Engagement in Multiple Steps of the Advance Care Planning Process: A Descriptive Study of Diverse Older Adults

Abstract

OBJECTIVES: To assess engagement in multiple steps of the advance care planning (ACP) process 6 months after exposure to an advance directive. In this study, ACP is conceptualized similarly to the behavior change model.

DESIGN: Descriptive study.

SETTINGS: County general medicine clinic in San Francisco.

PARTICIPANTS: One hundred seventy-three English or Spanish speakers, aged 50 and older (mean 61) given a standard (12th-grade reading level) and an easy-to-read (5th-grade reading level) advance directive.

MEASUREMENTS: Six months after exposure to two advance directives, self-reported ACP contemplation; discussions with family, friends discussions with clinicians; and documentation were measured. Associations were examined between ACP steps and between subject characteristics ACP engagement.

RESULTS: Most participants (73%) were nonwhite and 31% had less than a high school education. Sixty-one percent contemplated ACP, 56% discussed ACP with family or friends, 22% discussed ACP with clinicians, and 13% documented ACP wishes. Subjects who had discussed ACP with their family or friends were more likely to discuss ACP with their clinicians (36% vs 2%, P<.001) and document ACP wishes (18% vs 4%, P=.009) than those who had not. Latinos and subjects with less than a high school education discussed ACP more often with family or friends (P<.06) and clinicians (P<.03) than other ethnic groups and subjects with more education.

CONCLUSIONS: ACP involves distinct steps including contemplation, discussions, and documentation. The ACP paradigm should be broadened to include contemplation and discussions. Promoting discussions with family and friends may be one of the most important targets for ACP interventions, and literacy- and language-appropriate advance directives may help reverse patterns of sociodemographic disparities in ACP.

Advance care planning (ACP) is an iterative process whereby patients identify values and goals for future medical care.^1–3 Some clinicians and researchers have endorsed adapting the behavior change model (transtheoretical model) to describe the process by which patients identify, communicate, and document their treatment wishes.^4,5 Building on this work, this article presents a conceptual model of ACP that includes similar discrete steps of change (Figure 1). The steps include precontemplation (a stage in which the individual lacks awareness of or has no desire to engage in ACP planning), contemplation of one's values and future treatment wishes (a stage in which individuals understand the relevance of ACP to their own lives and begin to form intentions to engage in ACP), preparation and values clarification (a transitory stage that links contemplation to the action stages but can also link many of the action phases to one another), actions such as discussions with family, friends, and clinicians, and documentation (a stage in which individuals overtly engage in behaviors that make their ACP wishes known), and maintenance or reflection on one's choices (a stage in which individuals have made end-of-life choices and are in a position to reflect on these choices given changes in their life circumstances). Similar to behavior change, it was hypothesized that the ACP steps may, at times, follow a sequential trajectory, but that patients may also revert back to previous stages such as preparation.⁶ Unlike the behavior model, for ACP it may also be possible for patients to skip stages based on life circumstance and/or proactive clinicians.

Figure 1

Conceptual model of the Process of Advance Care Planning. Bolded boxes are the steps of the ACP process described in this study. Conceptual model adapted from Prochaska et al.⁴ and Pearlman et al.⁵

To the authors' knowledge, no studies have assessed engagement in multiple steps of the ACP process in response to an intervention. In contrast, many ACP intervention studies, including those demonstrating sociodemographic disparities in ACP, have focused solely on documentation of an advance directive or proxy form.^7–13 Fewer studies have assessed discussions,^14–17 and to the authors' knowledge, none have assessed engagement in a broader range of ACP steps, such as contemplation. Contemplation of one's end-of-life wishes and discussions with family and friends can be as important in guiding care at the bedside as discussions with physicians and completing advance directives.² Therefore, it is important to understand the extent to which interventions encourage patients to engage in each of these ACP steps.

