| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Polisher Research Institute, Abramson Center for Jewish Life (formerly Philadelphia Geriatric Center), Horsham, Pennsylvania.
2 Center for Policy Research, Syracuse University, New York.
3 Social Science Computing, University of Pennsylvania, Philadelphia.
Address correspondence to Vicki A. Freedman, PhD, Director, Polisher Research Institute, Abramson Center for Jewish Life (formerly Philadelphia Geriatric Center), 1425 Horsham Road, North Wales, PA 19454. E-mail: vfreedman{at}abramsoncenter.org
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. Using a national sample of older unmarried Americans, we examine changes in total hours, paid hours, and unpaid hours of care in response to declines and improvements in personal care (activities of daily living, or ADLs) and routine care (instrumental ADLs, or IADLs) disability. We model changes in the total hours of care received in the past month and jointly model changes in unpaid and paid hours, using Tobit models.
Results. Changes in the total hours of care received respond to both increases and decreases in the count of ADL limitations and appear close to symmetric. In contrast, responses to IADL disability dynamics appear to be far less symmetric: Although increases in the count of IADL limitations are met with substantial increases in the total hours of care, decreases are not met with correspondingly large declines in care. The same general pattern is found for unpaid and paid care, and for Medicaid-funded home care.
Discussion. Disability and care are not static constructs in old age. Older unmarried persons experience worsening, stabilizing, and recovery of function, and their care hours change accordingly. Evaluations of home care programs must be cognizant of such dynamic realities.
THE cost of providing home-based long-term-care services to older Americans with disabilities reached $37 billion in 2000, over 65% of which was paid by government programs (Congressional Budget Office, 1999
). There are also substantial uncompensated hidden costs associated with the many hours of "informal" care provided primarily by family members of persons with chronic disabilities (Ettner, 1996). Although married older persons with disabilities most often rely on a spouse to provide care (Spillman & Pezzin, 2000), those without a spouse are subject to a more fragmented—and potentially less stable—array of care choices.
Conservative projections suggest the number of older persons with disabilities could exceed 12 million by the year 2030 (Congressional Budget Office, 1999). Slower growth of potential family caregivers is also projected, in part because of fertility declines and increasing demands from work (Wolf, 2001). Thus, despite studies suggesting recent improvements in old-age functioning (Freedman, Martin, & Schoeni, 2002), the anticipated growth in the sheer number of older Americans with disabilities who require formal home care services remains a policy concern.
In light of these demographic trends, policy makers have shown interest in restraining the future growth in public spending for home-based care while supporting America's families. At the federal level, the Balanced Budget Act of 1997, for example, shifted the Medicare home health benefit from a cost-based to prospective payment system, thereby providing incentives to contain the length of time older Americans with disabilities receive home care services for medical conditions. Other recent federal initiatives have aimed to expand family-provided support through mechanisms such as tax deductions and respite programs (U.S. Department of Health and Human Services, 2001). States are also concerned about growth in long-term-care expenditures under the Medicaid program, and many have adopted cost-containment strategies to curb home care costs (Wiener, 1997).
Related to these policy initiatives are discussions of appropriate targeting of home care services to those in need (Kane, 1999). Evaluations of targeting often rely on cross-sectional evidence comparing the characteristics of those receiving home care services with those who do not. Such exercises ignore the inherently dynamic underpinnings of both the disablement process and care decisions. Here we provide additional insight into discussions of home care service allocation by considering the care process from a more dynamic perspective and by considering both paid and unpaid sources of care simultaneously. In this context, effective targeting implies not only getting people with few informal resources the services they need when they acquire a disability, but also increasing such services for those whose disability worsens and reducing such services for those who improve.
The framework we develop here is unique in that it allows us to explore whether responses to disability declines and improvements are symmetric; that is, we test whether reductions in response to disability improvements are of similar magnitude to increases in response to disability declines. In doing so we distinguish between paid and unpaid hours and between responses to changes in personal care (e.g., difficulty bathing, dressing, or eating) and routine care (e.g., difficulty managing money, shopping, or using the telephone). We also identify circumstances under which services remain in place despite improvements in functioning. Our analysis focuses on a subgroup of older persons most likely to experience shifts in care: those who are unmarried.
