| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Division of Geriatrics, San Francisco VA Medical Center and the University of California, San Francisco
b Departments of Psychiatry and Neurology, San Francisco VA Medical Center and the University of California, San Francisco
c On Lok Senior Health Services, San Francisco, California
d Division of Nephrology, MetroHealth Medical Center, Case Western Reserve University, Cleveland, Ohio
e Division of Geriatrics, University of South Carolina School of Medicine, Columbia
f Palmetto Health Alliance, Columbia, South Carolina
Kenneth E. Covinsky, San Francisco VA Medical Center (111G), 4150 Clement Street, Bldg. 1, San Francisco, CA 94121 E-mail: covinsky{at}medicine.ucsf.edu.
Decision Editor: John E. Morley, MB, BCh
 |
Abstract |
|---|
 Top Abstract MethodsResultsDiscussionReferences |
|---|
Methods. This was a cross-sectional study of 2806 patients (mean age 78, 73% women, 29% African American, 12% Hispanic, 54% with dementia) with at least one potentially working caregiver (defined as one who is either currently employed or who would have been employed if they had not been providing care) and their 4592 potentially working caregivers. Patients were enrollees at 11 sites of the Program of All-Inclusive Care for the Elderly (PACE). Social workers interviewed patients and caregivers at the time of PACE enrollment. Caregivers were asked if they had reduced the hours they worked or had stopped working to care for the patient. Nurses interviewed patients and caregivers to assess independence in activities of daily living (ADLs) and the presence of behavioral disturbances. Comorbid conditions were assessed by physicians during enrollment examinations.
Results. A total of 604 (22%) of the 2806 patients had at least one caregiver who either reduced the number of hours they worked or quit working to care for the patient. Patient characteristics independently associated with a caregiver reducing hours or quitting work were ethnicity 
, 95% confidence interval [CI] 1.14–1.78 for African American; 
, 95% CI 1.43–2.52 for Hispanic), ADL function below the median (
, 95% CI 1.44–2.15), a diagnosis of dementia (
, 95% 
–2.17 if associated with a behavioral disturbance; 
, 95% CI 1.06–1.63 if not associated with a behavioral disturbance), or a history of stroke (OR = 1.42, 95% CI 1.16–1.73). After controlling for these patient characteristics, caregiver characteristics associated with reducing work hours included being the daughter or daughter-in-law of the patient (OR = 1.69, 95% CI 1.37–2.08) and living with the patient (OR = 4.66, 95% CI 3.65–5.95 if no other caregiver lived at home, OR = 2.53, 95% CI 2.03–3.14 if another caregiver lived at home).
Conclusions. Many caregivers reduce the number of hours they work to care for frail elderly relatives. The burden of reduced employment is more likely to be incurred by the families of ethnic minorities and of patients with specific clinical characteristics. Daughters and caregivers who live with the patient are more likely to reduce work hours than other caregivers. Future research should examine the impact of lost caregiver employment on patients' families and the ways in which the societal responsibility of caring for frail elders can be equitably shared.
OVER half of older patients who meet eligibility criteria for nursing home care live at home (1)(2). The ability of these patients to remain at home is largely due to the unpaid care provided by caregivers (3). Arno and Levine estimated that the economic value of the informal care provided by family caregivers at 196 billion dollars in 1997 (4). This amount dwarfs national spending for formal home health care and nursing home care, and, if counted as part of national health care spending, would increase estimates of total spending by 20%. However, the cost or value of care provided by family members is often not considered in discussions of health care costs (5). Without such care, many frail patients who live at home would require nursing home care and the public costs of long-term care would increase.
The financial consequences to families of providing informal care are substantial (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). An important cause of these financial consequences is lost or reduced employment of caregivers (6)(8). As many caregivers report spending over 40 hours per week providing care, it is not surprising that caregiving frequently interferes with employment (4)(11)(13). Evidence that caregiving interferes with employment comes from studies of both terminally and chronically ill patients (6)(8). While reductions in employment because of caregiving are seldom considered in analyses of health care costs, they have significant implications for patients' families. Further, to the extent that decisions to reduce employment in order to provide care prevent or delay the public expense of nursing home care, such decisions may result in significant public cost savings. However, little is known about the characteristics of patients and families that are associated with reductions in caregiver employment. Developing a better understanding of these characteristics can provide insight into how this cost of caring for frail elders is distributed across society and may be helpful in targeting interventions to caregivers at risk of reduced employment.