Given low levels of engagement in ACP by minorities and patients with lower educational attainment,^18–24 an advance directive modified to patients' literacy (5th-grade reading level with graphics) and language (English or Spanish) needs was created.²⁵ Preferences between this form and a standard advance directive written at a 12th-grade reading level were compared in English- and Spanish-speaking older adults from an urban general medicine clinic. Six months after exposure to the advance directives, engagement in multiple steps of ACP, including contemplation, discussions with family or friends and physicians, and documentation, were assessed. It was hypothesized that ACP would occur in discrete steps, with more patients engaging in contemplation than other steps. Because it was hypothesized that minorities and patients with lower educational attainment would be less likely to engage in ACP,^18–24 associations between engagement in the ACP steps and subject characteristics were explored.

METHODS

Subjects and Study Design

For this descriptive study, telephone interview information was collected between February and July 2005, 6 months after subjects participated in a study assessing their preference for a standard or low-literacy advance directive. The complete details of the preference study have been published.²⁵ Briefly, a convenience sample of 205 subjects were enrolled from the General Medicine Clinic at San Francisco General Hospital, a large, urban county hospital. Patients in this clinic have been found to have high chronic disease burden, even at younger ages.²⁶ Physicians referred patients who met the inclusion criteria of being aged 50 and older, having a primary care physician, and self-reporting fluency in English or Spanish. Participants who were deaf, delirious, or diagnosed with dementia (as determined by subjects' clinicians) and whose measured visual acuity was less than 20/100 (as determined by study staff) were excluded.²⁷ These exclusion criteria were chosen, because these subjects may have had difficulty reviewing the forms or being reached for the follow-up telephone interviews. A consent process designed for low-literate populations was used,²⁸ and subjects were offered $20 for participation. The University of California, San Francisco, and San Francisco General Hospital institutional review boards approved the study.

Participants in the advance directive preference study reviewed both a standard form (written at the 12th-grade reading level without graphics) and a low-literacy form (written at the 5th-grade reading level with graphics and a clear layout, available at http://www.iha4health.org) in English or Spanish.²⁵ Seventy-three percent (n=149) of subjects preferred the redesigned form, 17% (n=35) preferred the standard form, and 10% (n=21) did not report a preference. Subjects were offered both forms to take home, but no further interventions regarding ACP were implemented.

During enrollment for the advance directive preference study, participants' age, sex, income, years of education, language spoken, self-rated health status, previous experience with end-of-life issues (including prior admission to the intensive care unit (ICU), having family or friends admitted to the ICU, or having helped someone fill out an advance directive), previous advance directive completion, and patient religiosity (very to extremely vs somewhat to not at all) were recorded. Because engagement in ACP may vary according to race or ethnicity,^18–24 self-identified race or ethnicity was determined in the categories of white, Latino, African American, Asian, or multiracial, multi-ethnic, and other. Literacy was assessed using the validated 36-item short form Test of Functional Health Literacy in Adults in English and Spanish.²⁹ By convention, scores of 22 or lower (approximately less than or equal to the 8th-grade reading level) were defined as limited literacy, and scores greater than 22 as adequate literacy.³⁰

Telephone Interviews at 6 Months

Six months after exposure to the advance directives, bilingual research assistants assessed participants' report of ACP engagement by telephone.

Outcomes

The main outcomes were self-reported engagement in the four ACP steps of contemplation, discussion with family or friends, discussion with physicians, and documentation. Each participant was asked: “Since the day you finished the study, have you (1) thought; (2) talked to your family; or (3) talked to your doctor about the type of medical care you might want if you were sick or near the end of your life?” and (4) “Have you filled out an advance directive form for yourself?” The response categories were yes, no, and I don't know. Responses of I don't know were classified as no for these analyses.

Data Analysis

Percentages were used to describe engagement in ACP and participant characteristics. Associations between the ACP steps were explored using chi-square tests. Associations between participant characteristics and engagement in the ACP steps were also assessed using chi-square tests and t-tests. Because there was interest in assessing differences in ACP engagement across race or ethnicity and because all but two Spanish speakers were Latino, only race or ethnicity, rather than language, was used in bivariate and multivariate analyses.

