| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Philip D. Sloane, MD, MPH, Program on Aging, Disability, and Long-Term Care, Cecil G. Sheps Center for Health Services Research, University of North Carolina, Chapel Hill, NC 27599. E-mail: psloane{at}med.unc.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Alzheimer's disease • Outcomes • Nursing homes • Assisted living • Residential care • Special care units
Despite the recent growth of RC/AL, to our knowledge no large study has examined whether dementia care in these settings differs from that provided by nursing homes. Proponents of RC/AL claim that it provides a homelike atmosphere that may help maintain function, promote independence, improve resident outcomes, enhance family and staff satisfaction, and improve quality of life (Saxton, Silverman, Ricci, Keane, & Deeley, 1998). Others have noted, however, that because RC/AL homes often lack professional staff, they may not adequately address functional status needs, depression, comorbid conditions, and other causes of morbidity and mortality among persons with dementia (Pruchno & Rose, 2000). As a further complicating factor, specialized dementia care units and facilities have proliferated in RC/AL, adding one more subquestion to the debate (Davis et al., 2000).
Given that RC/AL and nursing homes often serve dementia residents with similar characteristics (Davis et al., 2000), an important policy and practice question is the extent to which outcomes of care may differ across settings. Among the outcomes of interest are mortality, rates of decline in functional status, and the incidence of new health events such as decubiti and infection (Galasko et al., 1995; National Institute on Aging/National Institutes of Health [NIH], 1999). In this article we compare health and functional outcomes and health care utilization of persons at similar stages of dementia, using data from a large, multistate cohort study of RC/AL facilities and nursing homes.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
To ensure adequate representation of the range of RC/AL types across the study states, we randomly sampled facilities in four strata and selected residents by using a strategy that was designed to enroll approximately equal numbers of participants in each stratum. The four sampling strata were (a) small (< 16 beds) RC/AL facilities; (b) new-model RC/AL facilities (
16 beds, built after 1986, and with at least one of the following features of newer, health service intensive RC/AL: charges that varied by resident service need, having 20% of residents who required assistance in transfer, having 25% of residents who were incontinent daily, or having a licensed nurse on duty 24 hr/day); (c) "traditional" RC/AL facilities ( 16 beds, not meeting the new-model definition); and (d) nursing homes. We defined dementia special care units as areas within facilities (or entire facilities) that were self-designated as Alzheimer's or dementia units, and within which at least three fourths of the residents had a diagnosis of Alzheimer's disease or a related disorder. The facility recruitment rate was 59%; nonparticipating facilities did not differ from participating ones in proprietary status, age, size, or occupancy rate, or resident age or ethnicity–race. Further details of the study design are provided elsewhere (Zimmerman, Sloane, Eckert, Buie, et al., 2001).
Within study RC/AL facilities, we enrolled a random sample of residents. Within nursing homes, we drew a stratified random sample with the goal of enrolling approximately equal numbers of persons with mild dementia (i.e., who were ambulatory and had a dementia diagnosis) and with moderate to severe dementia (i.e., had dementia and required physical assistance with ambulation, transfer, or feeding); thus, enrollment criteria within nursing homes excluded individuals undergoing rehabilitation and oversampled ambulatory persons with dementia. Across facility types the recruitment rate for eligible resident participants was 92%. Each facility received an on-site data-collection visit on enrollment in the study (October 1997 through November 1998), during which a variety of facility- and resident-level data were obtained. Subsequently, we had each facility telephoned quarterly for 1 year to identify the incidence of new health events and changes in a variety of psychosocial and functional measures for each enrolled resident. Participants who were permanently transferred out of study facilities were censored at the time of discharge; in these cases, we gathered medical, functional, and health care utilization data about their status prior to discharge. This study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill.
