End-of-Life Care Planning: The Importance of Older Adults' Marital Status and Gender

Abstract

Background:

As the U.S. population ages, dramatic shifts are occurring in the proportion of older adults who are divorced and widowed. Health status and behaviors are known to differ across marital status groups, yet research on end-of-life (EOL) care planning has only compared married and unmarried persons, overlooking differences between divorced and widowed individuals, by gender.

Objective:

This study aimed to examine marital status differences in EOL care planning by comparing the likelihood of discussions about EOL care, designation of medical durable power of attorney (MDPOA) for health care decisions, and completion of a living will for married, divorced, and widowed older adults, by gender.

Methods:

Analyses used data from the U.S. Health and Retirement Study for 2243 adults (50 years of age and older), who died during the course of the study. Post-death, proxy respondents reported on the decedents' EOL care discussions, living will completion, and establishment of an MDPOA. Multivariate regressions were estimated to test differences in care planning across marital status groups, for men and women.

Results:

Divorced men were less likely than married men to have had care discussions and to have engaged in any of the three planning behaviors. Widowers were more likely to have established an MDPOA. Both divorced and widowed women were more likely to have performed any of these EOL planning activities than married women.

Conclusions:

Divorced men and married women are at risk for lacking EOL care planning. Practitioners are encouraged to discuss the importance of such planning as they encounter these at-risk groups.

Introduction

Population aging has generated concerns about quality of life in old age¹ and rising health care spending.² Notable among growing health care costs are those for treatment of chronic conditions near end of life (EOL),³ as 50% of all Medicare spending during beneficiaries' final year of life is directed to inpatient hospital care.⁴ EOL care planning, including completion of advance directives (ADs), is widely encouraged to provide individuals greater involvement in their EOL care, ensuring that care decisions are consistent with their goals and values⁵ and, secondarily, to save costs on unwanted or inappropriate medical care.⁶

Despite increased promotion of EOL care planning,^7,8 rates of AD completion remain low among U.S. adults (∼37%), even in the elderly population (45.6%).⁹ Moreover, research indicates that EOL care planning, like most health practices and disease outcomes, varies substantially across social groups. Specifically, AD completion and EOL financial planning are done more by adults with greater financial assets¹⁰ and education.¹¹ EOL care planning also is more common among Whites than Blacks,¹¹ and married than unmarried persons.^11–14 Findings are inconsistent regarding whether men do more EOL care planning than women,^12,15 although men are consistently shown to discuss their care preferences with others less often than women.^11,12

Understanding how marital status relates to EOL care discussions and advance care planning is critical because health status and behaviors are significantly influenced by family processes and widely differ for persons who are currently married and unmarried.^15–17 Unmarried adults generally experience more health risks and problems than married persons, including smoking and drinking,¹⁸ disability,¹⁹ depression and loneliness,²⁰ poor self-reported health,²¹ and mortality.²² Such risks differ for divorced and widowed persons, and by gender. Research on EOL care planning, however, has generally only compared married and unmarried persons, without attending to differences among types of unmarried persons.^11–15 This is a serious limitation because the largest groups of unmarried older persons—the widowed and divorced—differ in terms of life experiences, family relationships, and sociodemographic characteristics (e.g., education and income), suggesting distinct EOL care planning as well.¹⁰

Reports of marital status differences in EOL care planning also do not consider gender as a critical intersecting variable. Given gendered roles within later life marriages, especially regarding marital power and decision making,²³ differences in EOL planning between married men and women are likely. In addition, divorce and widowhood have distinct consequences for men and women,²⁴ especially in terms of their financial effects and influence on family relationships. Because both economic and family relationship factors influence engagement in EOL care planning,^10,13 examining EOL planning activities for these distinct marital status groups, by gender, is essential.

Finally, demographic trends highlight the need for greater attention to EOL care planning for older widowed and divorced persons. As the U.S. population ages, dramatic shifts are occurring in the marital status distribution of older adults. Between 1990 and 2015, the proportion of adults older than 50 years, who were divorced, jumped 75% and the percent who were widowed dropped sharply. One-third of persons born between 1946 and 1964 are currently unmarried.²⁵ Thus, considering the rapidly changing demographic shifts in the U.S. population, this study aims to examine gendered differences in the likelihood of EOL care planning for married, divorced, and widowed older adults. Accordingly, this study will examine three aspects of older adults' EOL care planning: discussions about EOL care preferences, establishing a medical durable power of attorney (MDPOA), and completion of a living will.

Methods

Data

This investigation used data from The Health and Retirement Study (HRS), an ongoing U.S. national survey of adults 50 years of age and older. The HRS has collected core data biennially since 1992, using a multistage, area-clustered, stratified sample. The core surveys gather background data for respondents (e.g., race and gender), and time-varying information such as self-rated health, assets, and marital status. A unique aspect of the HRS is its inclusion of exit interviews with proxy respondents following the deaths of a subset of HRS respondents. These interviews address EOL care planning activities, circumstances surrounding the respondent's death, and health care delivered at the time of death,²⁶ as reported by the proxy respondent. Spouses provided proxy information in most cases involving married decedents, and adult children were generally the proxy respondent for unmarried decedents.

Our analytic sample was constructed by combining data for all deceased HRS respondents for whom exit interview data were collected in 2012 (n = 1187) and 2014 (n = 1242). Of these 2429 cases, we omitted 45 because the respondent's death occurred before 2008 or the date of death was unreported. We then matched decedents' exit interview data with their latest core survey (2012, 2010, or 2008). We substituted mean values for missing values for those cases (<100) missing data on the independent variables (see below), with the exception of the marital status and race/ethnicity variables, which could not be estimated. After eliminating cases where marital status or race/ethnicity was indeterminable, the final analytic sample included 2243 decedent respondents (92.3% of the original exit interview sample).

Dependent variables

Proxy respondents provided data for three dependent variables regarding the HRS decedents' EOL care planning by responding to the three questions below. “Yes” responses were coded as 1 and “no” as 0 for each question.

EOL Care Discussions: “Did [R] ever discuss with you or anyone else the treatment or care he/she wanted to receive during the final days of his/her life?”

Living Wills: “Did [R] provide written instructions about the treatment or care he/she wanted to receive during the final days of his/her life?”

MDPOA: “Did [R] also make any legal arrangements for a specific person or persons to make decisions about his/her care or medical treatment if he/she could not make those decisions himself/herself?”

A fourth dependent variable was constructed to denote whether the decedent had done any of the above EOL planning activities. This “Any Care Planning” variable was coded to 0 if all three questions had “no” responses and to 1 if any of these planning activities was reported.

Independent variables

Sociodemographic characteristics of the deceased respondent were drawn from the last available core survey and used as predictor variables. These included the following: race (1 = non-Hispanic White and 0 = Hispanic or non-White), education (1 = education beyond high school and 0 = no education beyond high school), age (in years), low assets (1 = respondent's assets were in the lowest quartile of assets for this subsample and 0 = otherwise), poor health (respondent's self-rated health at last survey, coded 1 = excellent to 5 = poor), and marital status—coded as follows: married or partnered at time of death (1-0); divorced (1-0); or widowed (1-0). Gender (male/female) was reported in the core surveys. This variable was used as a key stratifying variable for all analyses, consistent with the research aim of identifying marital status differences in EOL care planning within gender groups.

Analytic approach

Analyses were conducted in Stata 13.0. Each dependent variable was analyzed using all cases with valid data on that variable; thus, group sizes vary slightly across analyses. T-tests and chi-square tests were used to first compare marital status groups, within gender, on all dependent and independent variables, using a significance level of p ≤ 0.05. Next, multivariate analyses were performed using logistic regression because the dependent variables were dichotomous. These models tested whether marital status group differences were significant, within gender groups, controlling for background variables. To compare differences between groups on a categorical variable such as marital status, logistic regression requires designation of one of the groups as the omitted comparison category. We thus selected the married group to be compared with the divorced and widowed groups in these analyses. Multivariate results are reported using significance levels of p ≤ 0.01, 0.05, and 0.10 to capture significance trends even in some of the relatively small marital status × gender groups. The regression results are reported using odds ratios (OR). An OR = 1.00 indicates no differences in the likelihood of EOL planning based on that predictor variable. An OR >1.00 reflects an increased likelihood of planning, whereas an OR <1.00 denotes a reduced likelihood of planning for the associated predictor variable.