To determine whether subject characteristics were associated with each ACP step, stepwise multivariate logistic regression analysis was conducted. Variables forced into the model included characteristics previously shown to be associated with ACP engagement, including age, race or ethnicity, and education.^18–24 Because years of education and literacy score were correlated (Pearson correlation coefficient=0.54), it was decided to include education, because, of the two, it was most highly associated with ACP in bivariate results (see Results). Literacy, along with other covariates listed in Table 1, were then added in a stepwise fashion. All variables associated with the outcome at P<.05 after backwards elimination were kept in the final model.

Table 1. Participant Characteristics (N=173)

Characteristic	Value
Age, mean ± SD	61 ± 8
Female, %	54
Race or ethnicity, %
White	27
Latino	29
African American	25
Asian	8
Multiracial, multiethnic, other	11
Income <$10,000/year, %^*	51
Education
Years, mean ± SD	12 ± 5
<High school education, %	31
Literacy^†
Literacy score, mean ± SD	25 ± 11
Limited literacy, %	35
Spanish speaking, %	28
Very to extremely religious, %	49
Fair to poor self-rated health, %	69
Personal end-of-life experience, %
Previously filled out an advance directive^‡	15
Previously admitted to intensive care unit	40
Had friends/family admitted to intensive care unit	64
Helped others fill out advance directive	16

• ^* Income data available for only 144 participants (83%).
• ^† Literacy was assessed using the short form Test of Functional Health Literacy in Adults, a 36-item, timed reading comprehension test. Participants with scores ≤22 (possible range 0–36) were considered to have limited literacy or an approximate reading level of the 8th grade or less.
• ^‡ Because sensitivity analysis was performed by excluding participants who reported completing an advance directive before the preference study, this variable was not included in the composite variable of “personal end-of-life experience” used in subsequent analyses.
• SD=standard deviation.

When assessing discussions with doctors, sensitivity analyses were performed by excluding participants who had not seen their doctor in the previous 6 months (19 participants). Because it was also desired to assess ACP engagement in response to the advance directive intervention, an additional sensitivity analysis was performed by excluding participants who reported completing an advance directive before the preference study (n=26). Intercooled Stata version 8 software was used for all analyses (Stata Corp., College Station, TX).

RESULTS

Participant Characteristics

Of the 205 participants enrolled in the advance directive preference study, 173 (84%) participated in the 6-month follow-up. Six participants refused telephone follow-up, 14 were unreachable, nine did not have working telephones, and three had died. The 32 participants lost to follow-up did not differ from the remaining sample except that they were more likely to have limited versus adequate literacy (65% vs 35%, P=.001).

The mean age±standard deviation of participants at the 6-month follow-up was 61±8, one-quarter were white, most had limited financial resources, one-third had not completed high school, one-third were Spanish speaking, and more than one-third had limited literacy (Table 1). More than two-thirds of subjects reported fair to poor health, and more than one-third had previously been admitted to the ICU.

The ACP Process

Six months after reviewing two advance directives, most subjects reported contemplating ACP (61%) or discussing ACP with their family or friends (56%). Fewer subjects reported discussing ACP with their physicians (22%) or documenting their ACP wishes in an advance directive (13%) (Figure 2).

Figure 2

Percentage of participants who completed each step of the advance care planning process (N=173). Note: These results did not change appreciably after excluding 19 participants who had not seen their doctors in the 6 months since the advance directive preference study or the 26 participants who had filled out an advance directive before the preference study. ^*Data missing for one participant.

Participants who had contemplated ACP in the previous 6 months were three times as likely to have discussed ACP with their family or friends (P<.001) and their clinicians (P=.001) as those who had not (Table 2). Contemplation was not significantly associated with documentation (P=.10). Participants who discussed ACP with their family or friends were 12 times as likely to discuss ACP with their doctor (P<.001) and four times as likely to document their ACP wishes (P=.009) as those who had not. Participants who discussed ACP with their doctor were twice as likely to document ACP wishes (P=.08).

