Abstract
Background:
Assisted living (AL) is an increasingly common, place of care for dying persons. However, it remains unclear to what extent residents are able to age in place or if AL represents an additional transition before death.
Objectives:
Examine the sociodemographic characteristics, comorbidities, health care utilization, and end-of-life care pathways of AL residents before death.
Research Design:
A national cohort study of fee-for-service Medicare beneficiaries residing in large AL communities (25+ beds) during the month of January 2017 with 3 years of follow-up, using administrative claims data.
Subjects:
268,812 AL residents.
Measures:
Sociodemographic characteristics, comorbidities, and health care utilization at the end of life.
Results:
Between 2017 and 2019, 35.1% of the study cohort died. Decedents were more likely than the overall AL population to be 85 years old or older (76.5% vs. 59.5%), and diagnosed with Alzheimer’s disease and related dementia (70.3% vs. 51.6%). Most decedents (96.2%) had some presence in AL during the last year of life, but over 1 in 5 left AL before the last month of life. Among those in AL on day 30 before death, nearly half (46.4%) died in place without any health care transition, while 13.2% had 3 or more transfers before dying.
Conclusions:
AL is an important place of care for dying persons, especially for those with dementia. These findings indicate a need to assess existing policies and processes guiding the care of the frail and vulnerable population of dying AL residents.
Keywords: assisted living, end of life, health care utilization, health care transitions, place of death
An increasing number of older adults reside in assisted living (AL) and remain in this care setting until the end of life.1 Much of what we know about residents’ end-of-life experiences comes from retrospective cohort studies and surveys. Retrospective decedent cohorts focus on those who die in AL or reside in this setting toward the end of their life, systematically excluding residents who may have considered AL their home, but may have had to relocate due to the unavailability of certain services or potential interaction between state regulations and growing medical care needs. Surveys provide a valuable snapshot of AL residents and their potential care needs,2 but lack insight into care transitions that may follow residential care placement. There remains a knowledge gap about the needs and characteristics of dying persons who reside in AL toward the end of their lives, particularly given that transitions near death can be especially burdensome for older adults and their families.3,4
The overall health care utilization and major end-of-life care pathways among AL residents as their health declines is an important indicator of potential unmet needs and lack of advance care planning that may prompt transitions. In addition, knowing how the overall population of AL residents compares with the subset of decedents who either die in place or transfer to another setting for end-of-life care is key to assessing the delivery of care to terminally ill residents in this setting. The proportion of residents who are able to age and die in place also has methodological implications in determining the appropriate look-back window to capture the outcomes for a majority of dying AL residents. A better understanding of the organizational processes and regulations that may facilitate the delivery of end-of-life care in AL to meet residents’ growing medical care needs in place also has policy implications. In this descriptive cohort study, our objectives were to (1) document the proportion of AL residents who die during a 3-year follow-up and compare the sociodemographic characteristics and comorbidities of decedents with the overall population of AL residents, (2) describe the overall health care utilization of decedents across disease groups, and (3) track the different care pathways of persons residing in AL on day 30 before death.
METHODS
We conducted a cohort study of fee-for-service Medicare beneficiaries age 65 and older who resided in an AL community with 25 or more beds during the month of January 2017, based on their home addresses being a validated 9-digit ZIP code indicating residence in an AL community.5 We then drew on data from the residential history file,6 which utilizes a variety of Medicare administrative data (ie, claims and mandatory assessment data) to provide a daily location for Medicare beneficiaries. In order of certainty, we generated daily indicators determining whether a given resident had been hospitalized on any given day based on having a MedPAR hospitalization record, or whether they had been admitted to a nursing home based on the presence of either a claim or having a nursing home Minimum Data Set assessment or had a hospice claim indicating place of service. Residents were otherwise considered to have remained in AL. While health care claims data contain a record of all services billed to the Center for Medicare and Medicaid Services, the place of hospice service is not an auditable field and there are some inconsistencies in reporting (eg, beneficiaries with validated AL ZIP code whose hospice claims indicate that they are receiving services at home).
