| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to David R. Mehr, MD, MS, Department of Family and Community Medicine, University of Missouri-Columbia, M228 Medical Sciences Building, 1 Hospital Drive, Columbia, MO 65212. E-mail: Mehrd{at}health.missouri.edu
 |
Abstract |
|---|
 TOP Abstract Palliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
Key Words: Nursing homes • Aged • Dementia • Palliative care • Lower respiratory infection
According to a recent meta-analysis, at least 1.9 million Americans over age 65 have Alzheimer's disease—the most common form of dementia—with 1.4 million moderately to severely affected (U.S. General Accounting Office, 1998). Most of these individuals end their lives in nursing homes (Hux et al., 1998), and 60% or more of U.S. nursing home residents are cognitively impaired (Fries et al., 1997; Ribbe et al., 1997). End-of-life care for nursing home residents with dementia is thus a major issue.
In the last several years, there has been widespread recognition of the importance of palliative care, that is care directed at symptom management rather than curing disease or extending life. However, there is no systematic approach in the United States to identify residents with dementia who should be treated palliatively, particularly with respect to conditions that are often the immediate cause of death, such as infections, malnutrition, and dehydration. For example, among elderly patients with pneumonia admitted to a New York hospital, those with severe dementia had a 53% six-month mortality, compared with 13% for cognitively intact patients (Morrison & Siu, 2000). However, invasive procedures were quite similar in both groups. In another study, 23% of long-stay nursing home residents with moderate-to-severe dementia were hospitalized within a 6-month period (Fried & Mor, 1997). Furthermore, some nursing homes frequently hospitalize terminally ill residents just before death (Smith, Kellerman, & Brown, 1995).
In late dementia, neurological impairments frequently include immobility and swallowing disorders, both of which predispose patients to develop pneumonia (Marrie & Slayter, 1996). Lower respiratory infections (LRIs), predominantly pneumonia, may be the immediate cause of death for one third of all nursing home residents (Gross, Neufeld, Libow, Gerber, & Rodstein, 1988) and for 50%–70% of patients with dementia (Beard et al., 1996; Burns, Jacoby, Luthert, & Levy, 1990; Molsa, Marttila, & Rinne, 1986; Thomas, Starr, & Whalley, 1997). We took advantage of contemporaneous prospective studies in Missouri and the Netherlands to compare treatment and mortality of nursing home residents with dementia who develop an LRI. We approached this analysis from the perspective of attempting to understand how very different approaches to palliative care for patients with advanced dementia would affect treatment decisions.
|
Palliative Care for Advanced Dementia in the United States and the Netherlands |
|---|
|
TOPAbstractPalliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
Travis, Loving, McClanahan, and Bernard (2001), in a mixed-method study, found that for nursing home residents the transition to palliative care was often problematic and punctuated by episodes of aggressive care. In further work, they identified a hierarchy of barriers to palliation and end-of-life care: failure to address treatment futility (operationalized to include poor prognosis for survival or rehabilitation), lack of communication among decision makers, no agreement on a course of care, and failure to implement a timely plan of care (Travis et al., 2002). Hanson, Danis, and Garrett (1997) have studied perspectives of bereaved family members and of nursing home physicians and staff. Families reported particular dissatisfaction with physician communication. Nurse's aides, nurses, and physicians identified lack of training, a regulatory emphasis on rehabilitation, and limited resources as barriers to good end-of-life care (Hanson, Henderson, & Manon, 2002).
Except for many studies on advance directives, other research on end-of-life care in nursing homes in the United States is quite limited. Keay and colleagues (Keay, Fredman, Taler, Datta, & Levenson, 1994; Keay, Taler, Fredman, & Levenson, 1997) have proposed guidelines for assessing palliative care quality and have reviewed care in four Baltimore facilities. Engle (1998) has argued for providing quality care during a "living-dying interval," which may last months or years, rather than the all-too-frequent focus on the last 2 to 3 weeks of life.
In the Netherlands, palliative care is common in nursing homes at the end of life. In 53% of all deaths in Dutch nursing homes, some form of treatment was limited; in 23% of all deaths, artificial administration of food and/or fluids was specifically limited (van der Heide et al., 1997). These differences in care occur in the context of important differences between the United States and the Netherlands. These differences exist on at least two levels: a different social and legal understanding of end-of-life decisions and a different organization of care in nursing homes. Many are aware that, in the Netherlands, physician-assisted death (physician-assisted suicide and euthanasia) has been decriminalized if carried out in accordance with specific regulations. Previously, it was tolerated for years under limited circumstances (van der Maas et al., 1996; van der Wal et al., 1996). However, these studies also showed that physician-assisted death is very uncommon in nursing homes, in part because patient competence to make decisions is required. Despite this limitation, the widespread acceptance of physician-assisted death in the Netherlands undoubtedly is accompanied by a common attitude that life-sustaining treatment may be foregone.
