Nursing Home Alzheimer’s Special Care Units: Geographic Location Matters

Abstract

Objectives:

Limited data suggest nursing home (NH) Alzheimer’s special care units (ASCUs) may improve care and outcomes among residents with dementia. Unfortunately, information describing NH characteristics related to presence of ASCUs is lacking, especially whether location and neighborhood resources influence their presence. We examined locations of NHs with ASCUs and assessed whether neighborhood socioeconomic deprivation, region, and levels of rurality were associated with NH ASCUs.

Design:

Cross-sectional.

Setting and Participants:

Contiguous U.S. We used 2017 LTCfocus and NH Compare data to identify free-standing NHs and obtain addresses (N=13,207 NHs).

Methods:

NH ZIP+4 codes were linked to the Area Deprivation Index (ADI) (within-state ranking of neighborhood deprivation). The nine census-defined regions of the U.S. and Rural Urban Continuum codes categorized location. Descriptive analyses and binary logistic regression models, adjusting for NH characteristics, examined associations between NH ASCUs and location.

Results:

Nearly 15% of NHs had ASCUs. In adjusted models, odds of NH ASCUs were 58–69% lower in Pacific, Middle Atlantic, and Southern regions compared to the East North Central region (p-values<0.001). Odds of NH ASCUs increased 25–47% as rurality increased relative to NHs in the most metropolitan areas (p-values:<0.01); however, odds of NH ASCUs decreased 63% in the most rural areas (p-value:<0.001). ADI was not significantly associated with NH ASCUs. For-profit NHs had 42% lower and chain-affiliated NHs 34% higher odds of ASCUs (p-values<0.001). NHs with higher total staffing hours had 29% higher odds of ASCUs; odds of ASCUs were 46% lower in NHs with more RN staffing hours (p-values<0.001).

Conclusions and Implications:

Using a robust sample, region, rurality, ownership, and nursing hours significantly predicted NH ASCUs while ADI did not. Geographically tailored interventions should be considered to promote use of NH-based ASCUs.

Brief Summary:

Alzheimer’s special care units are unevenly distributed across the U.S. Region and rurality, but not socioeconomic deprivation, influence availability. New approaches are needed to promote equitable access to these units.

INTRODUCTION

Despite increasing national efforts to prevent or cure Alzheimer’s disease and related dementias (ADRD), 5.8 million Americans age 65 years and older are currently diagnosed with these degenerative brain diseases, with expected increases to nearly 14 million by 2050.¹ States located in the Western and Southeastern U.S. are projected to be especially impacted with ADRD by 2025, as prevalence of these diseases is expected to increase nearly 35% compared to 2020 prevalence estimates.¹

While care for persons with ADRD usually begins in communities, use of nursing home (NH) care is commonplace as diseases progress and complex symptoms require advanced medical care provided by NH staff. Each year, approximately 30% of Americans age 65 years and older (more than 630,000 Americans) spend their last days in NHs and nearly 70% have a diagnosis of ADRD.² Quality of NH care is a long-standing concern among families and policy makers and is an essential research area given rapid increases in populations of persons with ADRD transitioning into NHs.

Limited data suggest that NH Alzheimer’s special care units (ASCUs) improve care and resident outcomes,^3–5 but evidence is mixed.^6,7 Unfortunately, information describing general characteristics of ASCUs, as well as structures, processes, and mechanisms of action potentially contributing to beneficial outcomes is lacking. Furthermore, no standard definition of NH ASCUs exists^6,8 and reasons why NHs establish ASCUs are unclear; they may be established as an attempt to address resident needs or to increase marketability – or perhaps combinations of both.^3,9,10 Key limitations in prior studies of NH ASCUs include small samples and failure to meaningfully incorporate geography and neighborhood resources (factors known to impact health and healthcare utilization).^11–17 NH priorities may differ depending on patient populations, local competition, and neighborhood characteristics. Understanding associations between location, neighborhood resources, and NH ASCUs may provide critical and innovative insights, bringing clarity to currently unknown features of NH care that influence care for residents with ADRD.