A number of descriptive analyses were conducted. First, mortality was determined using the Medicare Beneficiary Summary File over 3 years of follow-up. We compared the overall population of AL residents and decedents in terms of sociodemographic characteristics and medical diagnoses. Second, health care utilization and site of death were characterized among those who died during the study period, examining differences between disease groups. Health care utilization comprised various indicators including the presence in AL during the last 12 months, 90 days, and 30 days. Hospice utilization, inpatient hospital admissions, and nursing home use were also examined. The place of death was classified as being either in AL, a nursing home, an inpatient hospital, or a hospice facility.
Given extensive multimorbidity in the study population, mutually exclusive disease groups were created using claims data to examine health care utilization among subgroups with established terminal care pathways of cancer, frailty including memory issues, or organ failure.7 End-of-year flags from the Chronic Condition Warehouse capture a variety of claims data sources during the prior 12 calendar months or the number of these months until death. We first determined if a beneficiary had a diagnosis of the 5 deadliest cancers in the United States (lung, colorectal, prostate, breast, or pancreatic), because we consider that these diagnoses provide a distinct and more predictable disease trajectory. In the absence of these cancers, beneficiaries with a diagnosis of Alzheimer’s disease and related dementia (ADRD) were classified together because memory issues are expected to inform clinical management of other chronic conditions. In the absence of these diagnoses, beneficiaries with chronic obstructive pulmonary disease (COPD) or heart failure (HF) were grouped under an organ failure disease profile of terminal decline. Health care utilization was examined across disease groups to detect any specific patterns in utilization. We also report these descriptive characteristics and health care utilization among those with and without ADRD, which was the most common life-limiting diagnosis among AL decedents. Supplementary Materials, Supplemental Digital Content 1, http://links.lww.com/MLR/C776 also provide these statistics stratified by the presence/absence of the 5 deadliest cancers, and COPD and/or HF (Tables S1-S4, Supplemental Digital Content 1, http://links.lww.com/MLR/C776).
Finally, specific end-of-life care pathways and transitions were tracked during the last 30 days of life among those present in AL on day 30 before death. Any transfer between settings was counted, so being sent to the hospital from AL and then returning directly to AL after discharge would represent 2 care transitions. Enrollment in the Medicare Hospice Benefit (as indicated in hospice claims data) was not considered a transfer unless there was a change of care setting (eg, transfer to a hospice facility). Unless a given AL community has specific care processes in place discouraging hospice enrollment among their residents, Routine Home Care and the more intensive Continuous Home Care hospice can both be provided in any home setting, including residential care. General Inpatient hospice, however, cannot be provided in AL and requires a transfer to an inpatient setting.
RESULTS
Decedents’ Sociodemographic Characteristics and Comorbidities
The study population represents a national cohort of 268,812 fee-for-service Medicare beneficiaries residing in large AL communities in January 2017. From 2017 to the end of 2019, over one third of AL residents died (35.1%; N = 94,455). As shown in Table 1, compared with the overall AL resident population, decedents were more likely to be 85 years old or older (76.5% vs. 59.5%), and to have ever had a diagnosis of ADRD (70.3% vs. 51.6%) or HF/COPD (69.4% vs. 53.7%) as an end-of-year diagnosis, painting a picture of a severely multimorbid decedent population. Table 2 provides additional descriptive statistics about AL residents and decedents with and without a diagnosis of ADRD.
TABLE 1.