Furthermore, discussions by the authors with Dutch nursing home physicians suggest that they use a much broader interpretation of medical futility than in the United States. Consistent with this impression, Dutch nursing home physicians responding to a survey on decision making concluded that 30% of 228 residents diagnosed with pneumonia did not have a medical indication for curative treatment (van der Steen, 2002). In contrast, in the United States, medical futility, although extensively discussed, has not found wide applicability in decision making (Helft, Siegler, & Lantos, 2000). Codes of ethics declare that physicians do not have a duty to render futile care, but there is little consensus on what constitutes futility.
Turning to nursing home organization, a recent international comparison found the Netherlands had 325 nursing homes with 53,800 beds (26 beds per 1,000 elderly persons; Ribbe et al., 1997), about half the U.S. prevalence of 53 beds per 1,000 elderly persons. Units are divided into somatic and psychogeriatric, the latter mostly residents with dementia. Costs per bed are 1
times the U.S. cost (Ribbe et al., 1997), and Dutch residents are much less commonly hospitalized than in the United States (Frijters et al., 1997). A smaller number of nursing home beds and a higher cost per bed reflect that Dutch nursing homes tend to admit individuals with skilled nursing needs (somatic units) or severe behavioral or functional problems from dementia (psychogeriatric units). In the Netherlands, lesser institutional care needs are provided for in "homes for the aged," which encompass some individuals who would be in U.S. nursing homes, as well as many who might reside in residential care facilities in the United States. Furthermore, in the Netherlands, on-site physicians (one full-time equivalent physician per 100 residents) are specialists in nursing home medicine with a distinctive postgraduate training program (Hoek, Penninx, Ligthart, & Ribbe, 2000). Having physicians constantly in nursing homes means that they will often have ample opportunity to get to know residents and their families and to develop these relationships over time. This continuing contact may facilitate decision making at the time of illness and lead, in at least some cases, to advance orders for palliative treatment of infections. In contrast, few U.S. nursing homes have physicians on-site for other than occasional visits.
These large differences between the United States and the Netherlands in approach to end-of-life care and organization of nursing home care suggest that there should be significant differences in decision making about and treatment of LRIs in nursing home residents with dementia. The Dutch Pneumonia Study—conducted on psychogeriatric nursing home units and further discussed in the current paper—found that 23% of residents with pneumonia had antibiotics withheld for palliative reasons (van der Steen, Ooms, Adér, Ribbe, & van der Wal, 2002). These residents had extraordinarily high mortality (90% at 30 days). Comparing residents with dementia and an LRI in each country, the treatments they receive, and associated mortality outcomes has the potential to clarify how different approaches to care affect dementia patients at the end of life.
|
Data and Methods |
|---|
|
TOPAbstractPalliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
The Dutch Pneumonia Study prospectively identified subjects from psychogeriatric units in 61 nursing homes located throughout the country. Between October 1996 and July 1998, nursing home physicians diagnosed 706 consecutive residents with pneumonia who also met the following criteria: (1) psychogeriatric disease (97% explicitly had a dementia diagnosis) and (2) presence in the nursing home for at least 4 weeks. Diagnosis was on the basis of physicians' clinical judgment and usually did not include consideration of radiographic or laboratory data. A resident could be included only once even if a second episode of pneumonia occurred during the study period. The study protocol was approved by the Medical Ethics Committee of the VU University Medical Center.
In the Missouri LRI Study, 1,406 episodes of LRI, predominantly pneumonia, in 1,044 residents, were prospectively identified in subjects residing in 36 nursing homes in central Missouri and the St. Louis area between August 1995 and September 1998. Subjects were required to be in the nursing home at least 2 weeks before the illness and to be at least 60 years old. Because the Missouri LRI Study was primarily focused on outcomes, we excluded the small number of residents who had an explicit "no antibiotics" order from their attending physicians: 40 of 2,790 exclusions (1.4%).
