State Regulations and Assisted Living Residents' Potentially Burdensome Transitions at the End of Life

Abstract

Background:

Potentially burdensome transitions at the end of life (e.g., repeated hospitalizations toward the end of life and/or health care transitions in the last three days of life) are common among residential care/assisted living (RC/AL) residents, and are associated with lower quality of end-of-life care reported by bereaved family members. We examined the association between state RC/AL regulations relevant to end-of-life care delivery and the likelihood of residents experiencing potentially burdensome transitions.

Methods:

Retrospective cohort study combining RC/AL registries of states' regulations with Medicare claims data for residents in large RC/ALs (i.e., 25+ beds) in the United States on the 120th day before death (N = 129,153), 2017–2019. Independent variables were state RC/AL regulations relevant to end-of-life care, including third-party services, staffing, and medication management. Analyses included: (1) separate logistic regression models for each RC/AL regulation, adjusting for sociodemographic covariates; (2) separate logistic regression models with a Medicare fee-for-service (FFS) subgroup to control for comorbidities, and (3) multivariable regression analysis, including all regulations in both the overall sample and the Medicare FFS subgroup.

Results:

We found a lack of associations between potentially burdensome transitions and regulations regarding third-party services and staffing. There were small associations found between regulations related to medication management (i.e., requiring regular medication reviews, permitting direct care workers for injections, requiring/not requiring licensed nursing staff for injections) and potentially burdensome transitions.

Conclusions:

In this cross-sectional study, the associations of RC/AL regulations with potentially burdensome transitions were either small or not statistically significant, calling for more studies to explain the wide variation observed in end-of-life outcomes among RC/AL residents.

Introduction

Little is known about the quality of end-of-life care delivered in residential care/assisted living (RC/ALs), although they are an increasingly common site of end-of-life care. From 2011 to 2015, the percentage of community-dwelling hospice beneficiaries who died in RC/ALs increased from 11% to 18%.¹ There is also increasing demand for complex medical care due to increasing comorbidities among RC/AL residents. From 2007 to 2018, the percentage of RC/AL residents with more than six chronic conditions increased from 46.5% to 60.5%.² Over 80% of RC/AL residents need medication assistance.³ There are concerns about the quality of end-of-life care in RC/ALs due to their lack of capacity to meet complex care needs. Compared with those receiving hospice care at home, bereaved next of kin of RC/AL decedents report lower quality of care.⁴ An analysis using Medicare Current Beneficiary Survey data reported that almost one quarter of RC/AL residents who died between 2002 and 2018 lived in communities with a lack of services for care needs like skilled nursing or medication assistance.⁵ Poorly timed or inadequate care for frail older adults can contribute to poor end-of-life care outcomes.

Potentially burdensome transitions at the end of life are defined as three or more hospitalizations in the last 90 days of life, two or more hospitalizations for urinary tract infections, sepsis, pneumonia, or dehydration in the last 120 days of life, or any health care transition in the last 3 days of life.^1,6,7 These transitions provide an important quality indicator for end-of-life care, as research shows associations between shortened survival time and repeated hospitalizations for the same medical condition,^8,9 the poor health outcome of repeated hospitalizations at the end of life,¹⁰ and that health care transitions in the last three days of life were associated with lower satisfaction among bereaved next of kin.¹¹ Potentially burdensome transitions were also associated with other indicators of poor-quality end-of-life care in nursing homes (e.g., tube feeding insertion).⁷ Approximately one in five RC/AL decedents experienced potentially burdensome transitions in 2018,⁶ a rate similar to nursing home residents.⁷ There was also wide variation in these transitions at the state level.⁶ However, it is unclear how state- and license-level RC/AL regulations are related to the likelihood of residents experiencing these transitions.

Unlike nursing homes, RC/ALs are solely regulated and licensed by states. Prior research has illustrated wide within- and between-state variation in care provision regulations,¹² potentially shaping end-of-life care delivery. For instance, although hospice plays a crucial role in RC/ALs in facilitating hospice utilization¹³ and dying in place,^14–17 state regulations vary in their support for residents enrolled in hospice.^13,15 RC/AL staffing impacts resident-centered outcomes like hospitalizations and service utilization.^2,18 In 2018, although almost all RC/ALs operated under licenses with staffing requirement for direct care workers (DCWs), only two thirds were required to have registered nurses (RNs) and less than one third were required to have licensed practical nurses (LPNs) or licensed vocational nurses (LVNs).²

Medication management was listed as one of the top three concerns regarding quality of care in RC/ALs by the General Accounting Office.¹⁹ Different from nursing homes, medication administration in a majority of RC/ALs relies on DCWs,^19–21 and 36 states permitted DCWs to do so in 2014.²² In this article, DCWs are defined as DCWs without licensure or certification, not including certified nurse aides or certified medication aides.

