| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
|
|
|
|||||||
|
|||||||||
Correspondence: Address correspondence to Steven M. Albert, PhD, MSc, Gertrude H. Sergievsky Center, Columbia University, PH-19, 630 West 168th Street, New York, NY 10032. E-mail: sma10{at}columbia.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Survival • Home attendant care • Medicaid • Population-based study • Dementia
These conclusions, however, should be considered in light of important limitations: (a) "paid help" includes a great variety of programs (e.g., in-home nursing and interdisciplinary care teams, homemaker or home attendant care, adult day health care, hospice, and programs that mix these elements), making it difficult to isolate features of programs that might affect mortality risk; (b) the "dose" of formal care in most cases is quite low; for example, in one study of homemaker services, the intervention group received, on average, only 1 hr of extra service each day (Weissert, Wan, Liviertos, & Pellegrino, 1980); (c) studies, in some cases, have not been able to control adequately for differences in medical and cognitive status, a particular problem in observational studies, in which people in poorer health are more likely to receive services; and, finally, (d) many studies were limited in sample size and duration of follow-up.
Recently, a variety of home-care interventions, most involving medical or social service linkages, have shown survival benefits for vulnerable elders. Shapiro and Taylor (2002) reported that elders receiving a case-management intervention had a reduced risk of death over an 18-month period compared with a waiting-list comparison group (2.5% vs 6.1%). Home care interventions for patients discharged from hospitals also have been shown to reduce mortality risk in the case of heart disease (Inglis et al., 2004; Stewart, Marley, & Horowitz, 1999) and cancer (McCorkle et al., 2000). In the case of frail elderly people more generally, preventive home visits by nurses resulted in a lower combined risk of death or nursing home admission (10% vs 5.8%) over 14 months, though this difference did not achieve statistical significance (Dalby et al., 2000).
Accordingly, it is appropriate to reconsider the question of formal care and mortality risk. Given the success of formal care programs targeted to particular patient groups, supportive care programs may offer important benefits to frail elderly people more generally. These elders are at risk for excess mortality because they may be unable to meet such basic needs as provisioning, hygiene, medication management, nutrition, health monitoring, and social contact without assistance, and because emerging medical problems require timely evaluation and treatment.
The New York City (NYC) Medicaid Home Care Services Program, administered by New York City's Human Resources Administration, provides supportive care with regular nursing assessments. Low-intensity service includes help with light housecleaning, provisioning, and preparing of meals, whereas high-intensity service may include 24-hr-a-day personal assistance care in addition to housekeeping. A registered nurse or a nurse of that level visits the client's home at least once every 90 days, at which time paraprofessional care is reviewed and the client's functional status evaluated. Any Medicaid-eligible elder can apply for services in the program if a physician attests to need for assistance to remain at home. Hours of home care are assigned according to professional assessments of disability, with elders receiving 4, 8, 12, or 24 hr of service daily.
In the NYC program, each client also has a caseworker responsible for ongoing oversight of the case, including service reauthorization at appropriate intervals. The number of patients per caseworker averages 160. The typical duration of care is several years, with some clients receiving services for 15 to 20 years. Once admitted, it is extremely rare for elders to be dropped from the service. Even with hospitalization or admission to a nursing home for rehabilitation, elders are entitled to return to the program on discharge. The program currently serves approximately 55,000 clients daily through 75 vendors, each of whom has an agreed-on Medicaid caseload in a specific borough, ranging from 200 to 1,400 cases.
In northern Manhattan, older adults using the program received an average of 33.9 hr of home care each week in 1996, which is two to three times higher than the national average (Albert, Brassard, Simone, & Stern, 2004). The average duration of service was about 4.5 years. Although home care service is an optional Medicaid benefit that has been adopted by 26 states, 15 states have set limits on hours of home care service and 10 have set cost caps (LeBlanc, Tonner, & Harrington, 2001; Wiener, Tilly, & Alecxih, 2002). New York is the only state that has not placed limitations on the benefit of home care services, and it accounted for $1.66 billion of the $3.44 billion spent nationally in 1998 for the Medicaid personal care service state option (LeBlanc et al.; General Accounting Office, 1999).