Trained project nurses evaluated Missouri residents who had symptoms compatible with an LRI, including either respiratory or general symptoms. The study definition of LRI, which includes pneumonia and other LRI, was a modification of a surveillance definition for long-term care facilities (McGeer et al., 1991). The definition incorporated six signs and symptoms specific for respiratory problems (e.g., new cough, new sputum production) or general illness indicators (e.g., fever, acute decline in cognitive function) and chest radiographic findings if available. Most evaluations included a chest radiograph, complete blood count, and a chemistry panel. Details of identification, evaluation, and the precise LRI definition are described elsewhere (Mehr et al., 2001a, 2001b). Institutional Review Boards at the University of Missouri-Columbia Health Sciences Center and Washington University School of Medicine approved the protocol. Additionally, several hospital Institutional Review Boards conducted their own evaluations before granting permission to review medical records, and two nursing homes had ethical review panels that approved the project.
Data Collection and Measures
In both studies, extensive data were collected at initial enrollment, and residents were followed for 3 months. In the Dutch study, enrollment was at the time of the treatment decision by the nursing home physicians and, in the U.S. study, at the time of the resident's evaluation by a project nurse. Hereafter, in both cases, we will refer to this point as "the time of diagnosis." In the Dutch study, nursing home physicians recorded all data at specified intervals. At the time of diagnosis, this included specific diagnoses, treatments, and whether residents were being treated without antibiotics with palliation as the goal of therapy, with antibiotics but still with palliation as the goal, or with antibiotics with cure as the goal. The physicians also indicated their ratings for the Bedford Alzheimer Nursing Severity Scale (BANS-S) 2 weeks earlier (Volicer, L., Hurley, Lathi, & Kowall, 1994).
In the Missouri study, project nurses recorded clinical and patient assessment data, including collecting a new nursing home MDS Version 2.0 (Health Care Financing Administration, 1995). From the MDS, we obtained activities of daily living (ADLs) data and the Cognitive Performance Scale (CPS; Morris et al., 1994), which measures cognitive impairment. The Missouri investigators also abstracted information from medical records, including hospital records if the resident were hospitalized, to identify diagnoses, treatments, and medications. All antibiotic use between 30 days before and 30 days after the time of diagnosis was recorded. Radiology reports were rated as negative, possible, or probable for pneumonia using defined criteria. A brief survey of participating attending physicians was used to collect provider data, such as age and clinical training. Facility size was identified from public documents. The prevalence of advance directives during late 1997 was determined in a systematic sample of 25% of records in each facility.
To create a common ADL scale, we equated the four-level ADL items of the BANS-S and the five-level MDS self-performance scale by combining the middle categories to create three-level scales distinguishing independence (0), need for assistance (1), or full dependency (2). Using eating, walking, and dressing—ADL items available in both instruments—we created a summary ADL scale with a range of 0–6 by summing the recoded scores for the individual items. We also used this ADL scale in developing a common definition for severe dementia in both U.S. and Dutch subjects. For Missouri subjects, we required a CPS of 5 or 6, combined with an ADL score of 4 or more. We attempted to equate CPS scores and BANS-S scores using a small sample of Dutch subjects in another study who had an MDS and a BANS-S completed at the same time (D. L. Gerritsen, unpublished data). We defined severe dementia in the Dutch subjects as having a BANS-S of greater than 15 and meeting the same ADL standard of a score of 4 or more on the summary scale. Thus, all residents in both countries classified as having severe dementia were expected to have profound cognitive deficits and significant ADL impairment, including full dependency in at least one ADL.
Although data collection approaches and instruments clearly differed in the two studies, there were many common variables collected that permit joint analysis. These variables included age, gender, illness symptoms and signs (e.g., cough, decreased alertness, fever, respiratory rate, pulse), data on physical functioning (e.g., three ADLs, continence, swallowing problems), chronic diseases, and most treatments. Furthermore, even where directly comparable data do not exist, frequently data can be compared on a conceptual level. For example, although chest x-rays were rarely performed in the Netherlands, we report results for U.S. residents to shed light on treatment decisions in U.S. residents in comparison with the clear evidence that residents with dementia who are more severely ill are less likely to receive antibiotics in the Netherlands (van der Steen et al., 2002).
