assignment for reflection

Recommendations for end-of-life care in the intensive care unit: The Ethics Committee of the Society of Critical Care Medicine

Robert D. Truog, MD; Alexandra F. M. Cist, MD; Sharon E. Brackett, RN, BSN; Jeffrey P. Burns, MD; Martha A. Q. Curley, RN, PhD, CCNS, FAAN; Marion Danis, MD; Michael A. DeVita, MD; Stanley H. Rosenbaum, MD; David M. Rothenberg, MD; Charles L. Sprung, MD; Sally A. Webb, MD; Ginger S. Wlody, RN, EdD, FCCM; William E. Hurford, MD

KEY WORDS: palliative care; intensive care; end-of-life care

T hese recommendations are in-tended to provide informationand advice for clinicians whodeliver end-of-life care in in- tensive care units (ICUs). The number of deaths that occur in the ICU after the withdrawal of life support is increasing, with one recent survey finding that 90% of patients who die in ICUs now do so after a decision to limit therapy (1). Al- though there is significant variability in the frequency of withdrawal of life sup- port both within countries (2) and among cultures (3), the general trend is interna- tional in scope (4). Nevertheless, most evidence indicates that patients and fam- ilies remain dissatisfied with the care they receive once a decision has been made to withdraw life support (5). Al- though intensive care clinicians tradi- tionally have seen their goals as curing disease and restoring health and func- tion, these goals must now expand when necessary to also include assuring pa-

tients of a “good death.” Just as develop- ments in knowledge and technology have dramatically enhanced our ability to re- store patients to health, similar develop- ments now make it possible for almost all patients to have a death that is dignified and free from pain.

The management of patients at the end of life can be divided into two phases. The first concerns the process of shared decision-making that leads from the pur- suit of cure or recovery to the pursuit of comfort and freedom from pain. The sec- ond concerns the actions that are taken once this shift in goals has been made and focuses on both the humanistic and technical skills that must be enlisted to ensure that the needs of the patient and family are met. Although both of these issues are critically important in end-of- life care, the decision-making process is not unique to the ICU environment and has been addressed by others (6 –11). These recommendations, therefore, do not deal primarily with the process that leads to the decision to forego life- prolonging treatments but rather focus on the implementation of that decision, with particular emphasis on the ICU en- vironment.

This division of the process into two phases is necessarily somewhat artificial. Patients and families do not suddenly switch from the hope for survival and cure to the acceptance of death and pur- suit of comfort. This process happens gradually over varying periods of time ranging from hours to weeks. Similarly, the forgoing of life-sustaining treatments rarely happens all at once and is likewise a stepwise process that parallels the shift in goals. Although acknowledging the re- lationship between the process of deci-

sion-making and the corresponding ac- tions, these guidelines will focus on the latter.

These recommendations are written from the emerging perspective that pal- liative care and intensive care are not mutually exclusive options but rather should be coexistent (12–14). All inten- sive care patients are at an increased risk of mortality and can benefit from inclu- sion of the principles of palliative care in their management. The degree to which treatments are focused on cure vs. palli- ation depends on the clinical situation, but in principle both are always present to some degree. Figure 1 illustrates a useful paradigm for the integration of palliative care and curative care over the course of a patient’s illness.

Although many patients are best served by transfer to other environments (e.g., home, hospice, or ward) that may be more conducive to palliative care, some patients are so dependent on ICU technology at the end of life that transfer is not possible. For those who are ex- pected to survive for only a short time after the removal of life-sustaining tech- nology, transfer of the patient to a new environment with new caregivers is awk- ward and may disrupt the patient’s med- ical care. For these reasons, among oth- ers, intensive care clinicians must become as skilled and knowledgeable at forgoing life-sustaining treatments as they are at delivering care aimed at sur- vival and cure.

Preparation of the Patient, the Family, and the Clinical Team

As the decision to forego further use of life-sustaining treatments is being made,

From the Ethics Committee, American College of Critical Care Medicine.

The American College of Critical Care Medicine (ACCM), which honors individuals for their achieve- ments and contributions to multidisciplinary critical care medicine, is the consultative body of the Society of Critical Care Medicine (SCCM) that possesses rec- ognized expertise in the practice of critical care. The ACCM has developed administrative guidelines and clinical practice parameters for the critical care prac- titioner. New guidelines and practice parameters are continually developed, and current ones are system- atically reviewed and revised.

Supported, in part, by grants from the Argosy Foundation and the Harvard Risk Management Foun- dation.

Address requests for reprints to: Robert D. Truog, MD, Professor of Anaesthesia & Medical Ethics, Har- vard Medical School, Director, MICU, FA-108, Chil- dren’s Hospital, Boston, MA 02115.

Copyright © 2001 by Lippincott Williams & Wilkins

2332 Crit Care Med 2001 Vol. 29, No. 12

the family and clinical team must be pre- pared for what is to follow. As familiar as many clinicians may be with the process of withdrawing life support, it is a singu- lar event in the life of the patient and often is unprecedented for family mem- bers. Therefore, they may suffer great anxiety during the experience. Clear and explicit explanations on the part of the clinician may alleviate anxiety and refo- cus familial expectations.

Needs of the Patient. The healthcare team has an obligation to provide care that relieves suffering arising from phys- ical, emotional, social, and spiritual sources (7, 15–17). The patients in the study by Singer et al. (18) identified five domains of good end-of-life care: receiv- ing adequate pain and symptom manage- ment, avoiding inappropriate prolonga- tion of dying, achieving a sense of control, relieving burden, and strength- ening relationships with loved ones.

Most patients have already lost con- sciousness by the time life-sustaining treatments are removed (4, 19). Some, however, such as those with cervical quadriplegia or amyotrophic lateral scle- rosis, may be fully conscious. Whenever possible, patients should be prepared for the planned sequence of events and reas- sured about what they may experience.

Experience of hospice workers shows that the majority of dying patients fear pain and dyspnea (20). First and fore- most, patients should be assured that

management of their pain and distress will be the highest priority of their care- givers. Depending on personal prefer- ences and spiritual considerations, some patients will want to be more sedated than others. Patients should understand, however, that the clinicians will take their cues from the patient and will try to tailor the administration of sedation and analgesia to the individual needs and de- sires of the patient.

Closely related is the need to assure patients that they will be treated with respect and dignity, both during and after the dying process. A policy that explicitly allows and encourages the continuous presence of family and friends at the bed- side is one means of expressing this com- mitment. For patients who maintain re- lational capacity, the opportunity to say good-bye may be of paramount impor- tance.

Patients should know that their cul- tural beliefs are understood and that cul- tural expectations will be met (13). Clini- cians must plan ahead in this regard and be sure that they fully understand the relevant cultural expectations regarding the process of dying, the handling of the body after death, views about autopsy and organ donation, and cultural norms of grieving. Prior consultation with local representatives of cultural groups may be invaluable. Patients should be given every opportunity to experience spiritual mean- ing and fulfillment. Involvement of clergy

will often be desirable, and performance of religious services and rites at the bed- side should be encouraged (21). For chil- dren, cultural and spiritual observances should be oriented toward providing an age-appropriate understanding of dying, as well as providing the parents and fam- ily with meaningful rituals for coping with the death of a child.

Needs of the Family. Although the needs of the patient must be the primary focus of caregivers, there is growing con- sensus that a family-centered approach is particularly important in end-of-life care (22). Families of the dying need to be kept informed about what to expect and about what is happening during the dying process. Communication between clini- cians and grieving families may be diffi- cult in the absence of a prior relationship, as is frequently the case in the ICU. Pri- mary care providers and other more fa- miliar clinicians may be able to provide a helpful interface with the ICU team.

After conducting interviews, Hampe (23) identified eight needs of spouses of dying patients in the hospital setting: to be with the dying person; to be helpful; to be assured of the comfort of the dying person; to be informed of the person’s condition; to be informed of impending death; to ventilate emotions; to be com- forted and supported by family members; and to be accepted, supported, and com- forted by health professionals. Parents of children in pediatric intensive care units have identified their own needs, which Meyer et al. (24) arranged in a useful hierarchy: physical needs such as hunger and sleep; safety of their child; ready ac- cess to their child; access to optimal health care, accurate information from the healthcare team; participation in their child’s care; fulfillment of their pa- rental role; social support; and emotional consolidation and acceptance. Family members may neglect their own physical and emotional needs, to the detriment of their ability to participate in decision- making and care.

The needs of families have been as- sessed by a survey tool known as the Critical Care Family Needs Inventory (25). A meta-analysis of several studies that have used this tool identified the most important family needs, many of which focused on the desire to have on- going communication with the health- care team (26). Combining information from a number of studies leads to a sum- mary of the needs of families, as seen in Table 1 (23–25, 27, 28).

Figure 1. Palliative care within the experience of illness, bereavement, and risk. From Frank D. Ferris, MD, Medical Director, Palliative Care Standards/Outcomes, San Diego Hospice, 4311 Third Avenue, San Diego, CA, USA 92103–1407.

2333Crit Care Med 2001 Vol. 29, No. 12

Families need the opportunity to be with the dying person. Although not al- ways possible, a private room is the envi- ronment most conducive to emotional and physical intimacy and should be identified as a goal for excellent care of the dying (as well as a legitimate factor in justifying this cost to third-party payers). Usual restrictions on visitation should be relaxed as much as possible, especially with regard to restrictions on children (in some hospitals, even pets have been al- lowed for short visits) (29). This also may mean accepting and tolerating large groups of family and friends at the bed- side, which may be disconcerting to some clinicians. Whenever possible and within reason, withdrawal of life support should be timed to allow for the arrival of family members who must travel long distances. Not all families, however, want to be at the bedside at the time of the patient’s death. Notifying the family that death is imminent should not be linked with an expectation that the family will be present. Families need to be reassured that it is also acceptable for them to re- main at home.

Attention to detail can make an enor- mous difference. For example, providing the family with an electronic pager or cellular phone can allow them to break away for awhile without feeling out of contact. Clinicians can remind family members that they may want to contact clergy, friends, or others and can assist in making the calls if possible. Simple amenities like the presence of tissues, chairs, blankets, coffee, water, and a phone and general attention to the aes- thetics of the room can contribute sub- stantially to the family’s sense of well- being and peacefulness. After the death of the patient, attention to detail may be greatly appreciated, as in freshly shaving the face of a man or clothing a child in her own pajamas (23).