Background
Studies of paid and unpaid care for elderly persons cut across several disciplines. Policy makers have been largely concerned with recent increases in the utilization of paid care and in minimizing substitution of formal for unpaid family-provided care. To assist the medical community in targeting high-risk patients, epidemiologists and clinical researchers have identified risk factors associated with paid care use. Demographers have focused on the role of kin availability in constraining the choice of living and care arrangements. Social gerontologists have investigated family caregiving and the burdens associated with providing care to frail older relatives. Finally, economists have focused on the trade-offs between caregiving and work and on the role of market factors in the demand for care.
Despite the widespread interest in personal care arrangements, the literature is limited in two respects. First, most studies focus on care arrangements at a point in time and thus provide limited insight into the dynamic process driving changes in care arrangements for an individual. Second, most studies focus on the structure of care—for example, whether an older person with a disability uses community-based or institutional care, or whether the care is provided by an unpaid provider, paid provider, or some combination thereof—rather than the volume of care provided under a given the arrangement. However, the volume of service use, both in total and by type of provider, is probably the most crucial factor in calculating the costs of providing care to older persons with disabilities.
Although the literature on disability trajectories is growing (for a review see Wolinsky, Stump, Callahan, & Johnson, 1996), there are only a few studies using measures of changes in disability to predict care use (Langa, Chernew, Kabeto, & Katz, 2001; Miller & McFall, 1991; Miller, McFall, & Campbell, 1994; Stoller & Pugliesi, 1991; Worobey & Angel, 1990; Wolinsky, Callahan, Fitzgerald, & Johnson, 1993). These studies show that increases in limitations in activities of daily living (ADLs) and in instrumental ADLs (IADLs) result in a greater tendency to live with others and to use paid care or a "mixed" (paid and unpaid) care network. Studies of changes in community-based care have also been few in number. A study by Miller and colleagues (1994), using a national sample of elderly persons with chronic disability, suggests that care transitions could be quite frequent even over short periods of time. The authors found that, over a 2-year period, sources of care changed for at least 20% of persons initially receiving unpaid care and for over 50% of persons initially receiving both unpaid and paid care. Similarly, using a sample of 476 caregivers in Wisconsin, Seltzer and Li (2000) found that 36% of women with an elderly husband or father experienced a transition to or from a caregiving role over a 3-year period.
We know of only one past study that has included disability changes to predict changes in home care hours. Using the first and second waves of the Asset and Health Dynamics of the Oldest Old (AHEAD) study, Langa and colleagues (2001) found large increases in weekly hours of paid home care use between 1993 and 1995, and that changes in ADLs and IADLs were positively related to these changes. However, the authors ignored potentially important changes between the first and second waves of AHEAD in the survey questions used to obtain hours of care. Moreover, changes in ADLs and IADLs were included in their model as continuous variables ranging from -6 to +6 and from -5 to +5, respectively. This specification assumed symmetric responses; that is, increases in care hours in response to one additional ADL limitation were assumed to be equivalent to decreases in care hours in response to one fewer limitation. In addition, rather than considering in a single framework both unpaid and paid hours of care, the authors modeled only the receipt of paid home care and included patterns of unpaid home care as predictors of paid care, an approach that ignores the possibility that the two may influence one another.
Many behavioral phenomena exhibit asymmetric responses to opposing external forces. For example, in tight labor markets wages tend to rise, yet during recessions they may be "sticky" downward, failing to fall. Similarly, individual decision making seems to exhibit asymmetric reactions to a given level of risk attached to potential gains compared with potential losses. There are several reasons to anticipate that the hours of care used by older persons with disabilities will, as well, respond differently to comparably large declines and improvements in functional levels, other things being equal. This is particularly likely for unpaid IADL services provided by family members, which have substantial social and affective components. Time-use patterns connected to shared meals, housekeeping, or provision of transportation that are established in times of care needs may be maintained or only partly reduced if the care recipient recovers the capacity for self-care. If an older parent moves in with a child in response to ADL needs, then the coresidence may be maintained despite later recovery of function. Coresidence, in turn, fosters the provision of IADL assistance even if that assistance is not strictly necessary. Some studies have also documented some degree of unpaid ADL assistance received by older people who, according to their self-reported accounts, do not "need" such help (Bootsma-van der Wiel et al., 2001).