Descriptive Characteristics of Fee-for-service Medicare Beneficiaries Residing in AL January 2017, and Prospective Decedents Sub-cohorts by 2019
| Characteristics | AL residents on 1/2017 | AL residents on 1/2017 Deceased by 12/2019 |
AL residents on 1/2017 Deceased by 12/2019 Present in AL on day 30 before death |
|---|---|---|---|
| N (% of total AL residents) | 268,812 | 94,455 (35.1) | 67,577 (25.1) |
| Demographic characteristics by January 2017—N (%) | |||
| Age 85+ | 160,045 (59.5) | 72,230 (76.5) | 52,468 (77.6) |
| Female gender | 183,553 (68.3) | 61,375 (65.0) | 44,006 (65.1) |
| Non-Hispanic, Caucasian | 250,768 (93.3) | 89,805 (95.1) | 64,505 (95.5) |
| Time in AL by January 2017 | |||
| Median (IQR) | 3.0 (1.2–6.0) | 2.8 (1.1–5.5) | 2.8 (1.1–5.5) |
| Chronic conditions on January 1, 2017, using EOY 2016—N (%) | |||
| Total number, mean (SD) | 5.2 (3.1) | 6.2 (3.1) | 6.1 (3.1) |
| Cancer* | 27,855 (10.4) | 11,721 (12.4) | 8,459 (12.5) |
| ADRD | 85,978 (32.0) | 44,965 (47.6) | 32,413 (48.0) |
| COPD/HF | 90,715 (33.7) | 45,821 (48.5) | 32,213 (47.7) |
| Chronic conditions by December 2019, using EOY 2016–2019—N (%) | |||
| Total number, mean (SD) | 8.2 (3.4) | 9.1 (3.1) | 8.7 (3.2) |
| Cancer* | 47,839 (17.8) | 19,976 (21.1) | 14,031 (20.8) |
| ADRD | 138,573 (51.6) | 66,372 (70.3) | 45,548 (67.4) |
| COPD/HF | 144,322 (53.7) | 65,510 (69.4) | 45,107 (66.7) |
Cancers are limited to lung, colorectal, breast, prostate, and pancreatic cancers.
ADRD indicates Alzheimer’s disease and related dementia; AL, assisted living; COPD, chronic obstructive pulmonary disease; EOY, end of year; HF, heart failure; IQR, interquartile range; SD, standard deviation.
TABLE 2.
Descriptive Characteristics of Fee-for-service Medicare Beneficiaries Residing in AL January 2017, Deceased in 2017–2019, by ADRD Diagnosis Status
| Characteristics | Decedents without an ADRD diagnosis by the time of death |
Decedents with an ADRD diagnosis by the time of death |
|---|---|---|
| N (% of total decedents) | 28,083 (29.7) | 66,372 (70.3) |
| Demographic characteristics by January 2017—N (%) | ||
| Age 85+ | 20,236 (72.1) | 51,994 (78.3) |
| Female gender | 17,490 (62.3) | 43,885 (66.1) |
| Non-Hispanic, Caucasian | 26,929 (95.9) | 62,876 (94.7) |
| Time in AL by January 2017 | ||
| Median (IQR) | 3.0 (1.2–6.0) | 2.7 (1.0–5.2) |
| Chronic conditions on January 1, 2017, using EOY 2016—N (%) | ||
| Total number, mean (SD) | 5.6 (3.0) | 6.5 (3.1) |
| Cancer* | 4288 (15.3) | 7433 (11.2) |
| ADRD | — | 44,965 (67.8) |
| COPD/HF | 14,428 (51.4) | 31,393 (47.3) |
| Chronic conditions by December 2019, using EOY 2016–2019—N (%) | ||
| Total number, mean (SD) | 8.2 (3.1) | 9.5 (3.1) |
| Cancer* | 7212 (25.7) | 12,764 (19.2) |
| COPD/HF | 20,396 (72.6) | 45,114 (68.0) |
Cancers are limited to lung, colorectal, breast, prostate, and pancreatic cancers.
ADRD indicates Alzheimer’s disease and related dementia; AL, assisted living; COPD, chronic obstructive pulmonary disease; EOY, end of year; HF, heart failure; IQR, interquartile range.
Overall Health Care Utilization and Place of Death by Disease Group
As shown in Table 3, most decedents (96.2%) had some presence in AL during the last 12 months of life, but more than 1 in 5 (21.5%) were no longer in AL during the last month before dying. Hospice utilization was overall high in the last month of life (67.5%), but substantially lower among those in the organ failure group (58.3%) than those with cancer (69.9%) and ADRD (70.7%). Inpatient hospitalization occurred for 70.7% of decedents during the last year of life, and nursing home stays were also common during this period (49.6%). Over half of all decedents (50.5%) died in AL, while 19.6% died in the hospital and 23.1% died in a nursing home. Decedents in the organ failure group were more likely to die in hospital (28.2%), especially compared to those with ADRD (17.0%). Compared to decedents without a diagnosis of ADRD (Table 4), decedents with a diagnosis of ADRD were less likely to remain in AL in the last month of life (75.7% vs. 85.3%), and more likely to have a nursing home stay (52.6% vs. 42.6%) and to die there (25.1% vs. 18.6%).
