| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
Disability Statistics Rehabilitation Research and Training Center, University of California, San Francisco.
Requests for reprints or other information about this study should be directed to Robert Newcomer, University of California, 3333 California Street, Suite 455, San Francisco, CA 94118. E-mail: rjn{at}itsa.ucsf.edu
 |
Abstract |
|---|
 TOP Abstract Person-Environment TheoryMethodsResultsDiscussionReferences |
|---|
Methods. Data were obtained from the National Health Interview Survey, Supplement on Disability Followback Survey. Analyses consisted of a nationally representative sample of aged and nonaged adults with one or more ADL limitations.
Results. Slightly less than 1 in 5 subjects with a specific ADL limitation had unmet needs for that ADL. This was true across all ADLs. The likelihood of unmet ADL assistance increased with the number of ADL limitations and other health status indicators. It was at least 50% higher among those living in apartments than in houses and higher among Hispanics. There were no differences by age or gender.
Discussion. The P–E framework postulates that individuals seek settings matched to their capabilities, but findings suggest that many are at risk for adaptation at any one time. Specific risk factors are identified. Selection factors like preferences, expectations, and adaptation options available have not been directly measured.
More than 5.4 million American adults living in community settings need assistance from another person or special equipment to accomplish basic activities of daily living (ADLs) like bathing, dressing, toileting, transferring, and eating (National Center for Health Statistics [NCHS], 1998
). Growth in home- and community-based care during the 1990s (e.g., Spillman, Liu, & McGillard, 2002) has afforded the opportunity for many individuals to live in the community who might otherwise be in nursing homes or housing with services. Public concerns about the adequacy of assistance in community settings have paralleled this growth (e.g., U.S. General Accounting Office [U.S. GAO], 2003).
One definition of adequacy is the extent to which needed assistance is unavailable or insufficient. The prevalence of at least some unmet ADL assistance needs has been estimated. These rates range from 2.6% to 34.6% (see Desai, Lentzner, & Weeks, 2001; LaPlante, Kaye, Kang, & Harrington, 2004; Lima & Allen, 2001; Williams, Lyons, & Rowland, 1997, for an extensive review) among the aged. Much of this range results from whether incontinence is included in the definition of toileting difficulty. In all studies, rates of unmet need increased as the number of ADL limitations increased and if instrumental activities of daily living (IADLs; e.g., getting about inside, getting outside, getting places, shopping, preparing meals, doing light and heavy housework, using the telephone) defined tasks for which more assistance was needed.
These prevalence studies have found that living alone, the number or type of chronic health conditions, race and ethnicity (i.e., Black or Hispanic), marital status (i.e., divorced or never married), and limited insurance coverage were also associated with an increased likelihood of unmet need. Living with a spouse and use of equipment and assistive devices (e.g., canes, walkers, bathroom and/or kitchen modifications) were associated with reduced unmet need.
This article extends the examination of unmet needs in two ways. First, we use a person–environment (P–E) framework and broaden the factors considered, assessing the extent to which physical and social environmental factors are associated with the likelihood of unmet task assistance. This perspective recognizes that disability involves the match or fit between individual capabilities and the demands and supports available from the environment and that environment includes a host of physical and social dimensions. Second, we assess whether the risk for unmet need varies among housing types and among specific ADL limitations. The population studied had at least one ADL limitation.
|
PERSON–ENVIRONMENT THEORY |
|---|
|
TOPAbstractPerson-Environment TheoryMethodsResultsDiscussionReferences |
|---|
, Lawton and Nahemow (1973), and Carp and Carp (1984), among a number of others. The P–E framework guiding our work is that of Lawton and Nahemow. They postulate that individuals seek congruence between their capability and the demands made by their setting and that a match or congruence between capability and environmental demands (or supports) helps produce outcomes for the individual, such as life satisfaction and optimum performance, within that setting. Capability in this framework includes physical and cognitive functioning, economic resources, intellectual resources, and social skills. Environmental factors include physical features, services, and social components like roles, cultural norms, and social relationships. Mismatches, where demands exceed capability, are thought to produce stress and inadequate performance. Demands below capability are thought to produce understimulation and even an erosion of ability. As capability declines (e.g., due to changes in health status or cognitive functioning), other things being equal, an individual's adaptive range is thought to decline, with the individual becoming more sensitive to or affected by the demands from his or her environment. This is known as the environmental docility hypothesis (Lawton, 1969). We are concerned with identifying the extent of any mismatches, here reflected as unmet need for ADL task assistance.
