If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
High-quality care at the end of life supports freedom from pain and other potentially burdensome symptoms. Lowering symptom burden at the end of life is an urgent and achievable goal in delivering services in nursing home settings. Few published reports describe symptom burden among older adults in nursing homes; none examine links between symptom burden and modifiable features of nursing home organizational context (work environment).
Objectives
To examine the influence of organizational context on symptom burden and to compare symptom burden in the last year of life between nursing home residents with and without dementia.
Design
Retrospective analysis of longitudinal survey data.
Setting
A stratified random sample of 36 nursing homes in the Canadian provinces of Alberta, Manitoba, and Saskatchewan.
Participants
A total of 2635 residents with dementia and 1012 without dementia; 1381 front-line care staff.
Measurements
(1) Trajectories of 6 symptoms (dyspnea, pain, pressure ulcers, urinary tract infections, challenging behavior, delirium), assessed with the Resident Assessment Instrument-Minimum Data Set, version 2.0, between 2008 and 2012. All residents received assessments in each quarter of the year before death. (2) Modifiable organizational context, assessed with the Alberta Context Tool. Hierarchical mixed model, repeated measures regression, to simultaneously evaluate effects of time, dementia, and context on symptom trajectories.
Results
For all residents, prevalence of symptoms increased over time. In the last quarter before death, challenging behavior was the most frequent symptom in the dementia group (40.2%), delirium the most frequent symptom in the nondementia group (31.0%), and urinary tract infections least frequent (9.0% to 10.0%). Facilities with more favorable context had significantly higher prevalence of challenging behavior and delirium and significantly lower use of antipsychotics without diagnosis of psychosis.
Conclusion
Symptom burden increases as the end of life approaches but differs between high- and low-context facilities and between residents with and without dementia. Trajectories of treatable, burdensome symptoms at the end of life in nursing homes should be a priority focus for quality improvement. Modifiable features of organizational context that are linked to symptom burden offer new potential strategies and interventions for quality improvement.
Symptoms from diseases and chronic conditions put a significant physiologic and psychological burden of suffering on the rapidly growing population of seniors, particularly at the end of life. Of Canada's population of 35 million, 350,000 adults older than 65 years live in a special care facility (nursing home, chronic care, long-term care hospital, residence for seniors) at any given time.
Statistics Canada Living Arrangements of Seniors: Families, Households and Marital Status. Structural Type of Dwelling and Collectives, 2011 Census of Population.
As is typical of many developed countries, most residents will die in the nursing home. The most comprehensive work on burden of symptoms at the end of life was conducted by Mitchell and colleagues in the United States.
Dying in US nursing homes with advanced dementia: How does health care use differ for residents with, versus without, end-of-life medicare skilled nursing facility care?.
They reported that during the last year of life, the proportion of nursing home residents with dyspnea and pressure ulcers more than tripled to more than 30%, pain more than doubled to more than 25%, and the prevalence of aspiration increased from almost 10% to almost 40%.
These are unacceptably high rates of modifiable symptoms. They cause unnecessary distress and significantly increase health care costs. Strategies to reduce suffering and address this quality and cost issue must be based on better knowledge of modifiable symptom burden, and of the links between symptom burden and modifiable features of nursing home work environments.
Over the past decade, a key factor in increasing symptom burden is individuals entering nursing homes at more advanced ages and stages of disease trajectories, in particular the trajectory of dementia.
As the population ages, the prevalence of dementia also rises. At least two-thirds of residents in Canadian nursing homes have a diagnosis of dementia.
A number of researchers have studied single symptoms at the end of life for nursing home residents with dementia, but we identified only 10 studies addressing the broad range of symptoms that cause burden over time.
Dying in US nursing homes with advanced dementia: How does health care use differ for residents with, versus without, end-of-life medicare skilled nursing facility care?.
was the first to provide a comprehensive picture of symptom burden in advanced dementia at the end of life, based on a sample of 323 US nursing home residents.
Characteristics of an organization's context (work environment), such as leadership, culture, feedback mechanisms, and available information resources, are increasingly seen as important for quality and safety in health care.
These factors are modifiable. However, we know relatively little about the role of organizational context in the nursing home environment in general, and in end-of-life care specifically. We used data collected in the longitudinal Translating Research in Elder Care (TREC) study
conducted in the Canadian provinces of Alberta, Manitoba, and Saskatchewan. These data permitted a unique focus on the influence of nursing home organizational context on distressing symptoms in residents at the end of life.