Regarding paid care, especially to deal with ADL needs, we would expect third-party payers for such services to enforce declines in service levels in response to functional improvement and, accordingly, reduced levels of assessed need. However, the monitoring of changing care needs may be imperfect, occurring at large or irregularly spaced time intervals. Discretion on the part of case managers and home health aides can also lead to situations in which clients' functioning improves but services are not reduced (Jenkins, 1997). Thus, both unpaid and paid care services may respond asymmetrically to changes in the severity of disability. The responses exhibited by each type of care may differ from each other as well. We describe our empirical approach to these issues in the paragraphs that follow.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
. By the third (1998) wave of the study, the AHEAD study was integrated with its parent project, the Health and Retirement Study (HRS). The AHEAD study is uniquely suited to examine questions of how care arrangements are put into place over time. At each wave (hereafter, AHEAD93, AHEAD95, and HRS98), the AHEAD collects data on care arrangements, health status and transitions, kin networks, familial resources, and demographic characteristics. Although question changes preclude comparisons of hours of care and disability for 1993 and 1995, the AHEAD95 and HRS98 waves obtained information on disability and hours of care provided by nonspousal helpers in an identical fashion.
We restrict our analysis to those who were alive and unmarried at all three waves and who reported difficulty in 1995 with one or more out of six ADLs or with one or more out of five IADLs. We exclude those living in an institution in 1995 or 1998 in order to focus our analysis on home- and community-based care. Our final sample size consists of 746 unmarried community-based older Americans who report difficulty with one or more ADL or IADL activities. Sample characteristics are provided in Table 1. To compensate for differential probabilities of selection, we base estimates on weighted data.
|
). In effect, we view changes in care needs, whether improvements or declines and whether for ADL or IADL services, as "treatments" and the associated changes in care hours as "responses" to them. In order to isolate the effect of disability dynamics on changes in the volume of care, we model changes in the hours of care between 1995 and 1998 as a function of initial disability level in 1995 and the change in disability between 1995 and 1998.
We first address hypotheses about the effect of disability transitions on the total volume (measured in hours) received by older persons with disabilities. Because we measured hours of care in a continuous fashion by using hours of care per month, the change in monthly hours from t - 1 (1995) to t (1998)—denoted 
Vit-1,t—was also a continuous variable, which could take on both positive and negative values. In the population, the distribution of changes in care hours is, in effect, unbounded. At the individual level, however, two constraints on estimation must be noted. For those receiving no care at t - 1 (i.e., for those with Vit-1 = 0), Vit-1,t must be greater than or equal to 0. For individuals with Vit-1 > 0, declines cannot exceed Vit-1 (i.e., Vit-1,t 
-Vit-1). Such constraints manifest themselves in the data as heaping at -Vit-1, a problem also common to the analysis of cross-sectional measures of hours. This problem has been dealt with in the unpaid care literature (see, e.g., Wolf, Freedman, & Soldo, 1997) by use of a Tobit specification.
The two constraints described reduce to a single constraint: Changes in care hours are censored at -Vt-1. We therefore used a Tobit model of changes in hours with individual-specific censoring at -Vit-1. Because in the population care-hours responses are unbounded, as explained herein, the Tobit index function produces coefficients that allow for a test of symmetry. Symmetry exists if the effect of declines in disability on the total hours of care is the opposite of the effect of improvements in disability. That is, symmetry exists if the coefficient on a variable representing a given amount of functional improvement equals the negative of the coefficient on another variable representing the same amount of functional decline. To test for symmetry in unpaid and paid care hours, we extended the analysis by jointly modeling changes in the number of formal and informal hours by using a bivariate Tobit approach (Couch, Daly, & Wolf, 1999).
Outcomes
The outcomes in this analysis—change in total hours of care, and changes in the number of paid and unpaid hours of care—were calculated from information about the hours of care received in the month preceding the interview in 1995 and in 1998. Respondents were asked to provide information about caregivers by using three sets of questions, which were identical in both years. The first set of questions asked about any difficulty with ADLs (walking, bathing, dressing, eating, transferring, or toileting), the second set asked about any difficulty with IADLs (preparing meals, taking medicine, using the telephone, or shopping), and the third set (considered an IADL for our purposes) focused on difficulty with a single task: money management.
The same general approach was taken for each of the three sets of activities. For ADL activities, for example, respondents were first asked whether they had any difficulty, and, if so, whether someone helps, with each of the specific ADL items. After completing all six ADL items, respondents reporting that someone helps them with one or more activities were then asked to identify caregivers who helped in the past month with those activities for which they reported receiving help. This process was repeated for IADL and money management activities. After caregivers were identified for the three sets of activities, respondents were asked, regarding each helper (but excluding spouses and employees of nursing homes), how many days per week each caregiver provided help in the past month, how many hours per day that caregiver usually helped, the gender of the caregiver, and whether the helper was paid.
