End-of-Life Care for Critically Ill Patients

Abstract & Commentary

Synopsis: In survey data from 131 intensive care units in 38 states, life support was limited in 70% of all deaths with approximately one-half having either withholding or withdrawal of such support. Only one-quarter of patients who died underwent CPR.

Source: Prendergast TJ, et al. Am J Respir Crit Care Med 1998;158:1163-1167.

A number of recent surveys, including the "Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments (SUPPORT)," have demonstrated that physicians tend to provide more extensive treatment to critically ill patients than they might prefer.1 In addition, recent data have also demonstrated that significant economic expenditure is made for patients in the final year of life. There has been much discussion about the role of the pulmonary and critical care physician in providing palliative care (i.e., that component of care given to patients in whom there is little chance for recovery either of life or function).

To understand the evolution of palliative care within the intensive care unit (ICU), Prendergast and Luce first surveyed the practice of withholding and withdrawal of life support at two University of California-affiliated hospitals.2 They documented a significant increase over a five-year period in the practice of withholding and withdrawing life support from dying patients—from 51% having life support withdrawn in 1987-1988, to more than 90% in 1992-1993.

Expanding on this information, a survey was conducted in 131 ICUs at 110 institutions in 38 states. Units surveyed included those with active critical care and pulmonary teaching programs.

Of the 6303 deaths, a total of 393 were diagnosed with brain death. Of the remaining 5910 patients, only one-quarter received full critical care including unsuccessful cardiopulmonary resuscitation (CPR). Another 22% had intensive care but without CPR undertaken. Ten percent had life support withheld and nearly 40% had life support withdrawn. Overall, three-quarters had support either withheld or withdrawn. Neither the number of admissions nor ICU mortality in a variety of settings predicted the variability of practice in a given ICU. There were significant regional differences in these life support practices. In Missouri and New York, strict legal standards for withdrawal of life support by surrogate decision-makers resulted in a lower proportion of death preceded by withdrawal of life support.

Comment by Alan M. Fein, MD

Currently, there is significant national focus on palliative care. Physicians are questioning whether the intensive interventions provided in the past to all patients regardless of prognosis are reasonable, worthwhile, and even humane. The recent study by Prendergast and associates goes a long way in documenting a significant "social change in the practice of medicine." Nearly three-quarters of patients surveyed in this group of teaching hospitals had life support limited in some way—either withheld, withdrawn, or patients made do not resuscitate (DNR). This is up from significantly lower levels five years ago. There was, however, a major variation based on the type of ICU and the geographic location. Prendergast et al point out that there are as yet no standards of practice in the area of palliative care. Physicians are often uncertain about the legality of their actions when life support is withheld or withdrawn, even where the courts have supported the use of surrogate decision-making. Two major directions for palliative care have been 1) the recognition by physicians of the importance of patient autonomy in establishing advanced directives; and 2) the reliance on predesignated surrogate decision-makers using "healthcare proxies" when the patient is no longer competent to make decisions. In addition, the physician’s duty to relieve pain and suffering has become more pronounced in day-to-day practice. Concern about the immediate adverse effects of sedation and analgesia in the dying patient have been supplanted by the need to reduce dyspnea, pain, and suffering in terminally ill patients. The Supreme Court has supported the pre-eminence of pain relief even when it may secondarily increase the likelihood of death. Interestingly, New York and Missouri have been highlighted for their more stringent surrogate decision-making standards. In these states, if patients are unable to make their own end-of- life decisions, life support may be withdrawn only if a legally designated healthcare proxy or advance directive explicitly states that the patient wanted a specific intervention not to be given. This practice results in a more frequent continuation of care that the patient may not have wanted.

Prendergast et al also point out the role that managed care and financial incentives may be playing in withholding and withdrawing life support. They correctly argue that financial incentives may perversely cause physicians to provide too little rather than too much care. The issues of how to care for the critically ill patient who cannot be saved are becoming as important as the interventions to extend life.


    1. The SUPPORT Principal Investigators. JAMA 1995;274:1591-1598.

    2. Prendergast JT, Luce JM. Am J Respir Crit Care Med 1997;155:15-20.