|
|||||||
|
|||||||||
Correspondence: Address correspondence to Jennifer Cornman, Senior Research Scientist, Polisher Research Institute, Abramson Center for Jewish Life (formerly Philadelphia Geriatric Center), 1425 Horsham Road, North Wales, PA 19454. E-mail: jcornman{at}abramsoncenter.org
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Disability • Assistive devices • Survey design
At the same time, the relative importance of assistive devices (also referred to in the literature and surveys as assistive technology, special equipment, or aids) in meeting the needs of the older population has been increasing. Nearly two thirds of older Americans with an ADL disability use one or more assistive devices to meet their needs (Agree & Freedman, 2000), and this estimate increased during the 1980s and 1990s (Freedman et al., 2004; Manton, Corder, & Stallard, 1993; Spillman, 2004). The prevalence of the use of assistive devices is likely to rise further as the number and types of devices available increase. In the past 20 years alone, the number of assistive devices has expanded from 6,000 products (U.S. Congress Office of Technology Assessment, 1985) to over 29,000 products (National Institute for Disability and Rehabilitation Research [NIDRR], 2004).
Studies of disability trends have relied almost exclusively on self-reported measures from national surveys. Although potential sources of bias in disability trends have been explored in systematic reviews (e.g., Freedman et al., 2002), the various approaches to measuring assistive device use and their influence on disability prevalence rates has not been investigated. Our purpose in this article is to describe the range of approaches that national surveys employ to measure assistive device use; investigate how prevalence estimates of assistive device use by older Americans are influenced by survey design features; and explore how analytic decisions about the inclusion of assistive device measures may affect estimates of late-life-disability prevalence.
Background: Approaches to Defining and Measuring Disability
Unlike measures of disease, disability is socially defined, representing the intersection of an individual's abilities, her or his social and physical environment, and the demands of daily tasks (Agree, 1999). Conceptually, a person experiences a disability when the demands of a given context (environment and task) do not match her or his physical, cognitive, and sensory capabilities. Given these complexities, the lack of agreement about optimal measures of disability (Altman, 2001) and the wide variation of measures appearing in the late-life-disability literature are not surprising.
At least three general approaches to defining the prevalence of late-life disability have appeared in the literature. The first considers those who have difficulty with daily tasks, most often in the absence of accommodations (e.g., without help or equipment), as having a disability. The second approach classifies those needing or getting personal care as having a disability. Less common is a third approach that categorizes people who use any assistance—either personal help or assistive devices—in daily activities as having a disability.
Figure 1 illustrates how these various measurement approaches overlap. The largest circle represents the population that reports difficulty with daily activities in the absence of help or equipment. The circles drawn with dotted and dashed lines represent the populations using assistive devices and getting help from another person, respectively. Although the illustration here is for a hypothetical population, one can imagine that the size of the circles as well as the extent of overlap will vary by key demographic characteristics, particularly age.
|
The figure also illustrates that restricting the definition of disability to the population getting personal help excludes two potentially important groups receiving assistance: those who have difficulty and bridge that difficulty with only assistive devices and those who use devices for reasons other than difficulty (e.g., safety or convenience). Whether these so-called omitted groups have similar characteristics as those receiving personal help is unclear; one might hypothesize that they are likely to differ because they have accommodated their difficulty (or if not their difficulty, their situation) independently, without human assistance.
In sum, two common approaches to measuring disability—defined either as having difficulty or as receiving help—potentially omit relevant groups of individuals who use assistive devices to carry out their daily tasks. In this article, we explore both types of omissions and their implications for estimates of assistive device use and of disability prevalence.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
For item nonresponse (e.g., don't knows or refusals), we also assume that respondents are not using the devices in question. In the NHIS core, HRS, MEPS, NLTCS, and MCBS, this affects less than 0.7% of cases included in the samples from these surveys. In the NHIS disability supplement, about 2% of cases are missing on the device-use question, and we assume that these individuals are not using any devices. In the SIPP, there are no missing data in the public release file because an imputation procedure is used to assign values to missing data. Approximately 8% of cases have values imputed for the device-use questions. Results are unaffected if cases with item nonresponse or imputed data are dropped.
We use a similar strategy for dealing with missing data on the disability questions. If a respondent does not provide an answer to a disability question, we assume that the respondent does not have a disability. In the MEPS, NHIS, and HRS, less than 0.1% of cases are missing on the disability items and about 8% of cases in the SIPP have imputed data. In addition, in the HRS, respondents who report no functional limitations, or one functional limitation and no difficulty dressing, are skipped around questions about difficulty with walking or transferring, the measures of disability examined here. We assume that these respondents do not have difficulty with these activities (disability data from the NLTCS or MCBS are not examined). Again, results do not change if cases with either item nonresponse or imputed data are excluded.