Table 2. Associations Between Steps of the Advance Care Planning (ACP) Process (N=173)

ACP	Discussed ACP with Family^* n (%)		P -Value	Discussed ACP with Doctor n (%)		P -Value	Documented ACP Wishes^* n (%)		P -Value
	Yes	No		Yes	No		Yes	No
Contemplated ACP wishes
Yes	81 (77)	24 (23)	<.001	32 (30)	73 (70)	.001	17 (16)	88 (84)	.10
No	15 (22)	52 (78)		6 (9)	62 (91)		5 (7)	62 (93)
Discussed ACP with family or friends
Yes	—	—	—	36 (38)	60 (63)	<.001	18 (19)	78 (81)	.009
No	—	—	—	2 (3)	74 (97)		4 (5)	72 (95)
Discussed ACP with doctor^*
Yes	—	—	—	—	—	—	8 (21)	30 (79)	.08
No	—	—	—	—	—	—	14 (10)	120 (90)

• Note: The results did not change appreciably after excluding 19 participants who had not seen their doctors in the 6 months since the advance directive preference study or the 26 participants who had filled out an advance directive before the preference study.
• ^* Data missing for one participant.

Participant Characteristics Associated with the ACP Process

Latinos were more likely than other ethnic groups to discuss ACP with family and friends (73% vs ≤53% for all other groups, P=.06) and to discuss ACP with their doctors (37% vs ≤21% for all other groups, P=.03) (Table 3). Participants with less than a high school education were more likely than those with higher education to discuss ACP with family or friends (69% vs 50%, P=.03) and to discuss ACP with their doctors (35% vs 16%, P=.005). Participants who had lower incomes and no prior personal end-of-life experience (a composite variable that means helping another person fill out an advance directive or having personal experience or experience with family or friends being in the intensive care unit) were also more likely to discuss ACP with their doctors (26% vs 13%, P=.04 and 39% vs 17%, P=.003, respectively). Participants with fair to poor versus good to excellent health status and subjects with more versus less than a high school education were more likely to fill out an advance directive (16% vs 6%, P=.06, and 17% vs 4%, P=.02, respectively).

Table 3. Patient Characteristics Associated with Contemplating, Discussing, or Documenting Advance Care Planning (ACP)

Patient Characteristic	Contemplated ACP		Discussed ACP with Family or Friends		Discussed ACP with Doctor		Documented ACP Wishes
	%	P -Value	%	P -Value	%	P -Value	%	P -Value
Age
50–54 (n=43)	63		67		26		21
55–58 (n=43)	65	.81	58	.21	14	.54	14	.13
59–65 (n=42)	59		55		24		12
>65 (n=45)	56		44		24		4
Race or ethnicity
White (n=46)	50		51		13		9
Latino (n=51)	75		73		37		12
African American (n=43)	58	.13	44	.06	21	.03	19	.44
Asian (n=14)	50		50		14		21
Multiracial or multiethnicity, other (n=19)	63		53		11		5
Income, $/year
<10,000 (n=73)	63	.19	60	.08	26	.04	15	.48
≥10,000 (n=71)	52		45		13		11
Education
<High school (n=54)	70	.09	69	.03	35	.005	4	.02
≥High school (n=118)	57		50		16		17
Literacy^*
Limited (n=61)	57	.51	52	.42	31	.03	8	.20
Adequate (n=112)	63		58		17		15
Self-rated health status
Fair or poor (n=120)	62	.69	57	.60	22	.89	16	.06
Good or excellent (n=53)	58		53		23		6
Personal end-of-life experience^†
Yes (n=132)	63	.29	56	1.0	17	.003	14	.60
No (n=41)	54		56		39		10

• Note: With the exception of age, only participant characteristics that were statistically significantly associated with the ACP steps were included in this table. The results for sex and religiosity are not displayed.
• ^* Literacy was assessed using the short form Test of Functional Health Literacy in Adults, a 36-item, timed reading comprehension test. Participants with scores ≤22 (possible range 0–36) were considered to have limited literacy or an approximate reading level of the 8th grade or less.
• ^{† †} End-of-life experience is a composite variable that means helping another person fill out an advance directive or having personal experience or experience with family or friends being in the intensive care unit.