With RC/ALs now a common location for end-of-life care, it is important to examine regulations impacting their end-of-life care delivery. Our objective was to describe the extent to which regulations relevant to end-of-life care delivery (i.e., third-party services, staffing, and medication management) are associated with the likelihood of RC/AL residents experiencing potentially burdensome transitions toward the end of life.

Methods

Research design and study population

This is a retrospective cohort study of Medicare beneficiaries who died between January 1st, 2017 and December 31st, 2019, and whose enrollment 9-digit ZIP code corresponded to an RC/AL with 25 beds or more.²³ Based on a previously validated approach,^6,7 the sample was further restricted to residents present in RC/AL on the 120th day before death, to exclude residents who might have been transferred to other care settings (e.g., nursing homes). We focused on the contiguous United States and excluded Connecticut and Minnesota because these two states did not have sufficient data regarding licensed RC/AL location and capacity during the study period,²² preventing us from linking RC/AL residents to RC/ALs with known regulations based on ZIP codes.

Data sources

We retrieved publicly available RC/AL data from state licensing registries in 2017, including license type and ZIP code. These RC/AL registries were combined with an inventory of 2018 RC/AL regulations across license types.¹² For the outcome, repeated hospitalizations were identified with the Medicare Provider Analysis and Review (MedPAR) file, which captured a majority of the hospitalizations among Medicare beneficiaries, including Medicare Advantage beneficiaries.²⁴ Decedents' locations were inferred each day using multiple administrative claims and assessment records from the Center for Medicare and Medicaid Services (see Study Measures section).²⁵ We created several covariates from multiple data sources. RC/AL resident characteristics and comorbidities were obtained from the Medicare Beneficiary Enrollment Summary File and Chronic Condition Warehouse, respectively. Hospice enrollment was from Medicare hospice claims. The rate of hospital admission among Medicare decedents six months before death by hospital service areas (HSAs) (i.e., HSA hospitalization rates) was from the Dartmouth Atlas Project, as a geographic indicator for aggressive end-of-life care.^26,27

Study measures

Our primary outcome was the likelihood of RC/AL residents experiencing any type of potentially burdensome transitions in the last 120 days of life. Based on previous work,^6,7 three types of potentially burdensome transitions were considered: (1) three or more all-cause hospitalizations during the last 90 days of life, (2) two or more hospitalizations for urinary tract infections, sepsis, pneumonia, or dehydration during the last 120 days of life, and (3) health care transitions during the last 3 days of life.⁶ The first two types of transitions were identified using the MedPAR file. Transitions during the last three days of life were determined using the Residential History File and hospice claims, a validated method tracking the daily location of Medicare beneficiaries' health service utilization across multiple types of administrative claims and assessment data.²⁵

Our independent variables included RC/AL license-level state regulations in three care domains: third-party services (i.e., allow hospice/home health services, allow hiring private care aides), staffing (i.e., require RNs or LPN/LVNs, require minimum ratios/levels for DCWs), and medication management (i.e., require regular medication reviews, requirements regarding DCWs or RN/LPN/LVN for oral medication and injection administration). These domains were selected from 23 domains from a prior extensive policy analysis¹² and were identified as the most relevant to end-of-life care delivery in RC/ALs based on clinical input and empirical findings.^{2,18–21,28–30} Table 1 details each regulation and rationale for inclusion.

Table 1.

Residential Care/Assisted Living Regulations Relevant to End-of-Life Care Delivery

