Racial Disparity in End-of-Life Hospitalizations among Nursing Home Residents with Dementia

Abstract

Objective:

Explore within and across nursing home (NH) racial disparities in end-of-life (EOL) hospitalizations for residents with Alzheimer’s disease or related dementia (ADRD), and examine whether severe cognitive impairment influences these relationships.

Design:

Observational study merging, at the individual level, C2014–2017 national-level Minimum Data Set (MDS), Medicare Beneficiary Summary Files (MBSF), and Medicare Provider Analysis and Review (MedPAR). Nursing Home Compare (NHC) was also used.

Setting:

Long-stay residents who died in a NH or a hospital within 8 days of discharge.

Participants:

Analytical sample included 665,033 decedent residents with ADRD in 14,595 facilities.

Main Outcomes and Measures:

The outcome was hospitalization within 30 days of death. Key independent variables were race, severe cognitive impairment, and NH-level proportion of black residents. Other covariates included socio-demographics, dual eligibility, hospice enrollment, and chronic conditions. Facility-level characteristics were also included (e.g. profit status, staffing hours, etc.). We fit linear probability models with robust standard errors, fixed and random effects.

Results:

Compared to whites, black decedents had a significantly (p<0.01) higher risk of EOL hospitalizations (7.88%). Among those with severe cognitive impairment, whites showed a lower risk of hospitalizations (6.04%). But EOL hospitalization risk among blacks with severe cognitive impairment was still significantly elevated (β=0.0494; p<0.01). A comparison of the base model with the fixed and random-effects models showed statistically significant hospitalization risk by decedent’s race both within and across facilities.

Conclusions and Relevance:

We found disparities between black and white residents with ADRD both within and across facilities. The within-facility disparities may be due to residents’ preferences and/or NH practices that contribute to differential treatment. The across facility differences point to the overall quality of care disparities in homes with a higher prevalence of black residents. Persistence of such systemic disparities among the most vulnerable individuals is extremely troubling.

INTRODUCTION

As the prevalence of Alzheimer’s disease and related dementias (ADRD) continues to rapidly increase in the US, so has its presence in nursing homes (NH). In 2016, almost 50% of NH residents had a diagnosis of ADRD,¹ and 70% were expected to die there.² As the disease progresses, ADRD patients tend to experience increased cognitive impairment, memory problems as well as declines in ability to make decisions, and yet many may require particularly intensive care at this junction.³ Many ADRD residents experience eating problems that may lead to pneumonia, as well as fevers and infections⁴ resulting in hospital admissions. At the end-of-life (EOL) hospitalizations of NH residents are not infrequent although half are thought to be potentially avoidable.⁵ For residents with ADRD who are dying, hospitalizations are thought to be particularly burdensome as they contribute to confusion and stress for them and their family members, disrupt care plans and coordination, and result in poorer quality of care and life. ⁶

Recent research on NH residents has shown a significantly lower risk of hospitalizations, and particularly those considered to be potentially avoidable, among residents with ADRD compared to those without ADRD,⁷ perhaps indicating that NH staff recognize the increased risks associated with hospitalizations for these residents. Another study, focusing specifically on EOL residents showed that with increasing severity of ADRD, NH decedents were less likely to experience death in a hospital, again suggesting that as staff recognize disease severity they are better able to provide more appropriate and less burdensome levels of care.⁸

While it may be comforting to know that NHs are increasingly capable of recognizing and appropriately responding to the care needs of residents dying with ADRD, studies have also shown persistent and substantial variations in hospitalization patterns for EOL patients, including those with ADRD. ^8,9 Furthermore, ADRD residents are more likely to be admitted to lower quality homes, with fewer resources, including staffing,¹⁰ thus exacerbating disparities in care. Similarly, disparities in care have been documented for minority residents, especially African Americans¹¹ who tend to receive more intensive EOL care. Minority residents with ADRD also have been shown to receive care in homes with fewer resources and lower quality, compared to non-minority residents.¹²

To date, only a handful of studies have attempted to identify whether the observed disparities in EOL care are due to unequal treatment within the same facility or because of segregated placement in facilities providing poorer care quality. A study of NH decedents in New York State found no within facility disparities for in-hospital deaths or hospice use between blacks and whites.¹³ But another study, comparing a national population of NH residents with and without dementia, found the within but not across facility differences for in-hospital death.²

Motivated by the absence of clear evidence and the need to disentangle the potential source of disparities in EOL care quality, our study uses national level data to: 1) explore within and across NH racial disparities in EOL hospitalizations for residents diagnosed with ADRD, and 2) determine the extent to which severe cognitive impairment may influence these relationships.

