Racial and Ethnic Differences in Advance Care Planning: Results of a Statewide Population-Based Survey

Introduction

Advance care planning (ACP) is a process that guides individuals' medical care to be consistent with their goals and values if they are incapable of making decisions for themselves.^1–3 ACP can take many forms, including designation of a healthcare power of attorney (i.e., healthcare agent [HCA], surrogate); completion of documents such as an advance directive; and communication of values, goals, and preferences for end-of-life care with medical providers, a HCA, and other potential caregivers. ACP is not considered a one-time event but a series of activities that should occur over the lifespan, and may change over time as needed.¹

ACP can improve compliance with end-of-life wishes, enhance satisfaction with care, and result in more goal-concordant and less aggressive care.^4–6 ACP has also been associated with reduced anxiety and depression, higher quality of life, and increased satisfaction with end-of-life care for caregivers and surrogates.^5,7

Previous studies suggest differences in rates of ACP by race and ethnicity. Studies have documented less use of advance directives or other written ACP documents among African American compared to white individuals in several different population groups.^4,8–18 Although less research about ACP has been conducted with Hispanic individuals, available evidence suggests that Hispanics are also less likely than non-Hispanic whites to plan for end-of-life care.^8,9,19,20 Studies have also shown that racial/ethnic minorities tend to receive more intensive end-of-life care, including life-sustaining interventions,^12,21–27 which may be the result of less ACP.^12,28

To date, much of the research about racial and ethnic differences in ACP has focused on completion of written advanced directives (ADs),²⁹ included samples of primarily middle-aged and older adults or individuals with serious medical conditions, and recruited participants from clinical settings. Fewer studies have focused on racial/ethnic differences in ACP activities other than ADs among population-based samples of adults across the lifespan. Given this, as well as the extensive data demonstrating that written ADs alone do not typically facilitate goal-concordant care,^30–33 the goals of this study were to investigate racial and ethnic differences in (1) designation of a HCA; and (2) communication of values, goals, and preferences for end-of-life care with healthcare providers, a HCA, or other family members and friends among a randomly selected statewide sample of adults ≥18 years of age. In addition, we explored reasons for not engaging in ACP activities and the extent to which the reasons differed by race/ethnicity.

Methods

Study design and participants

Data were from a statewide random-digit dial (RDD) telephone survey. Telephone numbers were randomly generated for the State of Massachusetts from three sampling frames (RDD landline, RDD cell phone, listed landline), and individuals ≥18 years of age were randomly selected within households. The study was determined exempt by the university Institutional Review Board.

Data collection

Computer-Assisted Telephone Interviews were conducted by SSRS, Inc. in March to April 2016, using Survox survey software. Professional interviewers completed all data collection. Interviewer training included at least 15 hours of classroom sessions followed by project-specific training, mock interviews, and pretests. At least six call attempts were made to contact nonresponsive numbers by varying the times of day and the days of the week. Interviews were conducted in English or Spanish.

Measures

Outcomes

Naming a HCA

Participants were provided with definitions of a healthcare proxy and HCA and were then asked whether they had ever named someone as their HCA. Participants who had named a HCA were asked if they had talked to the person about their wishes, and if they had shared a copy of the healthcare proxy form with their doctor.

ACP conversations

Participants were asked whether they had ever had a conversation with a healthcare provider about wishes for care near the end of life.³⁴ They were also asked whether they ever had a conversation with at least one person other than a healthcare provider about end-of-life care wishes. Participants who reported having had such conversations with a healthcare provider, their HCA, or anyone else were also asked if the conversations included such topics as goals for care, abilities that were so critical to their life that they could not imagine living without them, or whether or not they wanted certain types of life-prolonging care. Those who endorsed having had at least one such conversation were asked with whom they had the conversation.

Reasons for not engaging in ACP activities

Participants who had not named a HCA, did not want to talk to a healthcare provider about end-of-life wishes, or had never talked to anyone other than a healthcare provider about their end-of-life wishes were asked reasons why they did not want or had not engaged in these activities. Participants selected from responses modeled from a 2013 national survey about end-of-life care conversations.³⁵

Race and ethnicity

Questions about race and ethnicity were similar to items in the 2014 Behavioral Risk Factor Surveillance System (BRFSS).³⁶ Self-reported responses were categorized into Hispanic, black/African American, non-Hispanic white, or non-Hispanic other.