To fill these knowledge gaps, our objectives were twofold: 1) examine locations of NHs with ASCUs across the U.S., providing first-ever geographic analyses of availability at the national level; and 2) assess whether neighborhood resources, region, and levels of rurality are associated with NH ASCUs.

METHODS

Data Sources and Sample

We used 2017 LTCfocus¹⁸ and Centers for Medicare & Medicaid Services’ Nursing Home Compare¹⁹ (NHC) datasets to identify free-standing Medicare and/or Medicaid-certified NHs and obtain addresses among NHs in the contiguous U.S. The LTCfocus dataset contains NH characteristics from NH-reported OSCAR/CASPER administrative data, including ASCU presence; the NHC dataset contains additional NH characteristics obtained through federally mandated reports and inspections. NH Zip+4 codes were linked with the Area Deprivation Index (ADI).¹⁵ The most recently available ADI (using 2015 U.S. Census data) includes 17 measures of neighborhood socioeconomic deprivation from 4 domains: education, employment, housing, and poverty; details of its construction are available elsewhere.¹⁴ We adopted nine census-defined regions of the U.S.²⁰ and used Rural Urban Continuum (RUC) codes²¹ to categorize county rurality.

Of 14,368 NHs matched between LTCfocus and NHC datasets, 155 NHs could not be merged with ADIs (e.g., invalid postal address) and 420 did not have ADI values (e.g., suppressed data). We excluded 586 hospital-based NHs because their care structures and processes are fundamentally different from free-standing NHs. The resulting study sample included 13,207 NHs.

This study protocol was approved by BLINDED FOR REVIEW’S institutional review board.

Outcome

We used a binary indicator for whether NHs have beds designated for residents with ADRD as our outcome variable. NHs reporting any beds for ADRD residents are considered as having ASCUs.

Key Covariates

Key covariates assessed NH location at three levels: census regions, county, and neighborhoods/census tracts. The census regions group states into nine categories: East North Central (reference), East South Central, Middle Atlantic, Mountain, New England, Pacific, South Atlantic, West North Central, and West South Central. RUC codes group counties into levels of increasing rurality: 1) counties in metro areas of 1 million population or more (reference); 2) counties in metro areas of 250,000–1 million population; 3) counties in metro areas of fewer than 250,000 population; 4) urban population of 20,000 or more, adjacent to metro area; 5) urban population of 20,000 or more, not adjacent to metro area; 6) urban population of 2,500–19,999, adjacent to metro area; 7) urban population of 2,500–19,999, not adjacent to metro area; 8) completely rural or less than 2,500 urban population, adjacent to metro area; and 9) completely rural or less than 2,500 urban population, not adjacent to metro area.

We included the ADI as a continuous measure: neighborhood scores of socioeconomic deprivation were ranked from lowest-highest within states and divided into deciles (i.e., 1=least disadvantaged, 10=most disadvantaged neighborhood).

Other Covariates

We included common NH-level covariates previously associated with NH performance and care patterns among residents with ADRD.^2,3,5,7–10 Covariates included ownership (for-profit vs. government owned/non-profit); chain affiliation (yes/no); number of beds; occupancy; RN hours per resident day (HPRD); total nurse staffing HPRD, % Medicare, Medicaid, female, and white residents; resident age; case mix acuity (measuring level of NH care need required for given resident populations); and county NH bed competition (Herfindahl-Hirschman Index; score from 0–1 with higher values indicating greater competition).

Statistical Analyses

We first examined geographic variation in locations of NH ASCUs across the U.S. through mapping. We then compared descriptive statistics across NHs with/without ASCUs and examined bivariate associations using chi-square and t-tests appropriate to respective measures. Finally, we estimated binary logistic regression models for presence of NH ASCUs, adjusting for NH characteristics. Analyses were performed using R Version 4.0.2.²² Alpha thresholds determining statistical significance were 0.05 for all analyses.