METHODS

Data.

The following CY2014–2017 national data were linked at the individual level: Minimum Data Set (MDS) 3.0, Medicare Beneficiary Summary File (MBSF), and Medicare Provider Analysis and Review (MedPAR). The MDS contains information on residents’ treatments, health and cognitive status, for all admitted to Medicaid and/or Medicare certified NHs. MBSF contains information on residents’ demographics and chronic conditions. The MedPAR file contains information on hospitalization events for all Medicare beneficiaries. We also linked the NH Compare (NHC) data to obtain facility-level characteristics (e.g. profit status, bed size, staffing hours).

Study population.

We included decedents with ADRD who spent their end of life in NHs during CY2014–2017. Following prior studies,^14,15 we identified NH decedents as residents whose death occurred within 1 day of an identified NH stay or 8 days of a hospital transfer from NH. The diagnosis of ADRD was based on the MBSF chronic condition files and the MDS diagnoses check box. Only long stay residents, i.e. those who had quarterly, annual, or significant correction to prior comprehensive assessments in the last quarter were included, because care provided to these residents is more relevant to the NH EOL care quality than for the post-acute residents. Our analytical sample included 665,033 decedent residents in 14,595 facilities.

Variables.

The outcome of interest was any hospitalization occurring within 30 days prior to death, defined as a dichotomous variable. Hospitalizations were identified using MedPAR.

The key independent variables of interest were residents’ race, presence of severe cognitive impairment, and the proportion of black residents in an NH. Race was categorized as white or black based on the MBSF. Cognitive impairment was measured using the cognitive function scale (CFS), which combined the Brief Interview for Mental Status (BIMS) and the Cognitive Performance Scale (CPS) from the MDS, categorizing residents as either severely impaired or not.¹⁶ Most long-stay residents have multiple assessments, which tend to increase at the EOL. We categorized the decedents as having severe cognitive impairment if two or more of their EOL assessments indicted this level of impairment. Fewer than 3% of the sample had only one assessment with a non-missing CFS value. For this group, we defined individuals as having severe cognitive impairment if it was so indicated on their only assessment. We categorized NHs into quartiles according to the proportion of black residents they had during the study period, with the first quartile including only NHs without any black residents.

We also accounted for many individual-level covariates that may be related to EOL hospitalizations, including demographic factors (e.g. gender and age), dual eligibility for Medicaid and Medicare, hospice enrollment, presence of aggressive behaviors, and individual chronic conditions, such as cardiovascular disease, mental illness, and others. Facility-level characteristics were also included to control for profit status, chain membership, bed size, occupancy rate, total staff hours, skilled care ratio (RN to other nursing staff), and proportion of Medicare and Medicaid residents.

Analysis.

All analyses were conducted at the individual level. We first compared EOL hospital use and individual characteristics by race of the residents and across NHs with different prevalence of minority residents. To investigate the relationship between race and EOL hospitalizations, we fit several linear probability models with robust standard errors. First, a base model (Model 1) was fit to examine the difference in this outcome between black and white residents, including the severity of cognitive impairment, facility-level percent of blacks, and year fixed effects. Second, we fit a NH fixed-effects model (Model 2), in which additional individual risk factors and year fixed-effects were included. In Model 3, we fit a random-effects model with a set of facility-level characteristics. In all three models, we included the interaction between race and severe cognitive impairment to examine whether and how the relationship between race and EOL hospitalizations was modified by decedents’ cognitive function.

The effect of race estimated from the fixed-effects models represents the within-facility difference in the risk of hospital use within 30 days of death between white and black residents. The difference in the effect of race estimates between Models 1 and 2, if found to be statistically significant, indicates that the outcome risk detected in the base model is partially due to the heterogeneity of facilities rather than to differential treatment of residents in the same facility. In the random-effects model (Model 3), the estimate of the facility-race mix indicates that across-facility variations may contribute to the difference in outcome prevalence between blacks and whites.

Another linear model was estimated to examine the relationship between facility characteristics and the percent of African-American residents.