Selection of Participants With Dementia
These analyses involved the 1,252 CS-LTC participants from 206 facilities who, on enrollment, met study criteria for dementia and whose level of dementia (mild, moderate, or severe) could be determined. We defined dementia to be present if, at baseline, (a) the medical record reported a diagnosis of Alzheimer's disease, senile dementia, organic brain syndrome, cerebral arteriosclerosis, multi-infarct dementia, subcortical dementia, Binswanger's disease, Pick's disease, Creutzfeldt-Jakob disease, Lewy body disease, or unspecified dementia; and (b) the participant received a score of 2 on the Minimum Data Set Cognition Scale (MDS-COGS; Hartmaier, Sloane, Guess, & Koch, 1994), which is roughly equivalent to a Mini-Mental State Examination score of < 23 (Gruber-Baldini, Zimmerman, Mortimore, & Magaziner, 2000; Hartmaier et al.). We assigned severity of dementia on the basis of impairment of three activities of daily living (ADLs): locomotion, transfer, and feeding. We classified participants meeting the study diagnosis of dementia who were independent in all three of these ADLs as having mild dementia; we classified those who required physical assistance with one or more of these ADLs as having moderate to severe dementia. This system of classification results in a moderate–severe category that roughly includes persons at stage 6 or 7 (severe or very severe cognitive impairment) on the Global Deterioration Scale (Reisberg, Ferris, de Leon, & Crook, 1982), stage 3 (severe dementia) on the Clinical Dementia Rating Scale (Hughes, Berg, Danzinger, Coben, & Martin, 1982), and a score of 5 or greater (moderate, severe, or very severe dementia) on the MDS-COGS (Hartmaier et al.). We excluded 30 enrolled participants with dementia from analyses because data were insufficient for us to determine the level of dementia.
Measures and Data Collection
Enrollment data were gathered on site by professional data collectors, the majority of whom were nurses. Follow-up data were gathered by telephone interview. All data collectors were trained and supervised by the study's principal investigators (P. Sloane and S. Zimmerman).
Data collectors collected basic demographic information on study participants, including age, gender, race or ethnicity, education, marital status, length of residence in the facility at baseline, and comorbid illness at baseline, from the resident or a staff member who was knowledgeable and familiar with the resident's health. For analysis purposes, we express comorbidity as a summary scale of the presence or absence of the following morbidities: peripheral vascular disease, congestive heart failure, hypertension, hypotension, myocardial infarction, other heart problems, Cushing's syndrome, diabetes, hyperparathyroidism, thyroid or other glandular disorders, pituitary disorders, blindness, glaucoma, cataracts, or macular degeneration, liver disease, chronic renal disease, stomach or intestinal ulcers, AIDS, arthritis, fracture, osteoporosis, Paget's disease, cerebrovascular disease, hemiplegia, paraplegia, dizziness or imbalance, seizures, mental retardation, Parkinson's disease, chronic pulmonary disease, cancer, and mental or psychiatric illness.
We had all measures for which change would be recorded over time collected at baseline from staff interviews, thereby allowing for the same individuals (or, in cases of staff turnover, their replacements) to serve as respondents for follow-up data collection. We used the following measures to assess change in participant function over time:
=.67 and internal consistency is 
=.84 (Alexopoulos, Abrams, Young, & Shamoian, 1988). We had baseline data collected during site visits to each of the study facilities. We had follow-up interviews conducted quarterly by telephone so that we could assess whether or not residents had died, been discharged (including date and destination of discharge), been hospitalized, or experienced any new or worsening morbidity during the quarter. We defined new or worsening morbidity as the incidence or worsening of fracture, infection, stroke or paralysis, bleeding from the stomach or bowel, diabetes, heart condition, or skin ulcer; we chose these morbidities because they are relatively common in long-term care and have risk factors that can be modified through medical or nursing treatment. At the 1-year interview, or when a participant had been discharged during the quarter, the interviewers also administered the study measures of ADLs, cognitive status, behavior, depressive symptoms, social function, and withdrawal, determining 1-year status or that immediately prior to discharge.
Analysis
We calculated descriptive statistics by using appropriate programs from the Statistical Analysis System (SAS Institute, 2000). We modeled rates of mortality, hospitalization, and the incidence of new or worsening morbidity by using repeated measures analysis, in which we used generalized estimating equations (Liang & Zeger, 1986) to fit a Poisson regression model. For analytical purposes, we combined all new or worsening morbidities. We used Cox proportional hazards models to generate p values for differences between mean time to death. We modeled functional change by using generalized estimating equations to estimate differences in means. In additional analyses we compared outcomes of dementia special care unit residents with non-unit residents in RC/AL and nursing facilities, controlling for baseline resident characteristics.
To adjust for baseline differences between the RC/AL and nursing home samples, we (a) analyzed all models in two strata by degree of ADL impairment (independent or with some impairment of locomotion, transfer, or feeding), and (b) adjusted all models for the following resident baseline characteristics: age, gender, race, marital status, education, length of stay, MDS-COGS, and number of comorbid conditions. To adjust for differential selection probabilities among the three facility types, we had all outcome analyses incorporate sampling weights (i.e., statewide proportions of residents and facilities in each sampling stratum, be it small, traditional, or new model).