TABLE 3.
Health Care Utilization Among AL Decedents by Disease Group
| Utilization | All | Cancer* | ADRD without cancer* | Organ failure COPD/HF, without cancer* or ADRD |
|---|---|---|---|---|
| Total | ||||
| N (% of decedents) | 94,455 | 19,976 (21.1) | 53,608 (56.8) | 15,097 (16.0) |
| Presence in AL† | ||||
| Last 12 mo of life | 90,891 (96.2) | 19,350 (96.9) | 50,975 (95.1) | 14,859 (98.4) |
| Last 90 d of life | 83,252 (88.1) | 17,776 (89.0) | 45,874 (85.6) | 14,083 (93.3) |
| Last 30 d of life | 74,193 (78.5) | 15,687 (78.5) | 40,750 (76.0) | 12,575 (83.3) |
| Hospice utilization | ||||
| Any in the last 30 d of life | 63,800 (67.5) | 13,955 (69.9) | 37,877 (70.7) | 8808 (58.3) |
| Inpatient hospital admission | ||||
| Any in last 12 mo of life | 66,798 (70.7) | 15,857 (79.4) | 36,628 (68.3) | 11,590 (76.8) |
| Any in last 30 d of life | 39,329 (41.6) | 9138 (45.7) | 20,919 (39.0) | 7510 (49.7) |
| Any in last 7 d of life | 25,854 (27.4) | 5974 (29.9) | 13,233 (24.7) | 5388 (35.7) |
| Nursing home utilization | ||||
| Any in last 12 mo of life | 46,875 (49.6) | 10,767 (53.9) | 27,468 (51.2) | 6982 (46.2) |
| Place of death | ||||
| Assisted living | 47,705 (50.5) | 9636 (48.2) | 27,668 (51.6) | 6953 (46.1) |
| Hospital | 18,508 (19.6) | 4014 (20.1) | 9123 (17.0) | 4253 (28.2) |
| Nursing home/SNF | 21,854 (23.1) | 4773 (23.9) | 13,286 (24.8) | 2914 (19.3) |
| Hospice facility | 6388 (6.8) | 1553 (7.8) | 3531 (6.6) | 977 (6.5) |
Cancers are limited to lung, colorectal, breast, prostate, and pancreatic cancers.
Presence in AL determined as not being admitted to a hospital or nursing home on any given day, based on administrative claims and assessment data.
ADRD indicates Alzheimer’s disease and related dementia; AL, assisted living; COPD, chronic obstructive pulmonary disease; HF, heart failure; SNF, skilled nursing facility.
TABLE 4.
Health Care Utilization Among AL Decedents With and Without ADRD
| Utilization | Decedents without ADRD |
Decedents with ADRD |
|---|---|---|
| Total | ||
| N (% of decedents) | 28,083 (29.7) | 66,372 (70.3) |
| Presence in AL* | ||
| Last 12 mo of life | 27,700 (98.6) | 63,191 (95.2) |
| Last 90 d of life | 26,440 (94.2) | 56,812 (85.6) |
| Last 30 d of life | 23,951 (85.3) | 50,242 (75.7) |
| Hospice utilization | ||
| Any in the last 30 d of life | 16,861 (60.0) | 46,939 (70.7) |
| Inpatient hospital admission | ||
| Any in last 12 mo of life | 20,029 (71.3) | 46,769 (70.5) |
| Any in last 30 d of life | 12,705 (45.2) | 26,624 (40.1) |
| Any in last 7 d of life | 8980 (32.0) | 16,874 (25.4) |
| Nursing home utilization | ||
| Any in last 12 mo of life | 11,972 (42.6) | 34,903 (52.6) |
| Place of death | ||
| Assisted living | 13,995 (49.7) | 33,750 (50.9) |
| Hospital | 7013 (25.0) | 11,495 (17.3) |
| Nursing home/SNF | 5210 (18.6) | 16,644 (25.1) |
| Hospice facility | 1905 (6.8) | 4483 (6.8) |
Presence in AL determined as not being admitted to a hospital or nursing home on any given day, based on administrative claims and assessment data.