Therefore, we analyzed reports of employment loss due to caregiving in a diverse multisite sample of community-living frail elders. Patients and caregivers were assessed at the time patients were enrolled in a community-based long-term care program. Our goals were to determine the demographic and clinical characteristics of patients and the characteristics of caregivers associated with a reduction in caregiver employment.
|
Methods |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
DataPACE, the datasource for these analyses, includes functional, clinical, and social data on patients that enrolled at the PACE demonstration sites. Sources of data include patients, caregivers, nurses, social workers, and physicians. Principles underlying PACE data collection include a consistent set of variables collected by all sites, consistent guidelines for recording data across sites, and centralized training and quality assurance procedures (17). Detailed procedure manuals outlined specific definitions and protocols for each data element, and data collection staff were trained to a standard of reliability (17). Staff from the coordinating center visited each PACE site on a yearly basis to monitor data collection and perform additional reliability checks.
Patients were eligible for this analysis if they had at least one potentially working caregiver at the time of enrollment. We defined a potentially working caregiver as a caregiver who either was currently working or would have been working if they were not providing care. During the eligible time window, 5062 patients enrolled in the 11 PACE sites. We excluded 1628 patients from the analytic sample because they did not have any caregiver (n = 633) or because all their caregivers reported they were not working and would not be working even if they were not caregivers (n = 995), resulting in a potential analytic sample of 3434. We excluded an additional 628 (18%) patients because of missing data about their caregivers, resulting in a final analytic sample of 2806 patients. This research was approved by the human subjects committee of the University of California, San Francisco.
Measures
Our primary dependent variable was whether or not a caregiver reduced the number of hours they worked or quit working in order to provide care for the patient. At the time the patient enrolled in PACE, social workers obtained information about household and nonhousehold caregivers providing unpaid assistance to the patient. Social workers generally interviewed both patients and caregivers simultaneously and combined both sources of information. Information about up to 3 household and 3 nonhousehold caregivers was obtained. If the caregiver was employed at the time of the patient's enrollment, the patient or caregiver was asked if the caregiver had to reduce the number of hours worked to care for the patient. If the caregiver was not currently employed, the caregiver or patient was asked if the caregiver quit a job to care for the patient. Caregivers who either reduced their hours or quit a job to care for a patient were classified as having reduced their work hours.
Demographic variables were recorded by the PACE intake worker at the time of enrollment. Data about dependence in five activities of daily living (ADL) and walking were obtained from the admission nursing evaluation (18). For each of these activities, patients were rated as independent (able to do without assistance of another person), partially dependent (needs help for part of the activity), or totally dependent (needs help for the entire activity). We created a score for ADL function by giving 2 points for each independent ADL and 1 point for each partially dependent ADL and categorized patients as having ADL function either better or worse than the median. Nurses also determined whether or not patients were continent of urine and stool and determined whether or not the patient had one of four problem behaviors (wandering, verbal disruption, aggressive behavior, and regressive behavior). Nurses or social workers measured cognitive function with the Short Portable Mental Status Examination (SPMSQ; (19)). The presence or absence of comorbid medical conditions was assessed by physicians on the basis of patients' enrollment history and examination. For dementia, we categorized patients as either not having dementia, having dementia without a behavioral disturbance, or having dementia with a behavioral disturbance. At the time of the enrollment evaluation, social workers also determined whether or not the patient used paid help for either personal care or home chore services or was enrolled in an adult day health program.
We categorized each working caregiver in terms of whether or not they lived with the patient and whether or not there was an additional caregiver living with the patient. We created three mutually exclusive categories: (i) working caregiver lives at the patient's home with no additional caregivers at the patient's home; (ii) working caregiver lives at the patient's home, with at least one additional caregiver at the patient's home (This additional caregiver could be either working or not working. Analyses that differentiated whether or not the additional caregiver was working did not affect the results.); or (iii) working caregiver does not live at the patient's home. The caregiver's relationship to the patient was assigned to one of four categories: (i) son or son-in-law; (ii) daughter or daughter-in-law; (iii) other male relative or friend; or (iv) other female relative or friend. We included spouses with other relatives because very few spouses stated that they were either currently working or that they would have been working if they were not providing care.
