Stages of Change for the Component Behaviors of Advance Care Planning

OBJECTIVES: To develop stages‐of‐change measures for advance care planning (ACP), conceptualized as a group of interrelated but separate behaviors, and to use these measures to characterize older persons' engagement in and factors associated with readiness to participate in ACP.

T he concept of the steps needed to accomplish advance care planning (ACP) is evolving. Traditionally, the goal of ACP has been to promote patient decision-making autonomy, by having individuals prespecify the care they wish to receive if they become unable to participate in treatment decision-making and the person who will make decisions regarding their care. The means of achieving this goal has been the completion of advance directive (AD) documents, including living wills and durable powers of attorney for health care. The original intent was that these documents were to be completed only after individuals deliberated about and clarified their goals of care and communicated with their loved ones and clinicians, but these additional steps frequently do not accompany the completion of ADs. [1][2][3][4] Moreover, it has been argued that these steps not only represent the means to promote patient autonomy, but also serve as important goals for ACP in and of themselves. 5 As a result, there is growing consensus that the components of ACP should include as distinct processes the facilitation of communication between patients, their potential surrogate decision-makers, and physicians and a consideration of patients' goals of care. [5][6][7] The concept of how individuals engage in ACP is also evolving. Traditional interventions have focused on the provision of standardized information and forms to patients and their surrogates, 8,9 but it is increasingly recognized that ACP may best be understood as a health behavior, for which individuals have highly varied motivation, barriers and facilitators, and self-efficacy regarding their participation. 10,11 It has been proposed that the most effective interventions to increase participation in ACP would provide tailored information according to participants' readiness for engagement. 8 The Transtheoretical Model (TTM) provides a useful framework for considering ACP as a process of behavior change. It characterizes behavior change over time through five distinct Stages of Change. These stages include precontemplation (no intention to change behavior in the near future), contemplation (thinking about changing behavior in the near future), preparation (commitment to changing behavior soon), action (a recent change in behavior), and maintenance (ongoing behavior change). As the organizing construct for the TTM, stages of change, when combined with additional behavior change constructs, including decisional balance, efficacy, and processes of change, serve as the basis for interventions adjusted to the needs of an individual. 12 Such individualized interventions have been found effective for a range of health behavior changes. 12,13 The application of a stages-of-change model to ACP has been previously described 14 using the concept of stages of change in an intervention that was demonstrated to increase rates of completion of advance directives. 15 There are three existing measures for evaluating stages of change in ACP. Two consider only the completion of ADs, 16,17 and the second asks about multiple components of ACP (communication with surrogates and physicians and completion of ADs) in a single item. 18 Several studies have recently proposed the application of a stages-of-change model to ACP consisting of multiple discrete steps, 11,19 but no measures currently exist to measure engagement in ACP in this way.
The purpose of this study was to bring together the concepts of ACP as a group of interrelated but separate healthcare behaviors. Its specific aims were to develop measures for the stages of change for these behaviors, to use these measures to characterize older persons' engagement in ACP, to examine the relationship between the stages across different behaviors, and to examine sociodemographic and health characteristics associated with stages of change.

Measurement Development
Measurement development began with identification of the set of behaviors that best characterizes current conceptions of ACP. This identification was accomplished through a literature review of expert opinion commentaries on ACP supplemented by the conduct of focus groups to understand older persons' experiences with these behaviors. 11 The literature review revealed a broad range of proposed objectives for ACP 3,[5][6][7]20 and the processes that best accomplish these objectives. 2,3,5,6,[20][21][22][23] Because several articles suggested a large number of potential objectives and processes that would yield an impractical number of behaviors, the literature review sought to identify a smaller set of shared objectives and processes. Even with this smaller set, several complex behaviors were identified that could be conceptualized as a series of separate behaviors rather than a single behavior composed of a number of discrete steps. In this case, when a decision needed to be made regarding ''lumping'' versus ''splitting,'' the guiding principle was whether the individual step was regarded in the literature as being sufficient to accomplish a necessary part of ACP.

Participants
Participants were 65 years of age and older recruited from two primary care practices and one senior center. These sites were purposefully selected to provide access to a group of older persons with diversity in race, socioeconomic status, and health status. In the primary care practices, letters were sent to sequential older persons whose physicians indicated that they did not have dementia. Persons who agreed (92% and 88% in the two practices) underwent a telephone screen to determine exclusion criteria, which included non-English speaker (7% and 2%), hearing loss precluding participation in interview (7% and 1%), nursing home resident (0% and 1%), acute episode of illness (8% and 4%), and cognitive impairment, defined as less than two-thirds recall on a test of short-term memory (7% and o1%). Of eligible participants, 83% and 80% completed interviews. In the senior center, volunteers were solicited for participation. All of the persons who volunteered were eligible for participation and completed interviews.