Results

Bivariate comparisons

Table 1 presents comparisons of background characteristics and EOL care planning based on marital status and gender. Examining results for males first, a few differences between marital status groups were noted. Widowers were the oldest group, with divorced men being the youngest. Divorced men were overrepresented by ethnic/racial minorities and had poorer health relative to the other groups. They also had the highest percentage in the low asset group, followed by widowers, and then married men. Regarding EOL care planning, divorced men were significantly less likely than widowed and married men to have done any planning; differences between married and widowed men were not significant. Divorced men were significantly less likely than both widowers and married men to have had informal discussions about EOL care and to have established a living will. Widowed men were significantly more likely than divorced and married men to have established an MDPOA.

Table 1.

Means (Standard Deviations) and Proportions on Key Independent and Dependent Variables, by Marital Status and Gender

Variable	Males			Females
Variable	Married (n = 742)	Divorced (n = 112)	Widowed (n = 220)	Married (n = 363)	Divorced (n = 149)	Widowed (n = 657)
Age at last survey	76.6^*^,† (10.1)	70.0^‡ (10.6)	84.4 (8.4)	73.7^† (10.5)	73.7^‡ (11.7)	85.0 (9.1)
% NH White	76.8^*	54.5^‡	78.2	73.6^*	63.1^‡	74.6
% with higher education	39.6	35.7	34.1	31.7	28.2	27.9
Poor health at last survey (1-5)	3.75^* (1.10)	3.92^‡ (1.03)	3.62 (1.11)	3.79 (1.11)	3.93^‡ (1.10)	3.70 (1.05)
% low assets	13.1^*^,†	50.0^‡	26.4	17.6^*^,†	50.0^‡	33.9
% any care planning	80.4^*	64.4^‡	86.5	76.9^†	82.6	86.8
% MDPOA	54.7^†	45.2^‡	74.8	54.3^†	56.9^‡	75.3
% living will	48.6^*	31.2^‡	52.3	39.0^†	44.8	53.7
% EOL care discussions	61.1^*	42.2^‡	59.5	53.2^*	66.9	58.9

Differences between married and divorced groups are significant at p ≤ 0.05.

†

Differences between married and widowed groups are significant at p ≤ 0.05.

‡

Differences between divorced and widowed groups are significant at p ≤ 0.05.

EOL, end of life; MDPOA, medical durable power of attorney; NH, non-Hispanic.

Marital status contrasts for females were similar to those of males for the background variables. Divorced women were most likely to have low assets, followed by widowed and then married women. The divorced women also were more likely to be racial/ethnic minorities and to have reported poor health, but the latter difference was significant in comparison to widows only. Widows were significantly older than married and divorced women. Regarding EOL care planning, widowed females were significantly more likely than the other groups to have done any planning. This same pattern applied to establishing an MDPOA, similar to results for males. Widows also were more likely than married women to have a living will. Finally, married women were less likely than divorced women to have had EOL care discussions. Because marital status groups differed on several background characteristics, in addition to EOL planning activities, multivariate analyses are required to isolate differences associated with marital status, controlling for other background differences.

Multivariate analyses of EOL care planning

Multivariate results estimating marital status effects on EOL care planning activities, controlling for background characteristics, are shown in Table 2. Considering background characteristics, similar findings were revealed for males and females. Across planning activities, a higher likelihood of planning was associated with greater age and education, and with being non-Hispanic White. Having low assets was significantly associated with reduced planning for females only, across all three planning activities. For males, low assets marginally predicted a reduced likelihood of having a living will.

Table 2.

Odds Ratios Predicting End-of-Life Care Planning, by Gender

Independent variables	Males				Females
Independent variables	Any care planning	MDPOA	Living will	EOL care discussions	Any care planning	MDPOA	Living will	EOL care discussions
Age at last survey	1.03^‡	1.04^‡	1.03^‡	1.00	1.04^‡	1.05^‡	1.03^‡	1.00
NH White (1-0)	2.66^‡	1.98^‡	2.52^‡	2.26^‡	3.57^‡	3.10^‡	2.77^‡	1.88^‡
Higher education (1-0)	1.63^‡	1.73^‡	2.23^‡	1.03	1.50^*	1.39^†	1.33^†	1.13
Poor health	1.12	1.06	1.14^†	1.04	1.05	1.03	1.09	0.98
Low assets (1-0)	0.88	0.78	0.73^*	0.93	0.46^‡	0.63^‡	0.59^‡	0.72^†
Divorced (1-0)^a	0.62^*	1.11	0.76	0.53^‡	2.40^‡	1.53^*	1.72^‡	2.20^‡
Widowed (1-0)^a	1.36	2.00^‡	0.99	0.97	1.62^†	1.89^‡	1.48^†	1.31^†
Constant	0.18^†	0.02^‡	0.02^‡	1.05	0.10^‡	0.02^‡	0.02^‡	0.67
Pseudo R²	0.087^‡	0.096^‡	0.098^‡	0.032^‡	0.140^‡	0.139^‡	0.090^‡	0.029^‡
n	1055	1043	1049	1059	1142	1131	1142	1153

Comparison category is married persons of the same gender.

p ≤ 0.10; ^†p ≤ 0.05; ^‡p ≤ 0.01.

Multivariate results concerning marital status differed across gender groups. For males, widowers stood out in terms of having established an MDPOA, with their chances of that activity being twice that of married men. Divorced men were distinct from married men in terms of engagement in EOL care discussions, with their chances of such conversations only half as high as those for married men. Overall, divorced men were 38% less likely to participate in any planning compared to married men.

More marital status group differences were revealed for females than males. The chances of completing each of the three planning activities were significantly greater for both divorced and widowed females relative to married women. Widows were almost twice as likely as married women to have established an MDPOA. EOL care discussions most distinguished the divorced women from their married counterparts, as they were twice as likely as married women to have participated in such conversations before death. Divorced and widowed females were ∼70% and 50% more likely, respectively, to have completed a living will than their married peers. Overall, married women were significantly less likely than both divorced and widowed women to have done any EOL care planning.

Discussion

This study advances research on EOL care planning by closely examining marital status differences for males and females separately, and by considering distinctions in EOL care planning for divorced and widowed older adults relative to their married peers. Prior research claimed few marital status differences in the likelihood of older adults' EOL care preparations,^11,13 or that unmarried persons engaged in fewer planning activities than married persons.^11,12 These studies, however, did not consider differences based on gender, nor did they differentiate widowed and divorced persons when comparing unmarried and married groups.

Our results demonstrate that such conclusions are misleading. We found marital status significantly affected EOL care planning, with marked variations by gender. Among male decedents, divorced men, but not widowers, were less likely than married men to have done any type of EOL planning. This distinction was strongest regarding EOL care discussions, wherein divorced men were nearly half as likely as married men to have had such conversations. Regarding establishment of an MDPOA, however, it was widowers who were distinct, with rates of such activity twice that of married men.

Among females, widowhood and divorce were robust positive predictors of EOL care planning—countering prior research where marital status comparisons were not conducted by gender.^11,13 As with males, among females, widows were distinct from their married peers in terms of designating an MDPOA, with their likelihood of such actions nearly double that of married women. Unlike the findings for males, divorced females had strikingly high odds of completing a living will and engaging in EOL care discussions relative to their married counterparts.

High rates of MDPOA designation among widowed males and females probably reflect activities prompted by spousal death. Others¹² speculate that a partner's death may familiarize widowed persons with important EOL preparations; while conducting the legal business (e.g., will and life insurance matters) that occurs following spousal death, widowed persons may be advised to designate an MDPOA if they do not have one. Conversely, older divorced persons are unlikely to experience these same legal discussions at the point of marital dissolution (because a death has not just occurred), especially if they divorced well in the past as young or middle-aged adults.

The lack of EOL planning uniquely documented for divorced relative to married men, which was most significant regarding EOL discussions, perhaps reflects the gendered nature of divorce. That divorced men were half as likely as married men to have had EOL care discussions, whereas divorced women had greater chances of such discussions than their married peers, may be due to gender differences in post-divorce social contacts. Unmarried persons generally have more nonkin ties than married persons.²⁷ These may be less intimate and less conducive to personal discussions, such as those about death. Moreover, because women depend on kin ties for post-divorce support more so than men do,²⁸ divorced women may have more opportunity than divorced men for conversations about EOL care. Such discussions significantly influence eventual AD completion.¹²

Overall, this study identifies divorced men and married women as most “at risk” for lacking critical EOL planning. We have argued that divorced men may lack the types of close relationships that promote EOL discussions and planning. In contrast, married women may minimize the importance of EOL care conversations and other planning activities because of the presence of a spousal relationship; indeed, they may assume their husbands know their wishes and will make appropriate EOL care choices for them—similar to other decision making in their marriages.²³ Unfortunately, proxy assumptions about spouse's preferred care reveal substantial inconsistencies with their dying partner's wishes.²⁹ Thus, for both divorced men and married women, the chances of them experiencing a “good death”³⁰ that is consistent with their EOL care wishes may be seriously compromised.