Analyses
We first report data on several treatments received in each country, including lack of antibiotic treatment, type of antibiotic used, if any, and hospitalization. From the Dutch data, we report antibiotics prescribed after diagnosis for treatment of the pneumonia, if any. Nontreatment with antibiotics was clearly indicated to be for palliative purposes. In the Missouri data, we lacked specific data on the intent of antibiotic treatment or nontreatment. Therefore, we defined antibiotic treatment for the LRI as any antibiotic prescribed from 2 days before diagnosis until 30 days after diagnosis. We reasoned that this would most clearly capture intent to treat infections with antibiotics.
We then performed bivariable analyses to describe differences in groups receiving antibiotics and not receiving antibiotics, as well as mortality within each country. We expected factors associated with lack of antibiotic use to differ substantially between the two countries. We also performed cross-national comparisons for these variables. T-tests were used to test differences in means for age and respiratory rate. Chi-square statistics were computed to test for differences in proportions (all other variables). Multivariable analyses in U.S. subjects used logistic regression with no antibiotic treatment as the dependent variable. We included as potential variables for inclusion resident characteristics, indicators of illness severity, provider characteristics, and facility characteristics. We confirmed our results treating nursing home as a random effect in a mixed model to address the clustering of residents within nursing homes. Because these results were almost identical, the simple logistic model is reported. Multivariable analyses for 635 of the Dutch subjects have previously been reported (van der Steen et al., 2002) and are related to new U.S. analyses in the "Results" and "Discussion" sections.
|
Results |
|---|
|
TOPAbstractPalliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
|
|
Figure 1 shows mortality by antibiotic treatment. Among the 23% of Dutch subjects not treated with antibiotics, 90.2% died within 30 days; mortality was 29.8% (p <.0001) in those treated with antibiotics. In contrast, 85% of U.S. subjects with dementia received antibiotics, and 30-day mortality was virtually identical for both those treated with and without antibiotics (16.7% and 17.5%, respectively, p =.85).
|
|
|
Discussion |
|---|
|
TOPAbstractPalliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
Over a century ago, Sir William Osler wrote, "Pneumonia may well be called the friend of the aged. Taken off by it in an acute, short, not often painful illness, the old man escapes those ‘cold gradations of decay’ so distressing to himself and to his friends" (Osler, 1899, p. 109). In Osler's day, care for pneumonia was only supportive, but today it can include antibiotics and even admission to an intensive care unit. Such options mandate careful consideration of when aggressive care is no longer indicated.
We found striking differences in management among residents with dementia who had an LRI in U.S. and Dutch nursing homes. Hospitalization and parenteral antibiotics were very common in the United States, although rare in the Netherlands. Whereas nontreatment with antibiotics was moderately more common in the Netherlands than in the United States, there were major differences between the two countries in who was treated and not treated, as well as in mortality.
Among Dutch subjects with a clinical diagnosis of pneumonia not treated with antibiotics, there was evidence of more preillness disability, more significant acute illness, and much higher mortality than those who received antibiotics. In contrast, U.S. subjects not treated with antibiotics had very similar mortality to those treated with antibiotics and frequently appeared to have a more equivocal or less severe acute illness. Although probably some U.S. subjects were deliberately not treated for palliative reasons, it is difficult to ascertain how many such individuals there might have been, and, given the low overall mortality, the number was probably small. For example, among the 280 individuals in the United States with severe dementia, only 55 (20%) did not receive treatment with antibiotics. In multivariable results from the United States, even residents with severe dementia were more likely to receive antibiotics if they exhibited clearer evidence of significant acute illness; such residents appeared less likely to receive antibiotics in the Dutch multivariable analysis.
These important differences undoubtedly reflect organizational differences in care in the two nations, as well as different cultural contexts for end-of-life decisions. In the Netherlands, having physicians in the nursing home all the time provides many more opportunities for families and physicians to discuss and plan care. In North Carolina, family members of deceased nursing home residents frequently reported dissatisfaction with physician communication (Hanson et al., 1997). Undoubtedly, one reason is relatively infrequent visits to nursing homes by most physicians in the United States. Many acute febrile illnesses are treated by physicians without in-person evaluation (Burton et al., 2001). Infrequent contact with families and lack of consensus standards for initiating discussions of treatment limits, such as for hospitalization or antibiotic therapy, make understandable the low prevalence of orders limiting such care. Furthermore, in the midst of a busy office or hospital practice, U.S. physicians have every incentive not to initiate discussions of treatment limitations at the time of an acute illness. Thus, indirectly, our findings emphasize the importance of future efforts to improve communication in the United States about end-of-life care in the nursing home. This conclusion is convergent with the finding of Travis and colleagues (2002) that the highest level barrier to initiating palliative care was a failure to identify the poor prognosis for rehabilitation or prolonged survival.