To remain consistent with the monthly reference period in the questions about hours of care, we calculated each caregiver's hours of care in the past month by multiplying for each helper the reported number of hours per day by the reported number of days per week, times 52 and then divided by 12. Summing the hours for each caregiver, we calculated the total hours of care received in the past month. We also summed hours provided by unpaid caregivers to yield unpaid hours in the past month and summed hours provided by paid caregivers to yield paid hours in the past month.
In a small number of cases, the hours of care were incomplete or missing entirely for a respondent because of missing information on one or more caregivers. For 42 respondents in 1995 and 31 respondents in 1998, we imputed information on hours per day or hours of care in the past month for one or more helpers. We assigned to caregivers with missing information the average number of hours of care provided by caregivers of the same type (male nonagency, female nonagency, or agency based) in networks with the same number of helpers (1, 2, or 3+). For 13 respondents in 1995 and 3 respondents in 1998, we also imputed probabilistically whether the caregiver was paid, based on the type of helper and the size of the network.
Predictors
The main predictors of interest are changes in the disability level of respondents. For both ADLs and IADLs, we included categorical indicators of the change in the count of limitations between 1995 and 1998 (e.g., increases of 1 or of 2 or more and decreases of 1 or of 2 or more compared with an omitted category of no change). As a further control for one's place in the disability continuum, we also included continuous variables indicating the number of each type of activity limitation in 1995.
Hours of care can respond to many factors besides changes in disability: Unpaid caregivers can experience events in other life domains that impinge on their availability; programmatic eligibility for paid services can be achieved or exhausted; private insurance stop-loss thresholds can be crossed; and so on. Accordingly, in all models we included a number of variables thought to reflect the preferences of older individuals with disabilities and their families and shown to be related to care hours in past studies. These covariates include age (Kemper, 1992; Miner, 1995), race and Hispanic origin (Marks, 1996; Tennstedt & Chang, 1998), and education (Kemper, 1992). Past studies have shown that hours received of both informal and formal care are influenced by the number of children that a person has, with different effects for sons and daughters (Picone & Wilson, 1999; Wolf et al., 1997). Accordingly, the potential supply of informal care hours was represented by a series of variables indicating the number of sons and daughters that respondents had. We also explored the effects of the number of sons-in-law and daughters-in-law but found no substantial effect. Although they were available in the AHEAD study, we omitted measures of coresidence with a child and distance to nearest child in our models, because decisions about living arrangements are likely to be made jointly with care arrangements (Pezzin, Kemper, & Reschovsky, 1996; Pezzin & Schone, 1999). Financial resources of the respondent were classified into low, medium, and high on the basis of 1993 income and liquid asset quintiles for the entire AHEAD sample.
Finally, the implications of asymmetric responses for policy makers will vary depending on the source of care. If unpaid caregivers demonstrate "sticky" behavior to improvements in health, policy makers may be less apt to intervene than if Medicaid providers do so. We investigated the source of payment for caregivers providing such services in two ways. Although sample sizes were too small to estimate complex models, we descriptively examine the mean changes in hours of care by source of payment for a caregiver (completely out of pocket vs. insured in part or in full) and mean changes in unpaid and paid hours for those reporting Medicaid coverage.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
A similar pattern was observed with respect to IADL tasks. In response to declines of two or more in the count of IADL limitations, older unmarried persons decreased their care hours by only 19 hr a month from approximately 143 hr in 1995 to 123 hr in 1998. The decline in care was the result of a trade-off between unpaid and paid care hours, with unpaid care hours declining by nearly one third and paid care hours increasing by 16 hr per month. Increases of two or more in the count of IADL limitations were met with substantial increases in total hours, tripling from 74 hr in 1995 to nearly 220 hr in 1998. The bulk of the additional hours came from unpaid sources (86 additional unpaid vs. 58 additional paid hours).