RESULTS

Among decedent residents with ADRD, the unadjusted prevalence of hospitalizations in the last 30 days of life was 29.53% among whites and 40.66% among blacks (Table 1). The overall prevalence increased with higher facility-level presence of black residents; from 16.8% in facilities with no blacks (first quartile) to 35.3% in the 4^th quartile NHs (29.6% of blacks). Furthermore, blacks and whites seemed to differ in many socio-demographic characteristics, with blacks being younger (82.6 versus 86.6 years) and more likely dually eligible for Medicare/Medicaid (91.1 % versus 72.6%), experiencing more severe cognitive impairment (32.5% versus 23.4%), but less aggressive behavior (84.1% none versus 79.8%). Black decedents were also more likely to have diabetes (48.4% vs. 30.5%), hypertension (87.2% vs. 77.9%), stroke (25.8% vs. 14.9%), and schizophrenia (7.5% vs. 3.9%). The average time period from the last MDS assessment to death was 36.7 days. Hospice enrollment prior to death appears to have been similar between whites (29.93%) and blacks (28.17%), but was higher in facilities with high minority presence, compared to none or low presence.

Table 1:

Characteristics of Nursing Home Decedents with ADRD, 2014–2017: By Race¹

	Resident’s Race			NH-level Proportion of Black Residents					Total Sample
	White	Black		1st quartile (0%)	2nd quartile (0.92%)	3rd quartile (5.38%)	4th quartile (29.61%)
Number of decedents (%)	598,502	66,531		75,401	208,746	198,812	182,074		665,033
	(90.00)	(10.00)		(11.34)	(31.39)	(29.90)	(27.38)		(100.00)
Outcome Variable
Hospitalization <30 days of death	25.93	40.66		16.83	23.07	28.72	35.30		27.40
Individual Characteristics
Age (SD)	86.65	82.64		88.32	87.49	86.17	84.06		86.25
	(9.17)	(11.05)		(8.28)	(8.67)	(9.34)	(10.40)		(9.45)
ADL impairment score (SD)	21.01	22.75		20.52	20.83	21.16	21.89		21.18
	(4.80)	(4.87)		(5.01)	(4.59)	(4.75)	(5.03)		(4.83)
% Female	68.81	61.25		69.64	69.80	68.34	65.10		68.06
% Medicare-Medicaid dual eligibility	72.61	91.09		63.10	70.36	75.44	82.81		74.46
% Enrolled in hospice at EOL	29.93	28.17		25.53	28.37	31.94	30.69		29.75
Marital Status
Married	21.30	15.80		22.45	21.93	20.89	18.55		20.75
Divorced/separated	10.66	16.12		7.78	9.58	11.68	13.98		11.21
Never married	9.15	21.67		6.72	7.75	9.56	15.88		10.40
Widowed	58.89	46.41		63.05	60.74	57.87	51.60		57.64
% With severe cognitive impairment (CFS)	23.37	32.46		22.23	22.67	24.05	27.22		24.28
% Aggressive behavior
None	79.82	84.13		76.98	79.22	80.92	82.05		80.25
Moderate	13.41	11.05		14.94	13.80	12.73	12.21		13.18
Severe	6.77	4.82		8.08	6.97	6.35	5.74		6.57
% Diagnosed with:
Cancer	5.15	6.49		5.34	5.06	5.22	5.57		5.28
CAD	13.89	12.66		13.62	14.27	14.18	12.80		13.76
Heart Failure	29.87	29.31		31.01	30.53	30.14	28.15		29.82
PVD	14.33	19.80		12.81	14.35	15.45	15.71		14.87
Diabetes	30.50	48.35		27.89	29.64	31.94	37.53		32.29
Hypertension	77.94	87.20		75.34	77.32	79.51	81.42		78.87
Asthma	26.30	23.27		24.29	25.85	26.83	25.96		25.99
Pneumonia	9.92	10.27		8.34	9.54	10.60	10.40		9.96
Septicemia	2.45	4.47		1.66	2.21	2.75	3.44		2.65
UTI	14.69	15.18		13.81	13.97	15.20	15.51		14.74
Wounds Infect	1.19	2.02		1.00	1.07	1.28	1.61		1.27
Hip Fracture	3.75	1.82		3.53	3.63	3.71	3.31		3.55
Other Fracture	4.68	2.27		4.41	4.66	4.65	3.99		4.44
Stroke	14.92	25.84		13.16	14.27	15.77	19.47		16.01
Anxiety	39.99	25.37		38.00	39.72	40.48	35.25		38.53
Manic Depression	4.11	3.30		2.50	3.25	4.44	5.11		4.03
Schizophrenia	3.87	7.54		2.16	2.70	4.22	6.86		4.23
Respiratory Failure	3.84	6.12		2.45	3.61	4.23	5.09		4.07
Days from last MDS assessment to death (SD)	37.06	33.03		38.33	38.12	36.65	34.29		36.66
	(62.41)	(65.84)		(62.54)	(62.02)	(63.06)	(63.34)		(62.77)
Facility Characteristics
Number of nursing homes				2042	4250	4134	4169		14595
% For profit				48.03	64.47	78.15	83.70		71.45
Multi-facility chain				47.57	59.97	62.04	62.81		59.61
Bed size (SD)				68.96	103.12	117.04	128.74		109.48
				(32.25)	(47.96)	(58.87)	(61.95)		(57.00)
Occupancy rate (SD)				79.67	81.32	81.22	81.89		81.22
				(15.80)	(14.34)	(13.83)	(13.75)		(14.27)
Total staffing hours/resident day (SD)				4.25	4.13	4.07	3.99		4.09
				(0.91)	(0.85)	(0.83)	(0.77)		(0.84)
RN to LPN+CNA staffing ratio (SD)				0.20	0.18	0.16	0.14		0.16
				(0.10)	(0.09)	(0.08)	(0.07)		(0.09)
% Medicaid residents (SD)				55.07	55.33	60.15	69.07		60.52
				(20.07)	(21.49)	(21.58)	(18.35)		(21.26)
% Medicare residents (SD)				9.32	14.84	14.78	12.57		13.41
				(8.21)	(11.82)	(12.12)	(10.01)		(11.14)