Other sample characteristics

Sociodemographic characteristics included age, gender, marital status, education, and household income. Participants were considered to have a regular source of care if they had one person they thought of as their personal doctor or healthcare provider. All sociodemographic characteristics and regular source of care were modeled from the 2014 BRFSS.³⁶ Having a serious health condition was defined as at least one hospital stay or serious illness, medical condition, injury, or disability requiring a lot of medical care in past 12 months.³⁷ Participants were asked whether they thought doctors should discuss end-of-life care issues with patients.³⁴ Finally, participants were asked how much they had thought about their own wishes for medical treatment if they were facing a health condition that made it hard to function in day-to-day activities. Health insurance status was not asked because <5% of Massachusetts residents were uninsured.³⁸

Statistical analyses

Participant responses were weighted to produce representative estimates of population parameters using a three-step process. First, we adjusted for likelihood of selection based on the probability of a phone number being included in the landline or cell phone sampling frame, likelihood a respondent was reached by landline or cell phone, and likelihood a respondent was selected if the household's landline phone was reached. Second, we conducted poststratification weighting using iterative proportional fitting to reflect the distribution of the adult Massachusetts population. We used the March 2015 Supplement of U.S. Census Bureau's Current Population Survey³⁹ for demographic benchmarks and the National Health Interview Survey⁴⁰ for estimates of cell phone only use. Finally, we constrained the weights to control the variance among them.

We assessed differences by race/ethnicity in the sample characteristics using Pearson's chi-squared tests. Second, we compared ACP activities by race/ethnicity using Person's chi-squared tests. Next, we computed multivariable logistic regression models to assess the relationship between race/ethnicity and ACP activities controlling for other sample characteristics. Before inclusion in multivariable models, we examined measures of association for all variables to determine potential problems with multicollinearity. To formally address multicollinearity, we computed model diagnostics specifically assessing the variance inflation factor (VIF) associated with each variable. A VIF above 2 was deemed indicative of potential multicollinearity and required additional examination. Finally, we used frequency distributions and Pearson's chi-squared tests to assess reasons for not engaging in ACP and the extent to which the endorsed reasons differed by race/ethnicity.

All statistical analyses were conducted using SAS/STAT software, version 9.4 of SAS System for Windows^© 2013, SAS Institute. Two-tailed p-values of <0.05 were considered statistically significant.

Results

Participant characteristics

A total of 1851 interviews were completed (American Association of Public Opinion Research Response rate 3 = 25%). The design effect was 1.56, and the margin of sampling error was ±2.28%.

Participants who identified as non-Hispanic white were more likely to be older, have more than a high school education, have incomes of ≥$50,000, and report having thought more about wishes for medical treatment than members of other racial/ethnic groups (Table 1). Compared with non-Hispanic whites, participants who identified as black/African American were less likely to report being partnered. No other sample characteristics differed by race/ethnicity.

Naming a HCA

Less than half (44%, 95% confidence interval [CI] = 41.3%–47.0%) of all participants had named a HCA. In multivariable analyses, participants who identified as Hispanic or non-Hispanic other were less likely than non-Hispanic whites to have named a HCA (Table 2, column 2). In addition, those ≥50 years (vs. 18–49), female (vs. male), partnered (vs. not), and reported having a serious health condition in the past 12 months (vs. not) were more likely to have named a HCA.

Among those who had named a HCA, the majority (86.2%, 95% CI = 83.3%–89.2%) had talked to the HCA about their end-of-life care wishes, but only half (54.2%, 95% CI = 50.0%–58.4%) had shared a copy of the healthcare proxy form with their doctor. There were no racial/ethnic differences in either activity.

Conversation about end-of-life care wishes

Only 14.5% (95% CI = 12.6%–16.5%) of all participants had ever had a conversation with a healthcare provider about their end-of-life care wishes, with no differences by race/ethnicity (Table 2, column 3). In multivariable analysis, older age, serious health condition in the past 12 months, and regular source of care were the only factors significantly associated with having had an end-of-life care conversation with a healthcare provider.

