Advance Directives and SUPPORT

The papers included in this issue of the Journal1-3 add to the wealth of scholarship generated by the Study to Understand Prognoses and Preferences for Outcomes on the Critically-Ill Hospitalized Adults (SUPPORT). Since publication of major findings in November 1995,4 this research initiative has greatly expanded our empirical knowledge of end of life care and the inadequacies of current practice for critically ill hospitalized patients. In this large multi-center study, investigators examined practice patterns before and after the implementation of the Patient Self-Determination Act (PSDA) of 1990.5 This federal legislation mandated patient education about advance directives and their documentation upon hospital admission. After PSDA passage, patients were randomized to receive either standard care or the SUPPORT intervention. In the intervention cohort, study nurses facilitated communication among patients, surrogates, and physicians and encouraged the use of advance directives. Neither the implementation of the PSDA nor the introduction of the experimental intervention improved end of life care. The papers collected here focus on the effectiveness of advance directives to influence care and promote savings at the end of life. Investigators conclude that advance directives fail to enhance communication and that their use does not result in a reduction of costs. When attempting to evaluate the findings presented in these three papers, a handy resource is What This Patient Needs is a Doctor, a tribute to Dr. Eugene A. Stead, former chief of Medicine at Duke University. The text contains many colorful quotations that crystallize truths about medical practice and clinical investigation. Commenting on the progress of a patient, Dr. Stead noted: I have nothing to add. Patient says she is feeling better and her chart seems to agree with this.6 With good humor, Dr. Stead cautions the clinician not to confuse clinical reality with our description of it. This admonition compels me to question the broad indictment of advance directives presented here and ask whether the investigators' conclusions reflect the limited value of advance directives or a shortcoming in study design.7, 8 Addressing this question is critical because both the lay press9, 10 and scholarly publications have interpreted SUPPORT's findings as evidence that advance directives do not work. See also pp 500, 508, and 513 Yet in a notable mistaking of research results for clinical reality, SUPPORT investigators dismiss the efficacy of documented advance directives among control patients who were not the object of the study intervention.2 As a group, these patients desired less aggressive end of life care and had a 23% savings compared to other patients. Although these data suggest the influence of advance directives on end of life care, investigators attributed these findings to a selection bias: … the control physicians documented advance directives among a highly biased group of patients, many of whom preferred a palliative approach to care, had DNR orders written early in their hospital course, and took other measures to limit the aggressiveness of their care. Because the randomized variable is the SUPPORT intervention, why discount the efficacy and savings seen in patients with documented advance directives in the control population? Attributing this to a sampling error fails to appreciate that an outcome of meaningful prospective planning is the cultivation of end of life preferences. We should expect patients whose advance directives surface naturally within the course of routine care to have articulated preferences. This does not reflect a bias. It indicates planning. Rather than dismiss the value of documented advance directives in the control population, this subset of patients warrants closer study. We should determine how the alternative of palliative care was presented and what led to the effective integration of their advance directives into care plans. Even more critically, we should delineate if these patients were more likely to have benefited from an on-going doctor-patient-surrogate relationship. This exploration could help to address whether practice patterns differ if a physician, known previously to the patient, remains involved in end of life decision-making.11 Although another SUPPORT study examining Do-Not-Resuscitate orders tells us the physician-specialty mix caring for patients,12 little is presented about continuity of care. We know that 12% of patients received advance directive counseling from their physicians, but we do not know if these physicians remained involved or if they were among the 48% of practitioners who thought the advance directive was helpful in planning end of life care.1 It is important to understand the physician's role because patients expect guidance from their physicians and seldom initiate prospective planning unilaterally.13 This suggests that advance care planning occurs in the context of relationships between the patient, the surrogates, and the physician. Looking at advance directive documentation without acknowledging the relationships that both sustain and inform them will lead us to confuse pre-existing documents for pre-existing relationships. The authors note rightly that “advance directives cannot be expected to function well unless they arise from effective communication between patient and physician.”1 Given this, it is not surprising that the intervention enhanced advance directive documentation but that these “documented” directives failed to influence care. These documents may not have been sustained by therapeutic relationships that have the power to direct care. Important life choices are not made in a formulaic or episodic manner but within the context of families and trusting relationships with clinicians. SUPPORT's findings suggest that preferences regarding end of life decisions require more than documentation to be integrated into care decisions.14 They need an institutional setting that promotes dialogue, fosters continuity of care at the end of life, encourages the discovery of pre-existing preferences, and acknowledges the legitimacy of palliation.15-17 Unfortunately, investigators have constructed a study design — and intervention — that makes good communication difficult. The way the patient is asked about resuscitation exemplifies how options were communicated: As you know, there are a number of things doctors can do to try to revive someone whose heart has stopped beating, which usually includes a machine to help breathing. Thinking of your current condition, what would you want your doctors to do if your heart ever stops beating? Would you want your doctors to try to revive you, or would you want your doctors not to try to revive you?2 The question is too narrowly cast. If the patient is simply asked whether she wants to be resuscitated, without being given an alternative palliative care option, the response is predictable. This reinforces the false belief that there are just two options at the end of life: resuscitation or “doing nothing.” Posing the question in this way will bias responses that will lead to decisions to do “something” over “nothing.” To truly assess patient preferences about resuscitation, patients need to be informed more fully about the likelihood of a successful resuscitation, its morbidity, and other care alternatives. When clinicians present resuscitation choices robustly, patients overwhelmingly decide to forgo resuscitation.18 SUPPORT investigators should be commended for undertaking this randomized clinical trial. They have demonstrated the value of rigorous empirical research in clinical ethics.19 Nonetheless, we should recognize that the object of their study is anthropologically complex.20 When we interpret quantitative findings about care at the end of life, we must examine critically both the cultural assumptions that inform study design and the conclusions. It may be tempting to distill the many deficiencies in end of life care identified in SUPPORT into a critique of advance directives. However, as clinicians, we should remind ourselves that the failure is ours and not the fault of advance directives. Shifting responsibility away from ourselves indicates our own denial of death21, 22 and avoidance of much needed dialogue at the bedside of dying patients. It is too easy to scapegoat advance directives when it is clear that advance directives are going to be only as good as our efforts to obtain them. Advance care planning will become effective only when we are more comfortable discussing end of life care and when we understand the cultural determinants that have made American medicine so hesitant to accept human finitude. For as Dr. Stead reminds us, “You have to know the culture of the system as well as the science of the system in order for things to work at all.”23