Methods
First, we discuss differences in the design and wording of the surveys. Following this, for each survey, we present the weighted percentage of those aged 65 or older and those aged 85 or older who use any assistive device, assistive devices for specific ADLs (mobility, bathing, toileting, transferring, eating, or dressing), and individual devices (wheelchair and walker, cane, or crutches).
Next, we explore the implication for estimates of assistive device use of restricting device-use questions to those reporting difficulty. Using data from the HRS, we test differences in device-use rates estimated with and without restriction on difficulty.
We then use four surveys to explore the implications for disability rates of excluding (a) those who use assistive devices but report no difficulty (SIPP and HRS) and (b) those who use devices but report no help (MEPS and NHIS). In both cases we show the unadjusted rate excluding the group in question and the adjusted rate when the omitted group is added to the estimate, and we use t tests to examine differences in rates.
Finally, we explore the characteristics of the two excluded groups by using chi-square tests to indicate significant differences in rates across demographic groups (age, gender, race, marital status, and education). We also use multinomial logistic regression models with SIPP and HRS data to estimate models with a three-category outcome: (a) reports no difficulty and does not use devices; (b) reports no difficulty but uses devices; and (c) reports difficulty (the reference group). We estimate a second set of models with the MEPS and NHIS for the outcome: (a) gets no personal help and does not use devices; (b) gets no personal help but does use devices; and (c) gets personal help (the reference group). Predictors include age (60–74, 75–84, and 85+), gender, race (Black, White, or other), marital status (married or unmarried), and education (less than high school, high school education, or more than high school). We exclude cases missing on these predictors from both the bivariate and the multivariate analyses. We drop approximately 0.5% of the cases in the HRS, 1.8% in the MEPS, and 1.7% in the NHIS (data not shown).
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
Question Focus: Activities
A second difference is in the activities on which questions focus. Three surveys (HRS, NLTCS, and MCBS) ask about device use for specific tasks, and the other three (NHIS, MEPS, and SIPP) focus on use more generally, without reference to specific activities. Two surveys (NLTCS and MCBS) ask about equipment used to help in performing all ADLs, and one (HRS) limits its questions about device use for walking and transferring. The global questions vary in terms of how devices are referred to and what kinds of activities are mentioned. (Mentioned activities are, e.g., usual activities in the NHIS disability supplement, personal care or everyday activities in the MEPS, and no specific activities in the SIPP or NHIS core; see Appendix A for details.)
Question Focus: Devices
Finally, surveys also differ in whether they record specific types of devices or any use. Three surveys (NHIS, MEPS, and MCBS) record any use (the latter for each activity). The other three surveys record more detailed information. The NLTCS and HRS, for example, ask respondents to name the type of device used for each activity, whereas the SIPP asks explicitly whether the respondent uses "a wheelchair or scooter" or "a walker, cane, or crutches." Among the surveys that record more detail, there is variation in the items that are recorded and whether items are grouped together.
Prevalence Estimates Across Surveys
The first three rows of Table 3 show the prevalence estimates of the use of any device. We derived these estimates in slightly different ways and are thus shown on different rows. Estimates from the NHIS (core and disability supplement) and the MEPS are from a single question about the use of any assistive device. We derive the HRS, NLTCS, and MCBS estimates by summing across the use of devices for specific ADL tasks. The SIPP estimate includes those that report using a wheelchair or scooter or a walker, cane, or crutch.
|
The middle panel of Table 3 provides task-specific estimates of device use. Estimates are in general quite similar across surveys for the population aged 65 and older, but they vary a bit more for some activities for the 85 and older population. For example, estimates of device use for bathing (6.5% to 8.5%), toileting (3.2% to 4.9%) and transferring (5.1% to 5.9%) are quite similar for the 65 and older population, but the percentage of the 85 and over population using devices ranges from 16% to 24% for bathing, from 8% to 15% for toileting, and from 12% to 20% for transferring. There are two exceptions for this general pattern. First, estimates vary for both age groups for indoor mobility (8.1% to 14.6% and 24.9% to 40.3%, for the 65 and older and 85 and older populations, respectively). The indoor mobility estimates come from only two surveys—the HRS and NLTCS—which use different screening approaches and different questions (see Table 2 and Appendix A for details). Second, estimates for using devices for eating and dressing are quite low (less than 1.5%) and quite similar within each age group.