Given current demographic trends, these results are highly relevant for practitioners. Although it is difficult to predict which patients will require surrogate decision makers at EOL,⁵ this study provides evidence regarding who may lack important AD materials. Therefore, at various touch points throughout the health care delivery process—primary care visits, transitions from home to nursing care, or at hospital admission/discharge—care team members should be aware of interactions with these at-risk groups and capitalize on opportunities to address advance care planning issues. Even if these interactions only result in brief conversations that merely start the planning process, the potential benefit of these initial conversations cannot be understated.³¹ Clinicians also should be familiar with the kinds of issues that create barriers to advance care planning,³¹ such as patients assuming that family members know their wishes, or patients having conflicted family relationships. Familiarity with key barriers will facilitate clinicians' preparation for EOL and advance care discussions. In addition, because many clinicians lack understanding of and experience dealing with the types of psychological and emotional issues that can interfere with patients' AD completion,³² these findings underscore the value of including clinical social workers trained in relational dynamics and communication techniques as part of the care team. The expertise of clinical social workers may benefit at-risk patients as they reflect on their EOL care wishes and consider who may best ensure that their EOL care goals will be met. For unmarried patients, divorced men, in particular, who are known to be at greater risk for estranged family relationships,³³ such discussions may prevent complicated future scenarios where the patient could be designated an “unrepresented patient,” lacking a voice to make critical medical decisions at EOL.³⁴

Despite the value of our findings, this study has some limitations. Like others,⁵ we recognize that proxy reports in the HRS exit interviews may not be entirely accurate regarding the EOL planning activities that decedents had completed. Yet, we see a benefit in having some indicators of planning—although imperfect ones—which were actually reported after the death of HRS decedents, rather than merely using respondents' own reports of EOL care planning they have or have not completed. These latter self-reports, which for some respondents are gathered years in advance of their death, may underestimate planning that is in place at the actual point of death.

In sum, this study contributes to a growing literature on adults' EOL care planning by adding greater specificity to the study of marital status influences. The results are especially relevant for practitioners because they identify specific groups of adults who are most at risk for limited EOL care planning, and promote heightened attention to meeting the unique planning needs of these at-risk groups.

Footnotes

Acknowledgment

We thank Dr. Angela Curl for her assistance with the HRS data management.

A version of this article was presented at the meetings of the International Association of Gerontology and Geriatrics, July 2017, San Francisco, CA.

Author Disclosure Statement

No competing financial interests exist.

References

Crimmins

: Lifespan and healthspan: Past, present, and promise. Gerontologist, 2015; 55:901–911.

May

, Garrido

, Cassel

, et al.: Palliative care teams' cost-saving effect is larger for cancer patients with higher numbers of comorbidities. Health Aff, 2016; 35:44.

Kerr

, Donohue

, Tangeman

, et al.: Cost savings and enhanced hospice enrollment with a home-based palliative care program implemented as a hospice-private payers partnership. J Palliat Med, 2014; 17:1328–1335.

Cubanski

, Neuman

, Griffin

, Damico

: Medicare spending at the end of life: A snapshot of beneficiaries who died in 2014 and the cost of their care. Kaiser Family Foundation, 2016. https://www.kff.org (last accessed January 28, 2019)

Silveira

, Kim

SYK

, Langa

: Advance directives and outcomes of surrogate decision making before death. N Engl J Med, 2010; 362:1211–1218.

Garrudio

, Balboni

, Maciejewski

, et al.: Quality of life and cost of care at the end of life: The role of advance directives. J Pain Symptom Manage, 2015; 49:828–835.

Rajda

, Wei

, Steinberg

: End-of-life planning education and advance directives completion rates. J Palliat Med, 2015; 18:825–826.