After multivariable analysis adjusted for age, race or ethnicity, and years of education, Latinos remained more likely than whites to contemplate ACP (adjusted odds ratio (AOR)=3.28, 95% confidence interval (CI)=1.21–8.88) and to discuss ACP with family or friends (AOR=2.68, 95% CI=1.00–7.35). Subjects without prior end-of-life experience also remained more likely to discuss ACP with their doctors (AOR=2.68, 95% CI=1.13–6.35). In addition, with each 1-year increase in age, subjects were less likely to discuss ACP with family or friends (AOR=0.95, 95% CI=0.91–0.99) or to fill out an advance directive (AOR=0.92, 95% CI=0.85–0.99).

The results did not change appreciably after excluding the 19 subjects who had not seen their doctor in the prior 6 months. However, after excluding subjects who had filled out an advance directive before enrollment in the preference study (n=26), age was no longer associated with filling out an advance directive (P=.08).

DISCUSSION

To the authors' knowledge, this is the first study to explore multiple steps in the ACP process (contemplation, discussion, and documentation) after exposure to an advance directive. Most subjects had engaged in contemplation and discussion of ACP with family and friends. Fewer subjects had discussed ACP with physicians or documented their ACP wishes. Subjects who contemplated ACP were more likely to discuss ACP with their family or friends and physicians. Subjects who discussed ACP with their family or friends were more likely to discuss ACP with their physicians and to document their ACP wishes.

Although others have described ACP in terms of the behavior change model and have incorporated these principles into interventions,^5,17 to the authors' knowledge, no studies have explored whether exposure to an advance directive resulted in engagement in many steps of the ACP process. In one of the only other studies to assess multiple steps of the ACP process, 22% of outpatients contemplated treatment preferences and were more likely to contemplate their wishes and discuss them with family than talk with physicians or document their wishes.³¹ This study was conducted in the absence of an intervention, and therefore, the results may not be directly comparable to those of the current study. Another study showed that patients who discussed ACP with their physicians were more likely to document their wishes.³² Although these studies and the current study were cross-sectional, the drop-off between individual steps suggests that the ACP process may occur in discrete steps along a trajectory, as described in the conceptual framework (Figure 1).

Although the mean age of the diverse study sample was 61, close to 70% reported fair to poor health. This suggests that the cohort was quite infirm and that these patients, even those younger than 65, were appropriate targets for ACP. Participants' age was not associated with engagement in the ACP steps, except that older patients were less likely than younger patients to engage in discussions with family and friends. This association remained after excluding subjects who had previously completed an advance directive. These results are somewhat discouraging, because the ACP step of discussions with family and friends appeared to be one of the most important in terms of engaging in other steps of the ACP process. Possible explanations may include fear of burdening their family or friends³³ and social isolation.³⁴ The results suggest that special attention and resources may be needed to facilitate discussions between older adults and their family and friends. For older adults who are socially isolated, clinicians may need to focus on other ACP steps.

Contrary to the hypothesis that minority status and lower education would be associated with lower levels of engagement in ACP,^18–24 it was found that Latinos and participants with lower education were more, rather than less, likely to engage in ACP. Specifically, Latinos were more likely to contemplate ACP and discuss ACP with family or friends and physicians. Subjects with lower education were also more likely to engage in discussions with family or friends and physicians. A number of factors could explain this reversal of expected demographic patterns. Studies have shown that patients can be motivated to take action in response to information that they can understand.³⁵ In the current study, Latinos and subjects with lower education may have been disproportionately activated to engage in ACP after exposure to an advance directive that was linguistically concordant and appropriate to their literacy level.²⁵ Prior studies have shown that patients of all education and literacy levels prefer written materials in an easy-to-read format.³⁶ However, the benefits of the easy-to-read advance directive may have particularly helped subjects with lower education, limited literacy and non-English speakers, because before the intervention, many of these subjects may never have been exposed to easy-to-read materials designed to help them engage in ACP.^35,37 Participants without prior end-of-life experience may have been similarly activated. Another explanation for the reversal of expected demographic patterns may be that, by broadening the paradigm of the ACP process to include contemplation and discussions, this study may be capturing the clinically relevant ACP steps in which minorities and those with lower education are more likely to engage.³¹