The bottom panel of Table 3 shows the frequency of use of specific devices. Although the HRS and NLTCS collect information on the use of additional devices, in the SIPP, wheelchairs, walkers, canes, and crutches are the only devices about which data are collected. (The NHIS, MEPS, and MCBS do not ask questions about specific devices.) Results show that from 2% (HRS) to 5% (SIPP) of the population aged 65 and older and from 6% (HRS) to 11% (SIPP and NLTCS) of the population aged 85 and older use wheelchairs. Between 11% (NLTCS) and 17% (SIPP) of those aged 65 and older use a walker, cane, or a crutch, and between 32% (NLTCS) and 41% (SIPP) of those aged 85 or older use one or more of these devices.
Restricting Assistive Device Use Questions: Effect on Device-Use Prevalence Estimates
The left-hand side of Table 4 cross-tabulates the HRS device-use questions and questions about difficulty with daily activities (see Appendix B for wording of difficulty questions). The right-hand side demonstrates how estimates of device use are affected when questions about device use are restricted to those reporting difficulty. The first row shows that 9% of those individuals aged 65 and over who report having no difficulty walking across a room also report using indoor-mobility equipment. If these respondents were excluded from the use questions (as they would be if the device-use questions were restricted to those reporting difficulty), the estimates of using equipment for walking across a room would be cut in half. That is, only 6.3% would be identified as using indoor-mobility devices compared with 14.6% when no restriction is imposed. The rate of using transferring equipment also would be cut nearly in half if transferring-device-use questions were restricted to those reporting difficulty transferring (2.6% with restriction vs 5.1% with no restriction).
|
|
Combining Assistive Device Use With Rates of Difficulty and Help
Table 5 shows the potential effects of combining estimates of assistive device use with rates of difficulty and of getting personal help. The left side illustrates the changes in disability rates that result from augmenting difficulty responses with information about assistive device use. In the SIPP, for example, 6.5% of the population aged 65 and older report having difficulty getting around inside. An additional 13.1% of individuals report using a device but have no difficulty. By including these respondents, the percentage of individuals with an indoor-mobility disability would triple. In the HRS, results for walking across the room are similar, with the combined estimate of having difficulty or using a device (16.3%) double the estimate of those reporting difficulty (8.0%). The combined estimate for transferring is 8.8% compared with 6.2% for the estimate based only on difficulty.
|
Identifying Those Reporting Device Use but no Difficulty or Help
The top panels of Table 6 shows, for several demographic groups, the percentage of individuals who report no difficulty but use devices for several mobility-related tasks. Older respondents, women, Blacks, those who are not married, and respondents with less than a high school education are more likely to be in this group for getting around inside (SIPP and HRS), walking (SIPP), and transferring (HRS). Similarly, the percentage of individuals who report using devices but not help (MEPS, NHIS) is higher for older respondents, women, Blacks, unmarried individuals, and those with less than a high school education (bottom panels, Table 6).
|
|
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In addition, both device use and disability estimates are affected when device-use questions are asked only of those reporting difficulty with daily tasks. For example, by restricting questions to those reporting difficulty with mobility-related tasks, thus omitting those individuals who use devices but report no difficulty, our results show that estimates of device use would be half as large. If estimates of disability were based on difficulty or device use, that is, included those individuals who use devices but report no difficulty, the prevalence of disability would be up to three times as large as estimates based on difficulty alone. Finally, results showed that individuals who report no difficulty but use devices for mobility tend to be highly educated, and the majority of these individuals report using a cane.
Estimates of disability based on receiving or needing help also are influenced by device use. Combining estimates of those individuals using devices with those getting help increases estimates nearly threefold compared with rates of those persons getting help alone. Furthermore, younger individuals (aged 65–74) and Whites have greater odds of using devices but not help compared with getting help.
This study has several limitations. First, we could not definitively assess the amount of bias introduced into assistive device use and disability estimates by restricting questions about device use to those reporting difficulty. Although we contrasted estimates of difficulty and device use for mobility-related activities, whether these differences hold for other activities is unclear. Moreover, in some cases we assumed that the devices (e.g., wheelchair or walker, cane, or crutch) were used for all mobility-related activities across settings. Finally, neither of the two surveys used to assess this bias unambiguously measures difficulty without the use of assistive devices: the SIPP question specifies that users of assistive technology report difficulty when they use their equipment, but without personal help (see Appendix B), and the HRS measures of difficulty do not make reference to assistance at all, leaving up to the respondents whether they report their difficulty with or without assistance (see Appendix B). This implies that some proportion of the gap between reports may include satisfied users of technology who experience no difficulty when using their devices. Taken together, these limitations suggest that the difference between the older population with disabilities defined purely with difficulty questions and the population defined with a combination of difficulty and device use is likely sizeable, but just how large is unknown. Had a true measure of underlying difficulty been available, the combined rates of difficulty and device use may not have been as large as shown here.