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Table 1 compares the RC/AL and nursing home samples in terms of selected demographic and health status characteristics measured on study enrollment. The RC/AL sample of individuals contained a higher proportion of participants who were White, widowed, had at least a high school education, and had mild dementia; nursing homes housed more African Americans, more persons who were divorced or separated, and more individuals with less than an eighth-grade education. The mean length of stay of the nursing home cohort at the time of study enrollment (896 days) was longer than that of the RC/AL participants (725 days). On average, the individuals in RC/AL, when compared with the individuals in the nursing homes, exhibited less severe cognitive impairment, fewer comorbid conditions, fewer impairments of ADLs, more depression, better social function, and less withdrawal; we noted no difference between the groups in the prevalence of problem behaviors.
|
|
|
Rates of Functional Change
The lower portion of Table 3 displays the rates of functional change over 1 year for all study participants, adjusted for baseline age, gender, race, education, marital status, length of stay on enrollment, cognitive status, and comorbidity. All participants were included, with change scores for discharged or deceased participants calculated on the basis of reported status immediately prior to discharge. Mean rates of decline in ADL, cognitive, and social function did not differ between the facilities and nursing homes for either the mild dementia cohort or the moderate to severe dementia cohort. Similarly, we noted no statistically significant difference in rates of change in behavior, depressive symptoms, or withdrawal.
Residence in Specialized Units
When we compared dementia special care units with nonspecialized settings, we observed few differences in outcomes (Table 4). In both RC/AL facilities and nursing homes, we noted no significant differences in mortality or in rates of change in cognition, behavioral problems, depression, social function, or withdrawal. Rates of hospitalization and of new or worsening morbidity were lower in nursing home special care units than in nonspecialized settings in nursing homes; however, we did not observe this relationship in RC/AL. ADL decline was more rapid in special care units than in nonspecialized settings in these facilities, largely as a result of an increased rate of decline among persons with moderate to severe dementia; however, we did not observe this association in nursing homes.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
In this study we compared 12-month outcomes of a cohort of RC/AL residents with dementia with those of similar residents in nursing homes, controlling for baseline differences in subject age, gender, race, education, marital status, length of stay, cognition, and comorbidity. These analyses indicate that, for both mild and moderate to severe dementia, health outcomes did not differ between these facilities and nursing homes. Adjusted models (Table 3) identified no significant difference in mortality, incidence of new or worsening health conditions, or rate of change in functional dependency, cognition, behavioral problems, depressive symptoms, social function, or social withdrawal. Thus, these results suggest that, in terms of the health outcomes studied, many persons with dementia can be served equally well in either setting.
The one area where outcomes clearly differed was hospitalization rates, which were 69% and 42% higher, respectively, for participants with mild and moderate to severe dementia in RC/AL than for similar participants in nursing homes (Table 3). Although they did not appear to cause an overall increase in mortality or functional decline, such persons were largely responsible for the nearly 25% annual rate of nursing home transfer among facility residents (Table 2). This finding suggests that RC/AL facilities tend to have difficulty managing residents with dementia who have or develop significant medical or nursing care needs, in spite of some claims that these facilities can allow aging in place through the addition of home health, hospice, and other services (Thompson & Marinaccio, 1997). This finding is consistent with the results of Carter and Porell (2003), who reported that underlying respiratory, genitourinary, or circulatory conditions among nursing home residents were associated with increased risk of hospitalization. Thus, it appears likely that persons with dementia who are medically stable may do well over long periods of time in residential care or assisted living facilities. For those who are medically unstable, however, nursing homes may be better able to avoid hospitalization and, therefore, to both reduce cost and prevent the functional deterioration that frequently follows hospitalization (Coleman, Barbaccia, & Croughan-Minihane, 1990).
These analyses also found little difference in outcomes of dementia special care units when they were compared with nonspecialized settings (Table 4). In both nursing homes and RC/AL facilities, rates of mortality and of change in cognition, behavioral problems, depressive symptoms, social participation, and withdrawal did not differ between special care and nonspecial care units. Nursing home special care units did, however, appear to have reduced morbidity and hospitalization rates, and units in RC/AL to have higher rates of ADL decline. Although worthy of follow-up study, these isolated findings do not appear to represent a consistent pattern and, therefore, could well be spurious, considering the use of multiple comparisons in these analyses. Thus, the overall impression is that these findings add to a growing body of evidence that placement in a special care unit does not alter health and functional outcomes (Day, Carreon, & Stump, 2000; Phillips et al., 1997; Sloane, Lindeman, Phillips, Moritz, & Koch, 1995). Our study did not, however, evaluate psychosocial outcomes such as overall quality of life and resident and family satisfaction, and it is possible that these may have differed between special care units and nonspecialized settings.
