ARTICLE

TOP ABSTRACT Article REFERENCES

Ten years ago, my colleagues and I undertook an observational study of the withholding and withdrawal of life support from critically ill patients in two intensive care units (ICUs) in hospitals affiliated with the University of California, San Francisco (UCSF). We did so because we believed that life-sustaining therapy was frequently forgone in our ICUs, and that this practice was ethically acceptable and clinically desirable if it reduced unnecessary suffering in patients who were unlikely to benefit from further treatment. We also hoped that physicians elsewhere might become more actively involved in managing the deaths of their patients after reading about our experience. We learned and subsequently reported (1) that withholding and withdrawal of life support occurred in 51% of the patients who died in our ICUs during a 1-yr period in 1987 and 1988, a larger percentage than we anticipated at the time.

Physicians whom we interviewed in this study recommended that life-sustaining therapy be forgone primarily because their patients were dead by neurologic criteria or had a poor prognosis as determined by the physicians' clinical experience and knowledge of the medical literature. All but 2% of patients or surrogates agreed with these recommendations, although reaching such agreement frequently required several days. Support was continued for those few patients or surrogates who insisted upon it. For the other patients, cardiopulmonary resuscitation (CPR), mechanical ventilation, and vasopressors were the therapies most frequently withheld and withdrawn. Of those patients who were not brain dead, 70% received analgesics and sedatives as other care was discontinued. Although our observations were limited to two institutions affiliated with UCSF, we speculated that forgoing life-sustaining therapy would become more common throughout the United States because policy makers and the public were increasingly concerned about the uncertain benefits, the possible harm, and the high cost of critical care.

Five years after our initial study, we performed a follow-up study in the same two ICUs. We found and then reported (2) that forgoing of life-sustaining therapy preceded death in 90% of patients who died in the ICUs during a 1-yr period in 1992 and 1993. Death according to neurologic criteria, and a poor prognosis, expressed in terms of futility, were again the reasons cited by physicians for recommending that life support be withheld or withdrawn. In 56% of cases in which futility was cited, physicians judged that patients had "absolutely" no chance to leave the ICU alive. However, in 44% of cases, physicians applied the concept of futility despite an estimate of ICU survival of between 1% and 50% (median: 5%). Again, almost all patients or surrogates agreed with physician recommendations to limit care, and care was provided for patients or surrogates who requested it. However, as was not the case in our previous study, do-not-resuscitate (DNR) orders were written without the consent of the surrogates of several patients for whom other kinds of life support had been continued. We considered the writing of these DNR orders justifiable, and we questioned whether some physicians were not too passive in permitting the exercise of technology to persist for so long in hopeless cases.

In a subsequent article (3), I argued that physicians do not have a responsibility to provide futile or unreasonable care even if a patient or surrogate insists on it. My argument was based on the belief that physicians need not merely furnish all services asked of them, but that they have a professional obligation and a social sanction to provide only therapies that are beneficial, to avoid harm, and to allocate medical resources wisely. Although physician pursuit of beneficence, nonmaleficence, and distributive justice may conflict with patient autonomy, such conflict has not prevented physicians from acting unilaterally in limiting life-sustaining treatment. This practice was demonstrated by Asch and colleagues (4), who surveyed members of the Critical Care Section of the American Thoracic Society (ATS) in 1990 and found that over 80% of respondents had withheld or withdrawn life support on the basis of futility. Many physicians reported forgoing care without the consent or knowledge of patients and surrogates, and some reported doing so over patients' or surrogates' objections.

In 1994, Thomas J. Prendergast, M.D., and I asked physicians in 167 institutions associated with fellowship programs in Pulmonary and Critical Care Medicine and just Critical Care Medicine to classify all deaths that occurred in their ICUs over a 6-mo period into one of five defined categories (5). With 126 ICUs at 107 of the 167 sites (64%) reporting, we amassed data for more than 71,000 admissions and some 6,100 deaths (9% overall mortality rate): 25% of deaths followed full ICU care and failed CPR, 23% of deaths occurred in patients who received full ICU care excluding attempted CPR, 13% of deaths occurred in patients who had some form of life support other than CPR withheld prior to their deaths, 33% of deaths followed the withdrawal of life support, and 6% of patients were brain dead. When the first and last categories were totaled, 69% of the ICU deaths were found to have followed the forgoing of CPR and other forms of life-sustaining therapy.

At first glance, the results of this recent study suggest that withholding and withdrawal of life support have become so commonplace in American ICUs that they are de facto standards of care. Nevertheless, significant variability was found among surveyed ICUs in terms of mortality rate (from 2% to 45%), failed CPR (from 4% to 79% of deaths), and willingness to withdraw support (from 0% in six ICUs comprising approximately 2,700 admissions and 200 deaths to 71% in other ICUs). This variability could not be explained by differences in the kind (e.g., medical versus surgical), size, or geographic location of the ICUs. Some of it might have been due to differences in the severity of illness of patients in the ICUs, which we did not measure. However, most of the variability probably related to differences in the beliefs and behavior of physicians, as has been suggested by another study (6). Whatever its explanation(s), the variability precludes the conclusion that true practice standards exist in the area of forgoing life-sustaining therapy, although such forgoing clearly has become common. It also raises the question of how to regard outliers at both ends of the clinical spectrum: those ICUs in which withholding and withdrawal of life support are customary and those in which they occur rarely if at all.

