Place of Death, Care Intensity, Symptom Prevalence, and Symptom Management for Patients with Stroke at End-of-Life

Abstract

BACKGROUND:

Each year in the United States, 795,000 people are diagnosed with a new or recurrent stroke, resulting in 165,393 deaths. However, little is known about the quality of end-of-life (EOL) care in this population, including the place of death, care intensity received, symptom experience, and management. This study sought to explore whether place of death and care intensity received in the last month of life were correlated with EOL symptom prevalence and management in patients with stroke.

METHODS:

Data for this secondary analysis were drawn from the National Health and Aging Trends Study, a representative sample of Medicare beneficiaries. Logistic regression was used to identify whether place of death and care intensity received in the last month of life were associated with symptom prevalence and symptom management in patients diagnosed with stroke, adjustive for demographic covariates.

RESULTS:

Of patients diagnosed with stroke (n=257), 161 (62.6%) reported pain, 130 (50.6%) reported dyspnea, and 148 (57.6%) reported emotional distress in the last month of life. Patients who died in the acute-care setting were half as likely to report pain (OR 0.51, 95% CI 0.27–0.95), but twice as likely to report dyspnea (OR 2.07, 95% CI 1.02–4.21) and adequate dyspnea management (OR 1.97, 95% CI 1.06–3.67). Place of death was not significantly associated with emotional distress (OR 1.62, 95% CI 0.88–2.97), pain management (OR 0.67, 95% CI 0.39–1.15), or emotional distress management (OR 0.97, 95% CI 0.55–1.68). Age (OR 0.56, 95% CI 0.36–0.88) and race/ethnicity (OR 4.80, 95% CI 1.15–19.99) were significantly associated with care intensity, but symptom prevalence and management were not associated.

CONCLUSIONS:

Results suggest that patients with stroke have distressing symptoms at EOL, but future research is needed in the evaluation and management of symptoms at EOL in patients with stroke.

Introduction

Globally, stroke is a leading cause of death¹ with 7.25 million stroke-related deaths reported worldwide in 2021.¹ In the United States, stroke is ranked as the fifth leading cause of death with one death occurring approximately every three minutes and eleven seconds.¹ Although the prognosis following stroke is variable, the high mortality rate and low average life expectancy make many patients with stroke appropriate candidates for palliative care soon after receiving a stroke diagnosis. Although there have been established American Heart Association (AHA)/ American Stroke Association (ASA) palliative and end-of-life (EOL) care stroke guidelines for the past decade,² palliative care is not a standard aspect of stroke care^3,4 with continued uncertainty around the EOL experience and outcomes for patients diagnosed with stroke.^4–9

A stroke presents with an acute onset and requires rapid intervention. Patients post-stroke are often admitted to an inpatient hospital unit for further management and observation.^9–12 Despite recent stroke research that has reported a decrease in the percentage of stroke deaths in the hospital setting,¹ very little is known about the care intensity of patients with stroke at EOL in the hospital setting and the impact of care intensity on the quality of EOL care. In research that has explored the prevalence and management of symptoms at EOL, there is a wide variability in symptom identification that can lead to unmet needs and distressful symptoms.⁸

Given the high post-stroke mortality rates and limited understanding of the quality of care for patients diagnosed with stroke, more research is needed to examine the care surrounding the EOL. To address this gap, this study aims to describe and identify whether place of death and care intensity received in the last month of life were associated with symptom prevalence and management in Medicare beneficiaries diagnosed with stroke.

Methods

This was a secondary analysis using data from the National Health and Aging Trends Study (NHATS).¹³ The NHATS is a longitudinal prospective cohort study of Medicare beneficiaries funded by the National Institute on Aging under a cooperative agreement with the Johns Hopkins University Bloomberg School of Public Health (U01AG032947) and the University of Michigan’s Institute for Social Research, with data collection by Westat.¹³ De-identified publicly accessible data were received from the NHATS and thus this study was deemed to be exempt from Institutional Review Board review of authors’ home institution.

Following development in 2011, NHAT surveys have been conducted annually. If the sample person is deceased, a last month of life interview is conducted with a proxy respondent. A proxy respondent is a person who is knowledgeable about the sample person’s daily routine and health and can be a relative, non-relative, or facility staff person.

Survey responses were included from proxy respondents (18≥ years) who are (a) responding on behalf of a deceased individual (65≥ years); (b) the deceased individual had a diagnosis of stroke within 2 years of death from 2011 to 2023 (NHAT rounds 1-12); and (c) had a completed last month of life interview (Supplementary Figure 1).