Our research objectives were to describe the trajectory of symptom burden at the end of life in a cohort of nursing home residents, and to determine if organizational context or dementia status influenced symptom prevalence. These are necessary first steps to improving quality and safety of care at the end of life in residential long-term care.
Methods
Our sample was 36 nursing homes in the 3 Canadian provinces that participated in the TREC (protocol described elsewhere).
Study protocol for the translating research in elder care (TREC): Building context—an organizational monitoring program in long-term care project (project one).
We measured organizational context with the Alberta Context Tool (ACT), a core component of a survey questionnaire administered to 1381 front-line care staff using structured in-person interviews with computer assistance.
Study protocol for the translating research in elder care (TREC): Building context—an organizational monitoring program in long-term care project (project one).
To construct trajectories of symptom burden, we used RAI-MDS 2.0 assessments (2008–2012). The RAI-MDS 2.0 has been examined extensively for reliability and validity, and is appropriate for research purposes (allowing for some weaknesses, notably around depression, which is not a focus here).
From the assessments, we built an analytic dataset of residents assessed in at least 4 quarterly periods before death. Date of death was determined by the administrative discharge record of the RAI-MDS 2.0 and could have occurred in the nursing home or in hospital. Nursing homes complete assessments only once in any calendar quarter, thus time varied between a resident's final assessment and death.
We extracted measures of selected symptoms in the 7-day period before assessment, unless otherwise noted. Dyspnea was recorded as present or not. Pain was recorded as daily pain that was at least moderate at its worst or that was horrible or excruciating at any time. The assessor used both resident reports of pain and other cues for residents not verbally communicative. Pressure ulcers of stage 2 or higher were recorded (partial loss of thickness in skin layers presenting clinically as an abrasion, blister, or shallow crater). Urinary tract infections in the 30 days before assessment were recorded. Challenging behavior was recorded as verbally abusive, physically abusive, or socially inappropriate/disruptive behavior on at least 1 day in the past week. Delirium was measured with 6 items (easily distracted, altered perception, disorganized speech, restlessness, lethargy, mental function varies over the day). Delirium was recorded if (1) any item was present and over the past 7 days appeared different from usual function or (2) any item was present and not of recent onset but differed from the previous assessment. Antipsychotic use without a diagnosis of psychosis was recorded if the resident either had no diagnosis of Huntington disease or schizophrenia, or had not experienced hallucinations in the past 7 days. A resident was classified as having dementia if either or both of the RAI-MDS 2.0 items, Alzheimer's disease or dementia other than Alzheimer's disease, were recorded at any time in the resident's assessment history.
Organizational Context
The ACT is a validated survey instrument for 10 modifiable elements of organizational context: (1) leadership, (2) culture, (3) evaluation (feedback processes), (4) formal team interactions, (5) informal team interactions, (6) social capital, (7) structural and electronic resources, (8) organization slack—time, (9) organization slack—staff, and (10) organization slack—space.
All items are scored on a 5-point Likert agreement scale (strongly disagree to strongly agree). The original validation was completed with 752 pediatric acute care nurses
and provided evidence for acceptability, internal consistency reliability (Cronbach α ≥0.70 for 10 of 13 concepts), and validity. We found statistically significant correlations between instrumental research utilization and all but one ACT concept. The psychometric testing of the nursing home version was based on responses from 645 care aides.
Results of confirmatory factor analyses were consistent with the factor structure theorized in the development of the ACT. For 8 ACT concepts, we found significant correlations with instrumental research utilization; internal consistency reliability (Cronbach's α ≥0.70 for 8 of 10 concepts) and acceptability (there was minimal missing data with 93.5% of the health care aides providing complete data on all ACT items and the time to complete the ACT survey was below target with a mean of 11.08 minutes and an SD of 2.93 minutes) were confirmed.
we used a hierarchical cluster analysis to aggregate by facility the 10 ACT measures captured. We applied k-means clustering to assign facilities as either low or high context, with higher-context facilities having more favorable work environments.
Analysis
To compare the various measures by time (4 periods), dementia (yes/no), and facility context (high/low), we carried out a mixed model, repeated measures regression to nest residents within facilities and facilities within level of context. In addition to the analytic cohort that was assessed in at least 4 consecutive quarters, we identified additional residents who were assessed in 1 to 3 quarters before death. We compared their last assessment to that of the 4-quarter cohort. SAS version 9.4 (SAS Institute, Inc, Cary, NC) was used for all analyses.