The Lawton–Nahemow framework and variations on it have been used in many studies, predominantly of elders. Among other things, these studies adjusted for individual capabilities and examined the appropriateness and availability of such environmental attributes as social supports and service assistance for their effect on individual outcomes like life satisfaction (Kahana, et al., 1995), social activity, and relocation (e.g., Gitlin, 2003; Longino, Jackson, Zimmerman, Bradsher, 1991; Speare, Avery, Lawton, 1991). The Lawton–Nahemow framework has also been used to examine moves among the nonaged with disabilities (Newcomer, Kang, Kaye, LaPlante, 2002).
P–E formulations beyond the Lawton–Nahemow model seek to incorporate individual preferences or predispositions along with individual capability and environmental components (see Lawton, Windley, Byerts, 1982; Wahl & Weisman, 2003, for a review of these perspectives). These P–E frameworks have few empirical applications and require measures not available to this investigation.
|
METHODS |
|---|
|
TOPAbstractPerson-Environment TheoryMethodsResultsDiscussionReferences |
|---|
). About 40,000 households are surveyed through personal interviews. Items are self-reported or reported by a household representative. NHIS-D respondents identified with an indication of disability in the interview (criteria included functional limitation, specific disabling diagnoses, perception of disability, and use of disability-related services) were reinterviewed 7 to 26 months later in Phase II of the NHIS-D, also known as the Disability Followback Survey. Median lag between interviews was 13.6 months. Included in the Followback was information on the individual's housing, the length of time he or she had lived in those quarters, and his or her current functional ability (as measured by ADLs and IADLs). The NHIS-D Phase I began in January 1994. Phase II concluded in April 1997.
There were 25,805 respondents to the NHIS-D Followback Survey. Of these, 9,646 (representing 15.1 million people) were living outside of nursing homes and receiving or needing help from another person (including supervisory help) with either ADLs or IADLs. Our analysis included the subset of persons (representing 5.4 million people) who reported receiving or needing hands-on and/or supervisory or standby help from another person in performing at least one ADL task. Inclusion of those needing supervisory or standby help with those needing hands-on help allowed for a widened margin for identifying persons with ADL limitations.
Measures
Items in the NHIS-D Followback Survey reflected a fairly complete array of individual and environmental measures representing dimensions of the Lawton–Nahemow framework. These are shown in Table 1 with sample counts and weighted population estimates. Here we discuss the expected theoretical relationship of each measure to the outcome of unmet assistance needs.
|
Self-reported health status and the ability to perform ADLs and IADLs are representative of self-maintenance abilities. The use of 5 ADLs (bathing, dressing, transferring, toileting, eating) and 10 IADLs (getting about inside, getting outside, transportation or getting places, shopping, preparing meals, doing light and heavy housework, using the telephone, medication management, money management) are commonly used to measure disability. We used these 15 items. Another consideration in defining task limitations was whether to include only individuals receiving or needing "hands-on" help or to also include those needing only supervision or standby help from another person in performing a task. The selection of subjects for this study was inclusive of need or receipt of both hands-on and supervisory ADL assistance. This was done to assure a wide net for identifying at-risk individuals.
Incontinence, the use of special equipment (i.e., wheelchairs or walkers, feeding tubes, catheters), and difficulty opening or reaching cabinets represent inabilities and describe individual self-maintenance limitations. The latter may also reflect some adaptations. Cognitive ability reflects limitations requiring special adaptation by the individual caregivers. Hearing and vision problems may be either life long or of recent onset. The varied nature of these limitations complicates the expected direction of association with the outcome of unmet need. In the context of cross-sectional data we might expect that higher frailty and complex care needs may be seen in a steady-state situation and thus may be positively associated with more attentive supports. For example, self-reported health status and financial resources are associated with home modifications and the use of assistive equipment (Mathieson, Kronenfeld, Keith, 2002). On the other hand, a number of studies (e.g., Desai et al., 2001; LaPlante et al., 2004; Lima & Allen, 2001; Williams et al, 1997) report that unmet need increases with the number of ADL–IADL limitations—conditions that may be more subject to change over the short term.