A related limitation is that we were unable to explore whether task-specific, global, or detailed device-use questions provide similar estimates of the prevalence of assistive device use. For this issue to be explored further, a survey that measures devices by using more than one approach is needed. A study now underway to create an instrument for national surveys to measure assistive-technology use and effectiveness is being funded by the Assistant Secretary for Planning and Evaluation in cooperation with the National Institute on Aging and the National Center for Health Statistics. The new instrument, developed with input from policy makers, researchers, and extensive cognitive testing by the National Center for Health Statistics, will be piloted with a purposive sample of 300 individuals aged 50 and older. Both global and detailed items will be included and use will be assessed across multiple tasks.
Despite these limitations, our results have important implications for survey designers interested in measuring device use and disability among older Americans. Our results suggest that a nontrivial proportion of older adults use devices but report no difficulty with daily tasks, and that these individuals are more highly educated than those reporting difficulty. Future studies might explore a number of possible explanations for such discrepancies, including the relationship among reports of device use, underlying disability dynamics, the physical, cognitive, and mental status of individuals, and perceptions about ability and independence. To facilitate these explorations, survey designers may want to consider asking questions about assistive technology independent of questions about difficulty and give researchers the option of studying further the personal characteristics and social and environmental conditions of this group. Qualitative interviews also may add to our understanding of how these individuals perceive their situations and why they report using devices but not difficulty.
Researchers interested in studying older Americans who use assistive devices may find reassuring the finding that national estimates fall within a relatively narrow range. However, despite the fact that the estimates are similar, it is possible that different approaches to identifying assistive device use may capture different groups of users. An important next step will be to investigate the reliability and validity of the various approaches identified here—global, task specific, or device specific—and to better understand whether the various approaches identify similar populations.
Finally, researchers who study disability may want to consider including those who report using devices but no difficulty or no help. These individuals are important to consider because, as a younger and more educated group, they may represent the patterns of adaptation new cohorts will make as they age. Whether this group should appropriately be included in estimates of the older population with disabilities, however, raises questions about where the clinical threshold for disability should be set in national estimates. It may be more appropriate to include this group along with those making behavioral accommodations in the population with preclinical disability (Fried et al., 2000).
Evidence suggests that the number of older adults using assistive devices has been steadily growing over time (Freedman et al., 2004; Manton et al., 1993; Spillman, 2004); whether this trend reflects changes in underlying health (e.g., decreasing levels of disability), changes in the propensity to turn to devices instead of human care (i.e., increasing acceptance of technology), changes in the availability of personal care, or changes in the desire to maintain independence is still unclear. What is clear is that the use of assistance and perceptions of difficulty and ability are inextricably interwoven, and more attention to the measurement of these concepts, and to the implications of measurement techniques for our estimates of disability in the older population is warranted.
|
Footnotes |
|---|
1 Polisher Research Institute, Madlyn and Leonard Abramson Center for Jewish Life, North Wales, PA.
2 Department of Population and Family Health Sciences, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD.
Decision Editor: Linda S. Noelker, PhD
Received for publication May 5, 2004. Accepted for publication August 19, 2004.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Kitchener, T. Ng, H. Y. Lee, and C. Harrington Assistive Technology in Medicaid Home- and Community-Based Waiver Programs Gerontologist, April 1, 2008; 48(2): 181 - 189. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. C. Cornman and V. A. Freedman Racial and Ethnic Disparities in Mobility Device Use in Late Life J. Gerontol. B. Psychol. Sci. Soc. Sci., January 1, 2008; 63(1): S34 - S41. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Sachs-Ericsson, C. Schatschneider, and D. G. Blazer Perception of unmet basic needs as a predictor of physical functioning among community-dwelling older adults. J Aging Health, December 1, 2006; 18(6): 852 - 868. [Abstract] [PDF]
|
|
|
|
|
|
V. A. Freedman, E. M. Agree, L. G. Martin, and J. C. Cornman Trends in the Use of Assistive Technology and Personal Care for Late-Life Disability, 1992-2001 Gerontologist, February 1, 2006; 46(1): 124 - 127. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| All GSA journals | Journals of Gerontology Series A: Biological Sciences and Medical Sciences | Journals of Gerontology Series B: Psychological Sciences and Social Sciences |