As a physician in an ICU at the "customary" end of the spectrum, I still believe that our approach to forgoing life-sustaining therapy at UCSF is generally acceptable and desirable. Nevertheless, I often ask myself whether we have gone too far in recommending that life support be withheld or withdrawn from some of our patients for whom further care would not be futile in a stringent sense of the word. I also wonder whether we have been too zealous in promoting these practices in other institutions, particulary because in the process we have challenged patient autonomy by suggesting that care could be forgone in certain circumstances without the agreement of patients or their surrogates. Challenging patient autonomy probably was important at one point in the evolution of our profession's view of limiting care for certain critically ill patients, because physicians were just becoming comfortable with the ideas that not all demands upon them must be met and that life, especially nonsentient life, need not be supported at all cost. Nevertheless, I sense that many of my colleagues in San Francisco and other cities have realized that we must be very cautious in exercising our influence, if not authority, over patients and their surrogates in prompting the forgoing of life-sustaining therapy in the ICU.

Some of the colleagues I refer to have been members of one or both of two bioethics task forces established by the ATS in recent years. The first task force was convened in 1989 and chaired by Paul N. Lanken, M.D. After considerable deliberation, it recommended that patient requests to withhold or withdraw life support should be respected, and that surrogates should be identified to help make decisions about life-sustaining therapy when patients lack decision-making capacity. In addition, however, the task force stated that a life-sustaining medical intervention can be limited without the consent of patient or surrogate when the intervention is judged to be futile, and that health-care institutions have a responsibility to promote ethically sound decision making regarding life-sustaining therapy. Included among these responsibilities are written policies defining the circumstances in which it is acceptable to withhold or withdraw life support, and those in which there may be limits to a patient's or surrogate's directives either to forgo or to request life-sustaining interventions. The task force's recommendations, which were adopted by the ATS Board of Directors in 1991, were subsequently published in the American Review of Respiratory Diseases (7) and the Annals of Internal Medicine (8). They remain the boldest statement by any professional medical society of which I am aware on the prerogatives of physicians and health-care institutions in restricting the application of critical care.

In 1993, a second ATS Bioethics Task Force was convened under the leadership of Dr. Lanken and Peter B. Terry, M.D., to develop recommendations for the fair allocation of ICU resources. Much of this second task force's first meeting, in December 1993, was devoted to a discussion of how President Clinton's plans for health-care reform would affect ICU resource allocation. Several articles were distributed as background reading, including publications on medical triage and a draft of an article I had prepared in advance of the meeting. In this draft and in the Clinical Commentary (9) that was subsequently published in the American Journal of Respiratory and Critical Care Medicine, I argued that the President's goals of cost containment and universal access to basic health care could only be realized by restricting the use of marginally beneficial services, including some ICU care, and by balancing physicians' unrestrained advocacy for individual patients with a proportional advocacy for the entire population. I do not know whether the task force members accepted my ideas at the time. However, I remember clearly that the task force considered rationing of ICU services likely, and perhaps desirable if marginally beneficial ICU care was restricted, and wanted to provide an ethical and practical framework for such rationing if and when it was required.

More than 3 years have passed since that first meeting, and the task force's recommendations, which were adopted by the ATS Board of Directors in March 1997, appear in the October 1997 issue of the American Journal of Respiratory and Critical Care Medicine (10). Triage is mentioned in the recommendations, along with the idea that medically appropriate patients should be admitted on a first-come, first-served basis when demand for ICU beds exceeds supply. The recommendation also is made that marginally beneficial ICU care can be restricted on the basis of high cost relative to benefit, but that decisions to limit care should be made only by explicit institutional policies that reflect a social consensus in support of such limitations. Other recommendations are that patients should have equal access to ICU care regardless of their personal and behavioral characteristics, that ICU care should be equally available regardless of patients' ability to pay, and that patients and the public should be informed of financial incentives for limiting ICU care by physicians or health-care institutions. Among the basic principles underlying these strong recommendations are that each individual's life is valuable, and equally valuable with those of others, and that access to ICU care, when medically appropriate, is an essential part of a basic package of health-care services that should be available to all.

Once again, the ATS Bioethics Task Force has come up with a bold statement, but it is not what I would have predicted 4 years ago. What accounts for the statement's deemphasis on rationing critical-care services and its emphasis on securing equal access to the ICU? Although the cochairmen and other task force members may offer other explanations, I suspect that they would agree that changes in health care in the United States have inspired, if not mandated, the ATS's strong position in support of the right of all patients to receive medically appropriate ICU care. President Clinton's goal of universal access to health services has not been realized since the task force first met; in fact, more Americans lack health insurance today than ever before. Managed care, which many hoped would contain costs without compromising quality, has become broadly identified as a tool for increasing corporate profits by denying medical services. Whether this is true or not, the traditional physician-patient relationship has been increasingly complicated by the presence of a third party: the payor that seemingly seeks to restrict access to health care. As a result, patients have become increasingly suspicious of their health-care institutions and of those physicians who are rewarded for restricting access. In this environment, it is entirely appropriate that the ATS Bioethics Task Force and the ATS Board of Directors have taken their stand.