Social resources
Studies of personal assistance needs (such as those cited above) typically represent social resources by marital status, household size, the relationship of household members, and access to family and friends. The presence of a spouse and/or family members may increase the likelihood of informal assistance. Frequent contact with family and friends may be associated with better quality of assistance and possibly serve as additional sources of assistance. Paid providers may supplement family members or primary assistance for those who are not married or living with family members—decreasing unmet need. A physician seen regularly is another type of social resource. We have included this because of possible influences on disease management and quality assurance relative to the level of informal care.
Social roles and norms
Social roles and norms were represented in our analysis by demographic attributes rather than direct measurement. Gender may have represented an individual's experience in various IADL tasks (e.g., shopping, meal preparation, housecleaning) and expectations about themselves or others as to the amount of assistance that might be appropriate. Age served as a marker for possible generational differences in attitudes and expectations, including perhaps some difference in preference or expectation about self-sufficiency and self-directed care. Race and ethnicity may have represented cultural attitudes, such as preferences and expectations about self-sufficiency, filial responsibility, and even tolerance for various levels of discomfort or inconvenience. Race may also have been a proxy for barriers to access to alternative living situations.
Physical features
The physical features of one's living quarters and setting and how these are associated with unmet ADL–IADL need have been studied through work focused on housing modifications (e.g., to the kitchen or bathrooms), design features (e.g., single multistory units), or the presence of special equipment (e.g., ramps, handrails, and elevators; Verbrugge & Sevak, 2002). These features are thought to improve access and self-efficacy. In cross-sectional analysis the direction of this effect may be ambiguous, as adaptations may have been made for those who need them and not all adaptations reduce unmet need for assistance.
Another element of living quarters is the potentially intrinsic meaning of "home" (Gitlin, 2003). Relocation studies have drawn attention to this in noting adjustment consequences for those who have moved. Studies of community-dwelling residents have found strong preferences to remain in one's home and the presumed buffer this familiar setting may offer for the loss of personal autonomy and control (Rubinstein, 1989). The meaning of home is seldom measured directly. It is much more common to use proxy measures (as we have done here), such as a differentiation into owner versus renter, length of time in the residence, and housing type. The assumption is that extended residence engenders familiarity with one's home and community and a stronger social network. These in turn are thought to facilitate adaptation and accommodation. Owners may be more likely to modify the physical features to better meet their needs. On the other hand, ownership may constrain relocating to a setting more adapted to one's ability.
Apart from the intrinsic meaning of home, the rights and privileges of an individual regarding the amount and duration of assistance may be affected by housing type. For example, persons living in a house may have fewer constraints (such as from zoning, fire and safety codes, or concerns from management or fellow residents) on the services that may be brought into the setting than an individual living in an apartment. On the other hand, persons living in houses may experience higher demands (including IADLs) for the maintenance of the setting. The direction of effect on unmet need is unclear, but evidence from the relocation studies cited earlier suggests that moves are sometimes reflective of adjustments to declining ability, and/or a preference to remain living independently rather than with family members.
Both ownership and housing type were used, as the effect of ownership status would be confounded by the inclusion of individuals in service-rich settings without adjusting for housing type. For example, among those with ADL limitations in the NHIS-D, 79.6% of those in houses were owners, as were 30.5% of those in houses with services. The balance in both housing types were "renters" (with or without actually paying rent), as were those in apartments. We have included measures that distinguish houses, apartments, and housing with services. These were consolidated from the housing types in the NHIS-D (Phase II, Section A, question 9). Information about services available in a facility distinguished places providing supportive care (e.g., maid service, meal preparation, medication supervision, assistance in ADLs) from apartments.
The last element of the environment considered was the availability of community resources. This was measured indirectly using community size, with the assumption that larger communities may have had more services and providers available. Such availability was expected to reduce the prevalence of unmet need. More specific measures of locality are not available in the NHIS-D.