Ethics Approval and Informed Consent
Ethical approvals for TREC were obtained from the University of Alberta Health Research Ethics Board (File B-051007), University of Calgary Conjoint Health Research Ethics Board (Ethics ID E-21379), University of Saskatchewan Behavioural Research Ethics Board (BEH 08–165), and University of Manitoba Fort Garry Campus Research Ethics Board (Protocol E2008:010). Operational approvals were obtained from all relevant health care organizations.
Results
Characteristics of Residents in Low- and High-Context Facilities
A total of 3647 nursing home residents were assessed in at least 4 quarters before being discharged due to death (Table 1). We found small but statistically significant differences between the cohorts from high- and low-context facilities. Residents of high-context facilities had somewhat longer average length of stay, more unstable health and aggressive behavior, lower prevalence of a cancer diagnosis, and higher prevalence of dementia.
Table 1Characteristics of Nursing Home Residents at Last Assessment Before Death (n = 3647)
Characteristic
All Residents
Residents in Low-Context Facilities
Residents in High-Context Facilities
P Value (χ² or t test)
Sample size, n
3647
1901
1746
Age at death, y
88.0
88.2
87.7
.09
Length of stay at death, y
3.5
3.3
3.7
<.001
% Female
65.8
66.7
65.0
.28
Cognitive Performance Scale 4+
55.9
56.2
55.7
.76
Activities of Daily Living long 15+
82.0
82.1
81.8
.79
Depression Rating Scale 3+
37.8
36.6
39.2
.11
Changes in Health, End stage disease, and Symptoms and Signs (CHESS)
The CHESS scale32 is a measure to assess the severity and instability of a resident’s situation, and the risk of serious decline. It has been demonstrated to be a strong predictor of mortality and to be strongly associated with physician activity, complex medical procedures, and pain. It uses items measuring a recent decline in cognition or physical independence, end-stage disease, and symptoms and signs of medical problems.
is a measure to assess the severity and instability of a resident’s situation, and the risk of serious decline. It has been demonstrated to be a strong predictor of mortality and to be strongly associated with physician activity, complex medical procedures, and pain. It uses items measuring a recent decline in cognition or physical independence, end-stage disease, and symptoms and signs of medical problems.
† Verbally abusive, physically abusive, socially inappropriate or disruptive, or resisting care.
Symptom Burden in Residents With and Without Dementia
The prevalence of burdensome symptoms was generally high. Overall and in the dementia group, the most frequent symptom in the last quarter of life was challenging behavior, recorded in 40.2% of the residents with dementia and 20.6% of the residents without dementia. Delirium was the most frequent symptom in the nondementia group (31.0%) and had a similar prevalence in the dementia group (29.2%). The least frequent symptom, overall and in each group, was urinary tract infection: 9.1% in the dementia group and 10.0% in the nondementia group. Figure 1 shows the average prevalence of 5 measured symptoms and of the use of antipsychotics for each of the last 4 quarters of life. Statistical significance is noted for dementia status, organizational context, and time, which we controlled for simultaneously in the model. Pressure ulcer prevalence is not included in Figure 1, as it was not significantly associated with context.
Fig. 1Symptoms in the last 12 months of life in high- and low-context settings and among residents with and without dementia.
Except for challenging behavior, the prevalence of all symptoms for residents with or without dementia significantly increased as death approached. Persons without dementia experienced a higher prevalence of dyspnea, whereas persons with dementia experienced higher prevalence of challenging behavior and use of antipsychotics. Pressure ulcer prevalence increased over time in both the dementia and nondementia groups, from 4.6% to 9.5% and from 5.3% to 10.7%, respectively (PDem = .038, Ptime < .001).
Symptom Burden in Residents of Low- and High-Context Facilities
Figure 1 highlights the association of organizational context with symptom burden of residents in their last year of life. The prevalence of dyspnea, pain, and urinary tract infections is significantly lower in higher-context facilities. However, higher-context facilities have higher prevalence of challenging behavior and delirium. In higher-context facilities, the lower use of antipsychotics without a diagnosis of psychosis is statistically significant.
Symptom Burden in Residents With Fewer Than 4 Quarterly Assessments
Our measures of symptom burden preceding death may be conservative because our analysis included only residents assessed in at least 4 quarters before death (n = 3467). We compared those residents to residents assessed in 1 to 3 quarters (n = 2562) before death. Residents assessed in 4 quarters showed lower prevalence of pain (22.2% vs 26.8%, P < .0001) and pressure ulcers (9.8% vs 12.1%, P = .005) but higher prevalence of challenging behavior (34.8% vs 30.4%, P < .0001). Rates of the other 3 symptoms did not differ significantly.