Regulations	Coding based on regulation data governing the license type	Rationale
Third-party services
Allow hospice services	Allow = explicitly allow persons on hospice to be admitted OR explicitly allow residents to use hospice services; Silent = silent on allowing residents to use hospice services AND is either silent on allowing persons on hospice to be admitted OR does not allow persons on hospice to be admitted (2 license types, merged with silent because of small sample size of 635 residents)	Research shows a positive association between utilization of third-party services and retention in RC/ALs.13,28 Third-party services could meet the growing care needs of residents approaching the end of life
Allow private care aide services	Allow = explicitly allows residents to hire private care aides (does not include family care); Silent = silent regarding private care aides (does not include family care)
Allow home health services	Allow = explicitly allows residents to use home health services; Silent = silent pertaining to home health services
Staffing
Require RNs	Require = has a requirement for RNs (e.g., ratio, staffing level, specified responsibilities); Silent = silent regarding RN staffing	Recoded variables based on the distribution of staffing specificity measures developed by Thomas et al.2 The measures were shown to be associated with hospitalization rates. Specifically, increases in DCW specificity were associated with decreased hospitalization rates among RC/AL residents.2 In addition, research in nursing homes consistently shows the positive relationship between residents' outcomes and requirements on increasing staffing levels46,47
Require LPN/LVNs	Require = has a requirement for LPN/LVNs (e.g., ratio, staffing level, specified responsibilities); Silent = silent regarding LPN or LVN staffing
Require minimum ratios/levels for DCWs	Require minimum ratios/levels = specifies staffing ratios or minimum staffing levels for DCWs (i.e., DCWs without certification or licenses, not including certified nurse aides or medication aides); Other requirements or silent = silent pertaining to DCW staffing level OR requires DCWs, but does not specify levels or ratio OR specifies that DCW staffing levels or ratios are “sufficient” to meet acuity of the residents (a majority of the sample (72%) resided in RC/ALs with such regulation)
Medication management
Require regular medication reviews	Require = explicitly requires scheduled reviews of residents' medications; Silent = silent regarding regular medication reviews	Indicator for regular monitoring of medication management practices. This is important given that RC/AL residents take multiple medications on a daily basis,37 likely even more toward the end of life, and half of the medications are distributed by DCWs,42 who do not have formal training in assessing resident health conditions, monitoring potential adverse drug events and drug interactions.20,43 Depending on state regulations, regular medication reviews might cover a variety of topics, such as the type of staffing required for the review, reviewing medication inventory, safe medication storage, accurate dosage and timing for medication administration, etc.19,20,27
Require RN/LPN/LVNs to administer oral medication	Require = explicitly requires RNs OR LPN/LVNs to administer oral medication; Not require = explicitly does not require RNs OR LPN/LVNs to administer oral medication; Silent = silent pertaining to oral medication administration by BOTH RNs AND LPN/LVNs. Note: no license types have conflicting regulations regarding RNs vs. LPN/LVNs (e.g., explicitly requires RN but explicitly does not require LPN/LVN, or vice versa). The license types that do not require RNs to administer oral medication are also the licenses that do not require LPN/LVNs to administer oral medication	Although licensed nursing staff are more qualified to administer medications, having such a regulation in place could create barriers in timely medication administration given the lack of presence of RN/LPN/LVN in some RC/ALs, in addition to the costs for hiring nurses and potential nurse shortage in local markets.50 Permitting DCWs to administer medications may facilitate medication management, although DCWs may not have enough training to ensure safe medication administration.20,43 Note: CNA or medication aides are not included in this category. We did not include the regulation requiring CNAs to administer oral medications due to the low prevalence of this regulation in our sample (8%)
Permit DCWs to assist with oral medication	Permit = explicitly permits DCWs to prepare and help residents take oral medication; Prohibit = explicitly does not permit DCWs to prepare and help residents take oral medication; Silent = silent pertaining to oral medication by DCWs
Require RN/LPN/LVNs to administer injections	Require = explicitly requires RNs OR LPN/LVNs to administer oral medication; Not require = explicitly does not require RNs OR LPN/LVNs to administer oral medication; Silent = silent regarding oral medication administration by BOTH RNs AND LPN/LVNs. Note: Similar to the oral medication regulation, no license types have conflicting regulations regarding RNs VS LPN/LVNs (e.g., explicitly requires RN but explicitly does not require LPN/LVN, or vice versa)
Permit DCWs to administer injections	Permit = explicitly permits DCWs to administer injections; Prohibit = explicitly does not permit DCWs to administer injections; Silent = silent pertaining to injection administration by DCWs

AL, assisted living; CNA, certified nurse aides; DCW, direct care worker; LPN, licensed practical nurse; LVN, licensed vocational nurse; RC, residential care; RN, registered nurse.

Covariates included RC/AL decedent sociodemographic characteristics (i.e., age, sex, race/ethnicity, dual eligibility status), Medicare Advantage enrollment, hospice enrollment on the 120th day before death (i.e., early hospice enrollment), length of stay in RC/ALs, and HSA hospitalization rates. Early hospice enrollment accounted for personal preferences in the approach to care and that those with very early hospice enrollment are unlikely to have any potentially burdensome transitions. We did not use other hospice utilization measures toward end of life (e.g., hospice utilization during the last month of life) as it was not a clear indicator for personal preferences, given that short-term hospice might be accessed after hospitalizations.^31,32 Comorbidities were included as covariates for the Medicare fee-for-service (FFS) subgroup (see Table 5 footnote b for details).

Table 5.