In each round, NHATS participants were asked if a doctor diagnosed them with a stroke since the last interview. We included participants who answered yes and then died within the following two years. In the last month of life interview, proxy respondents were asked where the participant’s death took place. Responses were dichotomized into hospital or other (home, nursing home, hospice, and other). For participants who died in the hospital setting, the care intensity that the patient received during their hospitalization was dichotomized into high acuity (intensive care unit (ICU) or emergency room) or low acuity (palliative care unit, hospice unit, or other type of unit).

Symptom prevalence was measured by the proxy respondent’s self-reported response to the yes/no survey question that asked if the sample person experienced the symptoms (1) pain, (2) dyspnea, and (3) emotional distress in the last month of life. Each symptom was first dichotomized to missing or non-missing and then to a yes or no response. The symptom management for pain, dyspnea, and emotional distress were measured by the proxy respondent’s self-reported response to the survey question that asked if the sample person received more help, less help, or the right amount of help that was needed to manage their symptoms in the last month of life. Symptom management was dichotomized as adequate management (right amount of help) or inadequate management (more or less help). NHATS demographic data of age at death, gender, race and ethnicity, and proxy respondent relationship were included as covariates.

Descriptive statistics were used to describe baseline characteristics. Logistic regressions were used to identify whether place of death (other vs. inpatient hospital) and care intensity (low-acuity inpatient vs. high-acuity inpatient) received in the last month of life were associated with symptom prevalence and management in patients diagnosed with stroke. Separate regression equations were run for each outcome, with symptom prevalence, symptom management and demographic covariates (age, gender, race/ ethnicity, place of residence, and proxy relationship) entered into each initial model. Backwards model building was then used to determine final models in which all correlates were significant at the 0.05 level. Odds ratios were used to examine the strength of the association between the independent and dependent variables. To test for multicollinearity, a correlation matrix was performed. Post hoc testing included goodness of fit tests, variance inflation factors, and a post-hoc power analysis.

Results

A total of 257 patients diagnosed with stroke who died between rounds and completed a last month of life interview were identified (Supplementary Figure 1). The majority of patients (73.1%) were ≥80 years of age, (63%) female, and (56.4%) White non-Hispanic. Of the proxy respondents who completed the last month of life interviews, 125 (48.6%) were a child of the patient diagnosed with stroke (see Table 1).

Table 1.

Characteristics of Patients Diagnosed with Stroke (n=257)

Demographics	Age	65-69 (%)	9 (3.5%)
		70-74 (%)	19 (7.4%)
		75-79 (%)	41 (15.9%)
		80-84 (%)	53 (20.6%)
		85-89 (%)	62 (24.1%)
		90+ (%)	73 (28.4%)
	Gender	Male (%)	95 (37%)
		Female (%)	162 (63%)
	Race/Ethnicity	White non-Hispanic (%)	145 (56.4%)
		Black non-Hispanic (%)	79 (30.7%)
		Hispanic (%)	23 (8.9%)
		Other (%)	10 (3.9%)
	Proxy Respondent Relationship	Spouse/ Partner (%)	50 (19.5%)
		Child (%)	125 (48.6%)
		Other (%)	82 (31.9%)
Variables	Place of Death	Hospital (%)	63 (24.5%)
		Other (%)	194 (75.5%)
	Hospital Care Intensity	Low Acuity (%)	34 (54%)
		High Acuity (%)	29 (46%)
	Pain	Missing (%)	27 (10.5%)
		Yes (%)	161 (62.6%)
		No (%)	69 (26.8%)
	Dyspnea	Missing (%)	24 (9.3%)
		Yes (%)	130 (50.6%)
		No (%)	103 (40.1%)
	Emotional Distress	Missing (%)	29 (11.3%)
		Yes (%)	148 (57.6%)
		No (%)	80 (31.3%)
	Pain Management	Missing (%)	117 (45.5%)
		About Right Amount (%)	119 (46.3%)
		More/less than was needed (%)	21 (8.2%)
	Dyspnea Management	Missing (%)	147 (57.2%)
		About Right Amount (%)	98 (38.1%)
		More/less than was needed (%)	12 (4.7%)
	Emotional Distress Management	Missing (%)	161 (62.6%)
		About Right Amount (%)	78 (30.4%)
		More/less than was needed (%)	18 (7%)

Note. The analytic sample size for characteristics of patients diagnosed with stroke was 257.