Discussion
The cornerstone of high-quality care at end of life is ensuring that persons are free of pain and other potentially burdensome symptoms.
Our study reveals a clear trend toward worsening symptom burden in the last year of life for residents of nursing homes, regardless of dementia status. However, symptom prevalence differs remarkably between persons with and without dementia. Residents without dementia experience higher prevalence of pain, dyspnea, and pressure ulcers, whereas residents with dementia have higher prevalence of challenging behaviors.
An important finding in this study is the higher prevalence in higher-context facilities of responsive behaviors of dementia, combined with lower prevalence of antipsychotic use. Lower antipsychotic use can be expected to result in higher levels of responsive behaviors of dementia. Lowering antipsychotic use for residents not diagnosed with psychosis is a widespread imperative internationally.
Current high levels of antipsychotic medication use in nursing homes are disturbing because antipsychotics present known risks related to stroke and death, can mask symptoms of a treatable delirium, make clinical assessment of symptoms difficult, reduce residents' ability to communicate, and reduce residents' mobility.
In higher-context facilities, responsive behaviors may be more likely to be detected, correctly specified, documented, and appropriately managed. These facilities often have, for example, stronger patterns of staff engagement or higher rates of best practice use and thus be better prepared to deal with increased rates of responsive behaviors by applying appropriate nonpharmacologic behavior management strategies.
Strengths and Limitations of the Study
Our analysis was retrospective and focused on a sample of residents assessed in at least 4 quarters before death. Our study exploits the temporal dimension of the RAI-MDS 2.0 data to cover a full year before death. However, comparisons with symptom burden of residents assessed only in quarterly periods closer to death indicate that our primary analysis may slightly underestimate symptom prevalence overall. Our data source, the RAI-MDS 2.0, may also underreport symptoms. Its assessments are completed only quarterly and rely in part on the clinical record, which may underdocument. Reliability of this tool may be influenced by the variety of care staff who complete each assessment and their varied training.
Some symptoms for individuals with dementia are likely to be underdetected, especially as death approaches, because those with advanced dementia experience severe communication difficulties. For example, the recorded prevalence of dyspnea and pain is significantly higher in the nondementia group, suggesting that individuals with dementia require higher levels of vigilance for symptoms.
reported the progress of burdensome symptoms for nursing home residents in the 4 quarters before death, and only for residents with dementia. They report similar significant increases of burdensome symptoms (except for challenging behavior) as the end of life approaches, but report symptom prevalence of 5% to 20% higher than the dementia population in our study. One probable explanation for these differences lies in the different samples. We included persons with dementia of all stages, whereas Mitchell et al
focused on persons with severe dementia, who may have a higher prevalence of burdensome symptoms because of the severity of their disease. In addition, Mitchell et al
collected prospective data from chart reviews, interviews with nurses, and brief physical examinations of residents, while we retrospectively analyzed data collected by the RAI-MDS 2.0 completed by care staff. Mitchell et al
also included individuals with fewer than 4 quarters of observation in the near to death quarters, different from our consistent 4-quarter cohort. Therefore, our study may slightly underestimate prevalence of burdensome symptoms. Three additional studies
conducted in Europe and the United States compared symptom burden at the end of life between nursing home residents with and without dementia. Comparability to our study is limited, as those studies examined different populations (residents with dementia versus residents with terminal cancer,
) or used different assessment instruments and methods.
Practice and Policy Implications
Nursing homes play an increasingly central role in end-of-life care for older adults with moderate and advanced dementia. Nursing home care is by definition end-of-life care
and new policy should recognize and encourage the important role of nursing homes in determining whether comfort and dignity characterize the prolonged period of dying that is typical of dementia. Recognizing dementia as a chronic and terminal disease is a key element in such policies, as is focusing health care resources on optimizing end-of-life care.
Making room for dying: End of life care in nursing homes.
in: Jennings B. Kaebnick G.E. Murray T.H. Improving End of Life Care: Why Has It Been So Difficult? A Hastings Center special report. The Hastings Center,
Garrison, NY2005: S37-S41
Specifically, we must address issues such as (1) achieving meaningful person-centered care; (2) achieving adequate communication with persons with dementia so that they can express burdensome symptoms, other care needs, and preferences; (3) appropriate and timely surrogate decision making; (4) avoiding unwarranted treatment and transfer to acute care facilities and/or treatment with no prospect of improving quality of life
; and (5) optimizing end-of-life care for persons with dementia. These areas should be integral to care providers' education and training, central in policy formation, and central in interactions with family and other unpaid care givers.