The Association Between Residential Care/Assisted Living Regulations and Potentially Burdensome Transitions Among Residential Care/Assisted Living Decedents, A Subgroup of Decedents with Medicare Fee-For-Service Enrollment (N = 87,970, Adjusted Odds Ratio, 95% Confidence Interval)

	Model 1a	Model 2b	Model 3c
Third-party services
Allow hospice services (ref: silent)	0.99 (0.94, 1.04)	0.98 (0.93, 1.03)	0.97 (0.88, 1.06)
Allow home health services (ref: silent)	1.01 (0.96, 1.06)	1.00 (0.96, 1.05)	1.00 (0.92, 1.08)
Allow private care aide services (ref: silent)	0.97 (0.93, 1.01)	0.97 (0.93, 1.01)	1.00 (0.94, 1.06)
Staffing
Require RNs (ref: silent)	0.99 (0.95, 1.02)	0.99 (0.95, 1.03)	1.04 (0.98, 1.09)
Require LPN/LVNs (ref: silent)	1.01 (0.96, 1.05)	0.99 (0.95, 1.03)	0.97 (0.92, 1.04)
Require minimum ratios/levels for DCWs (ref: other requirements or silent)	1.05* (1.01, 1.10)	1.03 (0.98, 1.08)	1.08* (1.01, 1.15)
Medication management
Require regular medication reviews (ref: silent)	0.91* (0.87, 0.94)	0.95* (0.91, 0.99)	0.99 (0.94, 1.03)
Require RN/LPN/LVNs to administer oral medication
Require (ref: silent)	1.13* (1.08, 1.18)	1.09* (1.04, 1.14)	1.04 (0.98, 1.11)
Not require (ref: silent)	1.05 (0.98, 1.13)	1.01 (0.93, 1.08)	0.93 (0.81, 1.06)
Permit DCWsd to assist with oral medication administration
Permit (ref: silent)	0.93* (0.87, 0.99)	0.95 (0.88, 1.01)	0.94 (0.87, 1.02)
Prohibit (ref: silent)	1.05 (0.95, 1.15)	1.02 (0.93, 1.12)	0.90 (0.77, 1.04)
Require RN/LPN/LVNs to administer injections
Require (ref: silent)	1.05* (1.01, 1.10)	1.05* (1.01, 1.10)	1.11* (1.03, 1.20)
Not require (ref: silent)	1.12* (1.05, 1.20)	1.08* (1.01, 1.16)	1.26* (1.11, 1.45)
Permit DCWs to administer injections
Permit (ref: silent)	0.89* (0.84, 0.94)	0.90* (0.85, 0.95)	0.88* (0.81, 0.94)
Prohibit (ref: silent)	1.02 (0.98, 1.06)	1.01 (0.97, 1.05)	0.95 (0.89, 1.02)

^* Indicates statistical significance lower than 0.05.

Medicare FFS enrollment is defined as being enrolled as a Medicare FFS beneficiary for at least 12 months before death.

^a Model 1 reports regression model results of the association between each RC/AL regulation and potentially burdensome transitions without including comorbidities. Each model included the same set of covariates: RC/AL decedent sociodemographic characteristics (i.e., age, sex, race/ethnicity, dual eligibility), Medicare Advantage enrollment, early hospice enrollment, length-of-stay in RC/ALs, HSA hospitalization rate, and HSA random intercepts.

^b Model 2 reports regression model results of the association between each RC/AL regulation and potentially burdensome transitions, after controlling for comorbidities. Each model included comorbidities and the same set of covariates as specified in Model 1. Comorbidities were obtained from the Chronic Condition Warehouse, including Alzheimer's Disease and Related Dementia, Acute Myocardial Infarction, Asthma, Atrial Fibrillation, Congested Heart Failure, Hip Fracture, Chronic Obstructive Pulmonary Disease, Chronic Kidney Disease, Diabetes, Ischemic Heart Disease, Stroke, and a dummy variable indicating whether the individual had more than 5 chronic conditions.

^c Model 3 reports multivariable regression results of all RC/AL regulations and potentially burdensome transitions. The model included the same set of covariates as specified in Model 2.

^d DCWs refer to DCWs with minimal training, not including certified nurse aides or medication aides.

FFS, fee-for-service.

Statistical analyses

We first reported descriptive statistics of RC/AL decedents and regulations of interest. We took steps to estimate the association between RC/AL regulations and the likelihood of experiencing potentially burdensome transitions. First, to understand how each RC/AL regulation was related to the outcome, we examined the association in separate regression models, adjusting for sociodemographic covariates. Second, to test whether resident comorbidities influenced the association between the outcome and regulations, we conducted a subgroup analysis of decedents with continuous Medicare FFS enrollment in the last year of life. The sample restriction was added because comorbidities are underreported in the Medicare Advantage population. Finally, to account for potential interactions among RC/AL regulations, we conducted multivariable analyses, including all RC/AL regulations in one model, in both the overall sample and the Medicare FFS subgroup. All analyses were conducted at the RC/AL resident level and with HSA-level random intercepts. The random intercept accounted for regional clustering in the outcome among RC/AL decedents, potentially due to regional differences in preferences, case-mix, and local medical practice patterns.²⁶ Analyses were conducted with SAS version 9.4 and Stata version 17. This study relied on public use files and secondary administrative claims data, and was exempt from review according to the Brown University Institutional Review Board.