Data Collection
Trained research assistants interviewed participants in person. In addition to the items measuring ACP behaviors, the interview included measures of sociodemographic status (age, ethnicity, race, education, sufficiency of monthly income, 24 marital status, and household composition) and health status (self-rated health; 25 quality of life, assessed using a single-item global measure: ''How would you rate your overall quality of life?'' chronic conditions; instrumental activities of daily living; 26 depression 27 ). In the absence of available instruments in the literature, the survey also included items developed for the purposes of this study to assess participants' life experiences with life-threatening illness and decision-making (Have you ever had to make a medical decision for someone who was dying?), participation in activities to prepare for the future other than ACP, and knowledge regarding advance directives. The knowledge scale consisted of 15 true-or-false questions addressing basic facts regarding the purpose of and mechanisms of completing healthcare proxies and living wills, including such items as ''If a person's doctor believes a person should undergo a treatment, then that person needs to undergo it, even if he does not want to'' and ''The best place to keep a living will is in a safety deposit box.'' The full set of items is included in Appendix A.

Analysis
Univariate statistics were used to describe the study population and characterize the stages of change for the ACP behaviors. Characteristics associated with stages of change for two of the individual behaviors representing different aspects of ACP were examined in bivariate analysis using the chi-square test for categorical variables and analysis of variance for continuous variables.

Development of Items to Assess Stage of Change for ACP Behaviors
Traditionally, the objective of ACP has been to allow patients to specify the health care that they would want to receive in a time of decisional incapacity. A review of the literature revealed that the objectives of ACP have been greatly expanded. A number of objectives appeared repeatedly in commentaries reviewing the purpose of ACP: improving communication between patients, surrogates, and clinicians, 4,5,7,20-23 helping patients clarify their broad goals of care, [4][5][6][7]20,21,23 preparing patients and surrogates for the difficulties they may face at the end of life, 3,6 and decreasing family member and surrogate burden. 5,7,20 There was disagreement regarding the role of ADs. Although some authors argued that, because of their limitations, they should be abandoned, other authors argued that they had a continuing role.
It was decided to focus on objectives for ACP for which there were corresponding clearly identifiable and broadly applicable behaviors. The objectives of preparing patients and surrogates and decreasing burden appeared to depend upon understanding the needs of individual patients and family members and adjusting behaviors to these needs, so these objectives were not included. 1 The objectives of clarifying patients' goals of care, improving patient-surrogate communication, and fostering the completion of ADs were focused on. It was decided to include the latter despite their controversy because the focus groups revealed that this was the aspect of ACP most familiar to older persons. Clarification of goals of care and patient-surrogate communication were combined into a single objective, because patients' views regarding goals of care are a critical aspect of their communication with their surrogates. The challenge to the assessment of patients' clarification of goals is that this is a complex task, ideally encompassing patients' values as they relate to the acceptability of diminishing states of health, the trade-offs between the benefits and burdens of interventions, and the likelihood of different health outcomes, but the focus groups suggested that many older persons may not be familiar with one or more of these concepts. Because the goal of this study was to assess whether patients had engaged in values clarification, rather than actually engaging patients in this process, the notion of clarification of goals was simplified to include the most fundamental concepts most likely to be familiar to a diverse group of older persons. These included whether they had considered their attitudes toward the use of life-sustaining treatment and toward the trade-offs between quality and quantity of life. The behaviors corresponding to these ACP objectives were therefore completion of a living will, completion of a healthcare proxy, communication between patient and loved ones regarding patients' views about the use of life-sustaining treatment, communication between patient and loved ones regarding patients' views about quality versus quantity of life, communication between patient and clinicians regarding patients' views about the use of lifesustaining treatment, and communication between patient and clinicians regarding patients' views about quality versus quantity of life.
For each of these behaviors, patients were asked to choose from a fixed set of responses to indicate whether they had participated in the activity longer than 6 months before (maintenance), had participated in the activity within the past 6 months (action), were planning to complete the activity within the next 30 days (preparation), were thinking about completing the activity in the next 6 months (contemplation), or had not thought about or were not ready to participate in the activity (precontemplation). For the behaviors of communicating with loved ones and with clinicians, which represented behaviors consisting of several steps, participants could be in the precontemplation stage for a number of different reasons, including lack of awareness of life-sustaining treatments or of potential trade-offs between quantity and quality of life, never having thought about these issues, lack of readiness to talk about these issues, or having thought about these issues but lack of readiness to discuss them. Participants who indicated that they were not ready to participate in the behavior were asked about each of these reasons. The algorithm for asking about the behaviors of living will completion and communicating with loved ones about quality versus quantity of life are provided in Figure 1. The algorithms for the other behaviors followed similar logic.