The majority (75.5%) of patients died outside of the hospital setting (home, nursing home, hospice, or other). Of the 63 (24.5%) patients who were reported to have died in the hospital, 34 (54%) died in a low acuity inpatient setting (palliative care unit, hospice, or other type of unit). In the proxy response for the symptom prevalence of pain, dyspnea, and emotional distress in the last month of life, the majority reported the presence of pain (62.6%), dyspnea (50.6%), and emotional distress (57.6%). In the proxy response for the level of symptom management in the last month of life, 46.3% reported receiving adequate pain management, 38.1% reported receiving adequate dyspnea management, and 30.4% reported receiving adequate emotional distress management. Notably, the majority of data were reported as missing for the management of dyspnea (57.2%) and emotional distress (62.6%) with nearly half (45.5%) reported as missing for pain management (see Table 1).

The final model for place of death included race/ethnicity, pain, dyspnea, emotional distress and management of pain, dyspnea, and emotional distress. Patients who died in the acute care setting were half as likely to report pain (OR 0.51, 95% CI 0.27–0.95), but twice as likely to report dyspnea (OR 2.07, 95% CI 1.02–4.21). Patients who died in the acute care setting were also almost twice as likely to have adequate dyspnea management (OR 1.97, 95% CI 1.06–3.67). The symptom prevalence of emotional distress and symptom management of pain and emotional distress were not significantly associated with the place of death (see Table 2).

Table 2.

Factors Associated with Place of Death and Care Intensity

Factors Associated with Place of Death (Other vs. Inpatient Hospital)
Factor	OR	p-value
Race/Ethnicity	1.40	0.061
Pain*	0.51	0.033
Dyspnea*	2.07	0.044
Emotional Distress	1.62	0.120
Pain Management	0.67	0.145
Dyspnea Management*	1.97	0.032
Emotional distress management	0.97	0.912
Factors Associated with Care Intensity (High vs. Low)
Factor	OR	p-value
Age at Death*	0.56	0.012
Race/Ethnicity* (Black non-Hispanic)	4.80	0.031
Pain	2.90	0.092
Emotional Distress	0.86	0.829
Pain Management	1.21	0.753
Dyspnea Management	1.01	0.983
Emotional distress Management	0.31	0.095

Note. The analytic sample size for place of death was 257. The analytic sample size for care intensity was 63.

^*Statistically significant results.

The final model for care intensity included the patient’s age, race/ethnicity, pain, emotional distress, and management of pain, dyspnea, and emotional distress. Individuals who were older were half as likely to die in a high acuity inpatient setting (OR 0.56, 95% CI 0.36–0.88). Black non-Hispanic patients were over four times more likely to die in a high acuity inpatient setting (OR 4.8, 95% CI 1.15–19.99). Symptom prevalence and symptom management of pain, dyspnea, and emotional distress were not significantly associated with care intensity (see Table 2).

Discussion

This study examined whether place of death and care intensity received in the last month of life were correlated with EOL symptom prevalence and management in patients diagnosed with stroke from a nationally representative sample of Medicare beneficiaries. Four major finding emerged regarding symptom prevalence, place of death, care intensity, and symptom management.

First, the significant amount of missing data regarding symptom management despite the majority of patients having documented distressing symptoms (pain, dyspnea, and emotional distress) suggests that proxy respondents may not have a good sense of the amount of symptom management the patient wanted or received at EOL. These findings are similar to current literature that acknowledges the gap in knowledge surrounding symptom assessment and symptom management at EOL for patients with stroke.^4,8 This gap may be due to the acute sudden onset of stroke, lack of advance care planning, and inconsistent symptom assessment.

Second, the majority of patients were reported to have died outside the hospital setting, which is consistent with other research that also reported the majority (67%) of patients in the U.S. diagnosed with a stroke die outside of the hospital setting.¹⁴ In the current study’s evaluation of place of death, the symptom prevalence of pain and dyspnea, and symptom management of dyspnea were significantly associated. While it is known that pain, dyspnea, and emotional distress are commonly reported symptoms at EOL,⁸ results suggest that patients with uncontrolled symptoms may be more likely to go to the hospital. Additionally, the lack of association between the prevalence of emotional distress and management of pain and emotional distress with place of death suggests that proxy respondents had difficulty distinguishing between pain and emotional distress. These findings support that stroke-related barriers and inconsistent symptom assessment without the use of validated tools leads to questions regarding the adequate management of distressing symptoms at EOL.⁸