Outcome of unmet need
The outcome of the P–E match represented in our analysis is the reported perception of unmet need for personal assistance. The NHIS-D used branch questions iterating through each of the ADLs and IADLs to derive a measure of unmet need. Those receiving or needing help with a task were asked whether they need more hands-on or supervisory assistance with that task. Following a convention used by Lima & Allen (2001), a person who said they have a limitation but reported that he or she had enough help was classified as having "no unmet need." Those reporting that they were receiving help but that they needed more help were considered as having "inadequate help" or unmet need. Those who said they needed help from another person but did not receive it from any source were classified as having "no help," and in our study (due to the small number) they were included in the unmet need category. Unmet need as a dependent variable was defined in two ways: (1) those with unmet need for only hands-on help, and (2) those who had unmet need for either hands-on or supervision or standby help.
Analysis
Analyses were conducted using SUDAAN multivariate logistic regression software (Shah, Barnwell, & Bieler, 1997). Separate models were estimated for each ADL dimension. The two variations of the outcome measure were used in separate sets of models. Results were similar with either measure, but the goodness of fit was slightly higher using only the hands-on help outcome. Only these results are reported.
All models adjust for differences by the main effect of age groups, by housing type, and by whether the study subject reported for themselves or via a proxy interview. We selected the approach of representing age groups and housing types by dummy variables (instead of having separate models for these subgroups) after testing the association of age and housing dummy variable interactions with each of the other independent variable relative to unmet need. No such interactions were statistically significant. The presence of proxy respondents raised a concern that such respondents might overstate helper availability or otherwise have differential reporting on other risk factors relative to self-reports. The effect of proxy respondents was first evaluated using sets of models that systematically included and excluded these respondents. No meaningful differences in the models' risk factor coefficients were found. A second step reestimated the model using a proxy main effect and proxy by covariate interactions. Again, no statistically significant interactions were found. All dimensions of our P–E framework are represented by the measures shown in the models. All measures were entered in a single step.
|
RESULTS |
|---|
|
TOPAbstractPerson-Environment TheoryMethodsResultsDiscussionReferences |
|---|
Unmet Assistance Need
Table 2 shows the number of persons receiving or needing more help in each ADL and IADL and a corresponding percentage of those reporting unmet need in that activity. There is a substantial difference in the number of people in these comparisons depending on the criteria used to define need. For example, 4.4 million persons receive or need hands-on assistance in at least one ADL. This estimate increases to 5.4 million if need for supervision is included.
|
Correlates of Unmet ADL Assistance Needs
Table 3 shows logistic regression results for each of five ADLs using the outcome of unmet need for hands-on assistance. The sample in each model was limited to persons with that particular ADL limitation. Similar results (available from the authors) were obtained using the broader outcome of unmet need for either hands-on or supervisory assistance.
|
Several social resources variables (i.e., living with others, having a spouse provider or a paid helper, and having a proxy respondent) were associated with a reduced likelihood of unmet need. Most of these odds ratios were statistically significant. The frequency of getting together with or talking to friends and family members outside one's household tended to be associated with reduced unmet need risk, but only one of the models (eating) was statistically significant. Seeing a physician regularly and length of time in the current residence had no significant association with unmet need nor a consistent pattern of effects.
Among the social roles and norms measures, gender generally had odds ratios approaching 1.0 and no statistically significant associations. When compared with younger adults with disabilities, those aged 65 and older showed an inconsistent pattern of unmet need and no statistically significant odds ratios. Whites compared with the other racial groups showed a lower likelihood of unmet need for all but eating. These effects were statistically significant for bathing and dressing and for any unmet need. When compared with other racial/ethnic groups, Hispanics showed an increased likelihood of unmet need across all ADLs. These effects were at similar levels in all models and were statistically significant in all but the two least prevalent conditions—eating and toileting.
Physical environment included items representing features of the living unit and its geographic location. These items were evaluated for their independent effects on unmet need, but their most germane function was to adjust for features that might otherwise be confounded with the living quarters classification of the house, apartment, or housing with services. Being in urban areas of greater than 1 million persons and having ramps or a street level entrance were both associated with a reduced likelihood of unmet need. Half of these odds ratios were statistically significant. Features such as having more than a single floor, having widened doorways or hallways, and kitchen modifications tended to show increased risk of unmet need, but only two odds ratios were statistically significant. Automatic or easy-open doors and an elevator or stair glide had odds ratio tending toward 1.0. Bathroom modifications, while generally showing no association with unmet need, had a statistically significant association with reduced risk for both eating and transferring unmet need.