This study has important implications. First, the results indicate that ACP may be a process involving discrete steps, including, but not limited to, documentation. Although ACP steps may not always be linear, the results suggest that patients may move through the steps along a common trajectory. Understanding that ACP is a process comprising discrete steps can help clinicians recognize where patients are along the ACP trajectory and target interventions that will help patients move to the next appropriate step. For some patients, the next appropriate step or the appropriate final outcome may not be documentation. Because it is difficult to predict a patient's clinical course, and because advance directives often provide broad guidance rather than specific direction, helping patients engage in the contemplation step and in discussions with family and friends may be equally important as an advance directive in guiding clinical care at the beside.^2,7 In addition, patients may prefer to engage in some steps rather than others (e.g., discussions rather than documentation). Therefore, appropriate ACP endpoints for many patients may be contemplation or discussions with family, friends, or clinicians.

Second, the ACP paradigm, including what is considered a “successful” outcome to an ACP intervention, may need to be broadened to include a range of ACP steps such as contemplation, values clarification, and discussions and not focus solely on documentation. Future longitudinal studies should explore whether interventions that promote contemplation and discussions, even if they do not lead to increased documentation, result in medical care that is more consistent with patients' goals.

Third, because discussions with family and friends were associated with completing other ACP steps in this study, facilitating family discussions may be one of the most important targets of ACP interventions. Many patients, especially minorities, prefer to rely on family or friends to use their own judgment in making end-of-life decisions rather than being bound to a written document.^38,39 Because surrogate decision-makers are often left to make end-of-life decisions, including them in the ACP process is imperative. Involving proxies in the ACP discussion can help foster surrogate decision-makers' understanding of patients' values concerning medical care.⁴⁰ For patients who do not have surrogate decision-makers, discussions with physicians or documentation may be more appropriate ACP targets and should be facilitated.

Fourth, offering patients a literacy- and language-appropriate advance directive may have helped reverse the pattern of sociodemographic disparities often found in ACP. The easy-to-read advance directive may have also activated patients to engage in the steps of the ACP process with which they were most comfortable, namely contemplation and discussions with family and friends—steps that were detectable given the broadened ACP paradigm used in this study.

Fifth, this study suggests the importance of modifying federal or state regulations to legitimize ACP steps other than documentation. The focus on documentation may be related to the passing of the Patient Self-Determination Act, which requires healthcare organizations to provide information about advance directives and has, in some instances, been incorporated into physician performance measures.⁴¹ In addition, many state laws limit surrogate decision-making and the use of oral advance directives.^2,3,42 Even if ACP research studies include ACP outcomes such as contemplation, values clarification, and discussions with family as important end points, these changes may not be translated into clinical practice unless the broader paradigm is embraced in the current regulatory environment.

The ACP process is complex, iterative, and dynamic, and as with any model, there are caveats. Given a proactive clinician or life experiences, some patients may skip ACP steps, as was evidenced by the small proportion of subjects who did not complete earlier steps but completed later steps in the ACP process. In addition, discussions with family and friends may be equivalent to or supersede the contemplation step for some patients, such as Latinos, who are more likely to involve family and friends in their medical decision-making.⁴³ Furthermore, given the cross-sectional nature of this study and the possibility that subjects may have completed the early phases of ACP before exposure to two advance directives, it may have been possible to detect movement only along the later stages of the ACP process.

This study lacked additional information that might have helped further define the ACP process. For subjects who engaged in ACP, information concerning what triggered their activation was not available. The triggering event could have been the advance directive intervention; a proactive clinician, family member, or friend; or a change in clinical condition. For subjects who reported engaging in the contemplation step, information concerning the type of medical care subjects were considering was not available. Because contemplation was assessed only at one time period, it was not possible to further assess subjects' thought processes. In addition, it was not possible to explore the ACP step of precontemplation or preparation. To assess precontemplation, future studies could ask whether patients have knowledge of the components of ACP or find ACP personally relevant.⁴⁴ Future studies should also explore the best way to help patients clarify their values for health care and assess congruence between patients and their surrogates with respect to treatment wishes.