Although we conducted the study and its analyses to minimize bias, several aspects of the study methods deserve comment. The selection of states and facilities, although designed to capture a wide range of facility types and residents, did result in a sample from which the unadjusted rates (Tables 1 and 2) cannot be generalized to either the study states or the nation. However, our outcome analyses (Tables 3 and 4) did adjust for sampling weights and, therefore, should be generalizable at least to the study states. The 41% facility refusal rated could constitute a source of bias; however, a survey of nonrespondent facilities identified few differences between responders and nonresponders (Zimmerman, Sloane, Eckert, Buie, et al., 2001). Because the outcome data were reported by facility staff, and licensed nurses are absent from some RC/AL settings, it is possible that reporting of certain health conditions differed across settings; however, the study used standardized instruments and measured outcomes, the majority of which do not require advanced medical training to recognize. Attrition of participants was common in both settings; however, we minimized the impact of attrition by gathering retrospective data on status just prior to discharge, including all observations, and weighting analyses per unit time that the participant resided in the facility. Finally, regarding the abundance of negative findings (i.e., differences not statistically significant at the 5% level) for the comparisons of outcomes between RC/AL facilities and nursing homes, as well as those between special care units and nonspecialized care settings, it is important to consider the magnitude of differences that could have been detected with reasonable power (i.e., at least 80%). Where the outcomes involved incidence rates of death, hospitalization, and morbidity, the detectable differences were in the range of 5 to 15 events per 100 residents per quarter. Where the outcomes involved changes in physical, cognitive, and behavioral function, the detectable differences were in the range of 0.5 to 2 points on the various scales of measurement used, corresponding to standardized effect sizes in the range of 0.2 to 0.3. It therefore appears that the study was of sufficient size to detect outcome differences that are regarded to be relatively small in magnitude (Cohen, 1977).
In evaluating the difference between RC/AL and nursing home care, a key question is the extent to which health outcomes of chronic, progressive conditions such as Alzheimer's disease can be modified by the care setting. These analyses add additional weight to findings from other studies, which suggest that physiological factors have a far greater influence on health outcomes of long-term care populations than the care they receive, and that the influence of care settings on health and functioning is relatively minor. At the same time, they suggest that some facility types may be better able to reduce hospital use, thereby affecting overall care costs and limiting the need for residents and families to contend with transitions to another care setting.
Individual resident factors such as general physical health have successfully been used to create mortality risk indices that predict outcomes across facilities (Carlson et al., 2001). Studies also have shown that resident baseline characteristics significantly predict resident outcomes such as cognitive functioning, agitation, and social or language skills after 1 year (Chappell & Reid, 2000). In contrast, facility characteristics appear to have a minimal impact on such outcomes (Chafetz, 1991; Holmes et al., 1990; Webber, Breuer, & Lindeman, 1995). Thus, it appears that, although facility type and design may be important for resident and family satisfaction, their impact on resident outcomes such as mortality, morbidity, and functional status is relatively minor. Such findings reinforce the current trend to focus more on quality of life and less on health outcomes in evaluating quality of care.
Finally, it should be kept in mind that RC/AL facilities, although quite heterogeneous, are generally designed to serve a population that on average is less impaired than nursing home residents (Zimmerman et al., 2003). The data presented in this article document a high morbidity burden among the growing population of RC/AL residents with dementia. Given the high risk of hospitalization, transfer to nursing homes, and mortality demonstrated in these analyses, and the relative lack of research to date, the RC/AL setting may present promising opportunities for the development of interventions that could reduce future disability, morbidity, and mortality in a population that is still quite functional but at high risk for decline. One component of such interventions would have to be better methods for the prospective identification of individuals who are likely to become medically unstable, thereby improving the efficiency of placement of individuals with dementia in more or less medicalized settings.
|
Footnotes |
|---|
1 Cecil G. Sheps Center for Health Services Research, University of North Carolina, Chapel Hill.
2 Department of Family Medicine, University of North Carolina, Chapel Hill.
3 School of Social Work, University of North Carolina, Chapel Hill.
4 Department of Epidemiology and Preventive Medicine, University of Maryland, Baltimore.
Decision Editor: Richard Schulz, PhD
Received for publication September 28, 2004. Accepted for publication February 11, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S. Zimmerman, A. L. Gruber-Baldini, J. R. Hebel, L. Burton, K. Boockvar, G. Taler, C. C. Quinn, and J. Magaziner Nursing Home Characteristics Related to Medicare Costs for Residents With and Without Dementia American Journal of Alzheimer's Disease and Other Dementias, March 1, 2008; 23(1): 57 - 65. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