Apartment dwelling, compared with those living in houses, was generally statistically significant, with apartment dwellers having an increased risk for unmet need. The robustness of housing type relative to homeownership was tested using separate models both with and without housing type. Homeownership was not significant even when housing type was omitted. The odds ratios on most of the other covariates were similar with either the inclusion or omission of the dummy variable indicative of apartments and housing with services. These relationships remained testing apartment and design feature interactions (e.g., by race and unit modifications and features) and interactions among some features (e.g., number of floors by the presence of an elevator and unit modifications by difficulty using cabinets). These interactions (not shown) were not statistically significant. Housing with services, although suggesting an association with increased risk, was rarely statistically significant, perhaps due to the small sample.
Analyses of unmet IADLs need were conducted among our sample of persons with at least one ADL limitation. These results are available from the authors. Compared with the prediction of unmet need for ADL assistance, each attribute showed similar patterns in associations with unmet need, but fewer measures in the P–E model had statistically significant associations. Living in an apartment, as compared with living in a house, showed a statistically significant increased risk for unmet IADL assistance in walking assistance (OR = 2.1) and getting outside (OR = 2.6). Odds ratios approached 1.0 for the other tasks.
|
DISCUSSION |
|---|
|
TOPAbstractPerson-Environment TheoryMethodsResultsDiscussionReferences |
|---|
The P–E framework postulates a propensity to seek settings matched to one's capabilities, but it is notable that just under 1 in 5 of those with an ADL limitation had unmet needs for that ADL. This was true across all ADLs. The consequences of unmet personal assistance needs (e.g., housing changes, relocations, family interventions to change primary helpers or to augment helper assistance) cannot be determined from cross-sectional data, but the rate of unmet need suggests that many with ADL limitations may be seeking adaptations at any given time.
Among the several dimensions of the P–E framework, financial capability represented by Medicaid eligibility (and its potential access to public programs) was associated with reduced unmet need in all ADLs, with the exception of eating. As in earlier studies, the likelihood of unmet need increased as the number of ADLs increased. Other physical capability measures (e.g., fair or poor health, incontinence) were also evidence of an expected increased likelihood of unmet need, but there were important exceptions. Persons using special equipment and those with difficulty communicating had a lower risk of unmet need. Physical capability findings taken together (consistent among individual ADLs) suggest that at a point in time long-standing disabilities may be reasonably well accommodated and that conditions subject to change over a short period are more likely to be out of P–E match. A corollary is that if special needs have not been addressed, then those with these disabilities likely have not remained in the setting.
The inclusion of social and physical environmental features adds further insight into influences on unmet need. Several social resources (i.e., living with others, having a spouse or paid providers, and having a proxy respondent), in common with findings of unmet need in the literature, were associated with lower unmet need in ADLs. Living with others, at a minimum, likely affects the timeliness and availability of assistance. This effect seems to be further enhanced by having a spouse provider or a paid provider. Counter to expectations, social resources—like frequency of talking to or getting together with family and friends (that they are not living with), years in one's residence, and seeing a physician regularly—had no significant association with unmet need. When viewed within the context of the P–E framework, this finding may reflect the situational adjustments already in place. For example, the absence of an observed effect from the social contact measures and years in one's residence may be attributable to the high prevalence of social contacts among the sample and a function of individuals having previously substituted among social resources (including living with others) to obtain acceptable levels of social interaction. Physician effects, also nonsignificant, may be explained by the high prevalence of physician use among the study population—those with at least one ADL limitation.
Age, gender, and race were used to represent social roles and norms and had varied effects. Whites, when compared with other racial groups, tended to have lower risks for unmet ADL assistance, especially tasks of bathing and dressing—ADLs that are perhaps the most easily addressed and that are the least disruptive to a household. Hispanics tended to have higher risks for unmet needs than non-Hispanics. This pattern was present for all ADLs, with the exception of eating. Whether these findings are the result of cultural preferences or individual expectations about appropriate levels of care or differential access to assistance cannot be determined from the NHIS-D. Nevertheless, one might consider whether the lower prevalence of unmet need among Whites is due to the presence of more assistance in the home versus a prior period selection and relocation to alternative living settings (e.g., housing with services or nursing homes) where such assistance might be available. Conversely, higher unmet need prevalence among Hispanics may be a reflection of either a willingness on the part of caregivers (or the respondent) to accept higher levels of frailty among household members or an inability (including possible racial discrimination) to find suitable alternatives.