Third, in the patients who died in the hospital setting, the majority died in a low acuity inpatient setting. Consistent with current literature, the evaluation of the association of care intensity with symptom prevalence and symptom management found that the patient’s age was significantly associated with the acuity of inpatient setting. Older adults with advanced age have been reported to have longer hospitalizations, greater disability, and higher risk-adjusted mortality,¹⁴ which may speak to the need for higher acuity care following diagnosis. Additionally, patients who identified as Black non-Hispanic were over four times more likely to die in a high acuity inpatient setting, which demonstrates continued racial and ethnic health disparities. These disparities are noted by the AHA and National Institutes of Health, which report that Black non-Hispanic males and females have a higher age-adjusted death rate and a longer length of hospital stay compared to White non-Hispanic patients.^1,14 These results further support the need for further research on how to better support these underrepresented groups.

Fourth, symptom prevalence and symptom management of pain, dyspnea, and emotional distress were not significantly associated with care intensity. This suggests that symptom prevalence and management may not be better or worse in high versus low acuity inpatient settings. While death in the emergency room or ICU setting is less typically associated with high-quality EOL care when compared to death in the home or hospice setting,^15,16 high quality EOL care can be delivered in a high acuity inpatient setting.¹⁷ These results may also reflect improvements in EOL care in high acuity settings following the provision of AHA/ASA palliative care guidelines.²

Based on this study’s findings, healthcare providers should prioritize discussions regarding prognosis and goals of care with patients and family members soon after stroke diagnosis. Through clear communication from the healthcare team and disclosure of prognostic uncertainty, goals of care can be better defined earlier in care.⁴ Nurses are well positioned to initiate these discussions^9,18 with patients, family members, and the interdisciplinary healthcare team. Next, nurses should receive training in palliative and EOL care⁹ and adopt a comprehensive symptom assessment with the use of validated symptom assessment tools to better identify stroke-specific (dysphagia, dysarthria, spasticity)^2,8 and non-specific symptoms (pain, dyspnea, and emotional distress)⁸ at EOL and allow for adequate symptom management. Lastly, more research is needed in underrepresented racial and ethnic groups, and in the evaluation of symptom assessment and management.

Limitations

As with all secondary analyses, we were limited to what data has been collected and is available for analysis. The NHATS is a voluntary survey of Medicare beneficiaries and may not be representative of all patients who receive EOL care following a stroke. Notably, the last month of life interview is completed by proxy respondents and may not align with the sample person’s perception of symptom prevalence, need for symptom management, or adequacy of symptom management.

In this study, stroke was defined by patient self-report without access to data on the type or acuity of stroke. Data for direct cause of death were not available. The AHA/ASA guidelines specifically address the palliative care needs of patients and families diagnosed with a severe or life-threatening stroke and this definition may not have applied to all of our study participants. Additionally, patients diagnosed with stroke often face negative impacts on mobility and communication, which pose real challenges to symptom assessment and management and may have impacted the prevalence of findings. Notably, the population of interest was Medicare beneficiaries ≥65 years of age who often have multiple medical comorbidities. Therefore, future prospective research should adjust for co-morbidities.

Lastly, this study had a small sample size with significant missing data. While statistical significance was found, this study lacked power due to significant missing data. While the NHATS is a nationally representative sample of Medicare beneficiaries, our results may not be generalizable to all older adults with stroke. Therefore, future prospective research with a larger sample size is needed to further identify distressing symptoms and quality of symptom management at EOL in patients with stroke.

Conclusion

This study sought to describe the EOL experience and outcomes for patients with stroke and to evaluate if place of death and care intensity received in the last month of life were associated with symptom prevalence and symptom management. Results suggest that patients with stroke have distressing symptoms at EOL and that they are more likely to report dyspnea and received adequate symptom management if hospitalized. Additionally, symptom prevalence and symptom management may not be better or worse in high versus low acuity inpatient settings. Future research is needed in the assessment and management of symptoms at EOL in patients diagnosed with stroke. Specifically, more research is needed to develop a comprehensive EOL symptom assessment tool for patients diagnosed with stroke in all care settings. Comprehensive symptom assessment can lead to a better understanding of how nurses can identify symptoms that are distressing to patients diagnosed with stroke to ensure adequate symptom management. Lastly, this study supports the need for future research funding opportunities and policy development for the allocation of EOL resources.