Results

RC/AL resident characteristics

There were 129,153 RC/AL residents who died between 2017 and 2019 and were present in RC/AL on the 120th day before death, residing in 10,993 unique RC/ALs operating under 143 different licenses. The overall rate of potentially burdensome transitions was 20.6%. A majority of RC/AL decedents were 85 years or older (76.3%), female (63.5%), and White (94.6%) (Table 2). Approximately 15% of the sample was eligible for both Medicare and Medicaid, and 31.9% were enrolled in Medicare Advantage. Approximately 15% of RC/AL decedents had early hospice enrollment. The median RC/AL length of stay was 3.7 years (interquartile range 2.1–6.4).

Table 2.

Residential Care/Assisted Living Decedents Characteristics (N = 129,153)

	Frequency (%)
Age category
65–74	7797 (6.0)
75–84	22,817 (17.7)
85+	98,539 (76.3)
Sex
Male	47,163 (36.5)
Female	81,990 (63.5)
Race
White	122,120 (94.6)
African American	2984 (2.3)
Hispanic	2023 (1.6)
Other	2026 (1.6)
Medicare advantagea	41,183 (31.9)
Medicaid and Medicare eligibilityb	19,404 (15.0)
Early hospice enrollmentc	19,593 (15.2)
RC/AL length of stay, median [IQR]	3.7 [2.1, 6.4]

RC/AL decedents are defined as RC/AL residents who died 2017–2019, were present in RC/ALs and were not enrolled in hospice on the 120th day before death.

^a Medicare Advantage enrollment is defined as ever being enrolled into Medicare Advantage in the last 12 months of life.

^b Medicare and Medicaid eligibility is defined as ever eligible for Medicare and Medicaid in the last 12 months of life.

^c Early hospice enrollment is defined as hospice enrollment on the 120th day before death.

IQR, interquartile range.

The prevalence of RC/AL regulations

Table 3 reports the prevalence of each RC/AL regulation in the analytic sample, at the individual and license level, respectively. A majority of decedents lived in RC/ALs where state regulations explicitly allowed hospice services (84.8%) and home health services (79.0%), as opposed to states where regulations were silent on the use of these services. “Silent” was defined as an absence of RC/AL regulations enabling or restricting provider behavior specific to the provision examined.¹² The prevalence of regulations allowing private care aide services was comparatively low (35.7%). Regarding staffing, 60.9% and 35.7% of residents lived in communities requiring RNs and LPN/LVNs, respectively. Approximately one quarter resided in communities requiring minimum staffing ratios/levels for DCWs. Less than half of decedents (45.0%) lived in communities with licenses requiring regular medication reviews. A majority of decedents were in communities silent on whether RN/LPN/LPNs were required for oral medication administration (69.9%). About 82% of decedents lived in communities permitting DCWs to assist with oral medication administration.

Table 3.

Distribution of Residential Care/Assisted Living Regulations at the Decedent and License Level

	Decedent level (N = 129,153), Frequency (%)	License level (N = 143), Frequency (%)
Third-party services
Allow hospice services	109,544 (84.8)	105 (73.4)
Allow home health services	102,012 (79.0)	97 (67.8)
Allow private care aide services	46,094 (35.7)	55 (38.5)
Staffing
Require RNs	78,638 (60.9)	92 (64.3)
Require LPN/LVNs	46,078 (35.7)	55 (38.5)
Require minimum ratios/levels for DCWs	33,896 (26.2)	40 (28.0)
Medication management
Require regular medication reviews	58,166 (45.0)	76 (53.2)
Require RN/LPN/LVNs to administer oral medication
Require	28,216 (21.9)	30 (21.0)
Not require	10,557 (8.2)	8 (5.6)
Silent	90,380 (69.9)	105 (73.4)
Permit DCWs to assist with oral medication administration
Permit	106,072 (82.1)	118 (82.5)
Prohibit	10,034 (7.8)	7 (4.9)
Silent	13,424 (10.1)	18 (12.6)
Require RN/LPN/LVNs to administer injections
Require	38,204 (29.6)	49 (34.3)
Not require	12,841 (9.9)	9 (6.3)
Silent	78,485 (60.5)	85 (59.4)
Permit DCWs to administer injections
Permit	24,453 (18.9)	25 (17.5)
Prohibit	49,014 (38.0)	59 (41.3)
Silent	55,686 (43.1)	59 (41.3)