ADRD indicates Alzheimer’s disease and related dementia; AL, assisted living; SNF, skilled nursing facility.
Care Pathways in the Last 30 Days of Life
Figure 1 depicts the most common pathways for the cohort of decedents residing in AL on day 30 before death (N = 67,577). Nearly half (46.4%) died in AL without any health care transition. Among those present in AL on day 30, 15.3% had one transfer to hospital from AL and then died there, while 12.8% were admitted to hospital from AL but returned to AL to die (Fig. 1). Only a small percentage (2.2%) of decedents transferred directly from AL to a nursing home, while 1.8% of decedents transferred directly from AL to a free-standing inpatient hospice facility. Lastly, 13.2% of decedents had 3 or more transfers in the last 30 days of life (nearly all included hospitalizations), with 4.2% of the overall cohort eventually returning to AL to die after these transfers. Examining these trajectories across mutually exclusive disease groups (results not shown because of small cell suppression restrictions), those in the organ failure group are much less likely to die in place without transfers (37.9% vs. 49.8% among the ADRD group) and have the highest rate of transfers to hospital (42.4% direct from AL, and 15.4% with multiple transfers). If transferred to a hospital, those with ADRD are more likely to return to AL (39.1%) compared to those with organ failure (27.1%) or cancer (34.6%). For comparison, among AL residents in our cohort surviving through 2019 the percentage with an inpatient admission in 2019 was 24.8%, which provides a sense of how common hospital transfers are in this population throughout the study period.
FIGURE 1.
Care transitions and place of death of 2017–2019 decedents who were in assisted living on day 30 before death. All percentages are out of the total of 67,577 fee-for-service decedents who lived in assisted living in January 2017, died in 2017–2019, and were in assisted living on day 30 before death. *Lines without a percentage value added are <1%; the total of these lines is also <1%.
DISCUSSION
While AL represents an increasingly commonplace of care for older adults toward the end of life, there are few cohort studies examining the degree to which residents age and die in place despite many considering this setting to be their home.8 This study describes the characteristics of AL residents who died over a 3-year follow-up nationwide, as well as the terminal care pathways of those present in AL in the last month of life. Our results show that over a third of residents died by the end of the follow-up period and 49.5% of the overall decedent cohort did not die in AL. This represents an under-researched population among studies examining end-of-life care in AL, raising questions about the factors that led to their transfer out of AL such as the interaction between facility-level care processes for dying residents and the various state-level policies regulating care in this setting.9,10
We also found that although 46.4% of decedents who were present in AL on day 30 before death died in place without any health care transitions, hospitalizations and multiple transfers were especially common among decedents with COPD/HF, suggesting that emergency departments and inpatient hospitals remain an important setting for advance care planning and hospice referral for this population. Although direct transfers to nursing homes were relatively rare, we found extensive nursing home utilization after hospitalization particularly among those with ADRD, indicating that medical crises rather than thoughtful care plans largely triggered the relocation of AL residents to nursing homes. Such unexpected relocation can be stressful to both residents and family members, particularly when a majority of AL residents have cognitive impairment.3 While there were some differences in health care utilization, end-of-life care trajectories of those with and without ADRD were comparable, suggesting that AL is a residential care setting where memory issues do not radically change medical decisions at the end of life.
Findings also point to a highly multimorbid population with a majority of decedents having been diagnosed with ADRD before dying. Decedents had a higher prevalence of ever being diagnosed with ADRD than the overall 42% of AL residents with staff-recognized dementia previously documented in the 2010 National Survey of Residential Care Facilities (NSRCF),2 suggesting that dementia care is an important clinical feature of end-of-life care in AL. This finding has important implications for policymakers and AL administrators in terms of organizing end-of-life care for AL residents who are very likely to experience memory issues even outside of designated memory care.
