|Home | About | Journals | Submit | Contact Us | Français|
Rationale: Despite ongoing ethical debate concerning who should control decisions to discontinue life support for incapacitated, critically ill patients, the perspectives of surrogate decision makers are poorly understood.
Objectives: To determine (1) what degree of decisional authority surrogates prefer for value-sensitive life support decisions compared with more technical biomedical decisions, and (2) what predicts surrogates' preferences for more control over life support decisions.
Methods: This was a prospective study of 230 surrogate decision makers for incapacitated, mechanically ventilated patients at high risk of death. Surrogates reported their preferred degree of decisional authority using the Degner Control Preferences Scale for two types of decisions: a value-sensitive decision about whether to discontinue life support and a decision regarding which antibiotic to prescribe for an infection.
Measurements and Main Results: The majority of surrogates (55%, 127/230; 95% confidence interval, 49–62%) preferred to have final control over the value-sensitive life support decision; 40% (91/230) wished to share control equally with the physician; 5% (12/230) of surrogates wanted the physician to make the decision. Surrogates preferred significantly more control over the value-sensitive life support decision compared with the technical decision about choice of antibiotics (P < 0.0001). Factors independently associated with surrogates' preference for more control over the life support decision were: less trust in the intensive care unit physician, male sex, and non-Catholic religious affiliation.
Conclusions: Surrogates vary in their desire for decisional authority for value-sensitive life support decisions, but prefer substantially more authority for this type of decision compared with technical, medical judgments. Low trust in physicians is associated with surrogates preferring more control of life support decisions.
There are few empirical data on the perspective of surrogates about who should have authority for value-sensitive decisions regarding the use of life support. Prior studies report surrogates' preferred role in ICU decision-making in general, encompassing both end-of-life decisions and biomedical decisions.
Our objectives were to determine what degree of decisional authority surrogates prefer for value-sensitive life support decisions compared to more technical biomedical decisions, and what predicts surrogates' preferences for more control over life support decisions. These results suggest that surrogates may prefer more decisional control for value-laden decisions in ICUs than previously believed. Surrogates who had less trust in the treating physicians preferred more control over the final decision regarding withdrawal of life support.
Surrogate decision making in intensive care units (ICUs) is fraught with difficulty. Surrogates experience conflict with physicians (1, 2), difficulty identifying patients' treatment preferences (3), long-term psychiatric symptoms (4, 5), and lingering doubts about the decisions made (6). As clinicians seek to improve the quality of surrogate decision making, a key question arises: to whom should responsibility fall for decisions about whether to forego life-sustaining treatment? The 2003 Challenges in End-of-Life Care in the ICU consensus statement, generated from expert opinion, advocates a shared decision making model (7), in which surrogate decision makers and physicians share the responsibility of making decisions on behalf of incapacitated ICU patients.
There are few empirical data on the perspective of surrogates about who should have authority for value-sensitive decisions regarding the use of life support. The main work cited in professional society statements is that of Heyland and colleagues, who elicited surrogates' preferred role in ICU decision making in general, encompassing both end-of-life decisions and biomedical decisions (8). Anderson and colleagues also used general ICU medical decision making as a basis to obtain surrogates' decision-making preferences (9). However, there are conceptual differences between biomedical decisions, which largely depend on technical medical knowledge (e.g., which antibiotic is most effective in treating hospital-acquired pneumonia) and value-sensitive decisions, which hinge on both biomedical facts and on individuals' values and preferences about trade-offs between quantity and quality of life. It is therefore difficult to know whether the data cited in the 2003 consensus statement apply to the value-sensitive life support decisions that arise commonly in ICUs.
We therefore sought to determine (1) the degree of decisional authority preferred by surrogates for a paradigmatic value-sensitive life support decision, (2) whether decision-making preferences are sensitive to the type of decision (highly value-laden vs. predominantly biomedical decision), and (3) what factors predict surrogates' preference for higher degrees of authority for life support decisions.
From January 2006 to October 2007 we performed a prospective cohort study of surrogate decision makers for critically ill patients in four ICUs at the University of California San Francisco Medical Center. Study methods have been previously described (10).
Study investigators identified eligible patients and their surrogates by screening daily in each ICU. The four ICUs include mixed patient populations, with one neurologic ICU, one cardiac ICU, and two medical-surgical ICUs. Eligible subjects were the surrogate decision makers of incapacitated adult patients with respiratory failure requiring mechanical ventilation and an Acute Physiology and Chronic Health Evaluation (APACHE) II score greater than 25. The surrogates of patients dying within 48 hours of initiating mechanical ventilation were not eligible per Institutional Review Board requirements. If multiple individuals identified themselves as the surrogate for a patient, then we enrolled those who rated their role in decision making to be significant. Therefore, each patient could have more than one surrogate enrolled. To be considered, surrogates needed to be at least 18 years old and speak and read English well enough to not require an interpreter. Subjects provided written consent for all study procedures.
Subjects completed a written, self-administered questionnaire containing items about demographic information, past experiences as a surrogate, and whether the surrogate had discussed treatment preferences with the patient. We also assessed surrogates' perceptions of their communication with, trust in, and conflict with the patient's treating physician using the quality of communication scale (11), the validated physician trust scale (12), and a single-item conflict scale (13). Subjects also answered questions regarding religiosity about the perceived influence of God on their health, using the God Locus of Health Control Scale, with higher scores indicating a stronger belief in God's role in health (14, 15). Written depression screening was accomplished with the Patient Health Questionnaire-2 (16), along with measurement of dispositional optimism by means of the Revised Life Orientation Test (17). Higher Revised Life Orientation Test scores indicate more dispositional optimism. Subjects also completed validated measures of health literacy (Short Test of Functional Health Literacy in Adults) (18) and view of patient and doctor roles (Patient Provider Orientation Scale). The Patient Provider Orientation Scale is a mean score ranging from 1 to 6, with higher scores reflecting a preference toward a more “patient-centered” approach to healthcare compared with a more “biomedical model” (19). A measure of the patient's prehospitalization function was obtained using the Katz Activities of Daily Living score (range, 0–6). Higher scores indicate higher degrees of functional independence.
Subjects were presented with two clinical vignettes regarding treatment choices to be made for their loved one in a hypothetical clinical scenario: a life support decision and an antibiotic decision. The vignettes and response elements are provided in Figure 1.
Surrogates completed the validated Degner Control Preferences Scale to indicate their preferred level of involvement in the decision-making process for each scenario (20). The Control Preferences Scale is a tool used to assess what role individuals prefer for medical decision making and is rooted in the theory that there is a continuum of desired control among decision makers (21). The Control Preferences Scale is composed of five potential roles in clinical decision making. For both clinical vignettes, these five roles were available as choices on a 1 through 5 scale, with increasing physician control in the decision-making process assigned higher values on the scale (20).
We generated descriptive statistics for patients' and surrogates' demographics, covariates, and the main outcome measure (Degner Control Preferences Score) for the two clinical scenarios. The preferred level of decisional authority was initially measured as five separate categories, and for the multivariate analyses was dichotomized for ease of interpretation. Surrogates were adjudicated to prefer final authority over the decision if they chose either of the following two responses: “1. I prefer that I make the final decision about [wording varied based on vignette]” or “2. I prefer that I make the final decision after seriously considering the doctor's opinion” (see Figure 1). The other three responses on the scale were categorized as surrogate preference to have the physician involved in the decision making, which included the patient sharing the responsibility for the decision and the physician having ultimate authority for the decision. This analytic approach allows a focus on the clinically important question of what factors may lead surrogates to prefer a high degree of decisional authority.
We used the Wilcoxon signed-rank test to perform within-subject comparisons to determine if there were differences in subjects' preferred level of involvement in the life support decision compared with the antibiotic decision. We also applied a paired t test to calculate the P value for the difference in surrogates' preferred level of involvement between the two decisions.
To determine factors associated with surrogates' preferences to have final decision-making authority for life support decisions, we used a hierarchical logistic regression model, which accommodated the possibility of a clustering effect among surrogates nested within patients nested within physicians. We first fit a series of models with a single covariate to identify variables of interest. Variables with a P < 0.20 at this stage of modeling were considered for inclusion in models with multiple covariates. Final models were selected based on significance of covariates. In situations in which two covariates were highly correlated, only one of the two was retained in the model. All analyses were conducted using STATA 8.0 (College Station, TX).
Of 222 eligible patients, 20 were excluded at the attending physician's request and 27 of the patients' surrogates declined to participate. There were no significant differences in age, sex, race, or APACHE II score between enrolled patients and those who were eligible but not enrolled at the request of the physician or surrogate. For the remaining 175 patients, 230 surrogate decision makers were enrolled, with 40 patients having more than one self-identified surrogate (Figure 2).
Subjects' demographic characteristics and a summary of other covariates are shown in Table 1. More than half of the surrogates had prior experience as a surrogate decision maker (52%), and 60% had prior discussions with the patient about treatment preferences. Two-thirds (68%) of surrogates met criteria for possible depression on the Patient Health Questionnaire-2. The mean physician trust score was 20.9 (SD, 3.6; range, 9–25), with higher scores indicating higher trust in physician. The average quality of communication score was 83 (SD, 16.1; range, 0–100), with higher scores reflecting surrogates' perceptions of better communication by the ICU physician. Patient demographics are summarized in Table 2.
Figure 3 summarizes surrogates' preferred role in the value-laden life support decision. The majority of surrogates (55%, 127/230; 95% confidence interval, 49–62%) preferred to have final control over the value-sensitive life support decision, 40% (91/230) wished to share control equally with the physician, and 5% (12/230) wanted the physician to have final control. The Control Preferences Scale also provides information about whether surrogates wish to hear the physician's opinion about treatments. Ninety percent of subjects wished to receive the physician's recommendation (defined as selecting response choices 2–5) (Figure 3).
Figure 3 also summarizes surrogates' preferred role in the technical judgment about antibiotic selection. Surrogates preferred significantly more control for the value-sensitive life support decision compared with the technical medical decision about antibiotics; the mean score on the Control Preferences Scale for the value-laden life support decision was 2.4 ± 0.8 versus 3.9 ± 1.1 for the antibiotic decision (P < 0.0001), with a lower score indicating a preference for more decisional control.
Univariate analysis of factors associated with the preference for more decision control in value-laden decisions revealed six covariates to be significant: younger age, male sex, non-Catholic religion, a higher Patient Provider Orientation Scale score, a lower physician trust score, and a lower score for quality of physician communication. Table 3 summarizes the univariate odds ratios for predictor variables associated with preferences for maintaining a high decisional responsibility, defined as the surrogate preferring to decide alone or to decide after strongly considering the physician's opinion. Multivariate analysis revealed three factors significantly associated with surrogates' preference for a higher degree of decisional control for the value-sensitive decision: low levels of trust in their loved one's physician, male sex, and being a member of a non-Catholic religion (Table 4).
We found substantial variability in the role surrogates prefer in making value-sensitive life support decisions for incapacitated, critically ill patients, with a slight majority preferring to have final control of the decision. Surrogates with low levels of trust in the treating physicians were more likely to prefer to retain final authority for value-laden life support decisions.
Our results suggest that surrogates may prefer more decisional control for value-laden decisions in ICUs than previously believed. For example, Heyland and colleagues found that only 22% of subjects wished to maintain final authority over decisions in ICUs (8). Similarly, Anderson and colleagues reported that only 25% of surrogates preferred to be the person making the final medical decisions (9). Azoulay and colleagues found that 53% of surrogates did not wish any involvement in decisions to forego life support in ICUs (22). What may explain the different findings? Both Heyland and colleagues and Anderson and coworkers elicited surrogates' perspectives on decision making in general, combining both end-of-life decisions and technical medical decisions into their questions (8, 9). The data we report herein suggest that the type of decision influences surrogates' preferred role, with surrogates preferring more control over value-laden decisions compared with more technical medical decisions. Azoulay and colleagues studied a qualitatively different type of decision: whether to stop futile treatment when there was “no hope of recovery” (22). We studied decisions about what constitutes a state worse than death, which are qualitatively more value-laden and difficult than the decision to stop a clearly futile treatment. It is also possible that Azoulay and colleagues' findings from France are due to differences in prevailing cultural norms between Europe and the United States regarding medical decision making (23, 24). Our findings are in accord with a recent qualitative study by White and colleagues, which reported wide variability in surrogates' beliefs about what role physicians should play in value-sensitive life support decisions (25).
The results of the current study provide empirical support for the conceptual distinction between physicians sharing their opinion with surrogates and physicians having final authority over value-laden decisions. Specifically, although very few surrogates wished to cede all decisional authority to physicians, 90% wished to receive the physician's opinion about whether to forego life-sustaining treatment. Understanding the conceptual distinction between “who deliberates” and “who decides” may help physicians better match their practice to the preferences of individual surrogates.
We also found that surrogates who had less trust in the treating physicians preferred more control over the final decision regarding withdrawal of life support. Although this finding will not surprise clinicians who have experienced how loss of trust can undermine collaborative decision making, it is the first empirical evidence of this association among surrogate decision makers. These cross-sectional data cannot establish a causal association; however, they raise the possibility that surrogates' role preferences may be dynamic rather than static and constructed by their experiences with the health care team. Future prospective studies are needed to establish whether such a causal relationship exists. If so, it would strengthen the rationale for research on how to forge trusting relationships with families in ICUs. We speculate that a starting point may be for physicians to conceptualize trust building as an important goal in their interactions with surrogates and to structure their communication to accomplish this.
Our data suggest that physicians need to develop two skills that are not currently part of the core competencies of critical care clinicians. First, the variability in surrogates' role preferences suggest that physicians should develop the ability to elicit surrogates' preferred role in decision making. Existing evidence suggests that physician rarely inquire about surrogates' preferred role in decision making (26). Second, physicians should develop comfort with having different levels of authority for decisions based on the surrogates' preferences and the clinical context.
We wish to highlight our opinion that surrogates' preferred level of control over value-laden decisions is one among several considerations that are ethically relevant to the question of what role they should ultimately play in life support decisions. At least four other considerations are ethically relevant, including considerations of distributive justice, physicians' obligations to act for the good of their patients and to respect patients' previously stated treatment preferences, and cultural norms around medical decision making. Occasionally, one or more of these considerations may require physicians to assume more control over value-laden decisions than surrogates prefer. This step should not be taken without justification, however, because recent evidence suggest that surrogates are at higher risk of adverse psychiatric outcomes from the ICU experience when there is discordance between their preferred and actual role in decision making (4). We propose that physicians' default approach to value-laden decision-making should be to tailor their role to the preferences of the surrogate, and to depart from this only if compelled to by a stronger ethical obligation, such as those outlined above.
We also found that male subjects and non-Catholic subjects preferred significantly more control for value-laden decisions compared with female subjects and Catholic subjects, respectively. No previous studies of surrogate decision makers have examined these associations. However, several studies of patients (rather than surrogates) suggest that male sex is associated with a preference for less control in medical decision making (27–30). This discrepancy raises the possibility that the association between sex and decisional authority may be modified by whether one is acting as a surrogate or one is making decisions for oneself. The precise mechanism to explain this is unclear. Both associations should be interpreted with caution pending further research to replicate the finding and understand the explanatory mechanisms.
A somewhat surprising finding from this study is that a small minority of surrogates (12%) wished to retain final authority for the decision concerning antibiotic selection. However, these findings are qualitatively similar to those of other studies examining adult patients' preferred level of control over biomedical decisions. For example, in a large, population-based survey study in Canada, Levinson and colleagues found that a substantial minority of subjects disagreed with the statement that general medical decisions should be left up to doctors (31). Two studies using a similar metric to that used in the present study also found that a small minority of subjects desired to retain final control over largely technical medical judgments, including the decision about which antibiotic to use to treat an infection (30, 32). In our experience, very few physicians involve surrogates in seemingly routine clinical decisions in ICUs, and doing so would be logistically complex. Because this is the first study documenting this finding in surrogates, additional research is warranted to confirm it and to understand the reasons that underlie this preference.
This study has several strengths. We studied surrogates of actual patients at high risk of death who were actively engaged in the process of surrogate decision making; we speculate that this increases the likelihood that participants' responses represent considered judgments about their preferred role in decision making. The sample was diverse in terms of ethnicity, level of education, and prior experience as a surrogate. We used a validated outcome measure of preferences for decisional control. We also used two conceptually distinct types of decisions to assess how control preferences vary according to the nature of the decision.
This study has several limitations. We used written clinical scenarios to illustrate the types of decisions under study. Although we believe this is a methodological improvement over prior research that did not specify the type of decision under study, it is possible that surrogates' stated role preferences could differ in actual clinical situations. We found that nearly two-thirds of the subjects screened positive for possible depression, which is a higher prevalence than other studies in ICUs. This may be due to our use of a brief depression screening tool to measure depressive symptoms rather than a longer instrument, which maximized sensitivity at the expense of specificity. The study was conducted in one region of the United States and therefore may not be generalizable to areas in which there are different cultural perspectives on the physician–family relationship. Because we studied only surrogates of patients at high risk of death, it is possible that their attitudes may not represent those of surrogates of patients in less dire clinical circumstances. Nonetheless, it is arguably most important to understand the preferences of surrogates actively facing difficult decisions about life-sustaining treatment.
In conclusion, this report provides new empirical data to inform the debate about how physicians should approach the process of surrogate decision making in ICUs. The vast majority of surrogates wish to be active participants in the decision-making process, although not all wish to have complete authority for the final decision. The variability in surrogates' role preferences highlights the need to assess surrogate decision makers' preferences and to tailor the decision-making process accordingly.
Originally Published in Press as DOI: 10.1164/rccm.201008-1214OC on October 29, 2010
Author Disclosure: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
Supported by a Greenwall Foundation Bioethics Faculty Scholars Award (D.B.W.), a Paul Beeson Career Development Award from the National Institute on Aging (D.B.W.), and National Institutes of Health grant 1K23AG032875 (D.B.W.).
Some of the descriptive data from this study were previously presented in poster format: Bautista CA, Evans LR, Luce JM, Lo B, White DB. Are physicians' recommendations to limit life support beneficial or burdensome? Bringing empirical data to the debate. Presented at the Society of General Internal Medicine (SGIM) National Meeting. May 2009, Miami, FL.