In keeping with both ATS statements, in light of the current health-care environment, and in accord with my own changes in sentiment, I propose that physicians reconsider certain aspects of how they withhold and withdraw life support from the critically ill. I still believe that we can and should recommend that life-sustaining treatment be forgone if we consider it futile because the patient's prognosis is poor. Futility in this context need not be defined in narrow physiologic terms, such as an inability to raise the blood pressure by administering vasoactive agents. Rather, it should be seen as embracing a combination of quantitative (e.g., the probability of survival following a given medical intervention) and qualitative (e.g., survival should include an acceptable quality of life) dimensions, as advocated by Schneiderman and colleagues (12). This definition of futility is close to the concept of "inadvisable" treatments advanced by the Ethics Committee of the Society of Critical Care Medicine (11).

Curtis and associates (13) have demonstrated that even when a quantitative and qualitative definition of futility is used, physicians may misunderstand the probability of survival following an intervention such as CPR, and may fail to determine what quality of life is acceptable to their patients. These authors require that quantitative futility be based on a very low probability of survival, perhaps lower than 1%, and that the recommendation to limit care be preceded by a discussion of quality-of-life issues between physicians and patients or their surrogates. These requirements seem appropriate to me. I also believe that recommendations made after these requirements are met need not bear any relationship to rationing or resource allocation; they merely reflect the physician's obligation to seek benefit and do no harm.

In most instances, patients and surrogates accept physician recommendations to limit care if sufficient time is afforded them to consider these recommendations (1, 2). Nevertheless, on rare occasions, they insist that care be continued, and cannot be dissuaded from this position, presumably because they disagree with the prognosis we have provided them. When this occurs, I believe that we should not withhold or withdraw life support on the sole basis of our personal notions of futility, but instead should rely on institutional or multiinstitutional futility policies. Although their input is desirable, administrators, even if they are physicians, should not be allowed to impose futility policies that serve primarily to protect their businesses' bottom lines. Rather, the policies should be developed by practicing physicians, nurses, and other health professionals in concert with patients, community leaders, and, when appropriate, participants in managed-care organizations. This approach would replace payors in now tripartite physician- patient relationships with representatives of either the community in general or of the subscribers in an individual health plan. Futility policies crafted in such a fashion should specify which ICU interventions are beneficial and which are not. They should address potential conflicts of interest issues between physicians who are paid on a capitated basis and the patients for whom they care. Such policies should also be available in advance to persons who could use such information in selecting their care from a particular hospital, medical group, or health-care plan.

One model of a multiinstitutional policy (14) on futility has been developed by representatives of major hospitals in Houston, Texas. Under this policy, when a physician determines that an intervention is inappropriate but a patient or surrogate insists that it be provided, the physician must first discuss the reasons for his or her judgment, the possibility of transferring care without abandoning the patient, and alternative treatments such as palliative care. If agreement is not reached at this point, the physician must obtain a second opinion from another physician who has examined the patient, and must present the case before an institutional review body. If the review body affirms that the intervention is inappropriate, the intervention is terminated, a plan of alternative care is established, and interinstitutional transfer of the patient to another physician, in order to provide the intervention that has been deemed inappropriate, is not allowed. On the other hand, if the review body finds that the intervention is appropriate, orders to terminate the intervention are not recognized as valid without patient or surrogate assent.

The Houston collaborative policy is limited in that it was developed by institutional representatives without explicit community involvement, as its authors (14) acknowledge. The legal standing of such a policy is also uncertain, in light of court decisions such as that concerning Baby K (15), wherein the court mandated continuation of life support for an anencephalic infant in response to parental requests; or that involving Helen Wanglie (16), in which the court refused to appoint a surrogate other than the patient's husband who might agree to a limitation of the patient's care; and that of Catherine Gilgunn (17), whose physicians were absolved of liability after they withdrew life support over the objections of the woman's daughter. Case law is thus incomplete and contradictory in this area, and the medical profession cannot wait for such law to develop before clarifying its own values and methods for resolving conflict between patients and providers.

Furthermore, it is not clear whether the Houston collaborative futility policy is workable in the ICU environment, where decisions must often be made quickly, or that physicians will follow it rather than openly or surreptitiously limiting care on their own. Nevertheless, the policy represents a commendable approach in balancing patient autonomy and professional and institutional integrity, and its very presence seems to have prompted the resolution of disputes before institutional review bodies are required (B. A. Brody, personal communication, April 25, 1997). By shifting decision-making from the patient, physician, and payor to a group level, futility policies such as that adopted in Houston should move us closer to a social consensus on the issues of what is truly beneficial and how limited resources should be allocated. In the process, they should help restore public confidence in physicians and health-care institutions, a confidence that is severely threatened today.