Over half (53.9%, 95% CI = 51.0%–56.8%) of all participants reported having had a conversation with someone other than a healthcare provider about their end-of-life wishes. In multivariable analyses, non-Hispanic whites were more likely than other racial/ethnic groups to report having had such conversations (Table 2, column 4). In addition, those who were older, female, partnered, more highly educated, with higher incomes, and a serious health condition in the past 12 months were more likely to have had an end-of life care conversation with someone other than a healthcare provider.

Conversations about goals and preferences

Among participants who reported having had a conversation about their end-of-life care wishes with a healthcare provider, their HCA, or anyone else, 68.9% (95% CI = 65.7%–72.2%) stated that they had discussed their goals and preferences if they were facing a health condition that made it hard to function in day-to-day activities. Among those who had such a conversation, the discussion was with a healthcare provider for 27.5% (95% CI = 23.7%–31.2%) a HCA for 68.1% (95% CI = 63.3%–73.0%), and someone else for 47.4% (95% CI = 43.9%–50.9%) with no racial/ethnic differences for any of the measures.

Reasons for not engaging in ACP activities

Among the 55.9% (95% CI = 53.0%–58.7%) of participants who had not named a HCA, the most commonly endorsed reasons for not having done so were because they were healthy and did not need one yet, or felt that their spouse or family members would know what they wanted (Table 3, column 2). Two reasons differed by race/ethnicity. First, compared with non-Hispanic whites (12.0%, 95% CI = 8.8%–15.2%), Hispanics (36.7%, 95% CI = 24.5%–48.8%), black/African Americans (35.7%, 95% CI = 21.8%–49.6%), and non-Hispanic others (29.8%, 95% CI = 18.9–40.6; data not shown) were more likely to indicate that they did not have a person to choose as an agent. Second, compared with non-Hispanic whites (6.0%, 95% CI = 3.7%–8.3%), non-Hispanic others were more likely to indicate that if they chose one person, it would upset others (17.7%, 95% CI = 8.8%–26.7%; data not shown).

More than one-third (36.9%, 95% CI = 33.1%–38.7%) of participants had never had a conversation about end-of-life care wishes with a healthcare provider and did not want to do so. The most commonly endorsed reasons were because they preferred to speak first to family members about their wishes or because they were not sick and did not feel it was necessary (Table 3, column 3). One reason differed by race/ethnicity. Compared with non-Hispanic whites (15.3%, 95% CI = 10.7%–20.0%), Hispanics were more likely to indicate that they were waiting for a doctor to start the conversation (42.2%, 95% CI = 28.1%–56.2%; data not shown).

Among the 46.1% (95% CI = 43.2%–49.0%) of participants who had never had a conversation with someone other than a healthcare provider about end-of-life wishes, the most commonly endorsed reasons were because they were not sick and did not feel it was necessary, or they would trust their healthcare team to make decisions when needed (Table 3, column 4). There were no racial/ethnic differences in the reasons.

Discussion

In this statewide survey of adults ≥18 years of age, racial and ethnic minorities reported less participation in some but not all ACP activities for which they were questioned. Disparities in ACP were not the result of racial/ethnic minorities believing that such activities were unimportant; 86% of all participants believed that having conversations about end-of-life care wishes were important. Nonetheless, only 44% of all participants had named a HCA, and the odds of naming a HCA were 40%–60% lower for participants who identified as Hispanic or non-Hispanic other than for non-Hispanic whites after controlling for other sample characteristics. Lower rates of HCA may be partially explained by the fact that racial/ethnic minorities were more likely to indicate that they did not have a person to choose as an agent. Therefore, healthcare providers should explore whether there are potential nontraditional decision makers (e.g., fellow members of a religious organization)⁴¹ who might serve in this role for such patients. Racial/ethnic minority participants were also more likely than non-Hispanic whites to report that if they chose one person, it would upset others, and may particularly benefit from resources that provide guidance for explaining why a particular individual was chosen as the HCA.