Analyses
We did bivariate and multivariate analyses, each in two stages. For the first stage of each analysis, the patient was the unit of analysis. The goal of these analyses was to determine, among patients with one or more potentially working caregivers, the patient characteristics associated with at least one caregiver reducing the number of hours they worked in order to care for the patient. We used chi-square tests for the bivariate analyses. Continuous variables were divided into categories based on their distribution or clinically relevant cut-points. We used a stepwise logistic regression model (entry criteria, p < .20; stay criteria, p < .05) to determine patient characteristics independently associated with reducing work hours.
For the second stage of each analysis, the caregiver was the unit of analysis. The goal of these analyses was to determine the caregiver characteristics associated the caregiver reducing their hours to care for the patient. For the bivariate analysis, we used chi-square tests. For the multivariate analysis, we used generalized estimating equations. This approach accounts for the clustering of multiple caregivers with a single patient. This analysis controlled for the patient characteristics that were associated with reducing work hours in the first stage of the analysis.
After observing a strong relationship between African-American and Hispanic ethnicity and reduction of caregiver work hours, we performed additional analyses to determine whether the relation between ethnicity and work-hour reduction was explained by higher rates of paid formal care services among white patients. These analyses were limited to the 2283 patients with available data about the use of paid formal services. We repeated our patient level multivariable analyses including whether or not the patient received each of the formal services as a covariate. We compared the ß-coefficients for the ethnic categories in models with and without the formal services.
|
Results |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
The mean age of the patients was 78 years; 73% were women, and about half were white (Table 1 ). Most patients needed assistance with one or more ADL. A total of 54% of patients had dementia. There were 4592 potentially working caregivers (Table 2 ). The mean caregiver age was 45; 62% were women, and most were children of the patients. Thirty-six percent lived with the patient.
|
|
|
|
|
|
|
Discussion |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
We demonstrated that caregivers of African-American and Hispanic patients are more likely to reduce work to provide informal care. This may in part explain lower rates of nursing home use among minority patients and is consistent with other research demonstrating differing attitudes toward caregiving among minority patients (20)(21)(22)(23)(24)(25). For example, African-American caregivers report lower amounts of burden than white caregivers (22)(23)(24). Others have found that Hispanic-American patients with dementia are more likely to live with their children than white patients and that familial expectations may be greater in Hispanic families (25). The finding of higher rates of employment reduction in caregivers of African-American and Hispanic elders has important policy implications because the willingness to reduce employment hours to provide care may delay nursing home placement. While the economic consequences of reduced employment are borne by the family, the costs of nursing home care are often a public expense. African-American and Hispanic families may bear a disproportionate share of the total societal cost of caregiving.
Others have suggested that the decision to reduce work hours and provide care for a frail patient is in part determined by the opportunity costs of foregone wages (8). Because we have no data on the income of caregivers in our study, we do not know whether lower rates of work-hour reduction in white and Asian caregivers are explained by higher income levels. It is possible that caregivers with higher incomes choose to purchase services such as homecare and adult day health for their frail relatives in lieu of reducing employment. However, our data suggest that it is unlikely that this explains the ethnic variation in work-hour reduction. Caregivers of patients receiving these services were not less likely to reduce their employment hours, and adjusting for use of these services did not affect the relation between ethnicity and reduction in employment hours. Further, others have reported that paid home services generally supplement rather than replace the informal services provided by caregivers (26).
We found that specific clinical characteristics can identify patients whose caregivers are more likely to reduce employment. These include ADL dependence, a history of stroke, and a history of dementia, particularly if it is accompanied by behavioral disturbance. Each of these characteristics interferes with the patient's ability to care for themselves and imposes substantial demands upon caregivers in terms of physical work and the need for patient supervision. Since each of these conditions are common causes of nursing home placement, this finding also supports the hypothesis that the family's willingness to accept income loss may save the public the expense of nursing home care (27).