Families vary in their tolerance for uncertainty and ambiguity, but clini- cians, from the primary intensivist to the subspecialists to the nursing staff, should strive to deliver a consistent message. This may be facilitated by having all com- munication occur through the same per- son.

Families should clearly know the iden- tity of the attending physician, under- stand that this person is ultimately re- sponsible for the patient’s care, and be assured of his or her involvement. Clini- cians should avoid making firm predic- tions about the patient’s clinical course, because these are notoriously difficult to make, are often inaccurate, and may re- sult in a substantial loss of credibility when they are in error. Although clini- cians should be sensitive and compas- sionate in their communication, it is im- portant that they explain the physiologic process of dying and describe in concrete terms how the patient will die and what it will look like. At times it will be necessary for the clinicians to anticipate, ask, and answer questions that the family appears to be afraid or unable to verbalize. Fam- ilies may benefit from reassurance that the clinicians are focused on the patient’s comfort. Clinicians should earn the pa- tient’s and family’s confidence by contin- ually assessing the patient’s suffering and demonstrating that pain-relieving medi- cations and treatments are constantly available. Families should know that the caregivers are committed to having a presence at the bedside, even when the family members themselves are not able to be there. Finally, families often need to be reassured about the decisions they have already reached, emphasizing that the responsibility for these decisions is shared between the family and care team. This can help to dispel lingering doubts and potential feelings of guilt.

Families should have the opportunity to be helpful. They may be invited to participate in activities to relieve discom- fort, such as mouth care, bathing, and repositioning. They should be encour- aged to participate in assessment of the patient’s pain and suffering. This is espe- cially important in pediatrics and pro- vides parents with an opportunity to ex- press their nurturing role (16). Families also should be encouraged to bring in meaningful personal articles and be al- lowed to keep these articles at the pa- tient’s bedside.

Families should be encouraged to ex- press their emotions. Both before and

after the death of the patient, they should be given the opportunity to reflect on the patient’s life and to recall shared memo- ries. For neonates or young children, it may be necessary to create special mem- ories through spiritual rituals or cultural tradition.

During the withdrawal of life support, all distractions should be eliminated so that the family’s attention can be devoted entirely to the patient. In most cases, monitors should be turned off and the leads and cables should be removed from the patient. In some cases, catheters such as nasogastric tubes, urinary catheters, and arterial catheters also may be re- moved. In other situations, however, do- ing so may be more disruptive than ben- eficial. Even if there is the possibility that an autopsy may be required by the med- ical examiner, removal of catheters and tubes before death is not prohibited when this is done for the benefit of the patient and family (medical examiners may dis- courage or prohibit removal of lines and tubes after death, however). Bedrails can be lowered and restraints removed to al- low family members more intimate con- tact with the patient. Although some family members may not desire to be at the bedside through the process of with- drawal, they should be given the oppor- tunity to be present and possibly even to participate in the withdrawal of treat- ment. Finally, families should have pri- vate time to be with the patient after death and before removal of the body from the ICU.

Needs of the Clinical Team. Although all members of the clinical team should have active roles in providing end-of-life care, key aspects of this care should be performed and modeled by respected cli- nicians with leadership roles in the insti- tution. These individuals are in a unique position to reinforce the message that excellent care at the end of life is an institutional priority. Attendings should affirm their leadership by personally su- pervising critical aspects of this care. For example, only 64% of Society of Critical Care Medicine (SCCM) physician mem- bers who perform extubation at the end of life remove the endotracheal tube themselves; the remainder presumably leave this task to nurses and respiratory therapists (30). Although removal of an endotracheal tube is clearly not a techni- cally challenging procedure, personal in- volvement of the attending during this transitional event can send a powerful

Table 1. Ten most important needs of families of critically ill dying patients

To be with the person To be helpful to the dying person To be informed of the dying person’s changing

condition To understand what is being done to the

patient and why To be assured of the patient’s comfort To be comforted To ventilate emotions To be assured that their decisions were right To find meaning in the dying of their loved one To be fed, hydrated, and rested

2334 Crit Care Med 2001 Vol. 29, No. 12

message about the importance of end-of- life care.

The clinical team needs to be multi- disciplinary and committed to coopera- tion and clear communication. A recent survey by Asch (31) pointed to difficulties in this area, with critical care nurses re- portedly needing to engage in many co- vert practices that were in conflict with the physician’s orders. These included ad- ministering more opioid than ordered and concealing the action by falsifying the amount “wasted,” increasing doses of opioids when patients were already coma- tose, or only pretending to administer life-sustaining treatments that were or- dered, such as by substituting saline for a vasopressor infusion (31). The methodol- ogy of this study has been harshly criti- cized, and many doubt that it represents an accurate picture of current critical care practices (32, 33). Nevertheless, it does suggest that nurses are concerned about the overuse of life-sustaining tech- nology and the unresponsiveness of phy- sicians to address this concern as well as the patients’ pain and suffering. These concerns emphasize the need to develop a better consensus between physicians and nurses regarding the goals and strategies for providing end-of-life care in the ICU.

The Asch survey also pointed to the need for better education about end-of- life care and an institutional commit- ment to maintaining clinical compe- tence. This is aided by providing clinicians with opportunities to gain knowledge concerning intensive pallia- tive care. This education should focus on how to support and counsel families through the withdrawal process, ensure respect for various religious and cultural beliefs, and emphasize general communi- cation and teamwork skills. Educational efforts need to be ongoing so that new staff are continually oriented to these competencies (13).

Clinical teams need administrative support. This begins by affirming the value of intensive palliative care at the highest levels of the institution and con- tinues with protecting nursing staff from increased workloads when they are in- volved in delivering time-intensive pallia- tive care. Administrators also can support intensive palliative care by allowing clini- cians to minimize transfers of dying pa- tients from the ICU to unfamiliar staff and locations, unless this is in the best interests of the patient and family.

Clinical teams need to have opportu- nities for bereavement and debriefing.

One option is to have regularly scheduled meetings where staff can share their thoughts and experiences as well as cri- tique the quality of the care they pro- vided. This can be an opportunity to as- sess whether the patient experienced a “good death” and to discuss what went well and what could have gone better. These meetings also can be a forum for organizing a structured bereavement program that may include sympathy cards, follow-up phone calls, or distribu- tion of educational materials to help guide families through the grieving pro- cess.

Ensuring the Comfort of the Patient

Intensive care medicine is so thor- oughly grounded in the curative model of care that clinicians may have a difficult time “switching gears” and adopting a model focused primarily on symptom- atology. An important difference between these models is the criteria used to deter- mine whether a particular monitor, diag- nostic test, or therapeutic intervention is indicated. In the curative model, the cri- teria are related to the degree to which the procedure will contribute to the pa- tient’s recovery from illness. In the pal- liative model, the criteria are related to whether the intervention will improve symptom relief, improve functional sta- tus, or ameliorate emotional, psycholog- ical, or spiritual concerns (13, 34). Only interventions that are favorable in this analysis should be used.

The transition from the curative to the palliative model often occurs in a piece- meal fashion. Sometimes the patient may receive an inconsistent combination of therapies, some aimed at comfort and some aimed at cure. One practical solu- tion for dealing with this problem is to completely rewrite the patient’s orders and care plan, just as if the patient were being newly admitted to the ICU. Each monitor, test, or intervention should be evaluated in terms of the degree to which it furthers the patient’s goals before it is entered onto the order sheet. Some rou- tine procedures that usually are consid- ered an intrinsic part of ICU care, such as measuring vital signs, performing rou- tine laboratory tests and chest radio- grams, and endotracheal suctioning, may not contribute positively to the patient’s comfort and should be excluded. On the other hand, some therapeutic proce- dures, such as the intravenous infusion of

vasopressors or inotropes, may cause very little discomfort (requiring only the maintenance of intravenous access) but may substantially benefit the patient by maintaining perfusion of vital organs, thereby improving level of consciousness, renal and liver function, and gastrointes- tinal absorption. In some circumstances, such therapy might be reasonable, even in a terminally ill patient who is not re- ceiving other life-prolonging therapies (35).

One caveat to this approach is that clinicians must interpret the goals of treatment from the perspective of the pa- tient. For example, one study found that many cystic fibrosis patients were still taking vitamins on their last day of life, well after the point at which it was clear that they were very near death (36). Cer- tainly the vitamins were not providing any “medical” benefit at this point, yet the authors surmised that the vitamins may have been part of a routine of care that the patient found comforting, and that altering this pattern or ritual of care as the patient approached death would have caused more distress than comfort. In this sense, then, some treatments may be indicated because of the psychological benefits (rather than strictly medical ben- efits) that they confer on the patient.

In most cases, however, rewriting the orders at the time that the goals of care are revised should reduce the use of mon- itors, tests, and procedures. Campbell and Frank (37) estimated that implemen- tation of a comprehensive palliative care plan reduces the use of acute care inter- ventions by approximately 50%.

Assessment of Pain. Many patients die with treatable pain, even in intensive care units (5). One probable reason for this is the strong bias in medicine toward the treatment of diseases rather than symp- toms (e.g., the treatment for the acute abdominal pain of appendicitis is surgery, not morphine). Palliative care reverses these priorities and places symptom man- agement ahead of diagnosis and definitive treatment. Another reason why pain is inadequately recognized and treated is because it is inherently subjective (e.g., “pain is whatever the patient says it is”) and difficult to measure. Palliative care gives pain relief a high priority. The con- cept of pain as the “fifth vital sign” is one way of emphasizing the importance of treating pain assessment as a core ele- ment of patient care. The increased use of pain scales has provided for better quan- tification of the patient’s experience. Un-

2335Crit Care Med 2001 Vol. 29, No. 12

fortunately, pain scales may be better suited to postoperative and other forms of acute pain than they are to the chronic pain frequently experienced by dying pa- tients.

Assessment of pain in dying patients often relies primarily on evaluation of level of consciousness and awareness, breathing pattern, and hemodynamics. Consciousness can be assessed by the pa- tient’s response to stimuli, by the pa- tient’s agitation or motor activity, and by facial expression. Bispectral analysis, which uses a processed electroencephalo- graphic signal to assess a patient’s level of consciousness, has been used as an ad- junctive monitor for assessing patient comfort during the withdrawal of life support. Although this approach to pain assessment is at odds with the goal of reducing intrusive technology and mon- itoring at the end of life, in very rare circumstances it may have a role when assessment of distress is particularly dif- ficult, such as in patients who are receiv- ing neuromuscular blocking agents (see subsequent discussion) (19, 38).

