Abstract
This study explores the extent to which cognitive processes, specifically perceptions of one’s distance to death, are associated with informal and formal advance care planning (ACP) in a sample of older adults. Data come from the New Jersey End-of-Life (EOL) study, a survey of 305 adults aged 55–91. Binary logistic regression models evaluate the odds that someone conducted ACP. Persons who perceive remaining life span to be expansive or limited have significantly lower odds of formally planning for the EOL, relative to those in the middle category. Death anxiety and having a family confidante partially explain these associations. The association between future time perspective (FTP) and discussions is not statistically significant. Practitioners may consider individuals’ FTPs when discussing preferences for EOL medical care.
Introduction
Most people in the United States would prefer to die at home, free of pain, and with their affairs in order (Institute of Medicine [IOM], 2014; Morin, 1997). In actuality, two thirds die in an institutional setting, either in or shortly after leaving the hospital (Centers for Disease Control and Prevention [CDC], 2012). Roughly one fifth spend time in the intensive care unit before their deaths, receiving intensive medical intervention that may compromise quality of life (Angus et al., 2004; SUPPORT Principal Investigators, 1995). Moreover, many seriously ill individuals overestimate the amount of time they have left, which may cause them to delay preparing for death (Emanuel & Emanuel, 1998; IOM, 2014).
The medical community and policy makers have responded to these discrepancies between anticipated and actual end-of-life (EOL) experiences by encouraging formal advance care planning (ACP) consistent with individual values, preferences, and life stage (IOM, 2014). By executing a living will, individuals formally state their health care treatment preferences before such decisions are necessary, ideally while they are relatively healthy and prior to cognitive decline. Others also appoint a durable power of attorney for health care (DPAHC), who is responsible for making informed health care decisions on behalf of the dying patient. The 1990 Patient Self-Determination Act requires that any hospital receiving federal funds provide patients with an opportunity to complete formal ACP and encourage them to do so (Levin, 1990).
Seriously ill individuals endorse the use of formal ACP. Steinhauser and colleagues (2000) find that 81% of seriously ill individuals believe it is important to express treatment preferences in writing and 98% think it is important to name a DPAHC. Although some scholars question the effectiveness and appropriateness of ACP in fostering conditions necessary for a good death (Teno et al., 1997), most acknowledge it is a useful albeit imperfect framework for achieving care that is concordant with patients’ wishes (Collins, Parks, & Winter, 2006; Silveira, Kim, & Langa, 2010).
Medical professionals may also encourage individuals to informally discuss their preferences for EOL health care with family or physicians. Ongoing conversations about EOL health care preferences are an important component of EOL planning (IOM, 2014). Despite these efforts and attitudes, overall rates of formal ACP remain low. Estimates range from 23% to 58% for the adult population (IOM, 2014), although rates are as high as 60–70% among older adults (Silveira et al., 2010; Teno, Gruneir, Schwartz, Nanda, & Wetle, 2007).
The sociodemographic and health correlates of formal ACP are well documented, with most studies showing that persons of higher socioeconomic status (SES), Whites, and those in poorer health are more likely than lower SES persons, Blacks, Latinos, and healthier persons to do formal ACP (Carr, 2012b, 2012c; Carr & Khodyakov, 2007; Silveira et al., 2010; Teno et al., 1997). Among married respondents, Whites are more likely to informally discuss the EOL than Latinos; similar differences exist only between lower SES Whites and Blacks (Carr, 2012b). Relationship quality is also important for ACP: People who report higher quality, more supportive relationships are also more likely to complete both formal and informal ACP (Boerner, Carr, & Moorman, 2013; Carr, 2012b; Carr, Moorman, & Boerner, 2013).
However, few studies explore the ways that cognitive processes, especially future orientations, affect ACP. Some research suggests that those who have watched a loved one die a painful or traumatic death are particularly motivated to engage in ACP, presumably because the experience leads them to think about their own finitude (Carr, 2012a). However, no studies look at how individuals’ understanding of their own remaining life span relates to ACP behavior.
Socioemotional selectivity theory provides a framework for understanding how an individual’s perceptions of remaining life span, or future time perspective (FTP), influences the goals people select and the social interactions they intend to pursue. This theory posits that perceiving future time as truncated or finite prompts goals and social relations that provide immediate emotional rewards. Longer term future time perceptions result in knowledge-expanding goals and interactions. However, these studies tend to use experimental settings to measure intentions and do not explore the relationship between FTP and actual behavior.