However, recognizing and appropriately managing treatable symptoms as a foundational component of end-of-life care is more fundamental than policy. It goes to the core of the mission and values of individual and grouped nursing homes, to the core of professional practice, and to the responsibility of providers to ensure that professional and nonprofessional front-line workers have both the necessary knowledge and skills, and the conditions under which to use them. Consistent with these things, appropriately managing treatable symptoms should, we argue, be guaranteed.
A key and encouraging finding of this study is that modifiable characteristics of organizational context are associated with modifiable symptom burden. Previous work described the association between quality of care and staffing patterns,
However, no reports have focused explicitly on associations between organizational context and symptom burden or trajectories of symptom burden in the last year of life. We measured modifiable features of an organization's context, such as leadership, data feedback processes, staff interaction patterns, and resources. Our results offer a new avenue for intervention strategies, in addition to the usual strategies for reducing and ameliorating distressing symptoms, such as staff education. Using these contextual data, we have worked with nursing homes to first diagnose a performance gap: which of the 10 ACT elements are producing the greatest dampening effect on context scores, and on which units within a facility is this occurring? We have then advised on improvement strategies to address those gaps. Focusing on measurable and modifiable elements of organizational context in nursing homes is one promising direction to improving a nursing home's ability to manage symptom burden at the end of life.
Conclusion
Canadian seniors living in nursing homes, particularly those with dementia, are receiving suboptimal quality of care in the last year of life because treatable, distressing symptoms are not well addressed. As a result, these older adults are living and dying with unnecessary pain and suffering. Many factors combine to influence management of distressing symptoms at the end of life. Improving organizational context in nursing homes, by optimizing elements of context, such as leadership styles, communication patterns, data feedback, and resources, has the potential to improve symptom trajectories in the last months of life and thus reduce pain and suffering.
Acknowledgments
The authors acknowledge the TREC team for contributions to the larger TREC program. The TREC team (2007–2012) included Carole A. Estabrooks (PI); investigators Greta G. Cummings, Lesley Degner, Sue Dopson, Heather Laschinger, Kathy McGilton, Verena Menec, Debra Morgan, Peter Norton, Joanne Profetto-McGrath, Jo Rycroft-Malone, Malcolm Smith, Norma Stewart, and Gary Teare; decision-makers Caroline Clarke, Gretta Lynn Ell, Belle Gowriluk, Sue Neville, Corinne Schalm, Donna Stelmachovich, Gina Trinidad, Juanita Tremeer, and Luana Whitbread; and collaborators David Hogan, Chuck Humphrey, Michael Leiter, Charles Mather, Judy Birdsell, Phyllis Hempel (deceased), J. Ivan (Jack) Williams, and Dorothy Pringle (Chair, Scientific Advisory Committee).
References
Statistics Canada
Living Arrangements of Seniors: Families, Households and Marital Status. Structural Type of Dwelling and Collectives, 2011 Census of Population.
Dying in US nursing homes with advanced dementia: How does health care use differ for residents with, versus without, end-of-life medicare skilled nursing facility care?.
Study protocol for the translating research in elder care (TREC): Building context—an organizational monitoring program in long-term care project (project one).
Making room for dying: End of life care in nursing homes.
in: Jennings B. Kaebnick G.E. Murray T.H. Improving End of Life Care: Why Has It Been So Difficult? A Hastings Center special report. The Hastings Center,
Garrison, NY2005: S37-S41
This study was funded by a Canadian Institutes of Health Research (CIHR) grant-in-aid (MOP 53107). The funding agency had no role in the study design, data collection, data analysis, interpretation of the data, writing of the report, or decision to submit the article for publication.
Carole A. Estabrooks is the copyright owner for the Alberta Context Tool. The Alberta Context Tool is a royalty-free instrument distributed without financial interests.
Comprehensive geriatric assessment is not a novel concept and has been extensively demonstrated to be effective in a variety of inpatient and outpatient settings since the 1980s.1–3 A key feature that accounts for the effectiveness is the comprehensiveness, which means a holistic or multicomponent approach when assessing and/or managing the older patient. A familiar challenge exists, however, in attributing the benefits of individual components within the context of a multicomponent intervention.
00152-8&id=gr1.jpg){kind=link}