This study has additional limitations. Sixteen percent of the sample was lost to follow-up, which may have led to inadequate power to detect significant associations, specifically with the documentation step. Because subjects with cognitive impairment and limited vision were excluded, a disproportionate number of subjects in older age groups may have been excluded, limiting the ability to detect further associations between age and the ACP steps. Generalizability is limited, because the study was conducted at an urban general medicine clinic. Completion of the ACP steps was measured according to self-report, possibly introducing recall bias by overreporting engagement in ACP. There was no control group, and therefore the extent of activation to engage in ACP due directly to the advance directive interventions could not be assessed. Because engagement in ACP could have been the result of exposure to two advance directives, the results of this study may not approximate naturally occurring ACP in the absence of an intervention. Because almost all Latinos were Spanish speaking, it was not possible to disentangle the potential associations between race or ethnicity and language on ACP outcomes. Finally, information about the language discordance between subjects and providers, which might have contributed to communication difficulties, was not available.⁴⁵

CONCLUSION

After exposure to a standard and a literacy- and language-appropriate advance directives, diverse older outpatients reported engaging in multiple steps of the ACP process. ACP appeared to occur in discrete steps and to follow a trajectory from contemplation to discussions to documentation. More subjects contemplated and discussed ACP with family or friends than discussed ACP with clinicians or documented their treatment wishes. However, subjects who discussed ACP with family or friends were more likely to discuss ACP with clinicians and to document their wishes. Contrary to prior studies, Latinos and participants with lower education were more rather than less likely to engage in ACP. This may reflect exposure to a literacy- and language-appropriate advance directive or a broadening of the ACP paradigm to include steps upstream from the documentation step, such as contemplation and discussions.

This study suggests that important consideration should be given to broadening the paradigm of ACP in clinical practice, research studies, and policy to include all steps of the ACP process, including contemplation and discussions with family and friends; facilitating family discussions and offering extra resources for patients who lack surrogates; and offering patients literacy- and language-appropriate advance directives, because they may help to mitigate ACP disparities.

ACKNOWLEDGMENTS

Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the author and has determined that none of the authors have any financial or any other kind of personal conflicts with this manuscript.

Dr. Sudore is funded in part by the Pfizer Foundation through the Clear Health Communication Fellowship. This Foundation has no financial incentives associated with the results of this study, nor does Dr. Sudore. Dr. Sudore also created the redesigned advance directive for use in this study; the redesigned advance directive is offered for free on the Institute for Health Care Advancement's Web site (http://iha4health.org), and therefore, Dr. Sudore does not receive any royalties or financial support from this form. Dr. Sudore and this study were supported by the American Medical Association Foundation; National Institutes of Health (NIH), National Institute on Aging (NIA) Grant K07 AG000912; National Institutes of Health Research Training in Geriatric Medicine Grant AG000212; the Pfizer Fellowship in Clear Health Communication; the NIH Diversity Investigator Supplement 5R01AG023626-02; and NIA Mentored Clinical Scientist Award K-23 AG030344-01. Dr. Schillinger was supported by an NIH Clinical and Translational Science Award UL1 RR024131.

Author Contributions: Dr. Sudore had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sudore, Landefeld, Schillinger. Acquisition of data: Sudore, Schickedanz, Williams. Analysis and interpretation of data: Sudore, Schickedanz, Landefeld, Williams, Lindquist, Pantilat, Schillinger. Drafting the manuscript: Sudore. Critical revision of the manuscript for important intellectual content: Sudore, Schickedanz, Landefeld, Williams, Lindquist, Pantilat, Schillinger. Statistical Analysis: Sudore, Schickedanz, Lindquist. Obtained funding: Sudore. Administrative, technical, or material support: Sudore, Schickedanz, Williams. Study supervision: Sudore, Landefeld, Schillinger. Final approval of the manuscript: Sudore, Schickedanz, Landefeld, Williams, Lindquist, Pantilat, Schillinger.

Sponsor's Role: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript.