No abstract

To determine whether limits to life-sustaining care are becoming more common, we attempted to quantify the incidence of recommendations to withhold or withdraw life support from critically ill patients, to describe how patients respond to these recommendations, and to examine how conflicts over these recommendations are resolved. In 1992 and 1993 we prospectively enrolled 179 consecutive patients from two intensive care units (ICUs) for whom a recommendation was made to withhold or withdraw life support. Where possible, we compared results with data collected in the same units over a similar time period in 1987 and 1988. Recommendations to withhold or withdraw life support preceded 179 of 200 deaths (90%) in 1992 and 1993, compared with 114 of 224 deaths (51%) in 1987 and 1988 (chi2 = 73.76, p < 0.001]. Cardiopulmonary resuscitation was initiated in 10% of deaths in 1992 and 1993 as compared with 49% in 1987 and 1988. Ninety percent of patients agreed within less than 5 d, and only eight patients (4%) refused physicians' recommendations to limit life support. In cases of conflict, physicians in 1992 and 1993 deferred to patients with one exception: physicians were willing to refuse surrogate requests for resuscitation of patients they considered hopelessly ill. We conclude that 90% of patients who die in these ICUs now do so following a decision to limit therapy, that this represents a major change in practice in these institutions over a period of 5 yr, that most patients and surrogates accept an appropriate recommendation to withhold or withdraw life support, and that physicians will refuse surrogate requests in certain circumstances.

In some intensive care units (ICUs), fewer patients who die now undergo attempts at cardiopulmonary resuscitation (CPR), and many more have life support actively withdrawn prior to death than did a decade ago. To determine the frequency of withdrawal of life support, we contacted every American postgraduate training program with significant clinical exposure to critical care medicine, asking them prospectively to classify patients who died into one of five mutually exclusive categories. We received data from 131 ICUs at 110 institutions in 38 states. There were 6,303 deaths, of which 393 patients were brain dead. Of the remaining 5,910 patients who died, 1,544 (23%) received full ICU care including failed cardiopulmonary resuscitation (CPR); 1,430 (22%) received full ICU care without CPR; 797 (10%) had life support withheld; and 2,139 (38%) had life support withdrawn. There was wide variation in practice among ICUs, with ranges of 4 to 79%, 0 to 83%, 0 to 67%, and 0 to 79% in these four categories, respectively. Variation was not related to ICU type, hospital type, number of admissions, or ICU mortality. We conclude that limitation of life support prior to death is the predominant practice in American ICUs associated with critical care training programs. There is wide variation in end-of-life care, and efforts are needed to understand practice patterns and to establish standards of care for patients dying in ICUs.

Cytokines affecting mononuclear phagocytes were screened for activation of human macrophages to secrete H2O2 and kill toxoplasmas. In contrast to recombinant interferon-gamma (rIFN gamma), the following factors, tested in partially or highly purified form and over a wide range of concentrations, did not augment these functions: native interferon-alpha (nIFN alpha), rIFN alpha A, rIFN alpha D, rIFN beta, colony stimulating factor (type 1) (CSF-1), CSF for granulocytes and macrophages (GM-CSF), pluripotent CSF (p-CSF), tumor necrosis factor (TNF), native interleukin 2 (nIL-2), and rIL-2. Partially purified migration inhibitory factor (MIF) enhanced H2O2-releasing capacity submaximally without inducing antitoxoplasma activity, and warrants further study.

We now have a decade of experience with advance directives since the Patient Self-Determination Act was signed into law in November 1990. With few exceptions, empirical studies have yielded disappointing results. Advance directives are recorded by medical personnel more often but are not completed by patients more frequently. The process of recording them does not enhance patient-physician communication. When available, advance directives do not change care or reduce hospital resources. The most ambitious study of advance care planning, the Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments, failed to show any change in outcomes after an extensive intervention. Investigators have attempted to identify the reasons why the optimism about the Patient Self-Determination Act has not been realized. Many interventions to facilitate advance care planning were focused on specific treatment decisions. Recent research suggests that preferences for care are not fixed but emerge in a clinical context from a process of discussion and feedback within the network of the patient's most important relationships. Clinical trials emphasizing this approach have been successful. The approach that emphasizes communication, building trust over time, and working within the patient's most important relationships offers a hopeful model for clinicians working in intensive care units.