RESULTS

Nearly 15% of NHs had ASCUs in 2017. Within states, Indiana had the highest ASCU prevalence (35.0%) and Tennessee the lowest (1.9%) (Figure 1, Panel A). NHs with ASCUs were more prevalent in North Central and Middle Atlantic regions; NH ASCUs varied across the U.S. from 0.2% (Idaho, Mississippi) to 8.6% (Indiana) (Figure 1, Panel B).

Figure 1:

Geographic distribution of nursing home Alzheimer’s special care units (ASCUs). Panel A: % nursing home ASCUs by state (N=13,207 nursing homes). Panel B: % nursing home ASCUs across the U.S. (N=1,934 nursing homes). Nursing homes in Alaska, Hawaii, and the District of Columbia were excluded because they could not be matched between LTCfocus and Nursing Home Compare datasets.

Multiple NH characteristics were associated with presence of ASCUs (Table 1). ASCUs were more prevalent in NHs that were non-profit and had more beds, higher occupancy, fewer RN HPRD, fewer Medicare residents, and more female, white, older, and higher-need residents. NHs located in East North Central, Middle Atlantic, Mountain, New England, and West North Central regions were more likely to have ASCUs. Regarding rurality, the most densely populated urban areas and the most rural areas were least likely to have NHs with ASCUs. There was no statistically significant difference in presence of NH ASCUs regarding ADI.

Table 1:

Sample Characteristics: Nursing homes with and without Alzheimer’s special care units.

Sample Characteristics	ASCU* N: 1,934 (14.6%)	No ASCU N: 11,273 (85.4%)	p-value (χ2 or t-test)
For-profit	57.3%	75.7%	<0.001
Chain affiliation	58.8%	60.0%	0.346
Number of beds	131.8±62.3	104.1±52.3	<0.001
Occupancy	81.6±13.7	80.4±14.7	<0.001
RN hours per resident day	0.7±0.3	0.8±0.4	<0.001
Total nurse staffing hours per resident day	4.1±0.8	4.1±0.9	0.848
% Medicare	10.3±7.8	14.2±12.9	<0.001
% Medicaid	60.9±20.0	60.1±23.1	0.103
% female	67.7±11.4	65.8±11.9	<0.001
% white	85.1±18.0	79.0±22.3	<0.001
Average resident age	80.8±5.4	79.0±7.2	<0.001
Case mix acuity	12.0±1.0	12.2±1.4	<0.001
County nursing home bed competition (1-Herfindahl- Hirschman Index) †	0.8±0.2	0.8±0.2	0.298
State Area Deprivation Index	5.8±2.7	5.8±2.7	0.423
Region			<0.001
East North Central	27.3%	19.2%
East South Central	4.4%	6.9%
Middle Atlantic	12.0%	10.3%
Mountain	6.7%	4.5%
New England	8.3%	6.1%
Pacific	3.8%	11.8%
South Atlantic	8.3%	16.0%
West North Central	21.1%	11.4%
West South Central	8.2%	14.0%
RUC Code‡			<0.001
1	35.0%	42.4%
2	20.9%	19.9%
3	12.5%	9.8%
4	7.4%	6.0%
5	2.1%	2.2%
6	11.0%	8.8%
7	6.6%	5.3%
8	1.1%	1.6%
9	1.0%	2.5%

^*: Alzheimer’s Special Care Unit

^†:Score from 0–1 with higher values indicating greater competition.

^‡: Rural-Urban Continuum Code: 1) counties in metro areas of 1 million population or more (reference); 2) counties in metro areas of 250,000 to 1 million population; 3) counties in metro areas of fewer than 250,000 population; 4) urban population of 20,000 or more, adjacent to a metro area; 5) urban population of 20,000 or more, not adjacent to a metro area; 6) urban population of 2,500 to 19,999, adjacent to a metro area; 7) urban population of 2,500 to 19,999, not adjacent to a metro area; 8) completely rural or less than 2,500 urban population, adjacent to a metro area; and 9) completely rural or less than 2,500 urban population, not adjacent to a metro area.

