Family Ties at End-of-Life: Characteristics of Nursing Home Decedents With and Without Family

Background: Little is known about nursing home (NH) residents' family characteristics despite the important role families play at end-of-life (EOL). Objective: To describe the size and composition of first-degree families (FDFs) of Utah NH residents who died 1998–2016 (n = 43,405). Methods: Using the Utah Population Caregiving Database, we linked NH decedents to their FDF (n = 124,419; spouses = 10.8%; children = 55.3%; siblings = 32.3%) and compared sociodemographic and death characteristics of those with and without FDF members (n = 9424). Results: Compared to NH decedents with FDF (78.3%), those without (21.7%) were more likely to be female (64.7% vs. 57.1%), non-White/Hispanic (11.2% vs. 4.2%), less educated (<9th grade; 41.1% vs. 32.4%), and die in a rural/frontier NH (25.3% vs. 24.0%, all p < 0.001). Despite similar levels of disease burden (Charlson Comorbidity score 3 + 37.7% vs. 38.0%), those without FDF were more likely to die from cancer (14.2% vs. 12.4%), Chronic Obstructive Pulmonary Disease (COPD) (6.0% vs. 4.0%), and dementia (17.1% vs. 16.6%, all p < 0.001), and were less likely to have 2+ hospitalizations at EOL (20.5% vs. 22.4%, p < 0.001). Conclusions: Among NH decedents, those with and without FDF have different sociodemographic and death characteristics—factors that may impact care at EOL. Understanding the nature of FDF relationship type on NH resident EOL care trajectories and outcomes is an important next step in clarifying the role of families of persons living and dying in NHs.

Keywords: death; end of life; family; hospitalization; nursing home; rural Introduction Nursing homes (NHs) serve as the primary site of endof-life (EOL) care for persons with chronic progressive illnesses. 1,2Studies find that one in five Americans, including nearly 70% of people with advanced dementia, die in NHs each year. 2,3By 2030, 40% of all U.S. deaths are projected to occur in NHs. 3,4Unfortunately, many NH residents have unmet needs for symptom management and emotional support, poor resident/familyprovider communication, and experience burdensome care transitions at EOL. [5][6][7][8] Families often play a critical role in person-centered care for NH residents, promoting quality of life and making EOL care decisions. 9,10They may visit more frequently and provide more hands-on support when concerned about the resident's well-being or when they lack confidence in the care provided. 9,10Additionally, families may commonly react to changes in resident condition as a crisis, leading them to advocate for care that may be physically and psychologically detrimental, 5,6 particularly during the final months of life.Many NH residents, however, do not have any family to advocate for their care, which may impact their EOL care. 11,12espite the influential role of families at EOL, little is known about the characteristics of families of those who die in a NH.The purpose of this brief report is to describe the sociodemographic and death characteristics of NH decedents with and without first-degree family (FDF) members.

Methods
This retrospective cohort study focused on 43,405 individuals in the Utah Caregiving Population Science (Utah C-PopS) cohort 13,14 who died in a Utah NH of natural causes between 1998 and 2016.The Utah C-PopS cohort is derived from the Utah Population Database [15][16][17][18] which is a statewide population health research database that links individual-level administrative and health data to spousal and FDF members' data through a population pedigree (genealogy) structure.FDF include children, siblings, and parents.Key variables and data sources used to measure decedent sociodemographic, EOL care, and death characteristics are described elsewhere. 13escriptive statistics, chi-square tests, and t-tests were performed to describe the NH decedent cohort and their FDF members, as well as to compare sociodemographic, socioeconomic, and death characteristics of NH decedents with and without FDF.All descriptive statistics analyses were performed using R statistical software (Core Team, 2021).This study was approved by the University of Utah's Resource for Genetic and Epidemiologic Research and Institutional Review Board.
Nearly 40% had moderate levels of comorbidity (Charlson Comorbidity Index [CCI] 3+), 19,20 43.3% had a diagnosis of dementia, and almost a quarter had two or more hospitalizations in their last six months of life.Males had more inpatient hospitalizations within the last month of life compared to females (30.6% vs. 23.2%,p < 0.0001).The most common primary causes of death were heart disease (21.1%), dementia (16.7%), and cancer (12.8%).