Nearly three in five decedents (60.5%) resided in communities that were silent on whether RN/LPN/LPNs were required for injection administration. About 4 in 10 decedents (43.1%) were in RC/ALs with silent regulations regarding DCWs for injection administration, whereas 18.9% and 38.0%, respectively, lived in communities permitting or prohibiting DCWs to administer injections. The prevalence of regulations at the license and decedent levels were similar. Figure 1 illustrates the between- and within-state variation in RC/AL regulations, using injection administration requirements pertaining to different types of staff as an example. DCW injection administration was permitted in 6 states and prohibited in 10 states. While 24 states were silent on this regulation, 10 states had inconsistent regulations across licenses (i.e., “permit + silent” or “prohibit + silent”).

FIG. 1.

State regulations pertaining to injection administration by staffing in RC/AL. “Inconsistent across licenses” could be a combination of regulations permitting DCWs to administer injections, or silent on this regulation, across different license types in one state (i.e., “permit+silent”). Another possible combination could be “prohibit+silent” pertaining to DCWs. Regarding regulations involving RN/LPN/LVNs for medication management, “inconsistent across licenses” could be “require+silent” or “not require+silent.” Regulations for Alaska, Hawaii, and Minnesota were shown on the map, but not included in the analysis. AL, assisted living; DCW, direct care worker; LPN, licensed practical nurse; LVN, licensed vocational nurse; RC, residential care; RN, registered nurse.

Associations between RC/AL regulations and potentially burdensome transitions in the overall sample

Table 4 presents adjusted odds ratios (AOR) and 95% confidence intervals (CI) for the association between RC/AL regulations and the outcome of interest, in separate regression models (Model 1) and then together in one model (Model 2). Most regulations used “silent” as the reference group (Table 1). In Model 1, the presence of regulations requiring regular medication reviews was associated with significantly lower odds of potentially burdensome transitions (AOR 0.92, CI 0.89–0.95). We estimated predicted probabilities to facilitate the interpretation of the magnitude, specifically, the average predicted probability of experiencing potentially burdensome transitions for residents who lived in RC/ALs operating under licenses requiring regular medication reviews was 19.9%, in comparison to 21.1% among those living in RC/ALs with licenses silent on this regulation. Regulations permitting DCWs to administer injections was associated with a lower likelihood of transitions (AOR 0.93, CI 0.89–0.98).

Table 4.

The Association Between Residential Care/Assisted Living Regulation and Potentially Burdensome Transitions Among Residential Care/Assisted Living Decedents (N = 129,153, Adjusted Odds Ratio, 95% Confidence Interval)

	Model 1a	Model 2b
Third-party services
Allow hospice services (ref: silent)	0.99 (0.95, 1.04)	0.97 (0.90, 1.06)
Allow home health services (ref: silent)	1.02 (0.98, 1.07)	1.02 (0.95, 1.09)
Allow private care aide services (ref: silent)	0.96 (0.93, 1.01)	0.98 (0.93, 1.03)
Staffing
Require RNs (ref: silent)	0.99 (0.96, 1.03)	1.04 (0.99, 1.09)
Require LPN/LVNs (ref: silent)	1.02 (0.99, 1.06)	1.00 (0.94, 1.05)
Require minimum ratios/levels for DCWs (ref: other requirements or silent)	1.05* (1.01, 1.09)	1.10* (1.04, 1.06)
Medication management
Require regular medication reviews (ref: silent)	0.92* (0.89, 0.95)	0.96* (0.92, 0.99)
Require RN/LPN/LVNs to administer oral medication
Require (ref: silent)	1.13* (1.08, 1.18)	1.08* (1.02, 1.13)
Not require (ref: silent)	1.10* (1.03, 1.17)	1.02 (0.91, 1.15)
Permit DCWsc to assist with oral medication administration
Permit (ref: silent)	0.96 (0.90, 1.02)	0.93* (0.87, 0.99)
Prohibit (ref: silent)	1.06 (0.97, 1.15)	0.92 (0.81, 1.05)
Require RN/LPN/LVNs to administer injections
Require (ref: silent)	1.05* (1.01, 1.09)	1.12* (1.05, 1.19)
Not require (ref: silent)	1.14* (1.08, 1.21)	1.19* (1.06, 1.33)
Permit DCWs to administer injections
Permit (ref: silent)	0.93* (0.89, 0.98)	0.90* (0.84, 0.96)
Prohibit (ref: silent)	1.03 (0.99, 1.07)	0.96 (0.91, 1.01)

^* Indicates statistical significance lower than 0.05.