We sought to determine if the more intensive level of support provided by the program and its integration with other services may reduce mortality in older adults. We investigated this issue in a sample of community-resident, Medicaid-eligible older adults assessed in the period from 1994 to 1996 and followed through the end of 1999. Data from this cohort were merged with files provided by the NYC Human Resources Administration to examine the effect of the Home Care Services Program. The Columbia University-Columbia Presbyterian Medical Center Institutional Review Board approved the research protocol.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Medicaid participants included 64 (7.4%) non-Hispanic Whites, 218 (25.2%) non-Hispanic African Americans, 576 (66.5%) Hispanics, and 8 (0.9%) from other groups. Dementia was more likely to be diagnosed among minority groups, though this difference did not achieve statistical significance (White, 9.5%; African American, 22.1%; Hispanic, 20.2%; p =.08). Race–ethnicity groups did not differ significantly in gender (percentage female: 75.0%, Whites; 80.7%, African Americans; 75.5%, Hispanic). Mean age in years was 78.9 in White, 79.6 in African American, and 78.0 in Hispanic elders (p =.01). Non-Hispanic Whites reported significantly more years of school (9.6 vs 8.7 in African Americans and 4.9 in Hispanics; p <.001).
Clinical and Epidemiologic Measures
As part of the WHICAP assessment, clinical data were reviewed in a consensus conference, at which point a diagnosis of dementia and Alzheimer's disease, if warranted, was made on the basis of a neuropsychological paradigm and criteria established by the National Institute on Neurological Diseases and Stroke–Alzheimer's Disease and Related Disorders Association (NINDS-ADRDA; Stern et al., 1992).
Mortality in the cohort was tracked from 1994 to 1999 through subsequent interviews every 1.5 years and submission of identifying information to the National Death Index. Between 1994 and 1999, 21.8% (189/866) of the cohort died. The dependent measure in survival models was time between the 1994–1996 assessment and death or a participant's last assessment. Although age also can be used in modeling survival, we used study follow-up as our outcome because of the restricted follow-up period and different ages at which people began to use Medicaid home care. All models adjusted for age.
We considered a variety of sociodemographic, medical, functional, and caregiving indicators in examining the relationship between Medicaid home care service and mortality. Sociodemographic variables included age, gender, years of school, living situation (alone or living with others), and race or ethnicity based on 1990 U.S. Census designations. Medical indicators included dementia diagnosis and number of diseases elicited in the physician interview by use of a modified Charlson index. Medical conditions recorded included any heart condition, hypertension, other lung or chest conditions, diabetes, arthritis or joint or muscle pain, Parkinson's disease, mental or emotional conditions, or other conditions. Disability was based on a count of any reported difficulty with five activities of daily living (ADLs): bathing, dressing, using the toilet, eating, and personal grooming. For indicators of caregiving support apart from the Medicaid Home Care Services Program, participants were asked if they were receiving regular help with preparing meals, taking medications, cutting toenails, being mobile indoors, and going outdoors. These reports do not distinguish the source of such help, that is, whether it was provided by informal or formal providers; hence we could not introduce this measure into survival models to distinguish potentially independent effects of formal and informal care on mortality risk. Still, the measures are useful for establishing basic levels of caregiving support in people not receiving Medicaid home care services.
Administrative Claims Data
Medicaid Home Care Program Administrative Claims
Data from the NYC Home Care Services Program were available beginning in 1994. We submitted the list of WHICAP cohort participants to the NYC Human Resources Administration, which administers the Home Care Services Program. Matches were made by social security number, name, and birth date, and then inspected manually. For each individual identified in the two data systems, we retrieved all service periods of home care service and the number of care hours billed for each period. We then computed the mean number of hours across episodes for each individual.
Among Medicaid participants in the Washington Heights–Inwood cohort, 288 (33.3%) participated in the Home Care Services Program in 1994–1996. Participants received a mean of 35.1 (SD = 34.5) billed hours per week and a median of 20. The mean duration of service was 4.9 (±4.3) years through 1996. We considered anyone receiving Medicaid home care hours in the 1994–1996 period, when the WHICAP follow-up survey was conducted, to be a user of the service. Between 1997 and 1999, an additional 48 people began to receive home care in the Medicaid program. Given our focus on the prevalence period of 1994–1996, we consider these people nonusers in analyses. This is appropriate because these later users did not differ from nonusers in ADL disability at baseline or in mortality risk over follow-up.
Medicare Claims
If improved survival is associated with home care provided in the Medicaid program, one reason for this survival advantage may be that users are more likely to receive medical care as a consequence of participating in the program. To investigate this issue, we examined Medicare hospitalization and ambulatory medical claims for the cohort. We obtained 1996 Medicare files for residents in the zip codes that define the Washington Heights–Inwood area and matched WHICAP participants to claims data by name, gender, and birth date. Each match was again manually checked to make sure that spouses' claims (which can be made on the same social security number) were distinguished from participant claims. Here, we were limited to the subsample of participants who survived through 1995 (n = 821) and had an identified Medicare claim (n = 617/821, or 75.2%, a match rate similar to that for the full WHICAP sample; see Albert, Glied, Andrews, Stern, & Mayeux, 2002). The subsample included 223 Medicaid Home Care Service users and 394 nonusers.
To assess whether the 75% match rate might represent a biased selection from the full sample, we compared people with matches in the Medicare files with remaining participants. In chi-square tests, the two groups did not differ in sociodemographic indicators, ADL disability, or medical status. Lack of matches appears to be random and largely due to missing social security numbers and primary residence outside the Washington Heights–Inwood area.
Relationship Between Medicaid Prevalence Period and Survey Year
Because the WHICAP survey was conducted in 1994–1996, we identified all users of the Medicaid service in this time interval and compared them with people not using the service. Bias in survival estimates would be a concern if service users were more likely to be interviewed toward the end of this prevalence period. However, users and nonusers of home care services were equally likely to receive WHICAP assessments in each year of the prevalence period. In 1994, 37.3% of the nonusers and 35.6% of the users were assessed; the remainder were assessed in 1995 (53.3% of the nonusers and 55.8% of the users) and 1996 (9.4% of the nonusers and 8.6% of the users; p =.74).
Analyses
We compared mortality risk in Medicaid home care users and nonusers in proportional hazards regression models. Models adjusted for a variety of covariates associated with receipt of home care and mortality to try to identify the extent to which home care was a significant, independent predictor of earlier or later death. We checked the proportional hazards assumption of these models visually by using log-minus-log survival plots, and we also examined Martingale residuals to check for influential outliers (Norusis, 1993). In the latter, we noted one influential case, which, on removal, did not alter results. Because participation in the program was strongly related to ADL status, we developed a second model for people with ADL disability only.
A concern in all such analyses is how best to handle nursing home admissions. We developed proportional hazards models for time to nursing home admission and also report these results in the paragraphs that follow. Our strategy was not to combine the endpoints but rather to include nursing home admission as an additional predictor in the proportional hazards models for time to death. In this way, the model pays due regard to the potentially greater risk of nursing home placement (and the increased mortality associated with placement) among formal care users.
We also examined 1996 Medicare claims data to determine if users of the Medicaid Home Care Services were more likely to have a greater number of hospitalizations and ambulatory and hospital outpatient care visits. To explore this hypothesis, we compared the proportion of hospitalizations and mean number of Medicare claims (ambulatory and outpatient) among users and nonusers within categories defined by ADL status. In these analyses, we used a two-way analysis of variance for continuous measures and a chi-square test for differences in proportions.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
Differences were more apparent in the non-ADL disabled group. Users of the home care service in this group were significantly more likely to be older, live alone, have a dementia diagnosis, and have two or more medical conditions. They also reported receiving significantly more help in caregiving tasks.
Time to Nursing Home Placement by Medicaid Home Care Program Status
Over follow-up, 46 people entered nursing homes for long-term care, which constituted 2.2% of nonusers and 11.5% of users (p <.001 by chi-square). In the proportional hazards model for this outcome, the relative risk (RR) for nursing home placement associated with Medicaid home care was 4.2 (95% confidence interval or CI = 1.9, 9.7), after adjustment for sociodemographic indicators, medical status, and ADL disability at baseline.
Survival by Medicaid Home Care Program Status
Table 2 summarizes results from proportional hazards models for survival. One model includes the full sample of community-resident Medicaid elders, with an interaction term for ADL disability and receipt of Medicaid home care; a second model is restricted to the subset with ADL disability. Significant predictors of shorter survival time for the full sample of individuals included older age, male gender, dementia diagnosis, a greater number of medical conditions, and presence of ADL disability. Education, living alone, race–ethnicity, and nursing home admission were not significant predictors in this model. Although receipt of Medicaid home care services was not a significant predictor in the model, the interaction term for ADL disability and Medicaid home care was significant (RR = 0.45; 95% CI = 0.24–0.87). This suggests that mortality risk, although elevated for people with ADL disability, is lower among people with ADL disability who receive Medicaid home care services.
|
|
By contrast, if we restrict the sample to people receiving Medicaid home care, the number of weekly hours of home care was not a significant predictor of survival. Together, these findings suggest that the survival benefit appears to be linked to getting people with ADL disability into the program, rather than providing additional hours to people receiving the service.