Note: Facility characteristics represent an average aggregated value for CY2014–2017 for each unique nursing home.

¹Comparisons of covariates, between black and white residents and across NH-level proportions of black residents, are all statistically significant at p<0.01

Racial Disparity and Risk of EOL Hospitalizations

In Table 2, we depict the results from the base linear probability regression (Model 1). Overall, among those who did not have severe cognitive impairment, black residents with ADRD had a 7.9 percent point higher risk of hospitalizations in the last 30 days of life than white residents (p<0.01). White residents with severe cognitive impairment had a 6.0 percent point lower hospitalization risk compared to residents without severe impairment (p<0.01), but this did not hold true among blacks whose hospitalization risk was still significantly elevated (β=0.0494; p<0.01).

Table 2:

Hospital Use 30 Days before Death among ADRD Decedents: Linear Probability Models

	Model 1 Base β-coefficient (SE)	Model 2 Fixed Effects β-coefficient (SE)	Model 3 Random-Effects β-coefficient (SE)
Race (Ref: White)
Black	0.0788*** (0.0026)	0.0405*** (0.0026)	0.0438*** (0.0025)
Cognitive Performance Scale (Ref: Non-severe impairment)
Severe Impairment	-0.0604*** (0.0013)	-0.0207*** (0.0014)	-0.0203*** (0.0014)
Black × Severe Impairment	0.0494*** (0.0043)	0.0280*** (0.0042)	0.0274*** (0.0040)
Proportion of Black Residents in NHs (Ref: NH without any black residents)
Low Proportion of Black NH	0.0618*** (0.0016)		0.0517*** (0.0034)
Middle Proportion of Black NH	0.115*** (0.0017)		0.0959*** (0.0034)
High Proportion of Black NH	0.160*** (0.0019)		0.133*** (0.0038)
Year fixed effect	Y	Y	Y
Facility characteristics	NA	NA	Y
Individual characteristics	NA	Y	Y
Facility fixed effect	NA	Y	NA
Facility random effect	NA	NA	Y
N of observations	665033	665033	664848
N of NHs		14595	14595

Notes:

^*p < 0.10

^**p < 0.05

^***p < 0.01

Model 2: Individual characteristics included: age, gender, ADL score, Medicare/Medicaid dual status, hospice use, marital status, aggressive behaviors, and chronic conditions (e.g. hypertension, asthma, cancer); full model presented in Supplementary Table S1.

Model 3: Facility characteristics included: profit status, bed size, chain membership, occupancy, total staff hours, skilled nursing staff ratio; proportion of Medicare and Medicaid residents; full model presented in Supplementary Table S1.