The adjusted logistic regression model was overall significant (p-value:<0.001); the c-statistic of 0.77 indicates a good model fit. After controlling for NH covariates, for-profit NHs had 42% lower odds of ASCUs; chain-affiliated NHs had 34% higher odds of ASCUs (p-values<0.001) (Table 2). NHs with higher total staffing hours had 29% higher odds of ASCUs, while NHs with more RN staffing hours had 46% lower odds of ASCUs (p-values<0.001). NHs with higher percentages of white and older residents were 1–5% more likely to have ASCUs (p-values:<0.001). NHs located in Pacific, Middle Atlantic, and Southern regions had 58–69% lower odds of ASCUs (p-values<0.001) compared to NHs in the East North Central region. Compared to NHs in the most metropolitan areas, NHs in or near metropolitan counties with fewer than 1 million population had 25–47% higher odds of ASCUs (p-values<0.01); NHs in counties with urban population of 2,500–19,999 adjacent to metro areas had 44% higher odds of ASCUs (pvalue:0.002). In the most rural areas, odds of NH ASCUs decreased 63% (p-value:<0.001). ADI was not statistically significantly associated with presence of NH ASCUs.

Table 2:

Logistic regression results for presence of nursing home Alzheimer’s special care units.

	Odds Ratio	95% Confidence Interval	p-value
Key Covariates
State Area Deprivation Index	1.00	(0.98, 1.02)	0.854
Region (ref=East North Central):
East South Central	0.42	(0.32, 0.55)	<0.001
Middle Atlantic	0.42	(0.34, 0.52)	<0.001
Mountain	1.37	(1.07, 1.74)	0.011
New England	0.81	(0.65, 1.01)	0.061
Pacific	0.31	(0.23, 0.41)	<0.001
South Atlantic	0.31	(0.24, 0.38)	<0.001
West North Central	1.31	(1.10, 1.56)	0.003
West South Central	0.40	(0.32, 0.50)	<0.001
RUC Code (ref=1)*
2	1.25	(1.07, 1.45)	0.004
3	1.47	(1.22, 1.78)	<0.001
4	1.39	(1.10, 1.74)	0.005
5	1.05	(0.71, 1.52)	0.799
6	1.44	(1.15, 1.81)	0.002
7	1.23	(0.94, 1.61)	0.136
8	0.65	(0.38, 1.09)	0.113
9	0.37	(0.21, 0.61)	<0.001
Nursing Home-Level Covariates
For-profit	0.58	(0.52, 0.66)	<0.001
Chain affiliation	1.34	(1.19, 1.51)	<0.001
Number of beds	1.01	(1.01, 1.01)	<0.001
Occupancy	1.01	(1.01, 1.02)	<0.001
RN hours per resident day	0.54	(0.43, 0.68)	<0.001
Total nurse staffing hours per resident day	1.29	(1.19, 1.40)	<0.001
% Medicare	0.97	(0.96, 0.97)	<0.001
% Medicaid	1.00	(1.00, 1.01)	0.085
% female	1.00	(0.99, 1.00)	0.337
% white	1.01	(1.01, 1.01)	<0.001
Average resident age	1.05	(1.03, 1.06)	<0.001
Case mix acuity	0.99	(0.94, 1.04)	0.638
County nursing home bed competition (1-Herfindahl-Hirschman Index) †	0.81	(0.58, 1.12)	0.196
Observations		12,343
c-statistic		0.77
${\chi }_{30}^2$		<0.001

^*: Rural-Urban Continuum Code: 1) counties in metro areas of 1 million population or more (reference); 2) counties in metro areas of 250,000 to 1 million population; 3) counties in metro areas of fewer than 250,000 population; 4) urban population of 20,000 or more, adjacent to a metro area; 5) urban population of 20,000 or more, not adjacent to a metro area; 6) urban population of 2,500 to 19,999, adjacent to a metro area; 7) urban population of 2,500 to 19,999, not adjacent to a metro area; 8) completely rural or less than 2,500 urban population, adjacent to a metro area; and 9) completely rural or less than 2,500 urban population, not adjacent to a metro area.