This article investigates the potential of FTP to predict formal and informal ACP behavior in older adults and has the following research goals: (a) to determine the extent to which perceived remaining life span predicts ACP behavior and the extent to which the relationship persists net of social, psychological, and demographic factors associated with ACP and/or FTP; (b) to investigate the extent to which death anxiety and having a confidante might explain the relationship between FTP and ACP; and (c) to explore the potential for family confidantes to encourage ACP, regardless of FTP. The study uses a survey of 305 community-dwelling adults in New Jersey, aged 55 and older. Understanding the role of cognition in decision making around ACP may enable policy makers and clinicians to design more effective initiatives to encourage ACP.
Literature Review
FTP and ACP
Sociological and psychological perspectives on cognition emphasize that how people envision themselves in the future influences how they see themselves today, the social categories they perceive themselves as belonging to, the actions they take in the present, and the plans they make for the future (Hendricks, 2001; Mische, 2009; Schutz, 1967). Socioemotional selectivity theory, a life span theory of motivation, theorizes that an individual’s FTP, typically measured as one’s perceived distance from death, influences the social goals he or she sets and the types of individuals he or she seeks out in achieving those goals (Carstensen, Isaacowitz, & Charles, 1999). Perceived time horizons are “expansive” or “limited.” A person with an expansive FTP understands his or her death to be far in the future. He or she is motivated to acquire new information and knowledge that may help him or her achieve long-term goals and seeks out social contact in the present that may prove to be useful in the distant future (Carstensen et al., 1999; Lang & Carstensen, 2002). Conversely, an individual with limited FTP perceives time to be finite and death to be closer at hand. He or she focuses on the present (as there is less future time to consider) and pursues interactions that provide immediate emotional satisfaction (Carstensen, Fung, & Charles, 2003). He or she has an increasing desire to spend time with close loved ones, provide comfort to others, and take steps so that he or she might be positively remembered after he or she dies (Lang & Carstensen, 2002).
Applying this concept of FTP to reported actions, this study explores how having an expansive or limited FTP relates to informal (discussions) and formal (executing a living will and/or naming a DPAHC) ACP behavior. A person with an expansive FTP perceives his or her death as far off and therefore may be less likely to consider ACP in the present. He or she pursues interactions that expand his or her knowledge and experience base, rather than goals that focus on the final stages of life, such as ACP. In contrast, individuals with limited FTP may complete ACP as a means of protecting his or her loved ones from difficult decisions in the future. Having a limited FTP also encourages individuals to pursue emotionally meaningful goals with increased emotional regulation (having more positive and/or less negative affect; Carstensen, Pasupathi, Mayr, & Nesselroade, 2000). Greater emotional regulation may make it easier for people with limited FTP to undertake the potentially unpleasant task of ACP. Individuals with a limited FTP may be more likely to plan for the EOL because of the cognitive processes that accompany limited FTP and which may encourage ACP, namely, increased contemplation of death, a desire to protect loved ones from future pain, and heightened ability to regulate emotions during unpleasant events.
Formal Versus Informal ACP
Most scholarship on ACP focuses on formal planning mechanisms (living wills and DPAHCs) rather than informal discussions about the EOL. While discussing the EOL may be upsetting for some, barriers to holding a conversation are lower than completing formal ACP documentation. Individuals consistently discuss EOL preferences at higher rates than they complete formal ACP (Carr, 2012b, 2012c; IOM, 2014), which may make it difficult to obtain variation in samples of people who do or do not discuss the EOL. Moreover, compared to formal ACP, the concept of a discussion about the EOL is vague, with vast potential for variation in content. Formal ACP consists of well-defined, legally binding documentation that can be verified, whereas an informal discussion could consist of an intentional, focused, and structured conversation with health care professionals or merely an off-the-cuff response to a spouse or child. This study examines the relationship between FTP and both informal and formal ACP. Because of the potentially lower barriers to participation and diverse nature of discussions, the relationship between FTP and informal discussions may not be as strong as with formal ACP.
