There is a large and growing literature about patients' and physicians' preferences and medical practices at the end of life. This literature includes studies that aim to understand these preferences and practices better, as well as others that test educational and clinical interventions designed to improve the end-of-life experience. One can draw at least two generalizations from this literature. First, patients' preferences for how they want to die frequently do not accord with what eventually happens to them. Second, interventions aimed at solving this problem, from attempts to increase the use of advance directives to more costly and intense interventions such as using nurses as patient advocates, have generally not closed the gap between stated preferences and actual events.
One way to view this literature is to recognize that it is typical of the reporting of early efforts to solve previously unsolvable important problems. Indeed, the increasing number of studies in this area suggest that investigators and funding agencies have concluded that past efforts have failed and that new approaches are warranted. Implicit in this conclusion is the view that in time the problem will be solved by new and better methods. Less frequently articulated is an alternative conclusion—that the problem is not as tractable as we would like it to be. Simply put, the gap between patient preference and reality might in large part be a reflection of an understandable ambivalence about dying and contemporary medical responses to death.
It is our clinical impression that most patients and their families are not at all sure of “what they want,” want different things at the same or different times, or want something they cannot have or are unlikely to have. Patients and their families fear painful, uncomfortable, and undignified deaths at the same time they may hope for improvement. Patients, their families, and their clinicians have understandable fears about medical interventions for the very sick and aged; while medical interventions may relieve suffering and delay death, they can also bring pain and humiliation, and the life they prolong may be without meaning. This ambivalence is compounded by the fact that clinicians generally cannot offer patients and their families prognostic information that is detailed and reliable enough for making good decisions about the type, duration, and balance of curative and palliative care.
The movement to study and improve end-of-life care is so logically consistent and the problem so important that it may seem both unwise and callous to question its standard assumptions and goals. So while we tread lightly here, it is important to reflect on the possibility that contemporary interventions aimed at closing the gap between patient preferences and what actually happens at the end of life might themselves be futile. In addition, it seems prudent to remind ourselves of some troubling implications of current research and policy efforts, such as the potential to create guilt among clinicians and families when a “good” death is not achieved.
The prevailing view that patients have preferences for how they die and physicians often do not follow these preferences seems simplistic and unfair to all when it is more plausible to believe that dying patients are often not sure of what they want, want different things at different times, or want something they cannot have. How can the profession ever satisfy patients in the setting of such ambivalence? Yet, in this setting, patients and clinicians frequently fault themselves for not working harder at communication and for not obtaining and using advance directives.
It may very well be that the methods that have been used so far to determine patients' preferences have generally been too thin for the task of understanding individuals' complex and often ambivalent feelings about their life, death, and medical care at the end of life. Patients, families, and clinicians in these studies have generally answered focused questions about their attitudes, beliefs, and knowledge. Is it really plausible, for example, for a grieving family member to meaningfully answer questions about whether death was managed in accordance with a loved one's wishes in a way that is not confounded by loss itself ? The situation may call for a more intense, open-ended, and time consuming observation and analysis, such as gained through an ethnography of an intensive care unit or hospice. But regardless of the progress that might be made by new methodological approaches, there is likely to remain an irreducible core problem that will be difficult to solve.
A common response to “negative” interventional studies in end-of-life care is to acknowledge the flaws in current approaches and to redouble efforts to find new and better ones. This positivist response is not only consistent with the importance of the problem, our faith in clinical and policy solutions, and our belief in the importance of patient autonomy and choice, but is also motivated by the legitimate view that the perfect should not become the enemy of the good. But we should recognize that the profession's difficulty with end-of-life care is in part a reflection of the mixed signals and confusion that naturally accompany any situation in which there are no good outcomes. In the end, the persuasive motivations to improve care at the end of life may not translate into comprehensive solutions if it is our ambivalence about death in late 20th century American society and medical settings, more than inadequate knowledge and interventions, that underlies the problems we perceive.
Acknowledgements
Dr. Asch is the recipient of a Health Services Research and Career Development Award from the Department of Veterans Affairs. Dr. Aronowitz is a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar.