Structuring healthcare advance directives: Evidence from Chinese end‐of‐life cancer patients' treatment preferences

Abstract Background Patients' treatment decisions may be influenced by the ways in which treatment options are presented. There is little evidence on how patients with advanced cancer choose preferences for advance directives (ADs) in China. Informed by behavioural economics, we assess whether end‐of‐life (EOL) cancer patients held deep‐seated preferences for their health care and whether default options and order effects influenced their decision‐making. Methods We collected data on 179 advanced cancer patients who were randomly assigned to complete one of the four types of ADs: comfort‐oriented care (CC) AD (comfort default AD); a life extension (LE)‐oriented care option (LE default AD); CC (standard CC AD) and LE‐oriented (standard LE AD). Analysis of variance test was used. Results In terms of the general goal of care, 32.6% of patients in the comfort default AD group retained the comfort‐oriented choice, twice as many as in the standard CC group without default options. Order effect was significant in only two individual‐specific palliative care choices. Most patients (65.9%) appointed their children to make EOL care decisions, but patients choosing the CC goal were twice as likely to ask their family members to adhere to their choices than patients who chose the LE goal. Conclusion Patients with advanced cancer did not hold deep‐seated preferences for EOL care. Default options shaped decisions between CC and LE‐oriented care. Order effect only shaped decisions in some specific treatment targets. The structure of ADs matters and influence different treatment outcomes, including the role of palliative care. Patient or Public Contribution Between August and November 2018, from 640 cancer hospital medical records fitting the selection criteria at a 3A level hospital in Shandong Province, we randomly selected 188 terminal EOL advanced cancer patients using a random generator programme to ensure all eligible patients had an equal chance of selection. Each respondent completes one of the four AD surveys. While respondents might require support in making their healthcare choices, they were informed about the purpose of our research study, and that their survey choices would not affect their actual treatment plan. Patients who did not agree to participate were not surveyed.

While respondents might require support in making their healthcare choices, they were informed about the purpose of our research study, and that their survey choices would not affect their actual treatment plan. Patients who did not agree to participate were not surveyed.  1 Routine life-extending hospital treatments frequently involve severe pain, which is especially controversial when advanced cancer patients in interviews prefer comfort and dignity over LE interventions. 2 Previous studies found that treatment decisions are not made solely by the advanced cancer patient, but made, or influenced, by family caregivers, who had a greater willingness than patients themselves to choose life-extending treatments. 3 The view of life and death in Chinese traditional culture advocates the view of letting nature take its course, hoping that patients themselves can be understood and respected, and maintain their dignity of life. 4 Advance directives (ADs) enable patients to express their healthcare preferences, including no treatment or active interventions, before starting their end-of-life (EOL) care when they may not be able to communicate their treatment wishes. [5][6][7][8] ADs are tools or instruments that aim to ensure patients' wishes and preferences are respected, but ADs themselves do not necessarily 'respect the care wishes' of patients because patients' care wishes may not be clear or are overridden by family caregivers or doctors. In some countries, such as America, Canada, England, Spain, German and France, countries' national policies and practices encourage the completion of ADs. [9][10][11][12] Degenholtz et al. and Silveira et al. found that American elderly patients who completed ADs were less likely to die in nursing homes or hospitals and more often receive care consistent with their preferences. 13,14 Halpern et al. found that ADs provide seriously ill US patients with an opportunity to counter the tendency to administer treatments to extend life. 15 Given the importance of treatment choices embedded in ADs, it is important to understand how the structure of ADs might affect patient preferences. Previous studies found that using default options, or an already checked treatment choice, shaped patients' choices, and it follows that patients might change their choices with different, or no, default options. 15 Default options studies further suggest that people may not have well-articulated and firm views about the types of care that best promote their values at the EOL. 16 By being easily influenced by default options, seriously ill patients may express no clear preference for, say, hospice care or invasive treatment. Behavioural economics suggests that the preference for hospice care or invasive treatment might be 'constructed' when people are asked to select from a predetermined list of options, rather than reflecting a patient's deep-rooted preferences. 17 Although the effect of the default option on patients has been verified in several studies, [5][6][7][8][13][14][15] its effect on patients with advanced cancer remains unstudied. Unlike other types of seriously ill patients, patients with advanced cancer are in an irreversible and painful healthcare stage and EOL cancer care can only delay death for a short period of time.
For patients receiving active treatment, life is usually only extended for 1 or 2 years, with little hope of short-term improvement and no hope of long-term recovery. 18 We conducted a survey to determine whether default options influenced the ADs choices of a large sample of Chinese advanced cancer patients. 15 Second, we set up two standard AD groups to verify the position-dependent order effects, where the order of options may affect the choice of options given that the first option presented usually dominates. 19 20 but the pain produced in the process is far less than the first eight treatments, so it is 'comfort'. The difference in price between the two palliative treatments is to test the willingness of cancer patients to pay for palliative treatment. Copies of the ADs survey can be found in the online Supporting Information: 1. We also collected information on respondents' background characteristics, comprising sex, age, cancer type and cancer stage and information on the respondents' family members when the patient was unable to make their own healthcare decisions without support. While respondents might require support in making their healthcare choices, they were informed about the purpose of our research study, and that their survey choices would not affect their actual treatment plan. Patients who did not agree to participate were not surveyed.

| Randomization and interventions
Using the randomly generated list of participants, we allocated the four ADs in repetitive order to respondents. Each respondent complete one of the four ADs surveys: a comfort-oriented care William Knaus' prognoses and preferences for outcomes and risks of treatments. 21 We developed the AD forms for this research. When the eligible patients agreed to participate in the survey, one of four versions of ADs was provided to patients randomly.

| Outcomes
Given the propensity of the healthcare system to try to extend life in the absence of an alternative directive, one key outcome was the proportion of cancer patients across the four AD groups who selected a comfort-oriented healthcare goal versus the LE goal.
Secondary outcomes included the choices cancer patients made in 11 specific treatment options.

| Statistical analysis
Our analysis assessed the efficacy of default options in ADs among cancer patients who completed the questionnaire. Data analysis was conducted using Stata version 15. Statistical significance was set at p < .05. The noninferiority tests were analysis of variance (ANOVA).

| Patient and public involvement
No patients were actively involved in setting the research question, or outcome measures nor involved in the design of the study.
Patients were not involved in the interpretation or write-up of the results, nor are there plans for the results to be disseminated to the patient community affected by this research.

| Study sample
The characteristics of the 179 patients are shown in Table 1. Patients were on average 61 years old, mostly males (64%) and married (90%).
According to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, among the 179 patients, 53.6% of patients were diagnosed with urological cancer (renal, bladder and prostatic), 19.7% with digestive system cancer (gastric, colorectal and liver cancer) and 16.4% with lung cancer. As shown in Table 2 (and for specific significance online Supporting Information 2: eTable 1), there were no significant differences between the four AD groups.

| General healthcare goals
As shown in Figures 1 and 2

| Patients' expectations about their family members
EOL cancer patients were also asked about their choice of family member to determine their healthcare when they could no longer make care decisions themselves. As shown in Table 2   In previous research, order effects influenced decisions in domains as diverse as consumer preferences, 28,29 quality-of-life, 30 and personality impression. 31 However, order effects have been proven to be highly unstable in previous studies and to have little effect partly due to survey format, 32 reference effect, 30 familiarity and sensitivity. 29 The influence of the first-order effect was much weaker than the default option in our study. In the two standard AD groups, the order effect only showed significant differences in two individual-specific palliative care options. We infer that for advanced cancer patients whose goal is to prolong life, the default option can affect their preferences, but the order effect is not obvious.
Our study also found that most patients chose to appoint their

CONFLICT OF INTEREST STATEMENT
The authors declare no conflict of interest

DATA AVAILABILITY STATEMENT
The datasets used during the current study are available from the corresponding author.