There were no differences by race/ethnicity in having had conversations with a healthcare provider, believing that doctors should discuss end-of-life care issues with patients, or in wanting to have a conversation about end-of-life care wishes with a healthcare provider if this had not been done. Only 15% of respondents reported having had a conversation with a healthcare provider about wishes for end-of-life care and less than one-third had discussed what was important to them if they were facing a health condition that made it hard to function in day-to-day activities. Therefore, very few Massachusetts residents have articulated their goals and preferences to clinicians in ways that would allow for the provision of goal-concordant care if it was needed. Given the low rates of conversations with clinicians for all participants, clinical care protocols should be considered for enhancing routine and regularly updated ACP conversations between patients and their providers. Promising resources to aide in this communication include Honoring Choices,⁴² The Conversation Project,⁴³ Five Wishes,⁴⁴ Caring Conversations,⁴⁵ and the Serious Illness Care Program.⁴⁶ Clinical care protocols may be particularly helpful for Hispanics, given that they were more likely to report not having had a conversation about end-of-life care wishes because they were waiting for a healthcare provider to initiate the conversation.

More than half (54%) of all participants reported having had a conversation with someone other than a healthcare provider about their wishes for end-of-life care. However, similar to findings by Carr,¹⁹ the rates for such conversations were significantly lower for racial/ethnic minorities compared to non-Hispanic whites. Unfortunately, we do not have data to explain these disparities. Nevertheless, we found no racial/ethnic differences in reasons for not having had these conversations. While at least one recent population-based survey of U.S. adults ≥50 years of age found that disparities in ACP were not explained by sociocultural factors,⁴⁷ future studies should continue to explore topics such as culture, religion, personal health values, preferences for life-sustaining treatments, and distrust of healthcare settings as potential explanations for differences observed.^48–52

In addition to disparities by race/ethnicity, we also observed that men and individuals with fewer years of education were less likely to have had a conversation with someone other than a healthcare provider about end-of-life care wishes. Lack of experience in having conversations about values, goals, and preferences with family and friends may make it more difficult to address these issues in a healthcare context and may contribute to care disparities. Further exploration and confirmation of these differences by gender and education are needed.

Taken together, our findings suggest that health literacy may be an important factor in explaining disparities in ACP. Low health literacy has been documented as an obstacle to communicating values and preferences in decision making.^53–55 Therefore, while there may be some cultural issues specific to racial and ethnic minorities such as not having anyone to appoint as a HCA and perceptions of family disappointment in the choice of agent, health literacy is a plausible explanation for some of the findings we observed and should be considered in any ACP aides.

There are a number of study limitations. First, the cross-sectional study design limits our ability to make causal inferences. Second, similar to other population-based studies on similar topics,^{19,47,56–58} we did not verify self-reported data. It is possible that participants may have over-reported ACP activities. However, it is unlikely that there was differential reporting by race/ethnicity. Third, to include all participants in the analyses, we combined individuals who did not identify as Hispanic, black/African American, or white into a single group. The heterogeneity within this group likely masked important subgroups worthy of future studies with larger sample sizes. Similarly, we did not include individuals who were not fluent in English or Spanish and may have unique needs related to ACP. Next, the sample was limited to those with access to a telephone and able to communicate by telephone. Finally, some of the null findings may be due to small sample sizes.

Our study highlights that racial/ethnic minorities may be disadvantaged in the quality of care they receive if they have a serious illness and are unable to make decisions for themselves because most have not talked to anyone about their goals, values, or preferences for care. Since the main objective for ACP should be to prepare patients, clinicians, and surrogates to make the best possible in-the-moment medical decisions that are consistent with a patient's goals,^1,3 future work should focus on ensuring that all adults have named a HCA and have, at a minimum, spoken to their agent about their goals and preferences. This will move ACP efforts away from traditional methods of only targeting frail and older patients to normalizing the activities as part of high-quality, patient-centered healthcare.

Acknowledgments

The authors thank Andrew Dreyfus, Lachlan Forrow, MD, Ellen Dipaola, JD, Ellen Goodman, Megan Mclean, JD, MBA, and Harriet Warshaw for their comments on the survey instrument as well as David Dutwin, PhD, Tim Pokalsky, and Susan Sherr, PhD from SSRS, Inc. for their assistance with data collection. The survey was conducted for the Massachusetts Coalition for Serious Illness Care, made possible with funding from the Rx Foundation and Blue Cross Blue Shield of Massachusetts.

Author Disclosure Statement

No competing financial interests exist.