We also find asymmetric care dynamics when we restrict the analysis to those reporting Medicaid coverage. As shown in Table 4, although sample sizes are small for some cells, hours provided by paid caregivers increase dramatically in response to increases of two or more in the count of IADL limitations (+168 hr) but do not change appreciably in response to declines in the count of IADL limitations.
|
D = -ß-D indicate asymmetric responses. (For other model coefficients, see the first column of Table 6.)
|
|
In contrast, responses to IADL disability dynamics appear to be far less symmetric. Although increases in the count of IADL limitations lead to substantial increases in the total hours of care (e.g., ß = 261 for "+2 or more"), comparable decreases in the count are not followed by correspondingly large declines in hours (e.g., ß = -67 for "-2 or more"). In other words, responses to declines in IADL functioning are nearly 3 times the volume of responses to improvements (p <.05 for test of symmetry).
Are unpaid and paid care responses symmetric?
—In the last two columns of Table 5, we present coefficients on changes in ADL and IADL limitations from the bivariate Tobit models, in which changes in hours of unpaid and paid care are jointly estimated. Remaining coefficients are presented in the last two columns of Table 6. We find that unpaid care hours demonstrate a similar pattern to total care hours. That is, unpaid care hours appear to respond more to declines than to increases in the count of ADL limitations, but we cannot reject a formal test of symmetry even at p <.10. However, unpaid hours have an asymmetric response to a change of one in the count of IADLs. Paid hours display the same general pattern of symmetry with respect to ADL disability dynamics, but responses to IADL dynamics appear asymmetric. In particular, paid care hours increase with an increase of two or more in the count of IADL limitations but do not decline in response to declines of two or more in the count. Because of the small numbers of cases experiencing declines of two or more in the count of IADL limitations, statistical tests fail to reject the null hypothesis of symmetry.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
, Langa and colleagues (2001), and others, that both paid and unpaid care hours received by older unmarried Americans with disabilities increased during the 1990s. Unlike previous analyses, however, we find that for this population, care hours respond in different ways to shifts in ADL and IADL disability. Responses to ADL dynamics are relatively symmetric—increasing as disability worsens and decreasing as recovery ensues. In contrast, responses to IADL needs appear to be asymmetric. Unpaid care hours respond to both increases and decrease in the number of IADL limitations, but more so to the former than the latter. Paid care hours increase in response to declines in IADL functioning but do not decrease in response to IADL improvements. That is, once a paid arrangement has been put into place to meet the need for preparing meals, taking medicine, using the telephone, shopping, or managing money, the care remains in place despite improvements in ability.
Our study is limited in three important respects. First, the measure of disability dynamics we adopted has some inherent disadvantages. Perhaps most important, the measure does not capture improvements or declines in difficulty with specific activities, thus raising the possibility that asymmetric responses simply reflect declines and improvements in fundamentally different areas of functioning. We investigated whether declines in IADL functioning were more likely than improvements in IADL functioning to involve time-consuming activities (e.g., shopping) and found that this is not the case. We also found few statistically significant relationships between changes in difficulty with specific activities and in hours of care received. We therefore represented disability trajectories with changes in the number of activity limitations. Our measure of disability also does not capture changes in the degree of difficulty with a given activity over time. Indeed, answers to questions about difficulty may in part reflect getting help, because older unmarried persons were not instructed whether to consider help or equipment when answering questions about having difficulty with a given activity. At least one study suggests, however, that the questions used to elicit information about personal care are a reasonable indication of underlying difficulty without help or special equipment, if used (Freedman, 2000).
Second, because of data limitations, our analysis could not address how married older persons balance paid and unpaid care over time. Other studies suggest married older persons are less likely to use paid home care (Katz, Kabeto, & Langa, 2000; Langa et al., 2001), and that the spouse is first in line to serve as primary caregiver (Spillman & Pezzin, 2000) and seldom withdraws from this role (Jette, Tennstedt, & Branch, 1992). Certainly, a separate analysis of this group is warranted should the appropriate data become available. Until that time, it remains unclear how married older persons handle declines in health and what, if any, the repercussions are for joint care decisions.