Description of Participants and Their ACP Behaviors
A description of the participants is provided in Table 1. Completion of preparations for the future other than ACP ranged from 8% of participants having prepaid for their funeral to 84% having told someone where their important papers were. The distribution of stages of change varied widely across behaviors. Whereas approximately 50% to 60% of participants were in the action or maintenance stages for communicating with loved ones about lifesustaining treatment and completing a living will, 40% were in the precontemplation stage for communicating with loved ones about quantity versus quality of life. Only 34% were in the action or maintenance stages for completing a healthcare proxy, and only 5% to 10% were in these stages for communicating with physicians ( Table 2).
Individual participants' stages of change for each of two behaviors were compared, which revealed that participants were frequently in different stages for different behaviors. For example, even when action and maintenance stages were combined, 45% of participants were in a different stage for completing a living will and communicating with loved ones about quality versus quantity of life, and 51% of participants were in a different stage for completing a healthcare proxy and communicating with loved ones about life-sustaining treatment. There was no consistent ordering when participants were in different stages for a given behavior. For example, whereas 25% of participants were in a later stage for communication with loved ones about life-sustaining treatment than for living will completion, 16% were in an earlier stage. The only exception was for the behaviors of completing a living will and a healthcare proxy; of the participants completing a healthcare proxy, all but one had also completed a living will.
When factors associated with stages of change for completing a living will were examined, being nonwhite and never having made a medical decision for someone who was dying were associated with being in earlier stages of readiness, although the latter failed to reach statistical significance (Table 3). Those who were in later stages had greater knowledge regarding living wills and advance directives than participants in earlier stages, and they had participated in a greater number of other preparations for the future. Age, sex, education, and health status were not associated with stage of readiness for living will completion in this community-dwelling population. In contrast, when factors associated with stages of change for communication with loved ones regarding quality versus quantity of life, all of these factors were associated with stage of readiness, as was having had to make a medical decision for someone who was dying (Table 4).