Assessment of breathing patterns can be complicated in dying patients. Irregu- lar breathing patterns are a natural part of dying and may not be uncomfortable for the patient. Unfortunately, the irreg- ular pattern that accompanies dying is often referred to as “agonal,” which may imply to the family and other clinicians that the patient is in “agony.” Gasping is a medullary reflex and can occur in the absence of consciousness. Similarly, noisy respirations from airway secretions (the “death rattle”) are more likely to be distressing to the family and other ob- servers than they are to the patient. The

question of whether clinicians should ever treat the patient primarily to relieve the distress of the family is considered subsequently.

The hemodynamic status of the pa- tient (e.g., heart rate and blood pressure) is an unreliable indicator of pain, because tachycardia and hypertension can occur even in the absence of consciousness. Such hemodynamic signs may be more indicative of distress when they occur as part of a constellation of autonomic signs such as diaphoresis or lacrimation or when they occur in association with nox- ious stimuli.

The assessment of pain in neonates and small infants deserves special com- ment. Until recently, many clinicians be- lieved that these patients had diminished capacity to experience pain and suffering and that they were more prone to serious side effects from the use of potent anal- gesic and anesthetic medications. Recent studies suggest, however, that pain path- ways are functional from late gestation onward, and advances in anesthesiology and pediatrics have resulted in the devel- opment of safe anesthetic regimens and pain treatment protocols for patients of all ages (39 – 41). These insights extend the same emphasis on relief of pain and suffering that has become mandatory for adults to the clinical management of dy- ing newborns and children (42).

Assessment of Suffering. “Pain” and “suffering” are not synonymous, but nei- ther are they inherently distinct. In addi- tion to its neurobiologic dimensions, pain also has powerful psychological and cultural components. Suffering is a more global term and includes consideration of the existential pain that is an essential

part of the human condition. Some have argued that clinicians tend to be biased toward reductionistic interpretations of pain and suffering and often fail to attend to the broader and more difficult issues that may be of much greater importance to patients and families (43). The fact that there are not yet validated “suffering scales” does not diminish the importance of this dimension of the dying process.

Suffering may have profound mean- ings for patients that are unrelated to physical symptoms. Some patients, for example, may find redemptive meaning in their suffering and therefore may not want to avoid it entirely. By seeking to understand and appreciate these mean- ings, clinicians can individualize their care in ways that are responsive to these varying perspectives.

Nonpharmacologic Approaches to Pain and Suffering. “Dying in one’s sleep” has always been viewed as a natural way to depart from this life. There are many physiologic reasons to support this view. Respiratory depression during dy- ing may produce hypercarbia and hyp- oxia. Studies of alveolar anoxia suggest that the most rapid descent into uncon- sciousness with the least agitation occurs when hypoxia is allowed to progress in the face of normocarbia, a finding that could have relevance for approaches to ventilator withdrawal (see subsequent discussion) (44).

As cardiac activity decreases, hypoper- fusion will decrease cerebral function. Decreased oral intake will lead to dehy- dration and a similar decrease in cerebral function. “Starvation euphoria” is a rec- ognized phenomenon, possibly related to endogenous opioid production or the an-

Table 2. Possible physiologic consequences of forgoing specific therapies

System Intervention Effect of Withdrawal

Cardiovascular Vasopressors Vasodilation, hypotension (possible secondary tachycardia) Intra-aortic balloon pump Decreased coronary perfusion, decreased cardiac output Left ventricular assist device Decreased cardiac output Cardiac pacemaker Asystole, bradycardia, decreased cardiac output

Pulmonary Oxygen Hypoxia, possible sympathetic discharge and increased respiratory drive, followed by respiratory depression

Mechanical ventilation Hypercapnia, increased respiratory drive (brainstem), depressed consciousness

Positive end-expiratory pressure Decreased functional residual capacity, ventilation-perfusion mismatching, hypoxia

Extracorporeal membrane oxygenation and CO2 removal Hypoxia, hypercapnia, tachypnea, decreased cardiac output, tachycardia, bradycardia, asystole

Nitric oxide Pulmonary hypertension, hypoxia, decreased cardiac output Renal Dialysis Acidosis, uremia, fluid overload, hyperkalemia, lethargy, delirium Neurologic Cerebrospinal fluid drainage Increased intracranial pressure, leading to mechanical compression and

hypoperfusion of cerebral structures Nutritional Nutrition and hydration Lipolysis, ketosis, dehydration

2336 Crit Care Med 2001 Vol. 29, No. 12

algesic effects of ketosis (34). Table 2 summarizes the physiologic effects that accompany the foregoing of specific ther- apies and illustrates some of the ways that the withdrawal of treatments may actually contribute positively to the pa- tient’s comfort. Although these physio- logic effects probably contribute to the comfort of dying patients, they are not uniformly effective. Some may make the patient more uncomfortable before the patient’s consciousness diminishes. Ac- cordingly, these physiologic effects should be attenuated by other measures.

Environmental factors can play an im- portant role in promoting the patient’s comfort. As noted previously, there are pros and cons to having dying patients remain in the ICU. The advantages in- clude continuity of care and the greater availability of nurses and physicians. The benefits of leaving the ICU may include return to a more familiar (and possibly more private) setting, as well as less tech- nology and cost. In either location, much can be done to enhance the patient’s comfort, such as providing privacy and a comfortable bed, reducing lighting and noise, removing restraints, eliminating unnecessary monitors and machines, and providing the space and opportunity for interaction with the patient’s family and loved ones (45– 48). Beyond these simple measures, there may be cultural or spir- itual factors, such as the opportunity for ritual, prayer, or music, that can increase the patient’s comfort (49 –51).

Opioids. Opioids have been a mainstay for the treatment of pain and suffering in dying patients (Table 3). Opiates are �-re- ceptor agonists, and central �-receptors

invoke analgesia, sedation, respiratory depression, constipation, urinary reten- tion, nausea, and euphoria. Vasodilation may produce hypotension but also can have a therapeutic effect by decreasing venous return to the right heart, thereby decreasing filling pressures and relieving cardiogenic pulmonary edema. Practice parameters from the SCCM cite mor- phine as the preferred analgesic agent in the ICU, with hydromorphone and fenta- nyl as alternative agents (52).

Morphine is the most frequently used opioid analgesic in the United States, mainly because of its low cost, potency, analgesic efficacy, and euphoric effect. It has a half-life of 1.5–2 hrs in normal subjects after intravenous administra- tion, but the elimination half-life may be prolonged in patients with hepatic or re- nal dysfunction. Although allergic reac- tions to morphine have been reported, it is much more common for allergic symp- toms to be related to histamine release, especially when the medication is admin- istered rapidly (52).

Fentanyl is a synthetic opiate with 80 – 100 times the potency of morphine. Fen- tanyl does not cause histamine release, which may explain the reduced incidence of hypotension compared with morphine. It has less sedative and euphoric effects compared with morphine. It has a half- life of 30 – 60 mins because of rapid redis- tribution, but with prolonged administra- tion the elimination half-life increases to 9 –16 hrs, as the peripheral sites of redis- tribution become saturated. Because both fentanyl and morphine reach 90% of their peak effect within 5 mins of intra- venous administration, these medica-

tions can be safely redosed at 5-min in- tervals (53, 54). Hydromorphone is a semisynthetic morphine derivative, simi- lar to morphine but with more potent analgesic and sedative properties and sig- nificantly less euphoria (52).

SCCM practice parameters recom- mend against the routine use of meperi- dine. Normeperidine is an active metab- olite of meperidine that produces signs of central nervous system excitation such as apprehension, tremors, and/or seizures, especially in patients with renal insuffi- ciency (52). The Agency for Health Care Policy and Research has recommended that meperidine should not be used ex- cept for a brief course of treatment in otherwise healthy patients who have demonstrated an unusual reaction or al- lergic response to morphine (meperidine does not cross-react in morphine allergy) (55, 56).

When intravenous access is either not possible or not desired, alternative routes of administration should be considered, including oral, rectal, subcutaneous, and transdermal. Long-acting formulations of several opioids are also available. Because most patients dying in intensive care units have intravenous access, and be- cause these alternatives are extensively discussed in the palliative care literature, these other options for treatment are not reviewed here (57, 58).

Benzodiazepines. Benzodiazepines re- duce anxiety and cause amnesia, impor- tant in preventing recall or breakthrough suffering. In addition to having a desir- able synergistic sedative effect with opi- oids, benzodiazepines are anticonvul-

Table 3. Opioid analgesics

Medication Equianalgesic

Dosing, IV

Typical Starting Dose,

Adult, IV

Typical Starting Dose, Pediatric, IV

Duration, hrs

Typical Starting Infusion Rate Comments

Morphine 1 2–10 mg 0.1 mg/kg 3–4 0.05–0.1 mg�kg�1�hr�1 Histamine release (caution in asthma), vasodilation, hypotension

Hydromorphone 0.15 0.3–1.5 mg 3–4 Less pruritus, nausea, sedation, and euphoria than morphine

Fentanyl 0.01 50–100 �g 1–5 �g/kg 0.5–2.0 1–10 �g�kg�1�hr�1 Minimal hemodynamic effects, duration of action short when given by intermittent bolus, half- life prolonged when administered chronically

Meperidine 10 25–100 mg 1 mg/kg 2–4 Not recommended for chronic use; catastrophic interaction with MAO inhibitors; tachycardia; seizures

IV, intravenous; MAO, monoamine oxidase. From Refs. 52, 55, 56, 59, 60, 67, 127.

2337Crit Care Med 2001 Vol. 29, No. 12

sants and may help prevent the development of premorbid seizures.

Lorazepam is an intermediate-acting benzodiazepine that has a peak effect ap- proximately 30 mins after intravenous administration. In adults, elimination is not altered by renal or hepatic dysfunc- tion. The recommended starting dose is about 0.05 mg/kg every 2– 4 hrs when administered by intermittent bolus (52).

Midazolam is a short-acting benzodi- azepine. Because it is water soluble, it is not painful on peripheral injection. After intravenous administration, it undergoes a structural change to a lipophilic com- pound that rapidly penetrates the central nervous system and gives it an onset of action comparable to diazepam. It has a brief duration of action attributable to rapid redistribution, however, and ad- ministration by continuous infusion of- ten is required for the medication to have a sustained effect. Starting doses for adults are 1 mg intravenously or 1–5 mg/hr by continuous infusion. Starting doses for children are 0.1 mg/kg intrave- nously or 0.05– 0.10 mg·kg�1·hr�1 (52, 59 – 61).

Neuroleptics. Neuroleptics may be ef- fective when the patient is manifesting signs and symptoms of delirium. Delir- ium is an acute confusional state that can be difficult to differentiate from anxiety, yet the distinction is important, because the administration of opioids or benzodi- azepines as initial treatment for delirium can worsen the symptoms (52). Haloper- idol has proven efficacy in the manage- ment of delirium. Although the drug does not possess a significant sedative effect, patients whose delirium is ameliorated by haloperidol often require less sedation with other agents (52). In addition, in one study this agent was used at least occa- sionally as an adjunct to the discontinu- ation of life-sustaining measures by 24% of physicians (30).

Starting doses of haloperidol in adults range from 0.5 to 20 mg, depending on the severity of the patient’s delirium. Ad- ditional doses should be titrated at 30- min intervals until the patient’s symp- toms are controlled (62). Doses up to 50 or 60 mg may be required. Once delirium is controlled, patients often can be main- tained on 50% to 100% of this amount in divided doses over 24 hrs (52). Haloperi- dol also has been administered success- fully by continuous infusion, at doses ranging from 3 to 25 mg/hr (63).

Disadvantages of haloperidol include extrapyramidal symptoms, which are less

common when the drug is given intrave- nously as opposed to enterally. Extrapy- ramidal symptoms are more common in children, reducing the usefulness of this medication in the pediatric population (64).

Propofol. Propofol is a sedative and anesthetic agent that is attractive primar- ily because of its short half-life. In most studies of ICU sedation, it has had a com- parable effect to a continuous infusion of midazolam (52, 65). Low doses can be titrated to achieve varying planes of seda- tion or unconsciousness. A typical start- ing dose of propofol for both adults and children is 1 mg/kg, but some patients may become hypotensive with even this much, emphasizing the need to titrate to effect. When administered by infusion, a typical starting dose is 0.5 mg·kg�1·hr�1, with most patients requiring between 0.5 and 3.0 mg·kg�1·hr�1. The potential for drug incompatibility is a problem with propofol, because it requires that propo- fol be administered through a dedicated intravenous catheter. In addition, be- cause of the potential for contamination and infection, the manufacturer recom- mends that propofol infusion bottles and tubing be changed every 12 hrs and that solutions transferred from the original container be discarded every 6 hrs. Like diazepam, propofol is painful when ad- ministered via a peripheral vein (52).

Barbiturates. Barbiturates have both advantages and disadvantages when used at the end of life. Their disadvantages include an absence of analgesic effect, necessitating the concurrent administra- tion of analgesics (e.g., opioids) whenever the patient’s symptoms include pain. Bar- biturates also have been strongly linked to the practice of euthanasia, having been used for that purpose in the Netherlands and for the execution of prisoners by le- thal injection in the United States. Even when appropriately administered within existing guidelines, therefore, their use could be misinterpreted as the practice of euthanasia. Advantages of barbiturates include their ability to reliably and rap- idly cause unconsciousness, which may be necessary for the rare patient whose pain does not respond to any other ap- proach (66). In addition, because their mechanism of action differs from the opi- oids and benzodiazepines, they may be useful in patients who have developed extreme levels of tolerance to these other medications. On balance, although barbi- turates are very helpful in limited cir- cumstances, they are not in the first line

of medications that should be used in treating the terminally ill. Propofol offers many of the same advantages as the bar- biturates without the complicating fea- tures. A typical starting dose for pento- barbital, a barbiturate with a medium duration of action, is 150 mg intrave- nously for adults and 2– 6 mg/kg intrave- nously for children. For prolonged effect, the medication may be continued in doses of 3–5 mg·kg�1·hr�1. Because tol- erance develops rapidly, progressive esca- lation of the dose is often necessary (66, 67). These adjunctive agents are summa- rized in Table 4.

Principles for Dosing and Titration. Although starting doses for sedation and analgesia were discussed previously and included in the tables, in many cases these doses will be irrelevant, because most patients will have already received these agents and will have already devel- oped some tolerance to their effects at the time of withdrawal of life support. These agents should be titrated to effect, and the dose should not be limited solely on the basis of “recommended” or “suggest- ed” maximal doses. In most cases, pa- tients who do not respond to a given dose of an opioid or benzodiazepine will re- spond if the dose is increased—there is no theoretical or practical maximal dose. In rare cases, this generalization does not hold; in these patients, alternative classes of agents (like barbiturates or propofol) should be considered.

Current ethical and legal guidelines place importance on the intentions of cli- nicians in administering analgesics and sedatives at the end of life. Specifically, clinicians should administer doses that are intended to relieve pain and suffering but not intended to directly cause death. Because intentions are essentially subjec- tive and private, the only ways to infer the nature of an individual’s intentions are by self-report and by an analysis of his or her actions. Accordingly, documentation of one’s intentions in the patient’s chart is an important part of providing end-of-life care. When “p.r.n.” orders are written for analgesics and sedatives, the indication for administration should be stated clearly (e.g., pain, anxiety, shortness of breath). This reduces the likelihood of misinterpretation or abuse. With regard to actions, when a clinician titrates mor- phine in doses of 1, 5, or 10 mg every 10 or 20 mins, it is plausible to conclude that the clinician intends to make the patient comfortable and not to directly cause the patient’s death. On the other

2338 Crit Care Med 2001 Vol. 29, No. 12

hand, when a clinician administers 2 g of morphine acutely to a patient who is not profoundly tolerant, it is difficult not to conclude that the clinician did intend the death of the patient.

The concept of “anticipatory dosing” (as opposed to reactive dosing) also should guide clinicians in the use of se- dation and analgesia at the end of life. The rapid withdrawal of mechanical ven- tilation is an example of the need for anticipatory dosing. At the time of venti- lator withdrawal, the clinician can antic- ipate that there will be a sudden increase in dyspnea. It is not sufficient simply to respond to this distress with titrated doses of an opioid (reactive dosing). Rather, clinicians should anticipate this sudden event and provide adequate med- ication beforehand (anticipatory dosing). As a general rule, the doses of medication that the patient has been receiving hourly should be increased by two- or three-fold and administered acutely before with- drawing mechanical ventilation.

There are some data on the use of sedatives and opioids during the with- drawal of life support. In one study, non- comatose adult patients received analge- sia and sedation during withdrawal of life support, with an increase in benzodiaz- epine from a dose equivalent to 2.2 mg/hr of diazepam to 9.8 mg/hr and an increase in opioid from a dose equivalent to 3.3 mg/hr of morphine to 11.2 mg/hr at the time that life support was withdrawn (68). A retrospective study of three adult ICUs found that large doses of morphine (mean, 21 � 33 mg/hr) and benzodiaz- epines (equivalent to a mean diazepam dose of 8.6 � 11 mg/hr) were given dur-

ing the withdrawal of life support (69). A similar study performed in pediatric ICUs found an increase in diazepam equiva- lents from 0.26 to 0.68 mg·kg�1·hr�1 and an increase in morphine equivalents from 0.54 to 1.80 mg·kg�1·hr�1 during the withdrawal of ventilator support (70). In addition, a review of 121 neonatal deaths reported that most patients (84%) re- ceived analgesia as their life support was withdrawn, and that most of these pa- tients (64%) could be managed with doses of morphine in the usual pharma- cologic range (0.1– 0.2 mg/kg intrave- nously). Infants who were tolerant to morphine required larger doses, up to 1 mg/kg intravenously. Of particular note, there was no relationship between the dose of morphine used and the time until death after ventilator withdrawal (42).

Alleviation of Specific Symptoms. Campbell (29) called attention to many of the specific symptoms that may be expe- rienced by terminally ill patients. Dys- pnea is a form of suffering and is probably the most important symptom that must be relieved for patients dying in the ICU. The incidence of this problem is not well described, but data suggest that it is present in up to half of dying persons (29). Although dyspnea in patients dying of respiratory failure is almost always at- tributable to progression of their under- lying disease, clinicians should remem- ber that the differential diagnosis for dyspnea is extensive and includes many potentially treatable conditions such as reactive airway disease, infection, pneu- mothorax, congestive heart failure, and anxiety. The response to this sensation is both physiologic (e.g., tachycardia, tachy-

pnea) and psychological (e.g., panic, anx- iety, fear). Assessment should include an investigation for potentially treatable causes before focusing on symptom man- agement. Symptom severity scales, such as the modified Borg dyspnea scale and the Bizek agitation scale, can be used to assess symptoms associated with breath- lessness (29, 71–73).

Treatment of dyspnea may include pharmacologic and nonpharmacologic strategies. Simple positioning may be ef- fective. Patients with chronic obstructive pulmonary disease may be most comfort- able sitting up or leaning over a bedside table. Patients with unilateral lung dis- ease (e.g., pneumonia) may prefer lying on one side more than the other.

Pharmacologic approaches to dyspnea are varied. Oxygen may enhance patient comfort by relieving hypoxemia (74). However, one study of advanced cancer patients reported that oxygen was no bet- ter than air in relieving dyspnea (75). Sometimes patients experience symp- tomatic relief by having air from a fan blowing gently on their face and may have increased dyspnea from a feeling of claustrophobia associated with the ad- ministration of oxygen by a facemask. Opioids relieve dyspnea by depressing re- spiratory drive, producing sedation and euphoria, and causing vasodilation, which can reduce pulmonary vascular congestion. Patients also may benefit from the judicious use of bronchodilators and diuretics to relieve small airway ob- struction and pulmonary vascular con- gestion.

Nausea and vomiting are frequently reported at the end of life. As with dys-

Table 4. Adjunctive agents

Medication

Typical Starting Dose,

Adult, IV

Typical Starting Dose, Pediatric, IV

Duration, hrs

Typical Starting Infusion Rate, Adult

Typical Starting Infusion Rate, Pediatric Comments

Lorazepam 1–3 mg 0.05 mg/kg 2–4 0.025–0.05 mg�kg�1�hr�1 0.05–0.1 mg�kg�1�hr�1 Longer acting, ideal for long- term administration

Midazolam 1 mg 0.1 mg/kg 1.5–2 1–5 mg/hr 0.05–0.1 mg�kg�1�hr�1 Well tolerated but fairly expensive

Haloperidol 0.5–20 mg 2–4 3–5 mg/hr IV Not often used in pediatrics because extrapyramidal effects more frequent

Propofol 1 mg/kg 1 mg/kg 10–15min 0.5–3.0 mg�kg�1�hr�1 0.5–3.0 mg�kg�1�hr�1 Hypotension, lipid base lead to hyperlipidemia, painful on injection

Pentobarbital 150 mg 2–6 mg/kg 2–4 3–5 mg�kg�1�hr�1 3–5 mg�kg�1�hr�1 Propofol should replace pentobarbital in most end- of-life situations

IV, intravenous. From Refs. 52, 59 – 65, 71.

2339Crit Care Med 2001 Vol. 29, No. 12

pnea, potentially treatable causes should be investigated before resorting to symp- tomatic management. Most nausea and vomiting can be controlled with anti- emetic agents. Although nasogastric drainage is sometimes effective for relief from profound ileus or small bowel ob- struction, it may be more uncomfortable for the patient than occasional emesis.

Hunger and thirst are problematic concerns at the end of life. Some believe that the dying should always be given food and fluids and that this is a basic expression of our humanity and capacity for compassion (see “minority opinion” in Ref. 11). On this view, some caregivers believe that hunger and thirst should al- ways be treated and encourage placement of nasogastric or gastrostomy tubes in terminally ill patients to administer nu- trition when patients are no longer capa- ble of oral sustenance. Current palliative care practices, however, recognize that loss of hunger and thirst are normal physiologic responses to the dying pro- cess, and that forced nutrition and hydra- tion in this setting not only prolong the dying process but do not contribute to the patient’s comfort (76 –78). In addi- tion, the metabolic abnormalities associ- ated with dehydration tend to contribute to sedation and diminished conscious- ness rather than cause distress (76, 79). Although the symbolism associated with providing food and fluid should not be dismissed lightly, the majority view in the United States now holds that food and fluid should be provided if they are de- sired by the patient and contribute to the patient’s comfort; otherwise, they may be foregone (78, 80).

Skin ulceration may be caused by local tissue conditions, infection, or ischemia from hypoperfusion and localized pres- sure or edema. Even the best skin care regimens are unlikely to promote healing under these conditions. The frequent turning and dressing changes that are required can cause more pain and dis- comfort than benefit. Attention to keep- ing the patient clean, dry, and free from odor may be the best goal under some circumstances.

Fevers and infections frequently occur in critically ill and dying patients. Be- cause fever can be quite uncomfortable, antipyretics generally should be used. Ex- ternal cooling with ice packs, cooling blankets, or alcohol baths may create greater distress for the patient than the fever itself. Antibiotics may offer more benefit than burden for painful infec-

tions, such as otitis media, oral candidi- asis, or herpetic infections.

Anxiety and delirium often occur at the end of life. The use of physical re- straints should be avoided whenever pos- sible. Pharmacologic management should be gauged more toward the pa- tient’s comfort and peacefulness rather than toward resolution of the delirium.

Withdrawal of Life-Sustaining Treatments

The indications for any proposed in- tervention in a dying patient should be assessed in terms of the goals of the pa- tient. Any intervention that does not ad- vance the patient’s goals should be elim- inated. This simple advice is persuasive in concept yet difficult to follow. In reality, physicians have many biases and prefer- ences regarding the withdrawal of life- sustaining therapies that do not seem to be related to the needs or values of the patient. For example, a 1992 survey of SCCM physicians found that 15% almost never withdraw mechanical ventilation and that internists and pediatricians were more likely to withdraw mechanical ven- tilation than surgeons or anesthesiolo- gists (30). Unless these differences were attributable to underlying systematic dif- ferences in the patient populations they cared for, the origins of these variations in practice must rest primarily with the preferences of the physicians themselves (81).

Some of these preferences are related to culture and religious beliefs. Some Jewish clinicians, for example, have reli- gious reasons for believing that the with- drawal of life-sustaining treatments is “killing” and therefore is prohibited (4). In addition to these differences based on culture or religion, Christakis and Asch (82) reported that physicians prefer to withdraw therapy supporting organs that failed for natural vs. iatrogenic reasons, to withdraw recently instituted vs. long- standing interventions, to withdraw ther- apies leading to immediate death rather than delayed death, but to withdraw ther- apies leading to delayed death when faced with diagnostic uncertainty (82). There were also patterns in the preferences of physicians for the order in which treat- ments were withdrawn: first being blood products, followed by hemodialysis, vaso- pressors, mechanical ventilation, total parenteral nutrition, antibiotics, intrave- nous fluids, and finally tube feedings. There was an underlying trend toward

earlier withdrawal of treatments per- ceived as more artificial, scarce, or expen- sive (82– 84). Specialists have also been reported to prefer to withdraw the ther- apy with which they are most familiar; for example, pulmonologists withdraw me- chanical ventilation, nephrologists with- draw dialysis, and so forth (85). Decisions in pediatrics are also stereotyped, with deaths in most series almost always fol- lowing the withholding or withdrawal of either mechanical ventilation or extra- corporeal membrane oxygenation (86, 87).

In light of these (perhaps uncon- scious) biases, it is useful to review the wide range of life-sustaining treatments that are used in critical care medicine and to work toward an approach that is less centered on physician preferences and more focused on the unique situation and needs of the patient. Table 5 catalogs the types of life-sustaining treatments that may be withdrawn and illustrates the range of therapies that may be foregone, from measuring and recording vital signs to extracorporeal membrane oxygen- ation.

Terminal Extubation vs. Terminal Wean

Grenvik (88) was the first to describe a systematic approach to ventilator with- drawal at the end of life and advocated a gradual reduction in the ventilator set- tings over several hours. Since then, there has been an ongoing debate regard- ing the best method of withdrawing me- chanical ventilation. Although the early literature recommended blood gas mon- itoring during the withdrawal of ventila- tion, virtually all now agree that neither this nor noninvasive forms of respiratory monitoring are consistent with the pal- liative goals of promoting the patient’s comfort and reducing technology when- ever possible.

One recommended approach, com- monly referred to as “terminal extuba- tion,” involves removal of the endotra- cheal tube, usually after the administration of boluses of sedatives and/or analgesics. The second technique, known as a “terminal wean,” is performed by gradually reducing the FIO2 and/or the mandatory ventilator rate, leading to the progressive development of hypoxemia and hypercarbia. In the latter technique there is considerably variability in the pace of the process, with some complet- ing the wean over several minutes (19,

2340 Crit Care Med 2001 Vol. 29, No. 12

89 –91) and others stretching it over sev- eral days (92).

The preferred approach varies widely. A 1992 survey of SCCM physicians found that 33% preferred terminal weaning, 13% preferred extubation, and the re- mainder used both. These preferences were correlated with specialty: Surgeons and anesthesiologists were more likely to use terminal weaning, whereas internists and pediatricians were more likely to use extubation (p � .0001) (30).

The principle advantage of the termi- nal wean is that patients do not develop any signs of upper airway obstruction during the withdrawal of ventilation. They therefore do not develop distress from either stridor or oral secretions, and if the wean is performed slowly with the

administration of sedatives and analge- sics, they do not develop symptoms of acute air hunger. These advantages not only promote the comfort of the patient but reduce the anxiety of family and care- givers (93).

Another cited advantage of terminal weans is that they are perceived to dimin- ish the moral burden of the family and caregivers, presumably because the ter- minal wean is perceived as being less “active” than terminal extubation (30). Whether this is an advantage or disadvan- tage remains controversial. There is a risk that terminal weans may be per- ceived by families as bona fide attempts to have the patient successfully survive separation from the ventilator, even when this is not the expectation or intent

of the clinicians—particularly when the wean is prolonged over several days. Ter- minal weans therefore should not be adopted as a means of avoiding difficult conversations with families about the pa- tient’s condition and prognosis.

In contrast to terminal weans, termi- nal extubations have the principal advan- tages that they do not prolong the dying process and that they allow the patient to be free from an “unnatural” endotracheal tube (94). The process of terminal extu- bation also is morally transparent; the intentions of the clinicians are clear, and the process cannot be confused with a therapeutic wean (30).

Although these two concepts have be- come fairly well entrenched into the lex- icon of critical care medicine, we believe that the terminology of terminal weans and terminal extubations is confusing and should be replaced by more specific descriptions of the process. The use of the word terminal suggests that withdrawal will directly result in death of the patient. Occasionally, however, patients who are separated from the ventilator with the expectation of failure survive to be dis- charged from the intensive care unit or the hospital (95). Weaning generally re- fers to a therapeutic procedure that oc- curs when patients are improving and expected to survive. It may be unclear whether the process includes removal of the artificial airway, supplemental oxy- gen, or positive pressure ventilation. We believe it is preferable to use specific terms and to consider each of these ther- apies separately. An artificial airway may be removed (extubation), the patient may have supplemental oxygen discontinued, and/or positive pressure ventilation may be reduced or eliminated. These ap- proaches are not mutually exclusive. For example, withdrawal of the artificial air- way may occur simultaneously with the withdrawal of oxygenation and ventila- tion (terminal extubation). Ventilation and oxygenation also may be withdrawn rapidly (by transitioning to a T-piece) or slowly (by gradually reducing the FIO2 and/or ventilator rate). Then, as the pa- tient’s pharmacologic sedation is supple- mented by the effects of hypoventilation and hypoxia, the artificial airway may be withdrawn. It is conceivable that each therapy (artificial airway, supplemental oxygenation, and mechanical ventilation) may be continued or eliminated, depend- ing on the specific circumstances of the patient. In this way, decisions can be made more specifically and deliberately

Table 5. Treatments that can be withheld or withdrawn

Therapeutic Goal Therapy

Circulatory homeostasis Cardiopulmonary resuscitation Vasopressors and inotropic medication Antihypertensive medication External ventricular assist/replacement device Implantable ventricular assist/replacement device Pacemaker Implantable cardiac defibrillator Intra-aortic balloon counterpulsation Transfusion of blood products, albumin Intravenous crystalloid administration Invasive pressure monitoring

Respiratory homeostasis Mechanical ventilation Supplemental oxygen Artificial airway (endotracheal tube, tracheostomy tube, oral-

pharyngeal airway) Extra-corporeal membrane oxygenation or CO2 elimination Diaphragmatic pacing

Renal homeostasis Hemodialysis (continuous or intermittent) Hemofiltration Peritoneal dialysis

Neurologic homeostasis Cerebrospinal fluid drainage (may be palliative) Intracranial pressure monitoring Steroids, mannitol, hyperventilation Anticonvulsants (probably would continue for palliative reasons)

Endocrinologic homeostasis Steroids (may be palliative) Hormone supplementation or suppression (may be palliative)

Treatment of infection, inflammation, or neoplasm

Antibiotic, antifungal, antiparasitic, antiviral medications (may be palliative)

Anti-inflammatory medications (may be palliative) Immune “booster” medications Cytotoxic medication (may be palliative) Radiation therapy (may be palliative)

Nutritional homeostasis Total parenteral nutrition Enteral feeding via gastric or jejunal tube Intravenous dextrose

“Routine” measures Frequent phlebotomy for laboratory tests Frequent vital sign measurements Radiologic examinations Aggressive chest physiotherapy and endotracheal suctioning Placement of intravenous and intra-arterial lines Debridement of wounds

2341Crit Care Med 2001 Vol. 29, No. 12

than when the choices are only between terminal wean and terminal extubation.

Finally, the method of withdrawal has important implications for the adminis- tration of sedation and analgesia. Abrupt changes in the patient’s level of distress require the administration of anticipatory doses of analgesics and sedatives. If the decision is made to rapidly withdraw the artificial airway (extubation) or mechan- ical ventilation (transition to T-piece), for example, the patient generally should re- ceive medication before the withdrawal in anticipation of distress, with subsequent doses titrated to the patient’s level of comfort.

Withdrawal Prototypes

No two instances of the withdrawal of life support are ever identical, yet certain prototypes have a number of features in common. They depend on the clinical characteristics of the patient and the type of life support that is being withdrawn. These were discussed in more detail by Campbell (29).

Ventilator Withdrawal from Patients Declared Brain Dead. Patients who have been declared brain dead are dead. Re- moval of the ventilator is not the with- drawal of life support, because the venti- lator is not supporting life. The most straightforward approach to withdrawal of the ventilator in these circumstances is rapid removal of the artificial airway, ox- ygenation, and ventilation.

Clinicians should be aware, however, that brain dead patients may rarely ex- hibit dramatic movements, caused by the firing of spinal motor neurons, that are known as the Lazarus sign (96, 97). Such movements generally occur either during the apnea test or after the withdrawal of mechanical ventilation and are thought to be related to acute effects of hypoxia or ischemia on spinal motor neurons. The movements can be as extensive and com- plex as the patient sitting up in bed. Be- cause current brain death criteria do not require the loss of all spinal activity, these movements do not exclude the di- agnosis of brain death. If the patient’s family is to be at the bedside during ei- ther the apnea test or the withdrawal of mechanical ventilation, it is imperative that the clinicians prepare them for what they might see, so as not to alarm them with the fear that the diagnosis of brain death might have been in error.

Ventilator Withdrawal from Uncon- scious Patients Unlikely to Experience

Distress. This prototype includes patients who are comatose but who are not brain dead. Although patients who are truly comatose are not capable of experiencing anything, in some cases there may be doubt about whether the patient has any rudimentary capacity for experiencing pain or suffering. In these cases, clini- cians should err on the side of caution and provide an appropriate level of anal- gesia and sedation.

Withdrawal of life support usually can proceed rapidly in such cases, either by withdrawal of the artificial airway or by removing the mechanical ventilator. In either case, the patient may require an- ticipatory dosing with analgesics and/or sedatives and may require additional medication administered as necessary, ti- trated to the observed level of the pa- tient’s distress. Because some uncon- scious patients will not require the administration of any additional sedatives or analgesics, however, these should be given on an individualized basis accord- ing to need rather than dosed according to protocol (19).

Ventilator Withdrawal from the Con- scious or Semiconscious Patient Likely to Experience Distress. This prototype in- cludes patients who are definitely able to experience suffering, and the method of withdrawal needs to be tailored to mini- mize distress. In most cases, this will involve a more gradual withdrawal of both ventilator rate and supplemental ox- ygen. Although there is indirect evidence that patients may be more comfortable when supplemental oxygen is removed before ventilator rate (44), there are no clinical studies to support this approach. In any case, the gradual withdrawal of ventilator support allows clinicians the opportunity to carefully titrate sedatives and analgesics to the patient’s level of comfort, thereby ensuring that the pa- tient does not experience any treatable pain or suffering. Once the patient has lost consciousness from the combined ef- fect of the medications and hypoxia, then the artificial airway can be removed.

In some cases, such as those involving patients with cervical quadriplegia or those undergoing advanced life support, the patient may prefer the rapid with- drawal of ventilation while sedated to a sufficient depth to eliminate any possibil- ity of dyspnea or air hunger. This ap- proach is also acceptable but requires very close attention to the adequacy of the anticipatory dosing to make sure that the patient does not experience acute suf-

fering at the time of ventilator with- drawal. One technique for ensuring this is to use rapidly acting medications such as thiopental or propofol in sufficient doses to relieve the patient’s suffering (66).

Special Issues in Communicating with Families Near the Time of Death

Notification of Death. Breaking bad news is one of the most difficult tasks that physicians face but is a common necessity in the practice of critical care medicine. Little empirical research on this topic exists to ground recommenda- tions, however, and most suggestions are therefore based primarily on common sense, experience, and intuition. These deficiencies may explain in part why few clinicians have received formal training in how to deliver bad news. Even so, certain principles can be recommended (98 –102). Bad news should be delivered in person, whenever possible. The ideal location is in a private room that has seating available for everyone. Clinicians should be attentive to their appearance, especially if they appear disheveled from performing a resuscitation or other work in the ICU. They should learn how to demonstrate compassion and empathy, by beginning with words of condolence, maintaining eye contact, and extending a comforting touch when appropriate. Al- though well-intended, clichés like “He’s at peace now,” or “At least she lived a long and happy life” should be avoided, because these are often not well received and can be seen as offensive.

Clinicians often inadvertently use un- familiar jargon when talking with pa- tients and families. Words such as code, CPR, and vent should be avoided in favor of more clearly understood terms such as heart stopped, tried to start the heart, and breathing machine. In particular, cli- nicians should not be afraid to use the words died and death; saying only that resuscitation was unsuccessful or that the patient expired will often risk misun- derstanding (29). Development of these “bilingual” skills should be a priority for critical care clinicians.

The family frequently must be con- tacted by telephone if they are not present at the time of death. A Gallup poll of a sample of the U.S. adult population reported that when death of a family member was unexpected, most (64%) preferred to be told that the patient was

2342 Crit Care Med 2001 Vol. 29, No. 12

critically ill and to come to the hospital immediately (103). Only 26% preferred to be told over the telephone that the pa- tient had died. These findings were mir- rored in a companion survey of physician practices, which found that 72% of the physicians preferred to defer informing the family of the patient’s death until the family arrived at the hospital, whereas only 25% would relay the information immediately over the telephone. These preferences changed dramatically, how- ever, when the death of the patient was perceived as “expected.” In these circum- stances, only 13% of physicians would delay notification until the family’s ar- rival, with 83% informing the family di- rectly.

When the patient has been declared dead by neurologic criteria (“brain dead”), clinicians must be particularly careful with their words so as not to con- fuse the family. One of the most common mistakes is to say something like, “We have diagnosed your son as brain dead. He will die very quickly after he is re- moved from the ventilator.” Patients are declared dead at the time that the re- quirements for brain death are met. This is the time that should appear on the death certificate as the time of death. Removal of the ventilator at a later time should be seen as the removal of unnec- essary machines from a corpse. Although clinicians should be compassionate in the language that they use, they must take care to deliver an accurate and consistent message to the family and emphasize that bodily functions dependent on the brain are being artificially supported and will cease as soon as the machines are stopped. For example, a family could be told, “We tested your son and unfortu- nately we found that none of his brain is working. That means he is dead. He passed away at 6 o’clock.”

Permission for Autopsy. Physicians may sometimes have the opportunity to discuss the option of an autopsy with the patient or family before death, particu- larly in situations where death is expected and the patient or family has had an op- portunity to reflect on their wishes be- forehand. In most cases, however, discus- sions about autopsy occur within a short time after the patient’s death. Because this may coincide with the height of the family’s grief, many families may be un- able to cope with the complicated factors that must be considered in making this decision. This problem is compounded by the fact that education about the autopsy

procedure is perceived as inadequate in many residency programs (104), creating the risk of misinforming the family about the nature of the autopsy and possible alternatives. One frequent misconception is that the organs (or most of the organs) are customarily returned to the body af- ter they are examined. Another is that a limited autopsy (percutaneous biopsies or examination of a single organ, for exam- ple) is generally an acceptable substitute for a complete autopsy. Even although modern imaging and diagnostic tools have increased the accuracy of premor- tem diagnosis, complete autopsies con- tinue to provide answers to unresolved clinical questions and frequently reveal major unexpected factors that contrib- uted to the patient’s death (105).

Clinicians must be aware of local reg- ulations that require notification of the medical examiner after death. When re- quired, the medical examiner has author- ity to perform an autopsy without per- mission from the family. Clinicians should strive to maintain a supportive relationship with the family by emphasiz- ing the importance and necessity of med- icolegal examinations and that the clini- cal team typically has no influence over the medical examiner’s decision. Medical examiners may take religious reasons for opposing an autopsy into account in reaching their decision, but in most ju- risdictions they are under no obligation to do so. The medical examiner may not reach a decision concerning an autopsy until several hours after a patient’s death. Families should be informed that an eval- uation by the medical examiner’s office is pending so that they will not be surprised if the medical examiner chooses to per- form the autopsy. This is especially im- portant if the family would otherwise decide against having an autopsy per- formed, because they could feel betrayed if they believed that their wishes were being arbitrarily disregarded. A clinician might say, for example, “We will do ev- erything possible to respect your wishes regarding an autopsy, but you should know that the medical examiner is autho- rized by law to perform an autopsy, if he or she believes it is important for legal purposes.”

Organ Donation. Current federal reg- ulations require all institutions receiving Medicare or Medicaid funds to have the appropriate individual ask the family of every deceased patient for permission to procure tissues and organs (106). This discussion should occur separately from

notification of the patient’s death, and Health Care Financing Administration regulations now require that the request be made by someone specially trained in asking for organ and tissue donation. Critical care practitioners who are inter- ested in making these requests should therefore receive special training. Re- cently these federal regulations have been revised so that institutions are now re- quired to contact the local organ pro- curement organization concerning any death or impending death. When appro- priate, the organ procurement organiza- tion then sends a representative to the hospital to ensure that the family will be approached at the appropriate time by a professional skilled in presenting the op- tion of organ donation and in accurately answering the family’s questions and ad- dressing their concerns. Studies have documented that this approach enhances the likelihood that families will be asked to donate and might increase the chance that they choose to donate (107).

Although families of patients who have been declared brain dead commonly are asked to grant permission for organ donation, patients declared dead by car- diopulmonary criteria (so-called non- heart-beating organ donors) can also sometimes be suitable donors. Non- heart-beating cadavers have always been possible donors of skin, bone, corneas, and heart valves, but recent protocols have expanded the opportunities for some of these patients to donate kidneys, livers, and rarely even lungs and hearts. These solid organ procurements are performed under protocols that call for life-sustain- ing treatments to be withdrawn (usually mechanical ventilation) under controlled conditions (usually in the operating room), with death declared by cardiac criteria following 2–5 mins of pulseless- ness. Alternatively, non-heart-beating or- gan donation can proceed after a failed attempt at resuscitation. The solid organs then are either removed immediately or preserved in situ by infusing cold organ preservation solution through vascular cannulae before removal. This approach requires strict adherence to many ethical and technical details, and the procedure should never be performed on an ad hoc basis without a prospectively developed institutional protocol (108, 109).

Attending Funerals. Opinions about whether clinicians should attend funerals vary widely. Although it would be quite impractical for an intensive care clinician to attend funerals of patients regularly,

2343Crit Care Med 2001 Vol. 29, No. 12

attendance may be welcome and appro- priate when there has been a long- standing relationship between the clini- cian and the patient or family. Even when there has only been a brief opportunity for the clinicians to become acquainted with the patient or family, family mem- bers may feel a profound attachment to the ICU clinicians, perhaps because of the intensity of the ICU experience. Atten- dance at the funeral in these circum- stances may be highly valued by the fam- ily and could permit the clinician to release some of the grief and loss that is a part of working with critically ill and dy- ing patients. Striking a balance between the need to maintain a healthy emotional distance from patients and families and yet avoiding a destructive emotional de- tachment is a difficult yet important chal- lenge for ICU clinicians.

Bereavement Programs. The responsi- bilities of intensive care do not end when the patient is taken to the morgue. In addition to the issues about autopsy and organ donation outlined previously, fam- ilies may need assistance with choosing a funeral home and with making prelimi- nary arrangements for the disposition of the body. If a family has consented to an autopsy, the ICU should ensure that a physician (e.g., an intensivist, a subspe- cialist, or a primary physician) will notify the family and offer to meet with them as soon as results are available. By explicitly delegating this task to a specific clinician, the chances are reduced that this impor- tant follow-up will be overlooked. Spe- cific processes should be in place to en- sure rapid response to spiritual and psychological needs, as required by the Joint Commission on Accreditation of Healthcare Organizations. Bereavement programs can be structured to provide follow-up cards or notes to the family at set intervals (usually including the first anniversary) and can include sympathetic comments from nurses and doctors who were involved in the patient’s care. Sup- plemental information such as booklets or bibliographies to provide guidance and contact with support groups also can be provided (110, 111).

Special Ethical Issues

Terminal Sedation. Terminal sedation is a term that has been used to describe the practice of sedating patients to the point of unconsciousness, as a last resort and when all other methods of control- ling their suffering have failed. Typically,

either benzodiazepines or barbiturates are used as sedatives, although propofol could also be useful for this purpose (112). Once unconscious, patients typi- cally die of dehydration, starvation, or a complication of the treatment, with death usually occurring within several days (66, 113, 114).

This approach rarely is needed in the ICU environment, where patients sedated to the point of unconsciousness are gen- erally dependent on mechanical ventila- tion, with death following the withdrawal of that life-sustaining therapy. Occasion- ally, however, ICU patients who are not receiving mechanical ventilation will re- quire escalation of analgesics and seda- tives to the point of unconsciousness.

Some have argued that terminal seda- tion is merely a covert form of euthana- sia. Once the patient is unconscious, gen- erally no attempt is made to restore the patient to consciousness, and medical nutrition and hydration are terminated. Others have defended terminal sedation under the rule of double effect (115). In addition, the U.S. Supreme Court implic- itly endorsed the practice in two recent decisions concerning physician-assisted suicide, citing the technique as an alter- native to physician-assisted suicide that could ensure, at least theoretically, that no patient should die with “untreatable” pain. At least in part because of this legal endorsement, terminal sedation has be- come more widely practiced, although it remains controversial (116 –120).

Treating the Patient vs. Treating the Family. A standard principle in bioethics is that physicians should consider only the patient’s best interests and defend those interests against the potentially competing demands of third parties. This view may be a bit naïve. The interests of patients almost always are interwoven with those of family members and other loved ones, and physicians are often in the position of choosing which interests should prevail. This should not be sur- prising when one considers that family members make sacrifices for one another daily in everyday life; why should it be any different when it comes to making medical decisions? This tendency is espe- cially prominent in pediatrics, where pe- diatricians commonly see their role as “treating the family,” placing the best in- terests of the child within the context of the family’s resources and needs.

Attitudes about the proper role of the family’s interests vary widely. Some view the family’s wishes primarily as a conflict

of interest that needs to be blocked. Oth- ers allow the families’ wishes to enter into decision-making only with the ex- plicit permission of the patient, whereas others see the patients’ interests as being interdependent with those of the family and at times legitimately overridden by the needs of these others.

These issues take on a special signifi- cance at the end of life. Because the in- terests of the patient may be perceived as greatly diminished at this time, clinicians may be more likely to consider the needs of the family as more important. Con- sider, for example, the question of whether to perform a tracheostomy and initiate chronic ventilation for a severely demented elderly man who is primarily cared for by his daughter. Perhaps in this circumstance the needs and wishes of the daughter and her family should be con- sidered along with the best interests of the patient.

Similar issues arise in the use of sed- atives and analgesics at the end of life. Consider a patient who is near death and having “agonal” respirations. The family finds these very distressing, despite reas- surances from the clinicians that the pa- tient is unconscious and not experiencing any pain or suffering. Should the physi- cian administer additional opioid to the patient, with the intention of making the patient appear more peaceful for the ben- efit of the family? Both of these examples present relatively common dilemmas that are not well addressed by the standard principles and paradigms that currently exist in bioethics.

The Pharmacologically Paralyzed Pa- tient. Neuromuscular blocking agents (NMBAs) are required occasionally for the management of critically ill patients, pri- marily to facilitate the use of nonphysi- ologic ventilatory modes such as inverse- ratio ventilation and high-frequency oscillation. When a decision is made to withdraw ventilator support from a pa- tient who is paralyzed by these agents, there is a question as to whether the effects of the medication need to be re- versed or allowed to wear off before the ventilator is withdrawn.

This dilemma is not infrequent. For example, three of 33 patients (9%) in one study continued to receive NMBAs during the withdrawal of life support (68). One survey of physician members of SCCM reported that 6% have used NMBAs at the end of life at least occasionally (30), whereas another survey of pediatric in- tensive care specialists in the United

2344 Crit Care Med 2001 Vol. 29, No. 12

Kingdom reported that 12% would con- tinue NMBAs during ventilator with- drawal (121).

NMBAs possess no sedative or analge- sic activity and can provide no comfort to the patient when they are administered at the time of withdrawal of life support. Clinicians cannot plausibly maintain that their intention in administering these agents in these circumstances is to ben- efit the patient. Indeed, unless the patient is also treated with adequate sedation and analgesia, the NMBAs may mask the signs of acute air hunger associated with ven- tilator withdrawal, leaving the patient to endure the agony of suffocation in silence and isolation. Although it is true that families may be distressed while observ- ing a dying family member, the best way to relieve their suffering is by reassuring them of the patient’s comfort through the use of adequate sedation and analge- sia.

The same considerations apply to most patients who are receiving NMBAs at the time that the decision to withdraw life support is made. In most cases, the effect of these agents can be reversed or allowed to wear off within a short period of time, allowing for the withdrawal of mechanical ventilation in the absence of the confounding effects of paralysis. As a general rule, therefore, pharmacologic paralysis should be avoided at the end of life.

Patients who have been receiving NMBAs chronically for management of their ventilatory failure occasionally can present a more difficult ethical dilemma. In some situations, restoration of neuro- muscular function may not be possible for several days or even weeks, because of relative overdosage of the drug or the accumulation of active metabolites (122). When faced with this problem, the clini- cian must choose between withdrawal of the ventilator while the patient is para- lyzed vs. continuation of life support well beyond the point at which the patient and family have determined that the burdens of such treatments outweigh the probable benefits. In this circumstance, it may be preferable to proceed with withdrawal of life support despite the continued pres- ence of neuromuscular blockade. This recommendation is in accord with others who have commented on this issue (34, 123–125).

Before proceeding with the with- drawal of life support from a patient who is pharmacologically paralyzed, several issues must be carefully considered. First,

the clinicians must be quite certain that the patient is truly dependent on ventila- tor support for survival. This is not al- ways easy to do— 8% of “terminally weaned” patients from one study survived to hospital discharge (93). If there is a small but significant chance that the pa- tient could survive separation from the ventilator in the absence of the neuro- muscular blockade, then the effects of the blockade must be eliminated before ven- tilator withdrawal.

Second, clinicians must be aware that neuromuscular blockade will signifi- cantly impair their ability to assess the patient’s comfort. Paralyzed patients are unable to communicate any evidence of discomfort or distress during the process of withdrawal of life support. Autonomic signs such as hypertension and tachycar- dia are highly unreliable. The onus is on the clinicians to use medications in dos- ages sufficient to ensure the patient’s comfort despite the absence of the usual behavioral clues to the patient’s level of distress. This is certainly possible (it is done routinely by anesthesiologists car- ing for pharmacologically paralyzed pa- tients during anesthesia and surgery), but it does require sufficient knowledge, skill, and experience on the part of the ICU clinicians.