Abba

, Byrne

, Horton

, Lloyd-Williams

: Interventions to encourage discussion of end-of-life preferences between members of the general population and the people closest to them—A systematic literature review. BMC Palliat Care, 2013; 12:40–51.

Yadav

, Gabler

, Cooney

, et al.: Approximately one in three US adults completes any type of advance directive for end-of-life care. Health Aff, 2017; 36:1244–1251.

10.

Carr

: The social stratification of older adults' preparations for end of life health care. J Health Soc Behav, 2012; 53:297–312.

11.

Gerst

, Burr

: Planning for end-of-life care: Black-white differences in the completion of advance directives. Res Aging, 2008; 30:428–449.

12.

Carr

, Khodyakov

: End-of-life health care planning among young-old adults: An assessment of psychosocial influences. J Gerontol B Psychol Sci Soc Sci, 2007; 62:S135–S141.

13.

Carr

, Moorman

, Boerner

: End-of-life planning in a family context: Does relationship quality affect whether (and with whom) older adults plan?. J Gerontol B Psychol Sci Soc Sci, 2013; 68:586–592.

14.

Rao

, Anderson

, Lin

, Laux

: Completion of advance directives among U.S. consumers. Am J Prev Med, 2014; 46:65–70.

15.

Van Scoy

, Howrylak

, Nguyen

, et al.: Family structure, experiences with end-of-life decision making, and who asked about advance directives impacts advance directive completion rates. J Palliat Med, 2014; 17:1099–1106.

16.

Carr

, Springer

: Advances in families and health research in the 21st century. J Marr Fam, 2010; 72:743–761.

17.

Boerner

, Carr

, Moorman

: Family relationships and advance care planning: Do supportive and critical relations encourage or hinder planning?. J Gerontol B Psychol Sci Soc Sci, 2013; 68:246–256.

18.

Watt

, Heilmann

, Sabbah

, et al.: Social relationships and health related behaviors among older US adults. BMC Public Health, 2014; 14:533–552.

19.

Lin

, Brown

: Unmarried boomers confront old age: A national portrait. Gerontologist, 2012; 52:153–165.

20.

Wright

, Brown

: Psychological well-being among older adults: The role of partnership status. J Marr Fam, 2017; 79:833–849.

21.

Liu

, Umberson

: The times they are a changin': Marital status and health differentials from 1972 to 2003. J Health Soc Behav, 2008; 49:239–253.

22.

Dupre

, Beck

, Meadows

: Marital trajectories and mortality among US adults. Am J Epidemiol, 2009; 170:546–555.

23.

Bulanda

: Gender, marital power, and marital quality in later life. J Women Aging, 2011; 23:3–22.

24.

, Schimmele

: Uncoupling in late life. Generations, 2007; 31:41–46.

25.

Brown

, Wright

: Marriage, cohabitation, and divorce in later life. Innov Aging, 2017; 1:igx015.

26.

Servais

: Overview of HRS Public Data Files for Cross-Sectional and Longitudinal Analysis. Survey Research Center, Institute of Social Research, University of Michigan, 2004.

27.

Liebler

, Sandefur

: Gender differences in the exchange of social support with friends, neighbors, and co-workers at midlife. Soc Sci Res, 2002; 31:364–391.

28.

Gerstel

: Divorce, gender, and social integration. Gender Soc, 1988; 2:343–367.

29.

Moorman

, Hauser

, Carr

: Do older adults know their spouses' end-of-life treatment preferences?. Res Aging, 2009; 31:463–491.

30.

Frank

: Shared decision-making and its role in end of life care. Br J Nurs, 2009; 18:612–618.

31.

Lum

, Sudore

, Bekelman

: Advance care planning in the elderly. Med Clin N Am, 2015; 99:391–403.

32.

Aronson

, Kirby

: Improving knowledge and communication through an advance directives objective structured clinical examination. J Palliat Med, 2002; 5:916–919.

33.

Kalmijn

: Adult children's relationships with married parents, divorced parents, and stepparents: Biology, marriage, or residence?. J Marr Fam, 2013; 75:1181–1193.

34.

Bennett-Woods

, Abbott

, Glover

: Giving voice to the voiceless: The Colorado response to unrepresented patients. J Clin Ethics, 2017; 28:204–211.