We demonstrated that caregivers who lived with the patient are more likely to reduce work hours than caregivers who do not live with the patient. While there may not always be a causal relationship between a caregiver living with a patient and work-hour reduction, the decision of a caregiver to have a patient live with them likely reflects a strong commitment to caregiving. Others have shown that caregivers who live with the patient experience higher levels of stress and burden than caregivers who do not live with the patient (28). Our finding that the presence of an additional caregiver at home lessens the chance that a caregiver will reduce work hours is consistent with prior research demonstrating that the degree of support available to the caregiver is an important mediator of caregiver outcomes (22).
We found that daughters of patients were more likely to reduce work hours than sons. This was not surprising in light of consistent data from other studies demonstrating that women provide the majority of assistance with nonmedical care for frail relatives and generally report higher levels of subjective burden than male caregivers (22)(29). The relation between gender and reduction in work hours among caregivers of older patients parallels similar findings among caregivers of young children (30).
Our patient population may have implications for the generalizability of our results. All patients in this study were enrollees in one of the demonstration sites of the PACE program, a program designed to keep patients who would otherwise require nursing home care in the community. Since data were obtained at the time of enrollment, our results do not reflect the impact of the services provided by PACE. Nonetheless, patients who enroll in PACE are probably cared for by families that have strong propensities toward avoiding or delaying nursing home care. Our results may be most generalizable to patients and families with similar propensities. The fact that most frail elders who meet criteria for nursing home placement continue to live in the community (1)(2) suggests that this probably represents a large population.
Almost all of the patients in this study were covered by Medicaid. The generalizability of our results to patients that do not have Medicaid is uncertain. However, the Medicaid-eligible population is of considerable policy importance because they are eligible to have their nursing home costs covered by the state. Among the Medicaid eligible, the willingness of families to accept financial burdens in order to delay nursing home care helps conserve public resources.
Several additional methodologic issues should be considered in interpreting our results. First, we do not have data on the exact number of hours caregivers stopped working or the actual cost of those decreased hours. However, because about one-third of the caregivers who reported decreasing their hours gave up working entirely, it is likely the amount of employment loss is substantial. Second, we do not know what the consequences of the lost income were to the family. Future studies will need to elucidate these consequences.
It is important for providers and policy makers to fully recognize the economic and personal contributions of unpaid family caregivers (6)(13)(31). Most elders eligible to live in nursing homes continue to live in the community, in large part because of the unpaid efforts of their caregivers. However, almost no data are available on the economic consequences of caregiving, and the costs of caregiving are rarely considered in policy discussions of healthcare costs. The failure to consider these costs parallels the anecdotal reports that the health care system is not responsive to the needs of caregivers (32). It is important that future work examine methods of mitigating the economic and personal costs of caregiving and ways in which the societal responsibility of caring for frail elders can be equitably shared.
|
Acknowledgments |
|---|
Received October 26, 2000
Accepted October 30, 2000
|
References |
|---|
|
TopAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. G. Rossi, M. Beccaro, G. Miccinesi, P. Borgia, M. Costantini, F. Chini, D. Baiocchi, G. De Giacomi, M. Grimaldi, M. Montella, et al. Dying of cancer in Italy: impact on family and caregiver. The Italian Survey of Dying of Cancer J Epidemiol Community Health, June 1, 2007; 61(6): 547 - 554. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Wolff and J. D. Kasper Caregivers of Frail Elders: Updating a National Profile. Gerontologist, June 1, 2006; 46(3): 344 - 356. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Sink, K. F. Holden, and K. Yaffe Pharmacological Treatment of Neuropsychiatric Symptoms of Dementia: A Review of the Evidence JAMA, February 2, 2005; 293(5): 596 - 608. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. W. Rabow, J. M. Hauser, and J. Adams Supporting Family Caregivers at the End of Life: "They Don't Know What They Don't Know" JAMA, January 28, 2004; 291(4): 483 - 491. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. A. Banks and J. E. Morley Memories Are Made of This: Recent Advances in Understanding Cognitive Impairments and Dementia J. Gerontol. A Biol. Sci. Med. Sci., April 1, 2003; 58(4): M314 - 321. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