DISCUSSION
Conceptualizing ACP as a set of interrelated health behaviors provides a foundation for model-based health behavior change intervention development. Current thinking about advance care planning reveals it to consist of behaviors including clarification of values; communication between patients, surrogates, and clinicians; and completion of written directives. In this cohort of older community-dwelling persons recruited from physician offices and senior centers, there was a full spectrum of readiness to participate, or stages of change, for each of these behaviors. Stages of change varied widely across the behaviors, and individual participants were frequently in different stages of readiness for the different behaviors, without evidence of ordering or sequential completion. Only a small proportion of participants had communicated with their physicians regarding their values and preferences. Relatively few sociodemographic, Have you made out a living will?
How ready are you to complete a living will?
When did you do this?  health, or psychosocial factors were associated with stages of change for completing a living will, but a broader range of factors was associated with stages of change for communication with loved ones about values.
In the current study, the proportion of older persons who were in the action or maintenance stage for living will completion and healthcare proxy designation was substantially higher than the proportion reported in the control groups of multiple randomized controlled trials to have completed a living will and a durable power of attorney for health care. 9 Of a representative national sample of family members of people who had died, 71% had completed an advance directive. 28 Another recent study using a national sample found, in a national sample of adults aged 60 and older, that 68% of decedents who required decision-making at the end of life but lacked decision-making capacity had an advance directive. 29 Although it is possible that these directives were completed close to death, which would then account for the high prevalence of directives, several studies, including the present one, have failed to find an association between severity of illness and advance directive completion. It is therefore unclear what accounts for the disparate range of completion rates between these studies.
The additional behaviors related to communication examined in this study have only recently become a separately measured part of the advance care planning process, and therefore, little is known about the prevalence of participation in them. Given the sizable proportion of participants in the later stages of behavior change for the majority of behaviors, the large number of participants in the precontemplation stage for communication with their physicians was striking. These findings are consistent with several other studies demonstrating low rates of communication between patients with individual diseases and physicians regarding end-of-life care and multiple barriers to this communication. 30,31 These low rates highlight the lost opportunity in the portion of the (now defunct) House of Representative's healthcare reform bill that would have provided incentives for promoting patient-physician communication, 32 although the fact that participants were in the earliest stage of readiness for this communication suggests that physician incentives may be insufficient to address these low rates. Intervention efforts to increase rates of communication will also need to focus on the most fundamental aspects of this behavior from the patient perspective, including addressing older persons' awareness of the need for and increasing their awareness of the positive aspects of this communication.
One previous study found that ACP consisted of a series of sequential steps in which individuals were most likely to have thought about their wishes regarding lifesustaining treatment, with a sizable subset moving on to discuss their wishes with family members and smaller subsets discussing their wishes with physicians and documenting their wishes in an AD. 19 The results of the current study, in contrast, suggest that communication with loved ones about values and advance directive completion are not sequential activities but are behaviors for which individuals may be in any combination of readiness, or stage of change, to participate. Studies demonstrating that there are different sets of barriers to and facilitators of participation in the  different behaviors support such a model of ACP. 10,11 This model of ACP suggests that interventions designed to promote ACP need to assess and target stage of change for each of the behaviors individually. It also supports the feasibility of stage-tailored interventions. To the extent that facilitators of behavior change are also shared across behaviors, it may be possible to build upon the motivations an individual has for completing one aspect of ACP to support their participation in those remaining.
The associations between patient characteristics and stages of change for the different ACP behaviors provide external validation for the measures and provide new insights into ACP as a health behavior. In the current study, nonwhite ethnicity was strongly associated with stages of change for living will completion, similar to the findings in a number of previous studies examining the dichotomous outcome of having versus not having a living will. [33][34][35] The current study extends these findings by demonstrating that  nonwhite participants were not only more likely not to have completed a living will, but also more likely to be in the earliest stages of readiness to participate in the process. The current study failed to find an association between health status, measured in terms of self-rated health and number of comorbid conditions, and stage of change for AD completion. This lack of association between health and ADs has also been previously demonstrated. 36,37 Much less is known about factors associated with other aspects of ACP. The results of the current study demonstrate that communication with loved ones about values regarding quantity versus quality of life is associated with a broader range of an individual's health and psychosocial characteristics than living will completion. It is also associated with their experiences with their own and others' serious illness. The association between stage of change and knowledge regarding ADs supports the notion that participants use certain processes of change, including gathering information, to move through the stages of readiness. 11 Living will completion and communication were associated with participation in other, non-health-related, activities related to end-of-life preparation, which suggests that the predisposition to participate in ACP is, at least in part, related to the importance placed on and willingness to plan for the end of life. It also suggests the potential utility of approaches to promote ACP based on bringing together ACP with other advance planning activities outside of the medical realm. For example, older persons cite their lawyers as a source of information regarding and prompt for completing advance directives, 38 suggesting that certain components of ACP might be promoted as part of estate planning. Although not routinely done, the results raise the possibility of incorporating ACP as a part of funeral planning or purchasing of long-term care insurance, which represent times at which individuals may be more receptive to the difficult task of thinking about the health care they wish to receive or forgo at the end of life.
The study has a number of limitations. First, because of the many models that exist for ACP, there is likely to be disagreement regarding the precise set of behaviors that define the process of ACP. Until there are more data examining the relationship between these behaviors and patient outcomes, determining which behaviors should be included as part of ACP will be the subject of debate. The behaviors used in this study were not meant to be definitive; rather, they were chosen as a set that appeared to be representative of the existing literature. Second, participant engagement in these behaviors was assessed in a simplified fashion. To ensure that participants would understand what consideration of goals meant, they were about only the most fundamental aspects of this process, potentially overestimating their engagement in these behaviors. Third, although participants were included regardless of their health status to examine participation in ACP in a broad cross-section of older persons, it could be argued that it is most relevant to examine ACP engagement in persons who are at greater risk of decisional incapacity. Finally, the study was restricted to ACP as it pertained to treatment decisionmaking for physical, and not mental, health.
ACP can be conceptualized as a group of related healthcare behaviors. This study demonstrates the variability in readiness to participate, or stages of change, for these different behaviors in a cohort of older, community-living persons. It also illustrates that these persons are frequently in different stages for the different behaviors. Individualized assessment and interventions targeted to the stage of behavior change for each component of ACP may be an effective strategy to increase older persons' participation in ACP.