Our work is subject to several limitations. First, the Dutch study included residents throughout the Netherlands, but the Missouri LRI Study included only central Missouri and the St. Louis area. Whereas the Missouri subjects and nursing homes were similar to national samples (Mehr et al., 2001b), findings should be extrapolated with caution. In particular, nursing home residents in Oregon have low hospitalization rates at the end of life (Tolle et al., 1998), which is consistent with a different approach to palliative care.
A second limitation concerns subject comparability across the two studies. Residents of psychogeriatric units in the Netherlands overall have more severe dementia than the residents studied in the United States, and the Dutch study used the physician's clinical impression of pneumonia as an entry criterion rather than specific diagnostic criteria. This approach likely led to inclusion of residents with clearer illness who may have been more severely ill. However, it is not likely many Dutch residents did not have an LRI. Even making the conservative assumption that missing information for a symptom meant its absence, the vast majority of Dutch subjects met the Missouri LRI Study definition (82% and 84% of those treated and not treated with antibiotics, respectively; van der Steen, 2002). Moreover, subject differences are unlikely to account for the divergence we found in how illness severity relates to the choice of treatment with antibiotics or not. Magnitudes may be biased, but differences in direction of effect should not occur. Differing illness severity certainly influences mortality differences, and further work is in process to address this issue for residents treated with antibiotics. For residents not treated with antibiotics, the cross-national differences are so extreme that they are unlikely to be accounted for just by differing illness severity, and illness severity is not related to treatment choice in the same way (i.e., different direction of effect). This difference makes attempts to adjust for disease severity problematic.
A final limitation is that we did not collect direct data on physician decision making in Missouri. We recorded antibiotics prescribed in either the nursing home or the hospital, but we could not determine how many residents received limited treatment because the physician wanted to provide only palliative treatment. We also acknowledge the possibility that some use of antibiotics occurred even where palliative care was intended. Nonetheless, to the extent that many U.S. physicians regard antibiotic treatment as nearly compulsory for severe infections, this perspective suggests a significant problem in the transition to palliative care. The care patterns documented in this paper clearly demonstrate very limited use of strictly palliative care in the United States in residents with dementia who have an LRI.
In conclusion, nursing home residents with dementia in the Netherlands are treated much less aggressively than in Missouri, and withholding antibiotic treatment for severely ill residents is more common in the Netherlands. Although several factors are likely involved, these findings suggest problems in the transition to palliative care for U.S. nursing home residents. Further work on how such transitions occur and how they can be facilitated is needed.
|
Footnotes |
|---|
1 Department of Family and Community Medicine, University of Missouri-Columbia, Columbia, MO.
2 Institute for Research in Extramural Medicine (EMGO) and the Departments of Nursing Home Medicine and Social Medicine, Vrije Universiteit (VU) University Medical Center, Amsterdam, the Netherlands.
3 Institute for Research in Extramural Medicine (EMGO) and the Department of Nursing Home Medicine, VU University Medical Center, Amsterdam, the Netherlands.
4 Sinclair School of Nursing, University of Missouri-Columbia, Columbia, MO.
Decision Editor: Laurence G. Branch,, PhD
Received for publication May 14, 2002. Accepted for publication October 28, 2002.
|
References |
|---|
|
TOPAbstractPalliative Care for Advanced...Data and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. R. Helton, J. T. van der Steen, T. P. Daaleman, G. R. Gamble, and M. W. Ribbe A Cross-Cultural Study of Physician Treatment Decisions for Demented Nursing Home Patients Who Develop Pneumonia Ann. Fam. Med, May 1, 2006; 4(3): 221 - 227. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. L. Kruse, D. R. Mehr, J. T. van der Steen, M. E. Ooms, R. W. Madsen, A. K. Sherman, R. B. D'Agostino, G. van der Wal, and M. W. Ribbe Antibiotic Treatment and Survival of Nursing Home Patients With Lower Respiratory Tract Infection: A Cross-National Analysis Ann. Fam. Med, September 1, 2005; 3(5): 422 - 429. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Tuch Commentary on "Lower Respiratory Infections in Nursing Home Residents With Dementia: A Tale of Two Countries": Dying in the Nursing Home Gerontologist, April 1, 2003; 43(90002): 94 - 95. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