Other Influences on ACP and FTP
Confounding variables
This article controls for life transitions and social and demographic factors that may affect the relationship between FTP and ACP and then explores two pathways that may explain the relationship. This study includes a control for age (in years), as older age may be positively associated with both ACP and limited FTP (Carstensen et al., 2003; Lang & Carstensen, 2002; Moorman & Inoue, 2013). The study includes measures for experiencing the death of a loved one and self-rated health status, as individuals experiencing these life transitions may be more likely to complete ACP and have a more limited FTP (Carr, 2012a; Carstensen et al., 2003; Silveira et al., 2010; Teno et al., 1997). Factors that are associated with either FTP or ACP behavior are also controlled. Depressive symptoms may limit FTP (Achat, Kawachi, Spiro, DeMolles, & Sparrow, 2000). SES, marital status, and race and ethnicity are all associated with ACP. Higher SES and married and widowed individuals are more likely to plan for EOL (Carr, 2012c; Carr & Khodyakov, 2007). African Americans and Latinos are less likely to plan than Whites (Carr, 2012b).
Explanatory pathways
This article explores two potential pathways that may explain the relationship between FTP and ACP. First, people with limited FTP demonstrate increased emotional regulation, responding more positively and less negatively to potentially emotionally charged situations (Carstensen et al., 2000). Attitudes toward death more generally, and one’s own mortality in particular, motivate decision making and action (Becker, 1973; Berger & Luckmann, 1967). Anxiety about one’s death has also been linked to lower odds of ACP (Carr & Khodyakov, 2007). Decreased emotional reactivity may increase acceptance of one’s mortality and lower anxiety about this eventuality, reducing barriers to ACP. The study includes a measure of death denial, anticipating that less denial and greater acceptance of death may facilitate ACP.
Second, in socioemotional selectivity theory, advancing age influences how people think about the social actors with whom they engage in goal attainment (Carstensen et al., 1999; Fung, Carstensen, & Lutz, 1999; Lang & Carstensen, 2002). Beginning in middle adulthood, individuals start to limit their social networks to others who are likely to provide opportunities for emotionally predictable and positive interactions, and report increased satisfaction from contact within their smaller networks (Carstensen, 1992; Fung et al., 1999). Having a close personal confidante may increase the likelihood of ACP by providing the emotional support necessary to engage in the imaginably daunting task of ACP (Boerner et al., 2013; Carr et al., 2013; Moorman, Carr, & Boerner, 2014). Persons with a confidante may have higher odds of completing ACP. Moreover, having a confidante may moderate the effects of FTP, where even those with an expansive time horizon may have a greater likelihood of doing ACP if they have a close confidante. The study includes a supplemental analysis that explores the potential moderating effects of having a family confidante on ACP.
Research Design
Sample
The study uses data from the 2006–2008 New Jersey EOL Survey, a 90-min, face-to-face survey of 305 noninstitutionalized adults, aged 55–91 seeking care at one of three medical centers in New Jersey. Most participants (206) had been diagnosed with colorectal cancer, type II diabetes, and/or congestive heart failure, chronic illnesses that are similarly prevalent in men and women over 55 years. The data also include a control group of 99 “healthy” respondents, although many of these individuals report other health conditions such as heart disease and asthma. The survey contained questions about EOL planning, other health behaviors, and health and sociodemographic characteristics.
The general internal medicine department at the University of Medicine and Dentistry of New Jersey identified 1,146 patients as potential study participants, 575 of which met inclusion criteria. Participants had to speak English or Spanish and have no cognitive limitations to participate. Additional reasons for exclusion include invalid contact information, death, not meeting minimum age requirements (55), and having severe physical limitations. Of the individuals who met the inclusion criteria, 305 participated, a 53% response rate. Common reasons for refusal include reluctance to participate in research, frailty, and time constraints. On average, respondents were 69 years old and had 14 years of education, approximately three fifths were women, about a quarter identified as non-Hispanic Black, and nearly a fifth as Hispanic. Trained interviewers used computer-assisted personal interview technology to collect data. The study had approval from institutional review boards at the principal investigator’s university and each study site. Carr (2012a) describes these and additional study details.