Third, we were unable to control for the changing policy and price regimes for home care that occurred during this period. The paid home care hours recorded in our data reflect services provided by a variety of programs, including Medicaid-funded home health and personal care services, personal care covered through Medicare's home health benefit, and services reimbursed by private insurance and those paid out of pocket. These different programs are subject to very different supply, reimbursement, and regulatory provisions that may have changed during the 1995–1998 period. Expenditures for the Medicare home health benefit, for example, which represent an unknown proportion of the paid services in our data, increased by the order of 30% per year from 1995 through 1997 but then experienced contractions of approximately the same magnitude in 1998 (Government Accounting Office, 2000). The contractions have been attributed in part to the implementation of the Balanced Budget Amendment of 1997, which began the shift in Medicare's home health benefit from a cost-basis benefit to a prospective payment system. Given that our data span from 1995 to 1998, only 1 year after passage of the amendment and hence only months into implementation of the transitional interim payment scheme, our estimates are likely to reflect more of the former regime than the current prospective payment system. Certainly, a further evaluation of the consequences of implementation of a Medicare home health prospective payment for the hours and balance of care of older Americans with disabilities deserves attention.
The asymmetries in home care dynamics revealed by our analysis have both research and programmatic implications. Cross-sectional multivariate analyses of variations in the intensity of home care generally exhibit relatively low explanatory power. Our results suggest that part of the unexplained variation in home care usage results not from omitted current-period variables but from omitted prior-period variables, in particular variables describing the past trajectory of disability and of responses to it. In addition, to the extent that current-period variations in service use reflect prior-period variations in disability severity and service use, issues related to equity in the allocation of services are raised.
Our results neither address the question of appropriateness of service levels, either before or after a change in disability severity, nor suggest that individuals are receiving more or fewer of the formal services for which they might be eligible through publicly funded programs. However, asymmetric responses to changes in disability severity imply the existence of cross-sectional differences in service use that are unrelated to current circumstances. Our results suggest the possibility that some older persons with disabilities may be receiving different levels of publicly and privately funded home care services, particularly for routine care tasks such as shopping and housekeeping, for reasons unrelated to their current needs; this is a suggestion worth addressing in future research.
Our findings of asymmetry in changes in paid service use might be linked to programmatic intervals for assessing clients' functioning. During the period covered by the data analyzed herein, Medicare home health practices (which cover personal care only if a beneficiary also is homebound and requires skilled nursing on an intermittent basis), Medicaid regulations, and private insurance rules varied widely. According to a study of 72 Medicaid home care programs (Congressional Research Service, 1993), for example, 40% of existing programs required reassessments on an annual basis and another 50% once every 6 months, and the remaining 10% less frequently. In contrast, Medicare's home health prospective payment system requires, beginning in October 2000, a minimum of 60 days between evaluations. The dynamic framework adopted in this study could be used to explore optimal review intervals for public and private programs alike.
And what of the unpaid caregiver, most often the daughter, who continues to provide care with shopping, cooking, and cleaning after her parent's difficulties have eased? Although horizontal equity is not of programmatic concern here, why such behavior is occurring is an interesting question for future research. We can suggest several possibilities for further exploration. Perhaps permanent changes to work schedules made in anticipation of a long parental illness may free up a caregiver's time to continue routine care activities. Perhaps caregivers enjoy the nonmonetary rewards of participating in intergenerational transfers, or perhaps they are helping in anticipation of future monetary rewards or in response to past transfers (Henretta, Hill, Soldo, & Wolf, 1997; Walker, Martin, & Jones, 1992). It may be that continued coresidence of caregiver and care recipient fosters this helping behavior. In our sample, coresidence occurred in roughly 30% of cases; others have noted the complex interrelationships of care and living arrangement decisions (Pezzin & Schone, 1999; Pezzin et al., 1996). Finally, public incentives (e.g., tax credits or deductions for family caregivers) may also play some role (Gerald, 1993).
Finally, our analysis underscores the point that disability and care are not static constructs in old age. Older persons who experience disability worsen, stabilize, and recover, and their care hours change accordingly. Evaluations of home care programs must be cognizant of such dynamic realities. Moreover, although equitable service allocations are difficult to achieve and maintain in the presence of substantial heterogeneity in disability trajectories, home care programs and policies have to be built around the premise that old-age disability is a dynamic construct if policy makers are to take aim at such a goal.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication July 19, 2002. Accepted for publication July 3, 2003.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K.-L. Chou and J. C. B. Leung Disability Trends in Hong Kong Community-Dwelling Chinese Older Adults: 1996, 2000, and 2004 J Aging Health, June 1, 2008; 20(4): 385 - 404. [Abstract] [PDF]
|
|
|
|
 |
|
 L. W. Li Longitudinal Changes in the Amount of Informal Care Among Publicly Paid Home Care Recipients Gerontologist, August 1, 2005; 45(4): 465 - 473. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