This study had a few limitations. The analysis was restricted to large AL communities with 25 beds or more, which represented 84.1% of all licensed beds nationally in 2016.11 Residents were assumed to have remained in AL unless they appeared in other institutional settings, potentially including the outcomes of the small proportion of residents who were taken to a private family residence for terminal care. A small number of residents in independent living communities sharing a ZIP code with an AL community may also have been captured using our methodology,5 although we expect that these beneficiaries would be healthier and less likely to die during the follow-up period given a care setting with fewer services. Reliance on ZIP codes reported to the Center for Medicare and Medicaid Services also implies some limitations, including potentially missing very short AL stays where beneficiaries did not update their address before moving to another care setting.5 State registries do not provide information about whether specific AL communities were part of larger campuses such as Continuing Care Retirement Communities or owned by an entity including hospice agencies, which may influence terminal care trajectories. We were also unable to determine how closely health care utilization and trajectories were aligned with beneficiaries’ care preferences. However, the extent of previously documented state and geographic variation in these trajectories suggests that at least part of these patterns cannot be explained by patient preferences,12,13 and reflect unwarranted variation in end-of-life care delivery.14
In conclusion, our results provide insight into the different characteristics and health care utilization of AL residents who died over a 3-year period. There are methodological and policy implications to excluding the substantial proportion of AL residents who were unable to remain in place until the last month of life from analyses of the multilevel factors that shape their ability to age and die in place. Our findings also call for interventions to reduce transfers to hospitals from AL communities at the end of life, though not all of these transfers should be considered burdensome end-of-life transitions.12 Future research is needed to examine how the dying experience varies by care trajectories. More research is also needed into the processes and policies guiding the care of the frail and vulnerable population of dying AL residents.
Supplementary Material
Acknowledgments
The work submitted in this manuscript was sponsored by the National Institute on Aging, award number-R01AG066902. The funders played no part in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.lww-medicalcare.com.
The authors declare no conflict of interest.
REFERENCES
- 1.Teno JM, Gozalo P, Trivedi AN, et al. Site of death, place of care, and health care transitions among US Medicare beneficiaries, 2000-2015. JAMA. 2018;320:264–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Zimmerman S, Sloane PD, Reed D. Dementia prevalence and care in assisted living. Health Aff. 2014;33:658–666. [DOI] [PubMed] [Google Scholar]
- 3.Makaroun LK, Teno JM, Freedman VA, et al. Late transitions and bereaved family member perceptions of quality of end-of-life care. J Am Geriatr Soc. 2018;66:1730–1736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gozalo P, Teno JM, Mitchell SL, et al. End-of-life transitions among nursing home residents with cognitive issues. N Engl J Med. 2011;365:1212–1221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Thomas KS, Dosa D, Gozalo PL, et al. A methodology to identify a cohort of Medicare beneficiaries residing in large assisted living facilities using administrative data. Med Care. 2018;56:e10–e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Intrator O, Hiris J, Berg K, et al. The residential history file: studying nursing home residents’ long-term care histories(*). Health Serv Res. 2011;46(1 pt 1):120–137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lunney JR, Lynn J, Foley DJ, et al. Patterns of functional decline at the end of life. JAMA. 2003;289:2387–2392. [DOI] [PubMed] [Google Scholar]
- 8.Ball MM, Kemp CL, Hollingsworth C, et al. “This is our last stop”: negotiating end-of-life transitions in assisted living. J Aging Stud. 2014;30:1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Belanger E, Rosendaal N, Wang XJ, et al. Association between state regulations supportive of third-party services and likelihood of assisted living residents in the US dying in place. JAMA Health Forum. 2022;3:e223432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Temkin-Greener H, Guo W, Hua Y, et al. End-of-life care in assisted living communities: race and ethnicity, dual enrollment status, and state regulations. Health Aff. 2022;41:654–662. [DOI] [PubMed] [Google Scholar]
- 11.Harris-Kojetin L, Sengupta M, Lendon JP, et al. Long-term care providers and services users in the United States, 2015–2016. National Center for Health Statistics. Vital Health Stat. 2019;3. [PubMed] [Google Scholar]
- 12.Wang XJ, Teno JM, Gozalo PL, et al. State variation in potentially burdensome transitions among assisted living residents at the end of life. JAMA Intern Med. 2022;182:229–231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Thomas KS, Belanger E, Zhang W, et al. State variability in assisted living residents’ end-of-life care trajectories. J Am Med Dir Assoc. 2020;21:415–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wennberg JE. Unwarranted variations in healthcare delivery: implications for academic medical centres. BMJ. 2002;325:961–964. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.