Age and gender did not show systematic associations with unmet ADL needs. In the context of the P–E model, this suggests that there are few gender or age group differences in preferences or expectations about appropriate ADL assistance. Such a finding likely requires further investigation with direct measurement of preferences and expectations.
Physical environmental features were incorporated into the analysis to assess the extent to which they modulated and/or complemented the individual attributes and social resource affects commonly used in analysis of unmet ADL needs. Particular attention was given to possible differences between those living in houses versus those living in apartments. Those living in apartments were about 50% more likely to report unmet ADL needs than those living in houses. The analysis controlled for specific design features (number of floors, room and door modifications, presence of lifts and elevators) to partial out any effect of these features and not spuriously attribute them to housing type. Additional analyses (not shown) tested interactions of housing type with the individual capability measures and social resources. These tested whether there were any capability and resource differences associated with living quarters. These interactions were not significant.
The finding of higher unmet need among those living in apartments is consistent with the notion that apartment residents may have more restrictions on services or unit modifications than those residing in houses, but at least three other explanations are suggested by theory. One is that the familiarity and even meaning of "home" may be stronger among those living in houses than apartments. Some would suggest that such a connection may enhance the ability to accommodate to one's level of needs. Another interpretation is that selection factors may be differentiating those living in houses from those living in apartments, with apartment residents perhaps being more willing to "accept" certain levels of unmet need as a trade-off for living independently or because the alternative may be a transition into a higher (and undesired) level of care. A corollary to the selection hypothesis is that those living in houses have a lower threshold of unmet needs adaptation (perhaps due to the higher daily living demands of a house, things such as housekeeping and maintenance, and poorer proximity to shopping). A lower threshold would suggest that adaptations, such as obtaining higher assistance levels or relocation to alternative settings, may occur either more quickly or in anticipation of assistance needs.
While there are theory-based explanations for the finding of differences between apartment and house dwellers, there is still the risk that this difference may be attributable to unmeasured attributes. For example, even though we adjusted for income, the presence of paid providers, and household size in our models, there may be differentials we did not measure. One of these might be that those in houses have more disposable income to hire more paid help hours. This might be particularly true among homeowners with paid-off mortgages. Houses, too, might provide more room for family and other helpers to stay as needed, including short-term stays.
Finally, while the P–E framework suggests several plausible explanations for the risk of and/or reduction in unmet personal assistance needs, it is important to recognize that this study has not directly tested the adaptations that are attempted and/or made, nor examined the consequences of continued unmet need. Panel studies are needed to determine whether unmet need is protracted and whether those without support relocate out of the community. In this vein, we call particular attention to those with special equipment needs, cognitive impairment, or social isolation. The observed low prevalence of unmet need may belie avoidable circumstances that in a prior period may have "selected" against those with these limitations.
Similarly, physical features, including many unit modifications, were found to have little relationship to unmet need in the context of cross-sectional data. These findings should not be interpreted as evidence that design features have no value for those with limitations. The main function of these items in this study was to help achieve unbiased estimators on housing type and to minimize a spurious attribution of a housing type effect when the underlying factor might be a design feature. More refined and time-specific comparisons are required for an evaluation of efficacy. Our analysis measured the presence of a feature, but not whether it was appropriate or sufficient for the limitations of the individual in that setting at that time. An illustration of the potentially complex relationships is that of kitchen modifications and widened doorways and hallways. These tended to be associated with increased, rather than diminished, unmet ADL need. Such a counterintuitive relationship seems more plausibly explained by these modifications having occurred in a prior period than in causing the unmet need.
In addition, more work is needed for better understanding of the multiple factors that may be operating to influence the observed race and ethnicity differences. We suggest possible selection effects associated with preferences, expectations, structural barriers, and discrimination, but a full investigation of these issues requires direct measurement of these phenomena.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication March 3, 2003. Accepted for publication December 17, 2004.
|
References |
|---|
|
TOPAbstractPerson-Environment TheoryMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H.-W. Wahl, A. Fange, F. Oswald, L. N. Gitlin, and S. Iwarsson The Home Environment and Disability-Related Outcomes in Aging Individuals: What Is the Empirical Evidence? Gerontologist, June 1, 2009; 49(3): 355 - 367. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