Model 2 presents the results from the fixed-effects estimation. After controlling for individual-level characteristics and facility fixed effects, the association of blacks with hospitalization risk remained statistically significant (p<0.01), albeit somewhat attenuated; 4.1 percent point higher risk than for whites. This suggests that within-facility difference in the risk of EOL hospitalizations between white and black residents does exist.

After controlling for individual risk factors, hospitalization risk for the severely cognitively impaired white residents was still substantially reduced, showing a 2.1 percentage points lower risk (p<0.01) compared to whites without severe cognitive impairment. However, hospitalization risk among black residents with severe cognitive impairment was 0.73 percentage points higher than among those without (−2.07%+2.80%=0.73%), and was borderline statistically significant (p=0.062).

Furthermore, the difference in β-coefficients between models 1 and 2 was statistically significant (p<0.01), indicating that the differential EOL hospitalizations between whites and blacks were also due to the heterogeneity of facilities in which they resided. This is further supported by Model 3 (Table 2), depicting the results from the random-effects model. Consistent with the fixed-effects model, black residents with ADRD had increased risk of EOL hospitalizations (β=0.0438, p<0.01). After accounting for facility-level characteristics, facility race-mix was significantly independently correlated with the hospitalization risk. Compared to facilities without any black residents, homes with a low proportion of blacks (0.92%) had a 5.17 percentage points higher risk of hospitalizations; with the risk more than doubling at 13.3 percentage points in facilities with the highest penetration (29.6%) of black residents.

The coefficient estimates from the random (Model 2) and the fixed-effects (Model 3) models were overall similar, suggesting that the random-effects model did not suffer from any inconsistencies.

Racial Disparity in Facility Characteristics

NHs with higher proportion of blacks are significantly different in a number of characteristics, and some of these differences are far from trivial in effect size (Figure 1). For example, for profit facilities were associated with having a higher proportion of black residents (β=0.0343, p<0.01), while facilities with higher skilled staffing ratios were associated with having substantially fewer black residents (β= −0.188, p<0.01).

Figure 1:

Association of Facility-Level Characteristics with the Proportion of Black Residents

Note: Figure shows estimate coefficients and 95%CIs resulting from a facility-level linear regression, with year fixed effects. Dependent variable was the proportion of black residents in a facility. Independent variables included facility characteristics listed in the figure. Values are shown as percentage point changes; for example, for profit facilities were associated with having a 3.43 percentage point higher proportion of black residents than non-profit facilities.

DISCUSSION

In this study, we attempted to disentangle the within and across-facility differences in EOL hospitalizations among black and white residents with ADRD. We found that within the same NH black residents were more likely to be hospitalized within the last 30 days of life than their white counterparts, and that this within facility disparity persisted even after controlling for individual level characteristics, and increased among those with severe cognitive impairment. We also found that the disparity in this outcome measure persisted across facilities, and was significantly exacerbated with an increase in the prevalence of black residents.

To date, most studies of racial disparities in NHs have shown persistent across facility but largely no within-facility differences in treatment.¹⁷ We offer two possible reasons for the observed within-facility disparity in EOL hospitalizations. Prior studies have shown that minorities tend to receive higher-intensity life sustaining treatments,^18,19 in part because they are less likely to have advance care directives,^20–22 and also due to the underlying mistrust of the healthcare system. For example, a recent study has shown that disparities in the level of mistrust between black and white patients are exacerbated at the EOL.²³ The absence of advance care directives and higher level of mistrust among black residents and their family members may indeed contribute to the within-facility disparity in EOL hospitalizations. Another possibility is that NHs are more likely to hospitalize African American residents not because of their race but because they are more likely to be dually eligible for Medicare and Medicaid. Indeed, our findings show that residents who are dual at the EOL are incrementally more likely to be hospitalized, as are residents of NHs with a higher proportion of duals. This is supported by prior studies showing that facilities with a higher proportion of Medicaid residents have fewer internal resources to care for the acutely ill, and thus are more apt to hospitalize their residents.^24,25 Consistent with these findings, others have shown that Medicaid residents are more likely to be hospitalized both within and across facilities than their private pay counterparts, particularly for discretionary stays.²⁶ Furthermore, it has been suggested that NHs have no or little incentive to hospitalize their profitable private pay residents when such hospitalizations may be avoidable, but have significant incentives to hospitalize the less profitable Medicaid residents for similar conditions.²⁷