^†: Score from 0–1 with higher values indicating greater competition.

DISCUSSION

This is a first national study of geographic distributions of NH ASCUs and their characteristics. Despite increasing prevalence of ADRD in Western and Southern areas of the U.S.,¹ NHs with ASCUs are predominately located in North Central and Middle Atlantic regions. NH ASCUs may depend more on geographic region and NH characteristics than local neighborhood resources.

Regional variations in provider practices have been recognized since the 1970s.¹⁷ Recently, studies are including neighborhood-level resources in examinations of healthcare utilization using the ADI. Researchers determined that ADI is associated with cognitive function,¹⁶ 30-day rehospitalizations,^11,14 and multiple chronic conditions,¹² among others. Our study utilizes the ADI in a first-ever approach, examining neighborhood deprivation and its relation to specialized NH dementia care. While we did not find statistically significant associations, there are several potential explanations.

Related to regional practices, NHs appear responsive to local NH competition;⁹ therefore, if ASCUs are not in local markets, NHs may be less motivated to make additional investments. However, county NH bed competition was not statistically significant in our model. Prior studies identify rural-urban differences in care patterns and preferences among dying NH residents with ADRD, supporting our finding of regional variation of ASCUs: more intensive medical measures were less likely preferred and delivered to rural residents.^23,24 Further exploration is needed to understand whether geographic variations in NH resources (e.g., RN workforce, staff training in dementia care) are associated with resident outcomes. Presence or absence of NH ASCUs does not necessarily indicate the quality of care received by residents with ADRD. Perhaps some NHs are investing more in formal dementia care staff training or following other best practice recommendations;²⁵ others may be investing in culture change practices promoting resident-centered care.^26,27

Although heavily regulated in other areas, there are currently no federal regulations of NH ASCUs. However, assisted living facilities (ALFs) have similar special care units for residents living with ADRD (22%)²⁸ which are state regulated. State variations in frequency and stringency of regulations^28,29 impact end-of-life care trajectories among ALF residents.³⁰ State-level regulation may influence NHs’ willingness to establish ASCUs, therefore geographically-tailored interventions to address uptake of ASCUs may be needed.

We found NHs that were non-profit, chain-affiliated, and had higher total staffing hours were more likely to have ASCUs; surprisingly, more RN staffing hours was associated with lower odds of ASCUs. Perhaps ASCUs reduce RN workload burden or RNs are better equipped to care for residents with ADRD without formal ASCUs. Regarding ownership and chain affiliation, our findings suggest motivations for ASCUs may not be profit-oriented and may be more need-focused. Additionally, we identified potential disparities in access to NH ASCUs: NHs with more white and older residents were more likely to have ASCUs, while NHs with more Medicare residents were less likely to have ASCUs. These highlight potential influences of financial resources in implementing ASCUs: specialty dementia care may be more necessary for residents in later stages of the disease when they are more likely long-stay residents and not receiving Medicare-funded SNF care.

We note several limitations. First, our outcome variable does not indicate scope of ‘units’, only whether ADRD beds are designated. However, that NHs report specialized beds for residents with ADRD suggests attention to resident-centered care. Second, we were unable to incorporate the extent other long-term care options for dementia care (e.g., ALFs, home and community-based services) contribute to NHs’ intent for designating ASCUs. Nevertheless, we included local competition for NH beds to offset this concern. Regarding the ADI, our study is limited by shortcomings of Census data, including potential underrepresentation of population groups most in need of NH care.¹⁵

CONCLUSIONS AND IMPLICATIONS

Our findings demonstrate presence of NH ASCUs varies by region, rurality, and NH features including workforce characteristics more so than neighborhood resources. As debates over the value of NH ASCUs continue, our study appears to support notions that ASCUs have been established to address resident needs rather than NH marketability. We demonstrated the feasibility of using the ADI in health services research among long-term care settings and potential for meaningful contributions to policy and intervention development. Future research is needed to examine NH quality and presence of ASCUs through a geographic lens and identify associations with resident outcomes.