Measures
Dependent variables
Informal discussion is a dichotomous variable based on the question: “Have you discussed your future health care plans and preferences with any one? By future health care plans, we mean plans about the types of medical treatment you want or don’t want to receive if you become seriously ill in the future.” Those who answered “yes” were coded as 1; “no” responses were coded as 0. Formal ACP is a dichotomous variable based on the questions: “Do you have a living will or an advance directive? This is a set of written instructions about the type of medical treatment you would want to receive if you were unconscious or somehow unable to communicate.” and “Have you made any legal arrangements for someone to make decisions for you about your medical care, if you become unable to make those decisions for yourself? This person is sometimes called a Durable Power of Attorney for Health Care.” Respondents who answered yes to either or both questions were coded 1; those who answered no to both questions were coded 0.
Independent variable
The key independent variable of interest, FTP, is conceptualized as perceived distance to death. FTP is operationalized by combining measures of subjective life expectancy 5 years and 10 years into the future. Participants were asked “how likely do you think it is that you will live for another five years? For another ten years?” and responded to each question on a scale of 0 (no chance) to 10 (absolute certainty). In contrast to previous studies that present future time in general terms (Lang & Carstensen, 2002; Löckenhoff & Carstensen, 2007), the approach here prompts respondents to think about future time in a clearly specified manner (5- and 10-year increments) that is appropriate for the sample (older and seriously ill adults). This approach uses bounded perceived probability measures (from 0 to 10) rather than the 5- and 7-point Likert-type scales used in other studies. Following Lang (2000), the analysis categorizes respondents into three groups: expansive, limited, and intermediate (reference category). Respondents with an expansive FTP indicated absolute certainty of survival at 5 years and 10 years (10 to both questions). Those with a limited FTP indicated 5 or less for 5-year survival, a difference of 5 or more between 5- and 10-year survival, or answered 5-year survival but did not answer 10-year survival (three respondents). Those with an intermediate FTP indicated a high chance of 5-year survival (6 or higher) and did not alter their response much when predicting 10-year survival. Final variable categorization was based on best model fit (measured by pseudo R 2) across multiple categorizations of FTP (including as a continuous variable) and various treatments of age. This final categorization also performed consistently well in sensitivity tests with both independent variables.
Confounding variables
Variables that are likely to affect the relationship between FTP and ACP are also controlled. Age is measured continuously in years. Self-rated health is a dichotomous variable of roughly equal sizes in response to the question, “how would you rate your health at the present time?” Fair and poor responses were recoded as poor (reference category); good or better were recoded as good. This approach follows similar work on ACP (Carr, 2012b, 2012c; Moorman & Inoue, 2013). Death experience uses two questions to capture respondent experience with the loss of a loved one. Respondents were asked whether a close family member or friend had died in the last 10 years. The 84% of respondents who indicated yes were then asked how close they were to the deceased. To differentiate the potential emotional salience of the loss, responses were dummy coded as close to deceased (very close) not close to deceased (somewhat close, not very close, and not at all close), or no experience with death. Depressive symptoms are measured with a subset of items from the Center for Epidemiologic Studies Depression scale (Radloff, 1977). Scores are averaged responses to questions about on how many days in the past week (0–7) participants felt lonely, sad, depressed, everything was an effort, or they could not get going (α = .82). Higher scores indicate increased depressed affect. Marital status (currently married, divorced/separated, widowed, and never married), race and ethnicity (non-Hispanic White, non-Hispanic Black, and Hispanic), and education (years) are also controlled.
Explanatory pathways
The study explores two factors that may explain the relationship between FTP and ACP behavior. Greater death acceptance may accompany increased emotional regulation common at older ages, facilitating ACP. Death denial is the average of two questions about neutral acceptance of death (Wong, Reker, & Gesser, 1994). Using a 5-point scale (1 = strongly agree to 5 = strongly disagree), respondents rated their agreement that “death is simply part of the process of life” and “death should be viewed as a natural, undeniable, and unavoidable event” (α = .70). Higher scores indicate less acceptance of death or greater death denial.
According to socioemotional selectivity theory, as people age they increasingly pursue emotionally meaningful interactions with close social ties. The study explores the direct effects of family confidante on ACP and the extent to which having a confidante increases the likelihood of ACP, regardless of FTP, in a supplemental analysis. Family confidante is measured by the question: “Is there someone in your family with whom you can share your very private feelings and concerns?” where 1 represents yes and 0 represents no.
Analytic Strategy
Table 1 presents descriptive statistics for all variables by FTP. Analysis of variance (ANOVA) with post hoc comparisons and χ2 tests with p levels adjusted for three-category comparison indicate statistically significant differences between FTP categories. Binary logistic regressions are used to explore the relationship between FTP and ACP. Model 1 tests the unadjusted relationship between the two. Models 2–4 each add one confounding variable expected to affect both FTP and ACP: age, self-rated health, and death experience. Model 5 adds demographic controls: depressive symptoms, marital status, education, and race and ethnicity. Models 6–7 explore variables that might explain the relationship between FTP and ACP behavior: death anxiety and having a family confidante. The final logistic regression model includes all variables of interest. Supplemental analysis explores how family confidantes might facilitate ACP, regardless of FTP.
Descriptive Statistics and Subgroup Differences by FTP, New Jersey EOL Survey, 2006–2008.
Note. EOL = end of life; ACP = advance care planning. Means (and standard deviations) reported for continuous measures; proportions for categorical measures. Descriptive statistics for ACP refer to the 267 individuals who indicated whether or not they had an advance directive and/or a durable power of attorney for health care; all other rows based on 269 respondents. Subgroup comparisons conducted using two methods:
aAnalysis of variance with Scheffe test (p < .05) for continuous variables.
xχ2(p < .0167) for categorical variables. Subgroup differences denoted as follows: ab = expansive versus intermediate; ac = expansive versus limited; bc = intermediate versus limited.
Missing data
This analysis excludes 26 respondents who did not answer questions about perceived likelihood of survival (the key independent variable). These individuals reported significantly worse health and were less likely to discuss the EOL but otherwise did not differ from those who answered perceived survival likelihood questions. Ten respondents who reported “Other” race or ethnicity are also excluded. Mean and modal imputation on remaining missing independent and control variables preserves as much of the analytic sample as possible. Fewer than five (less than 2%) cases were missing data on any one variable and analysis using listwise deletion did not significantly change the results but did result in small cell sizes in some variable cross tabulations. Table 1 lists descriptive statistics for the 269 remaining respondents who indicated whether or not they discussed the EOL.
Results
Bivariate Analysis
Table 1 provides descriptive statistics for all variables used in the analysis. Overall, 71% of respondents reported informal ACP discussions; 52% reported a living will and/or DPAHC, which reflect higher rates of completion than in the adult population in general, but lower than found in other studies of older adults (IOM, 2014; Teno et al., 2007). Bivariate analysis using ANOVA and χ2 tests reveal statistically significant differences in important characteristics by FTP. All three groups vary significantly by age; on average, those with the most expansive FTP are youngest and those with limited FTP are oldest. Consistent with expectations outlined previously, compared to those with an intermediate FTP, respondents with an expansive FTP are significantly less likely to undertake any type of ACP (62% vs. 79% for discussions; 39% vs. 68% for formal ACP). They are also significantly less likely to identify as non-Hispanic White (47% vs. 71%). As theorized in socioemotional selectivity theory, poor health is associated with a more limited FTP in this sample: Respondents with a limited FTP report significantly worse health than those with an intermediate or expansive FTP (60% vs. 35% and 39%, respectively). They also report more depressive symptoms than those with either an intermediate or expansive FTP and are also significantly less likely to have a family confidante than their counterparts with an intermediate FTP (77% vs. 92%). Those with an intermediate FTP have significantly more education than respondents with expansive and limited FTPs (16 vs. 13 years).
Logistic Regression
Bivariate analyses identify significant differences in rates of ACP based on FTP. Table 2 presents results from a series of binomial logistic regression models that explore potential explanations for these differences. In the unadjusted logistic regressions (Model 1), only expansive FTP is significantly associated with ACP. Consistent with initial expectations, those who foresee a long future are less likely to undertake ACP. Compared to individuals with an intermediate FTP, those with an expansive view have 58% lower odds of discussing the EOL (p = .011). Similarly, having an expansive FTP decreases the odds of formal ACP by 70% (p = .00). Respondents with a limited FTP did not differ from those with an intermediate FTP in either category of ACP in the unadjusted model.