^a Model 1 reports regression model results of the association between each RC/AL regulation and potentially burdensome transitions. Each model included the same set of covariates: RC/AL decedent sociodemographic characteristics (i.e., age, sex, race/ethnicity, dual eligibility), Medicare Advantage enrollment, early hospice enrollment, length of stay in RC/ALs, HSA hospitalization rate, and HSA random intercepts.

^b Model 2 reports multivariable regression results of all RC/AL regulations and potentially burdensome transitions. The model included the same set of covariates specified above.

^c DCWs refer to DCWs with minimal training, not including certified nurse aides or medication aides.

HSA, hospital service area.

A significantly higher rate of these transitions was associated with regulations requiring minimum ratios/levels for DCWs (AOR 1.05, CI 1.01–1.09), regulations regarding RN/LPN/LVNs for oral medication administration (Require vs. Silent: AOR 1.13, CI 1.08–1.18; Not Require vs. Silent: AOR 1.10, CI 1.03–1.17) and injection administration (Require vs. Silent: AOR 1.05, CI 1.01–1.09; Not Require vs. Silent: AOR 1.14, CI 1.08–1.21). In Model 2, after accounting for interactions among regulations, the results were similar in significance and magnitude.

Associations between RC/AL regulations and potentially burdensome transitions in the Medicare FFS subgroup

Table 5 presents the association between RC/AL regulations and potentially burdensome transitions in different model specifications, among the Medicare FFS subgroup. Model 1 and Model 2 report AORs and CIs when RC/AL regulations were separately included in the model, without and with adjusting for comorbidities, respectively. Model 3 presents multivariable regression results, controlling for all covariates, including comorbidities. After controlling for comorbidities (Model 2 vs. Model 1), the regulations that remained significant were: requiring regular medication reviews (AOR 0.95, CI 0.91–0.99), requiring RN/LPN/LVNs to administer oral medications (AOR 1.09, CI 1.04–1.14), requiring RN/LPN/LVNs to administer injections (AOR 1.05, CI 1.01–1.10), not requiring RN/LPN/LVNs to administer injections (AOR 1.08, CI 1.01–1.16), and permitting DCWs to administer injections (AOR 0.90, CI 0.86–0.94).

After accounting for correlations, the regulations that remained significant in Model 3 were: requiring minimum ratios/levels for DCWs, requiring or not requiring RN/LPN/LVNs to administer injections, and permitting DCWs to administer injections. Although the positive associations pertaining to RN/LPN/LVN injection administration became stronger in multivariable analysis (Model 3 vs. Model 2), they could potentially be driven by certain states when holding other regulations constant. As Figure 1 showed, there were 17 states with some or all license types that explicitly required or did not require RN/LPN/LVN for injections.

Discussion

End-of-life care delivery in RC/AL necessitates a delicate balance between meeting residents' increasing care needs toward the end of life, respecting residents' desire to age in place, and complying with regulations.^15,33 Although most RC/AL residents prefer to die in what they consider their permanent home,²⁹ state regulations may present barriers when residents' care needs increase in the end of life, leading to potentially burdensome transitions. Previous work reports wide state-level variation in potentially burdensome transitions among RC/AL decedents, speculating that state regulations might influence these variations.⁶ In this cross-sectional study, the associations were either small or not statistically significant, suggesting a limited role of state RC/AL regulations in explaining variations in this end-of-life care outcome.

Regulations regarding third-party services or staffing

We found a lack of association between potentially burdensome transitions and regulations regarding third-party services or staffing. It might be that these regulations do not represent the actual level of service provided or utilized in RC/ALs, given that RC/AL residents living in residences with silent regulations could still utilize these services or that RC/ALs have internal policies regarding third-party services or staffing. Future studies are also needed to examine end-of-life care outcomes in RC/AL after accounting for the potential substitution effect between utilizing RC/AL staff and hospice/home health staff.³⁴

The relationship between minimum ratios/levels for DCWs and the outcome varied in different model specifications, especially in the Medicare FFS subgroup, suggesting potential interactions between this regulation and resident comorbidities, as well as other relevant regulations (e.g., medication management). The optimal regulation specification for DCWs remains unclear. Although requiring minimum DCW levels might be more specific than “sufficient” ratios/levels,² some argue that the latter offers administrators more flexibility in adjusting staffing levels based on residents' changing care needs.³⁵ A longitudinal study examining changes in these regulations and end-of-life care outcomes could shed light on this relationship.