Discussion
Families often play an important role in advocating for care at EOL; however, most existing research is limited by convenience samples with inadequate long-term follow-up data. 21To our knowledge, this is the first population-based study to explore the sociodemographic and death characteristics of NH decedents with and without FDF.This Utah C-PoPs study represents preliminary work deriving and describing this NH cohort and demonstrates that it is feasible to link FDF to NH decedents using a population pedigree registry.
Findings highlight gender, race, ethnicity, health insurance, socioeconomic, and geographic disparities in NH decedents without FDF-factors that may impact EOL care for an already vulnerable population. 22Interestingly, despite similar levels of disease burden (CCI 3+), those without FDF were less likely to have multiple hospitalizations at the EOL compared to those with FDF.Such a finding aligns with our prior mixedmethods studies, as well as other studies in the literature [7][8][9][10][11][12][13] suggesting that families may advocate for more aggressive care at the EOL.4][25][26][27] Future research  Marital status was defined as having a legal relationship with the decedent and was obtained from the death certificate.Unknown marital status was excluded due to small numbers.
e Education-maximum level of education for decedents as reported from the UPDB was obtained.Unknown education was excluded due to small numbers.
f Region-region was derived from zip code of decedents' residence as reported in death certificates and was categorized using 2018 Utah Department of Health classifications.Urban counties have a population density of >100 people per sq.mile; rural counties have a population density of <99 and >6 people per sq.mile, and Frontier counties have <6 people per sq.mile.g Insurance coverage was only available for individuals who had an inpatient or ambulatory surgery or emergency record in the UPDB within the last year of decedents' death.Please note that inpatient and ambulatory surgery records were from 1996 to 2016 and emergency department records were from 2000 to 2016.Medicare only means evidence of Medicare but no Medicaid; Medicaid only evidence of Medicaid but no Medicare.h Primary cause of death was obtained from death records and defined with ICD-9 and ICD-10 codes.i Comorbidity-Consistent with Charlson et al., 20 we used a weighted CCI that takes into account the number and seriousness of comorbid disease.Each comorbidity was weighted by a value, depending on the prognostic impact of a comorbid condition, ranging from 1 to 6 points with scores summed to calculate a CCI.Based on the CCI score, the severity of comorbidity was categorized into three grades: mild, with CCI scores of 1-2; moderate, with CCI scores of 3-4; and severe, with CCI scores ‡5.The CCI was based on diagnosis data obtained from inpatient, ambulatory surgery, and emergency department records within the last two years of decedents' death (Note: inpatient and ambulatory surgery records were from 1996 to 2016; emergency department records were from 2000 to 2016).should examine patterns of EOL health care utilization, including NH length of stay and use of palliative and hospice care to better understand the role families play at EOL.While 90.6% of the population in Utah lives in urban settings, *20% of Utah NHs are located in rural counties.Rural NHs are an important focus for policymakers given the lack of available postacute and long-term care services in these areas. 28,29Our study found that nearly 25% of all decedents died in a rural/frontier NH.Despite national policy initiatives to promote home-and community-based services over institutional care, 30 these efforts have been slow to be realized for individuals living in rural areas who have limited access to NH alternatives such as formal paid care, assisted living facilities, and adult day care centers. 31,32Future research is needed to better understand the impact of family, or lack thereof, on rural NH resident EOL care trajectories.
We acknowledge several study limitations.First, this study utilized data from one state; therefore, findings may not generalize to the national NH population.Nevertheless, study NH population demographics mirror national estimates that NH residents are predominately female, White, with at least a high school education. 33,34Second, this study is our first analysis of this NH cohort, and we only examined those who died in a NH and the binary association of FDF on EOL hospitalizations.Future work will link CMS Minimum Data Set, electronic health record, administrative claims, and other data for both NH residents and their FDF to examine the effect of types of family on NH resident EOL care outcomes, and how caring for a NH resident impacts family health.Third, not all family members serve as caregivers, or they may live out of state yet still provide substantial long-distance caregiving-factors not accounted for in this study.Other studies, suggest, however, that family geographic proximity, as well as their employment status and/or own health status may impact the role they want, or can, play in the care of NH residents. 9Despite possible limitations, this is the first study to find that among NH decedents, those with and without FDF have different sociodemographic and death characteristics-factors that may impact care at EOL.It further sets the stage for future research understanding the critical role that family plays in EOL care experiences of NH residents.

Conclusions
This study begins to illuminate vulnerable populations of NH residents (e.g., those without FDF and/or living in rural areas) for whom greater support may be needed to reduce health disparities and improve EOL care.These findings align with previous research suggesting NH families may need more anticipatory guidance and palliative care support to reduce potentially aggressive care at the EOL. 5,6,35,36Understanding the nature of FDF relationship type on NH resident EOL care trajectories and outcomes is an important next step in clarifying the role of families of persons living and dying in NHs.

FIG. 1 .
FIG. 1. NH decedent cohort with and without FDF members (n = 43,405) and linkages of the UPDB data sources.FDF, first-degree family; NH, nursing home; UPDB, Utah Population Database.

Table 1 .
Demographic, Socioeconomic, Context of Death, End-of-Life Health Care Utilization, and Size and Characteristics of First-Degree Family of Nursing Home Decedents

Table 1 .
(Continued)Sex as reported from UPDB.Unknown sex was excluded due to small numbers with the decedent and was obtained from the death certificate.Race and Ethnicity were based on the National Institute of Health definitions.Unknown race/ethnicity was excluded due to small numbers.
a b Age of decedent was referenced at the approximate time of the decedents' death.cd

Table 2 .
Demographic and Socioeconomic Characteristics of Nursing Home Decedents for First Degree Family Members **p < 0.01.a Race/ethnicity was utilized rather than race categories due to privacy concerns with cell sizes <10.

Table 3 .
Bivariate Analysis of Demographic, Socioeconomics, Context of Death, and End-of-Life Health Care Utilization Differences Between Nursing Home Decedents With and Without First-Degree Family Race/ethnicity was utilized rather than race categories due to privacy concerns with cell sizes <10.Stephens, et al.; Palliative Medicine Reports 2023, 4.1 http://online.liebertpub.com/doi/10.1089/pmr.2023.0023 **p < 0.01.a