Summary Table: Relative Odds of Engaging in Advance Care Planning by FTP, After Adjusting for Sociodemographic Characteristics and Potential Explanatory Variables, New Jersey End-of-Life (EOL) Survey, 2006–2008.
Note. ACP = advance care planning; df = degrees of freedom. Odds ratios presented. Omitted category is intermediate FTP.
†p < .10. *p < .05. ** p <.01. *** p <.001. ****p < .0001 or p = 0.
Adding potentially confounding variables in Models 2–4 (age, self-rated health, and death experience) does not change the relationship between expansive FTP and ACP, relative to intermediate FTP. Contrary to expectations, for limited FTP, the odds of formal ACP decrease and become marginally significant with the addition of these confounding variables to the logistic regression model (odds ratio [OR] = .50, p = .06). In contrast, adding sociodemographic controls (depressed affect, education, marital status, race, and ethnicity) to the logistic regression in Model 5 increases odds of ACP among those with an expansive FTP. The difference in rates of discussions between expansive and intermediate FTP disappears (OR = .69, p = .34), and is largely attenuated, although still statistically significant for formal ACP (OR = .42, p = .018). Including sociodemographic controls does not change the relationship between limited FTP and either type of ACP.
FTP differences in the odds of ACP remain largely unchanged when explanatory variables are individually controlled (Models 6 and 7), although distinctions between the two pathways emerge. Controlling for death anxiety increases the magnitude of difference between FTPs for all ACP. However, the difference is more statistically significant in one case: expansive and intermediate FTPs and formal ACP (OR = .36, p = .007). Controlling for family confidantes does not significantly alter FTP differences in ACP. However, unlike with death anxiety, the direction of the effect of family confidantes is different for informal and formal ACP. Family confidantes further decrease differences among FTPs in discussions but increase differences in the case of formal ACP.
Table 3 presents ORs from a logistic regression model including all variables of interest in this analysis. As expected, compared to intermediate FTP, having an expansive FTP is associated with lower odds of ACP, although the relationship is only significant for formal ACP. Compared to intermediate FTP, those with an expansive FTP are 65% less likely to complete a living will or name a health care proxy (p = .006). In contrast to expectations, those with a limited FTP are also less likely to plan for the EOL. Again, the relationship is only significant for formal ACP (OR = .41, p = .048). Having an FTP at either end of the spectrum (expansive or limited) is inversely related to EOL planning and is more important for formal ACP than discussions.
Binomial Logistic Regression for Advance Care Planning (ACP), New Jersey EOL Survey, 2006–2008.
Note. N = 269 for discussions; N = 267 for formal ACP; FTP = future time perspective; ACP = advance care planning; OR = odds ratio; CI = confidence interval.
†p < .10. *p < .05. **p < .01. ***p < .001. ****p < .0001 or p = 0.
Supplemental Analysis: Family Confidante
Table 4 provides three-way cross tabulations for ACP, FTP, and family confidante. Small cell sizes for respondents with no confidante preclude testing interaction terms for FTP and family confidante using regression. However, the cross tabulations show that having a family confidante facilitates discussions regardless of FTP, although a greater proportion of those with limited FTP discuss the EOL in the absence of a family confidante. Over 60% of individuals with a family confidante discuss the EOL, regardless of their FTP. Even though nearly three quarters of those with a limited FTP and confidante discuss the EOL, over four fifths with a limited FTP and no confidante do so. Having a family confidante does not have the same effect for ACP; smaller proportions of people with expansive or limited FTP and a confidante undertake formal ACP than those with the same FTPs and no family confidante (36% vs. 57% and 51% vs. 71%, respectively). Having a family confidante makes virtually no difference in formal ACP among those with an intermediate FTP.
Proportion of Individuals Who Plan for End-of-Life (EOL) by FTP and Family Confidante, New Jersey EOL Survey, 2006–2008.
Note. Pearson’s χ2 tests significant (p < .05) for discussions and formal advance care planning (ACP) among those with a confidante, where fewer respondents with expansive FTP and more with intermediate FTP discuss and formally plan for the end of life (EOL) than would be expected.