Our findings with regard to the across-facility differences between black and white decedents are wholly consistent with prior studies that have attributed such disparities to NH having a larger proportion of minority residents being resource and quality poorer,²⁸ and being located in disadvantaged communities where access to well trained staff is also an issue.^29,30 African-Americans with ADRD are even more likely to receive care in segregated, lower, quality homes then those without ADRD.¹² Furthermore, as we have shown NHs with higher prevalence of black residents are significantly more likely to be for-profit, and as such their mission and profit incentives may motivate them to hospitalize their residents more often. These NHs also appear to have significantly lower staffing ratios, which likely contribute to the relationship we observe between EOL hospitalizations and the proportion of blacks in a facility.

It has long been recognized that hospitalizations of NH residents, particularly at the EOL are burdensome and often potentially avoidable.^6,31 Indeed, the opportunity and the need to reduce such hospitalizations is greatest among residents faced with advanced chronic illness such as the ADRD. To that end, a number of initiatives aimed at reducing hospital transfers of NH residents have been undertaken following the implementation of the Patient Protection and Affordable Care Act (ACA). For example, in 2011 the Centers for Medicare and Medicaid Services (CMS) Innovation Center launched and initiative to reduce avoidable hospitalizations among NH residents, and have since reported a 31% decline between 2010 and 2015.³² We also found significant declines in hospitalizations occurring within 30 days of death between CY2014 and CY2017. However, the overall proportion of such hospitalizations remains quite high for residents with advanced dementia.^7,33 Moreover, while EOL hospitalizations for residents with ADRD and severe cognitive impairment would generally be considered even more unwarranted, we found a decrease in hospitalizations among whites but not among blacks.

Other initiatives, such as the Intervention to Reduce Acute Care Transfers (INTERACT)³⁴ have also been adopted, aiming to improve on-site evaluation and management of acute exacerbations through staff training and early detection, but the results have not been encouraging.³⁵ In 2018, CMS has initiated a value-based purchasing program for skilled nursing facilities aimed at reducing readmissions and thus reduce avoidable hospitalizations in the process. Early results suggest that SNFs are failing to achieve set benchmarks and, as a result, incur financial penalties.³⁶ Notwithstanding the success of such interventions, they seem to have little if any effect on systemic disparities in EOL hospitalizations among NH residents with ADRD, as shown in this study. Hospice enrollment, which may be able to reduce EOL hospitalizations is overall low, at less than 30%. Palliative care, which has the potential to both elicit patient preferences and guide treatment choices is absent from most US NHs,³⁷ and the prospects of a wide-scale palliative care implementation in NHs is dim without concerted policy efforts to incentivize the industry, and the regulatory changes to make uptake of palliative care sustainable.¹⁵

We would like to note several limitations of our study. First, we only present evidence of disparity in EOL hospitalizations, which cannot be generalized to other EOL outcomes. Based on prior studies, it is possible that within facility disparities occur in some but not in all outcome measures.³⁸ Second, the administrative data used in this study do not provide any information on patient/family preferences, and as of the end of 2010 the MDS no longer includes information on such patient wishes as cardiopulmonary resuscitations or do-not-hospitalize orders.³⁹

In conclusion, with regard to EOL hospitalizations we found significant disparities between black and white residents with ADRD occurring both within and across facilities. The reasons for within-facility disparity are not clear and may be due to residents’ preferences and/or NH practices that contribute to differential treatment. The across facility differences, which have been demonstrated before, point to the overall quality of care differences in homes with a higher prevalence of black residents. In either case, the persistence of such systemic disparities among the most vulnerable individuals is extremely troubling.

Supplementary Material

Supplementary Table S1

Supplementary Table S1: Hospital use 30 days before death among ADRD decedents in NHs 2014-2017

Key Points:

1. In the last 30 days of life, blacks with Alzheimer’s disease and related dementias (ADRD) who live in nursing homes experience significantly higher risk of hospitalizations compared to their white counterparts.

2. EOL hospitalization risk remains significantly elevated for blacks with ADRD and severe cognitive impairment.

3. Disparities between black and white nursing home residents with ADRD, in EOL hospitalizations, persist both within and across facilities.

Why does this paper matter?

Significant differences in EOL hospitalizations exist between black and white residents with ADRD both within and across nursing homes. The persistence of such systemic disparities among the most vulnerable individuals is troubling.