Medication management regulations: requiring regular medication reviews

We found a small and negative association between regulations requiring regular medication reviews and potentially burdensome transitions. Regulating routine medication reviews may indicate states' regulatory focus on safe medication management in RC/AL.^22,30,36 This is important for end-of-life care because polypharmacy is common among older adults in the last phase of life, making them particularly vulnerable to potential adverse events.³⁷ The small association found in this study could be a result of the wide variation in requirements on the frequency of reviews and the type of staffing.²² Prior studies suggest that states do not have consensus on what medication management review entails,³⁸ because it is a complex service involving tasks like maintaining appropriate inventory, safe storing, cueing residents, reading dosage instructions, and assessing/monitoring medication effects.^21,30,39

It is also unclear how regulations are related to the actual service provided regarding medication reviews in RC/AL.³⁹ Future studies should consider accounting for other regulations of relevance to medication reviews (i.e., frequencies and specific medication management activities required for the review), as well as the interaction between regulations and actual service provision.

Medication management regulations: medication administration by different types of staff

Our findings on medication management by DCWs have potential policy implications. Explicitly permitting DCWs to administer injections was associated with slightly lower rates of potentially burdensome transitions, which remained significant despite different sample and model specifications. Research suggests that family caregivers can perform complex tasks, including administering injections, with appropriate training and support from nurses.^40,41 In RC/AL, DCWs administer a majority of routine medications for residents.^20,42,43 Some states allow DCWs to administer as-needed medications (i.e., pro re nata),⁴⁴ which could be important for residents' acute or exacerbating need for symptom management in end of life, and therefore preventing unnecessary transitions.⁴⁵

However, more research is warranted on whether permitting DCWs to administer medications promotes better end-of-life care or carries other risks of adverse events, because DCWs would ideally be supported with appropriate training and ongoing supervision from licensed nursing staff.^38,42 Future work is needed to assess the effect of regulations pertaining to medication management involving DCWs in combination with other relevant regulations (e.g., Nurse Practice Act, DCW training/supervision requirements).

Explicitly requiring licensed nursing staff (i.e., require/not require vs. silent) to administer medications might ensure residents' safety, particularly in RC/ALs with higher-acuity residents, although such requirement may present additional barriers to service delivery in place.² Our findings indicate a small but positive association between the outcome and explicit medication administration regulations involving RN/LPN/LVNs, and that accounting for resident comorbidity does not entirely explain such positive association. Prior work shows mixed findings about the association between requirements increasing licensed nurse staffing and resident outcomes.^2,46,47 Additional work is needed to determine whether and how requirements about licensed nurse staffing could become a potential barrier to service delivery, after accounting for costs and availability of health care professionals in the local market.⁴³

Furthermore, this study uses silent regulations as the reference group in the analysis because RC/ALs operating under more explicit regulations (i.e., Require/Not Require vs. Silent, Permit/Prohibit vs. Silent) are, in general, under greater regulatory pressure than those operating under silent regulations.⁴⁸ In contrast, regulatory silence indicates a lack of action from state regulators.⁴⁸ However, it is also possible that RC/ALs in states with silent regulations already exceed minimum regulatory requirements because of consumer demand or other reasons.⁴⁹

Limitations

This study has several limitations. First, this cross-sectional study was descriptive in nature and did not allow us to test causality between these RC/AL regulations and potentially burdensome transitions. Such association could be influenced by several factors not captured in this study, such as additional levels of specification in regulations, other relevant policies (e.g., medication administration training/supervision requirements for DCWs, Nurse Practice Act²⁰), RC/AL resident acuity, on-site skilled nursing services, and family presence as important caregiving support. Second, our results are only generalizable to RC/AL residents in large RC/ALs (25+ beds). Our methodology of identifying RC/AL residents might capture individuals in other settings that share a 9-digit ZIP code with an RC/AL (e.g., independent living/home settings), although this is not a major concern with our focus on decedents, who have overall high levels of care needs.¹³ Despite these limitations, our study, to our knowledge, was the first to use a national RC/AL sample to examine the association between regulations relevant to end-of-life care delivery and the likelihood of residents experiencing potentially burdensome transitions.

Conclusion

We observed small or no associations between RC/AL regulations and potentially burdensome transitions in this cross-sectional analysis. More work is needed to disentangle the determinants of end-of-life care outcomes among RC/AL residents.

Authors' Contributions

X.W. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. X.W., J.M.T., N.R., L.S., K.S.T., D.D., P.L.G., P.C., and E.B.: Concept and design; acquisition, analysis, and interpretation of data; Critical revision of the article; statistical analysis; and administrative, technical, or material support. X.W.: Drafting the article. E.B.: Obtained funding. E.B.: Supervision.

Data Sharing Statement

The authors report that analytical syntax files can be accessed at https://doi.org/10.26300/hhy0-g366

Funding Information

This work was supported by the grant R01AG066902 from the National Institute on Aging. The funding had no influence on the design, data collection, analysis, and interpretation, nor the dissemination of the results of this study.

Author Disclosure Statement

No competing financial interests exist.