Finally, analyses limited to the first 12 months of follow-up also showed a significant difference in mortality. Among people with ADL disability, mortality was 5.5% in program users and 16.4% in nonusers (p <.01 by chi-square). In people without ADL disability, 1.6% of users and 2.3% of nonusers died (p = ns).
Medicaid Home Care Services and Medical Care Utilization
If participation in the Medicaid program is associated with a survival advantage, it may be because home attendant care, case management, and nursing oversight may increase the likelihood that changes in health are noted and treated. A reasonable proxy for increased medical vigilance is a greater number of physician visits and perhaps hospital admissions. We hypothesized that program users would have a greater number of Medicare claims. Because hospital outpatient care is often an additional source of primary care, we examined outpatient claims, as well as Part B ambulatory care and Part A hospitalization claims.
Table 3 shows that the mean number of ambulatory and outpatient care claims was significantly higher among Medicaid Home Care Service users. Within each ADL category, users had eight to nine more claims in the ambulatory care file and about three more claims in the outpatient file. Having a hospital admission was also more likely among users, but only in the non-ADL disabled group.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Why should formal care use in the NYC Medicaid program confer a survival advantage to older people with ADL disability? Our findings suggest two potential mechanisms. First, the greater use of ambulatory and outpatient medical care in people receiving Medicaid home care suggests that increased access to medical services may be the source of the survival advantage. In a survival model that included the number of Medicare claims and Medicaid program participation, the effect of program participation was attenuated but did not disappear. This suggests that the survival benefit may be due to greater medical service use among program participants but also to features specific to home care, such as better provisioning or home safety efforts that may reduce the likelihood of health problems. Second, we found that the survival benefit appears to be linked to getting people with ADL disability into the program, rather than providing additional hours to people receiving the service. This suggests that program participation is more than simply the number of formal care hours provided, and it points again to the benefits of medical and social service oversight built into the program.
The NYC Home Care Services Program has features that distinguish it from other community-based long-term-care programs. First, it provides more hours of home attendant care per week than other programs, as described herein. Second, home attendant care in the program is integrated with regular registered nurse visits to client homes and case-management oversight. Finally, the Home Care Services Program has developed a careful system of referral and evaluation that helps ensure targeting to frail older adults who are nevertheless medically stable and able to be maintained in the community. These differences should be kept in mind when the results are considered.
Still, it should be noted that, among people with ADL disability, nonusers and users of the service were quite similar in socioeconomic profiles and did not significantly differ in the proportion with dementia, with two or more medical conditions, or who were receiving at least some help with most caregiving tasks (Table 1). Thus, nonusers do not appear to represent a less medically stable population. The similarity in reports of caregiving support suggests that Medicaid formal care supplemented informal care. Our data unfortunately do not allow us to determine the relative role of formal and informal care in the observed survival benefit for home care users.
The supplementary hours provided by the program may help frail elderly persons better meet such basic needs as hygiene, nutrition, social support, and medication administration; and the regularly scheduled nurse visits may result in timelier interventions to prevent exacerbation of medical conditions. However, measures included in this research do not allow a direct test of potential health advantages associated with the NYC program. For example, our data do not allow a definitive test that home attendant care is associated with fewer falls or maintenance of weight in older people, or that periodic nursing assessments result in more timely medical care. This is clearly an area for future research. The greater number of ambulatory care claims, however, suggests increased medical vigilance and treatment among program participants.
Advantages of our study design include its careful diagnoses of dementia and use of administrative data. Because we began with a known population of older adults in a defined community, whose medical status was determined separately from administrative claims, we were able to avoid a number of biases characteristic of research limited to administrative claims only, such as the tendency to identify only people with the most severe disease (Newcomer, Clay, Luxenberg, & Miller, 1999; Leibson et al., 1999).
The survival benefit reported here was established in an observational study, not a randomized trial, and should be interpreted in this light. For this reason, we suggest that additional research be conducted to clarify effects of programs that combine personal assistance care with nursing and social service linkages.
|
Footnotes |
|---|
1 Gertrude H. Sergievsky Center & Taub Institute for Research on Alzheimer's Disease and the Aging Brain, Columbia University, New York, NY.
2 Department of Sociomedical Sciences, Columbia University, New York, NY.
3 New York City Human Resources Administration, New York, NY.
4 School of Science and Health, William Paterson University, Wayne, NJ.
Decision Editor: Linda S. Noelker, PhD
Received for publication April 23, 2004. Accepted for publication December 17, 2004.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
