Caring for the Dying in the Intensive Care Unit: Title and subTitle BreakIn Search of Clarity

Despite many scientific advances in critical care medicine and the continued best efforts of intensivists, the intensive care unit (ICU) remains a setting where, unfortunately, many patients will die. Among patients with chronic disease who die in the hospital, approximately half are cared for in an ICU within 3 days of their death and one third spend at least 10 days in the ICU during their final hospitalization.¹ In 1995, approximately 20% of deaths in the United States occurred in an ICU.² Many studies in the United States, Canada, and Europe consistently have shown that the majority of ICU deaths involve withholding or withdrawing life-sustaining treatments.^{3 - 10} Intensive care therefore involves difficult decisions about the use of life-sustaining treatments for critically ill patients who do not respond to critical care therapies. For these patients, an important goal is to achieve the best possible death and the most compassionate care possible for their families.

In this context, the article in this issue of THE JOURNAL by Sprung et al and the Ethicus investigators¹¹ describes end-of-life practices in 37 ICUs across 17 countries by documenting the process of continuing, withholding, or withdrawing life support. The investigators conducted a prospective, observational study of all patients admitted to these 37 ICUs over an 18-month period. The data were collected by the "senior intensivist responsible for patient care" at each institution. During the study period, 4248 patients died in participating ICUs, representing 13.5% of all ICU admissions. Withholding or withdrawing life-sustaining treatments preceded 73% of these deaths. The ICU mortality rate and the frequency with which life-sustaining treatment was withheld or withdrawn are consistent with previous studies from the United States, Canada, and Europe.^{3 - 10}

Since investigators enrolled a mean of only 2 ICUs in each country, conclusions about individual countries are limited. Instead, investigators divided Europe into 3 regions—northern, central, and southern; a higher proportion of deaths preceded by cardiopulmonary resuscitation (CPR) was found in the southern region, and a higher proportion of deaths preceded by withdrawing life-sustaining treatments was found in the northern region. As the authors point out, this study does not address the perspectives of patients and their families, a critically important aspect of end-of-life care. The investigators provide some information about a practice they designate as "shortening of the dying process." These 2 findings, the geographic variation and the prevalence of shortening of the dying process, merit further consideration.

Europe has been divided geographically, politically, and culturally for centuries. It is not surprising that end-of-life practices in the 3 regions vary. However, these findings should be viewed in the context of prior research investigating geographic variation in the provision of health care. Dramatic geographic variation exists in health care delivery for outpatient care, surgical procedures, hospital care, and intensive care.¹² Examples of geographic variation involving intensive care include withholding admission to the ICU,¹³ varying thresholds for red blood cell transfusion,¹⁴ differing criteria for diagnosing brain death,¹⁵ and variation in the proportion of deaths preceded by withholding or withdrawing life support.^{6 ,10} Variations in practice also occur within geographic regions and even within individual cities, raising the question of how much total variation can be explained by geography.^{16 - 17}

The Ethicus investigators note that variation in end-of-life care was associated with physician religion and suggest that physician religion may explain some of the geographic variability. Although physician religion may be a powerful explanatory variable, caution is needed when making such inference because the collinearity between religion and region prohibits multivariate analysis that might distinguish the influence of region from that of physician religion, patient religion, and other confounding variables. The many important differences in health care between northern, central, and southern Europe may provide alternative or additional explanations for this variation.

Some studies suggest that as the evidence for particular practices becomes more robust and as consensus regarding effective therapy increases, geographic variation decreases.^{18 - 19} It has been argued persuasively that large geographic variation in health care delivery is a call for increased implementation of existing research in clinical practice, creation of new evidence to guide practice, and development of consensus on best practices.¹² The Ethicus investigators speculate on whether treatment might be insufficiently intense in a region where rates of CPR are low and those of life support withdrawal are high. Conversely, they question whether treatment is prolonged unreasonably in regions where unsuccessful CPR precedes a higher proportion of deaths.

Physician biases influence willingness to withhold or withdraw life support.^{20 - 22} Recognition of these biases lend weight to the concern that extremes in geographic variation in end-of-life care in the ICU may be, by definition, evidence of suboptimal quality care in some places. Conversely, perhaps some of the geographic variation in end-of-life care in the ICU is a reflection of variation in cultural or religious values of physicians, patients, and families, and perhaps the best quality of care is different in different regions.²³ It seems likely that the geographic variation in end-of-life care across Europe reflects a combination of appropriate variation reflecting different cultures and inappropriate variation reflecting a shortage of evidence and consensus defining best practices. The challenges facing the critical care community worldwide are to differentiate these 2 types of variation, to identify and implement best practices that are independent of cultural differences, and to respect the important cultural distinctions that exist.

The Ethicus investigators describe one aspect of end-of-life care in the ICU that sets this study apart from others in the field. They provide some details of what they describe as shortening of the dying process. Investigators relied on the willingness of senior physicians in the participating ICUs to report these practices. Reports of practices that might be considered as acts of involuntary euthanasia were limited to 7 centers and concerned 2% of the patients that died (a total of 94 patients); a large proportion of this practice occurred in a single center. The Ethicus investigators speculate, on the basis of similar drug usage between life support withdrawal and shortening of the dying process, that clarity is lacking over what constitutes a shortening of the dying process. They imply that there may be a similar lack of clarity in the distinction between this process and withdrawal of life-sustaining therapies.

However, provision of narcotics or benzodiazepines to provide comfort in the setting of the withdrawal of unwanted or unsuccessful life-sustaining therapies is ethically and legally different from the use of these therapies or other drugs for the sole purpose of hastening death. If life-sustaining treatments are to be withdrawn, the use of narcotics or benzodiazepines to achieve patient comfort constitutes good-quality care. To withhold such palliative treatment, thus allowing patients to die in pain or distress, is poor-quality care. Administering medications for the sole purpose of hastening death, and without the consent of the patient, is illegal throughout Europe, North America, and many other parts of the world (even in Oregon and the Netherlands where patient consent is required for physician-assisted suicide). Compassionate critical care clinicians may often wish that death would come quickly in the setting of withholding or withdrawing life-sustaining treatments for the sake of patients and their families, but taking unilateral action to provide such hastening is a distinctly different phenomenon. Moreover, data on the withdrawal of life-sustaining treatments suggest that time to death is not associated with family members' reports of comfort and dignity for patients,²⁴ and thus is unlikely to be improved by hastening death. The "principle of double effect" provides a rationale for providing control of pain and other symptoms for dying patients, even if a potential adverse effect of the medications used might be to unintentionally hasten death.²⁵

While the distinction between high-quality palliative care in the setting of withholding or withdrawing life-sustaining treatments and active shortening of the dying process is clear, data from the Ethicus study and other studies cannot address the ethical question of what physicians should do. However, as the Ethicus investigators correctly point out, several important areas lack clarity. First, the results of the study suggest that the distinction between high-quality palliative care and active shortening of the dying process in the minds of some (albeit a minority) critical care clinicians may be lacking. Similarly, in a previous study, a minority of critical care nurses (17% of 852 nurses from adult ICUs in the United States) reported practices they considered to be euthanasia.²⁶ Second, it might not be possible to clearly distinguish these 2 phenomena by abstracting the medical records of patients who have died or by analyzing responses to surveys. Circumstances may occur in which it would not be possible to distinguish these 2 phenomena by observing the actions of critical care clinicians unless their intent also could be assessed accurately. Nonetheless, it is important to be clear that these phenomena are distinct.²⁷

The Ethicus study provides important insight into the variations in end-of-life practices in ICUs across Europe. The geographic variation by region of Europe raises important questions about the factors that determine this variation. This study also suggests that the practice of active shortening of the dying process in the ICU setting, albeit rare, seems to be the intent of some critical care physicians and seems to be isolated to a few institutions. The challenge this study raises for critical care clinicians throughout Europe and around the world is clear. Additional research is needed to guide best practices concerning end-of-life care in the ICU setting for patients and their families, define areas of end-of-life care that should be culturally determined, and develop consensus about those areas of practice that constitute good-quality care, regardless of region and culture.