Discussion
Bivariate analysis indicates that ACP behavior differs across groups with different FTPs in a sample of older adults. Binomial logistic regression further clarifies this relationship, indicating, as anticipated previously, that FTP matters differently for informal and formal ACP. Having an expansive FTP (perceiving death to be far off) is significantly associated with lower odds of all ACP in the unadjusted models, although the relationship is stronger for formal ACP than for discussions. The relationship only persists for formal ACP in analyses that include confounding and sociodemographic variables (age, health, experience with death, marital status, education, race, and ethnicity) and explanatory variables (death anxiety and family confidante). Death anxiety helps to explain the relationship between expansive FTP and formal ACP. Moreover, contrary to expectations, having a limited FTP (perceiving death to be closer at hand) is significantly associated with lower odds of formal ACP. Family confidantes potentially facilitate informal discussion, but not formal ACP.
Informal Discussions and FTP
In the unadjusted model, having an expansive FTP is associated with lower odds of discussing the EOL, compared to those with an intermediate FTP. However, FTP is not significantly associated with discussions in models that include additional social, psychological, and demographic control variables. Neither death anxiety nor having a family confidante explains the relationship between FTP and EOL discussions. These findings do not support the initial expectation that FTP is important for ACP discussions. The nonsignificant results found here may reflect two processes. First, they may reflect the nature of discussions. Communication is a relatively common social act, with low barriers to participation. While a potentially upsetting undertaking (California Health Care Foundation, 2012; IOM, 2014), conversations about the EOL may not be linked to social or economic resources, or to contextual factors (such as health, experiences with death, and quality of family relationships). Indeed, as nearly three quarters of respondents had discussed the EOL (Table 1), it is likely that most people, by the time they are 55 years old (the minimum age in the sample), have simply already experienced the opportunity to discuss the EOL. Among respondents who conducted only one type of EOL planning, 6 times as many discussed preferences as formally planned (61 vs. 10). Future research might explore whether FTP is related to such discussions among younger adults. In addition, as prior research shows, close relationships may play a facilitating role in EOL discussions (Boerner et al., 2013; Carr et al., 2013). As the logistic regression and supplemental analysis suggest, having a family confidante may affect the relationship between FTP and informal ACP. Having a confidante decreases the difference in discussions among different FTPs, and a slightly greater proportion of individuals with an expansive FTP discuss EOL if they have a family confidante than if they do not have one (63% vs. 57%).
Formal ACP and Expansive Time Perspective
Although FTP is not significantly associated with informal ACP in fully controlled models, individuals with an expansive FTP are less likely to undertake the proactive steps of executing a living will and/or appointing a DPAHC. For this group, moving beyond merely discussing the EOL to formalizing those discussions may not be a priority. Controlling for death anxiety further accentuates the relationship between expansive FTP and formal ACP. Interpreted through the socioemotional selectivity theory framework, for individuals with an expansive FTP, their death is far off and not something they worry about. If they consider their own mortality, they may think that they have plenty of time to formalize their discussions sometime in the future. Moreover, as these respondents are also younger and healthier than others, they likely have fewer encounters with the health care system, and therefore fewer opportunities to complete formal ACP in a context in which they are likely to do so. Supporting analysis shows that having a family confidante does not necessarily encourage formal ACP. A greater proportion of these individuals undertake formal ACP if they do not have a family confidante, which may indicate they do so in a health care setting. Future research might explore the locations and contexts in which people with expansive FTP complete formal ACP.
Formal ACP and Limited Time Perspective
Although the relationship between limited FTP and formal ACP is not statistically significant in the unadjusted model, controlling for various factors that affect FTP and ACP reveals that limited FTP is significantly associated with lower odds of formal ACP. This result is unexpected: as a group these respondents are older and sicker than the others, and both the socioemotional selectivity theory and ACP literatures would suggest these individuals would be more likely to undertake formal ACP. The fact that they do not could mean they do not consider formal ACP to be a meaningful interaction. Death anxiety may influence why those with a limited FTP do not undertake ACP. Although the FTP categories do not differ significantly with respect to their levels of death anxiety, when death anxiety is the final variable controlled, model fit significantly improves and odds of formal ACP for those with limited FTP decrease (significance level does not change). This result is consistent with research linking death anxiety and avoidance of formal ACP (Carr & Khodyakov, 2007).
Those with a limited FTP are also less likely than those with an intermediate FTP to have a close family confidante. Given prior work showing that close family relations increase the likelihood of engaging in ACP, the lack of this resource may partly explain the lower than expected levels of ACP among those with limited FTP (Boerner et al., 2013; Carr et al., 2013). However, a smaller proportion of individuals with a limited FTP complete formal ACP if they do have a confidante (51% vs. 71%), suggesting that these individuals may forego formal planning if they have someone they trust to make difficult decisions on their behalf.
Education could help explain why those with an intermediate FTP are most likely to undertake ACP. It is possible that these represent the “worried well” (Garfield, 1970), a relatively young and healthy cohort, knowledgeable about and with access to ACP. If this is the case, conceiving of ACP as a health behavior is consistent with health behavior models that indicate people make health decisions in anticipation of avoiding a potentially negative event (in this case, planning to avoid a “bad death”; Fried, Bullock, Iannone, & O’Leary, 2009).
Limitations
This analysis has some limitations. First, the data are cross sectional, so causal ordering of the processes and events under study cannot be determined. FTP is measured based on the individual’s perceived survival likelihood at the moment in which the respondent completed the survey, which may not be consistent with his or her FTP at the moment of planning. Using longitudinal data in future studies would help establish temporal ordering.
Subjective survival likelihood is a proxy for FTP, which may not be an appropriate proxy for the concept as understood in socioemotional selectivity theory (perceived distance to death), although prior studies have used similar measures (Lang, 2000). Including questions from Carstensen and Lang’s (1996) FTP Scale in studies addressing ACP may provide more accurate measures.
Socioemotional selectivity theory also presumes that having a limited FTP leads individuals to focus on close social relationships, such as family. However, not having a family confidante may in fact limit FTP. Future research should focus on the effect of lack of social support on FTP.
Another limitation is that the sampling frame (older adults in central New Jersey who agreed to participate in the survey) does not allow for generalization of conclusions to populations beyond the respondents. Moreover, the nonrandom sampling technique may introduce some degree of bias into the sample. Physicians identified most respondents as near the EOL. However, all were active recipients of health care services and had sufficient physical health and cognitive functioning to participate in a face-to-face encounter that lasted over an hour. The data do not address planning behaviors and characteristics of individuals near the EOL who were not well enough to participate or were unable to do so due to cognitive limitations. These limitations are widely acknowledged and commonly encountered in research focused on terminally ill, older populations (George, 2002).
Conclusion
This article explores how psychological constructs and cognitive factors such as FTP can provide insights into ACP behavior among older adults. This analysis demonstrates that FTP operates differently with regard to informal discussions and formal ACP and is more important for formal ACP than informal discussions. Together, these results may reflect the higher practical barriers to participation in formal ACP compared to discussions. Older adults may have had more opportunities to talk about the EOL, deemphasizing the importance of differences in FTP in whether people have these discussions. If this is the case, increasing exposure to the idea of ACP discussions in the broader population could increase the number of people who have them, particularly among older adults. As nearly everyone who undertook formal ACP also discussed the EOL, broaching the topic in conversations during health care encounters at all ages may make people more comfortable with the idea of discussing it and open a pathway to formal ACP (IOM, 2014).
Medical and policy leaders acknowledge the need for diverse approaches to formal ACP to increase persistently low participation rates, particularly among groups of people who do not typically complete formal ACP (IOM, 2014). Considering the role of FTP adds another facet to understanding what motivates people to undertake or forego formal ACP. Having physicians and other clinicians assess an individual’s FTP through a few short questions, and then frame ACP in a manner that is consistent with that perspective, could yield increased participation in formal ACP. If an individual has an expansive FTP, for example, formal planning might be explained as an opportunity to plan for his future health, even though he or she has plenty of time. If an individual has a limited FTP, formal ACP might be framed as an opportunity to protect family members from having to make difficult decisions and that discussions about ACP can be meaningful and emotionally rewarding (Byock, 1997). This approach might expand the perceived relevance of ACP to a broader range of individuals. It would also encourage ACP without restricting FTP or dampening optimism believed to be a protective factor at older ages.
Footnotes
Acknowledgments
Thank you to Dr. Deborah Carr, Dr. Sharon Bzostek, Dr. Ann Mische, and Dr. Julie Phillips for their ongoing support and feedback in the development and revision of this article.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.