Third, clinicians must balance the costs of waiting until the NMBAs can be reversed or wear off against the potential benefits. In addition to removing uncer- tainty about the prognosis and ensuring the availability of behavioral clues about the patient’s comfort, waiting until neu- romuscular function can be restored has the theoretical benefit of allowing the pa- tient to interact with family members and other loved ones both before and during the process of withdrawing life support.

In summary, in certain cases of pro- longed paralysis, it may be reasonable to proceed with removal of the ventilator provided the clinicians a) are highly cer- tain that the patient could not survive separation from the ventilator; b) proceed with careful regard for the patient’s com- fort; and c) have concluded that the ben- efits of waiting for the return of neuro- muscular function are not sufficient to outweigh the burdens.

Conclusions

The early years of critical care medi- cine were defined by remarkable discov- eries and innovations that dramatically reduced the morbidity and mortality of

disease. In recent years, critical care prac- titioners increasingly have recognized that our obligations to patients extend beyond our attempts to treat disease and include a commitment to providing pa- tients with a dignified and tolerable death.

Meeting these obligations will require that intensive care clinicians learn how to operate within a new paradigm or model of care. In the curative model, the “med- ical indications” for diagnostic and ther- apeutic procedures are judged relevant to the contribution they make toward cur- ing the patient. In the palliative model, however, these indications are judged relative to symptom relief, improved functional status, or the amelioration of emotional, psychological, or spiritual concerns. The former focuses on the treatment of diseases, the latter on the treatment of symptoms.

In this context, treatment of the pa- tient’s pain often becomes the highest priority. The notion of pain as the fifth vital sign is one way of signifying this importance. Critical care clinicians are in a unique position to affect this symptom. Not only are we expert in delivering med- ications to relieve suffering, but we also can provide leadership that will enhance our ability to provide palliative care in ways that go beyond medications. We should work toward developing a culture and physical environment in the ICU that enhance communication and facilitate the comfort of our patients.

Practical aspects of end-of-life care are inseparably wed to many intensely con- troversial ethical issues. Recommenda- tions such as these can only attempt to articulate practices that are based on sound ethical reasoning and that are con- sonant with current cultural and legal

R ecommendationssuch as these canonly attempt to articulate practices that are

based on sound ethical rea-

soning and that are conso-

nant with current cultural

and legal norms.

2345Crit Care Med 2001 Vol. 29, No. 12

norms. These norms are not static and undoubtedly will change over the years. What is unlikely to change, however, is a basic commitment to the comfort and well-being of the patient, regardless of whether the hopes for cure are high or nonexistent. As Albert Schweitzer noted several decades ago, “We all must die. But that I can save him from days of torture that is what I feel as my great and ever new privilege. Pain is a more terrible lord of mankind than even death itself” (126).

REFERENCES

1. Prendergast TJ, Luce JM: Increasing inci- dences of withholding and withdrawal of life support from the critically ill. Am J Respir Crit Care Med 1997; 155:15–20

2. McLean RF, Tarshis J, Mazer CD, et al: Death in two Canadian intensive care units: Institutional difference and changes over time. Crit Care Med 2000; 28:100 –103

3. Eidelman LA, Jakobson DJ, Pizov R, et al: Forgoing life-sustaining treatment in an Is- raeli ICU. Intensive Care Med 1998; 24: 162–166

4. Sprung CL, Eidelman LA: Worldwide simi- larities and differences in the forgoing of life-sustaining treatments. Intensive Care Med 1996; 22:1003–1005

5. SUPPORT Principal Investigators: A con- trolled trial to improve care for seriously ill hospitalized patients. The study to under- stand prognoses and preferences for out- comes and risks of treatments (SUPPORT). The SUPPORT Principal Investigators. JAMA 1995; 274:1591–1598

6. American Thoracic Society: Withholding and withdrawing life-sustaining therapy. Ann Intern Med 1991; 115:478 – 485

7. The Hastings Center: Guidelines on the Termination of Life-Sustaining Treatment and the Care of the Dying. Bloomington, Indiana University Press, 1987

8. American Academy of Pediatrics: Commit- tee on Bioethics: Guidelines on forgoing life-sustaining medical treatment. Pediat- rics 1994; 93:532–536

9. President’s Commission for the Study of Ethical Problems in Medicine and Biomed- ical and Behavioral Research: Deciding to Forego Life-Sustaining Treatment: Ethical, Medical, and Legal Issues in Treatment De- cisions. Washington, DC, U.S. Government Printing Office, 1983

10. ACCP/SCCM Consensus Panel: Ethical and moral guidelines for the initiation, contin- uation, and withdrawal of intensive care. Chest 1990; 97:949 –961

11. Society of Critical Care Medicine Ethics Committee: Consensus report on the ethics of foregoing life-sustaining treatments in the critically ill. Crit Care Med 1990; 18: 1435–1439

12. Danis M: Improving end-of-life care in the intensive care unit: What’s to be learned

from outcomes research? New Horizons 1998; 6:110 –118

13. Danis M, Federman D, Fins JJ, et al: Incor- porating palliative care into critical care education: Principles, challenges, and op- portunities. Crit Care Med 1999; 27: 2005–2013

14. Institute of Medicine: Committee on Care at the End of Life: Approaching Death: Im- proving Care at the End of Life. Washing- ton, DC, National Academy Press, 1997

15. Copp G: A review of current theories of death and dying. J Adv Nurs 1998; 28: 382–390

16. Fleischman AR, Nolan K, Dubler NN, et al: Caring for gravely ill children. Pediatrics 1994; 94:433– 439

17. Emanuel EJ, Emanuel LL: The promise of a good death. Lancet 1998; 351(Suppl 2): SII21–SII29

18. Singer PA, Martin DK, Kelner M: Quality end-of-life care: Patient’s perspectives. JAMA 1999; 281:163–168

19. Campbell ML, Bizek KS, Thill M: Patient responses during rapid terminal weaning from mechanical ventilation: A prospective study. Crit Care Med 1999; 27:73–77

20. Council on Scientific Affairs American Med- ical Association: Good care of the dying patient. JAMA 1996; 275:474 – 478

21. Irion PE: The agnostic and the religious: Their coping with death. In: Death and Min- istry: Pastoral Care of the Dying and Be- reaved. Bane JD, Kutsher AH, Neale RE, et al (Eds). New York, Seabury Press, 1975

22. Curley MAQ, Meyer EC: The impact of the critical care experience on the family. In: Critical Care Nursing of Infants and Chil- dren. Curley MAQ, Smith JB, Moloney- Harmon PA (Eds). Philadelphia, WB Saun- ders, 1999, pp 47– 67

23. Hampe SO: Needs of the grieving spouse in a hospital setting. Nurs Res 1975; 24: 113–120

24. Meyer EC, Snelling LK, Myren-Manbeck LK: Pediatric intensive care: The parents’ experience. AACN Clin Issues 1998; 9:64 –74

25. Leske JS: Needs of relatives of critically ill patients: A follow-up. Heart Lung 1986; 15: 189 –193

26. Hickey M: What are the needs of families of critically ill patients? A review of the liter- ature since 1976. Heart Lung 1990; 19: 401– 415

27. Coolican MB: Families: Facing the sudden death of a loved one. Crit Care Nurs Clin North Am 1994; 6:607– 612

28. Furukawa MM: Meeting the needs of the dying patient’s family. Crit Care Nurse 1996; 16:51–57

29. Campbell ML: Forgoing Life-Sustaining Therapy. Aliso Viejo, CA, American Associ- ation of Critical Care Nurses, 1998

30. Faber-Langendoen K: The clinical manage- ment of dying patients receiving mechani- cal ventilation: A survey of physician prac- tice. Chest 1994; 106:880 – 888

31. Asch DA: The role of critical care nurses in

euthanasia and assisted suicide. N Engl J Med 1996; 334:1374 –1379

32. Hoyt JW, Harvey MA, Wlody GS: The role of critical care nurses in euthanasia and as- sisted suicide. N Engl J Med 1996; 335: 973–974

33. Scanlon C: Euthanasia and nursing prac- tice—Right question, wrong answer. N Engl J Med 1996; 334:1401–1402

34. Brody H, Campbell ML, Faber-Langendoen K, et al: Withdrawing intensive life-sustain- ing treatment—Recommendations for compassionate clinical management. N Engl J Med 1997; 336:652– 657

35. Faber-Langendoen K, Bartels DM: Process of forgoing life-sustaining treatment in a university hospital. Crit Care Med 1992; 20: 570 –577

36. Robinson WM, Ravilly S, Berde C, et al: End-of-life care in cystic fibrosis. Pediatrics 1997; 100:205–209

37. Campbell ML, Frank RR: Experience with an end-of-life practice at a university hos- pital. Crit Care Med 1997; 25:197–202

38. Campbell ML, Bizek KS, Stewart R: Inte- grating technology with compassionate care: Withdrawal of ventilation in a con- scious patient with apnea. Am J Crit Care 1998; 7:85– 89

39. Anand KJS, Hickey PR: Pain and its effects in the human neonate and fetus. N Engl J Med 1987; 317:1321–1329

40. Walco GA, Cassidy RC, Schechter NL: Pain, hurt, and harm: The ethics of pain control in infants and children. N Engl J Med 1994; 331:541–544

41. Schechter NL: The undertreatment of pain in children: An overview. Pediatr Clin North Am 1989; 36:781–794

42. Partridge JC, Wall SN: Analgesia for dying infants whose life support is withdrawn or withheld. Pediatrics 1997; 99:76 –79

43. Cassell EJ: The nature of suffering and the goals of medicine. N Engl J Med 1982; 306: 639 – 645

44. Crippen D: Terminally weaning awake pa- tients from life-sustaining mechanical ven- tilation: The critical care physician’s role in comfort measures during the dying process. Clin Intensive Care 1992; 3:206 –212

45. Fontaine DK: Nonpharmacologic manage- ment of patient distress during mechanical ventilation. Crit Care Clin 1994; 10: 695–708

46. Jastremski CA, Harvey M: Making changes to improve the intensive care unit experi- ence for patients and their families. New Horiz 1998; 6:99 –109

47. Meyer TJ, Eveloff SE, Bauer MS, et al: Ad- verse environmental conditions in the re- spiratory and medical ICU settings. Chest 1994; 105:1211–1216

48. Krachman SL, D’Alonzo GE, Criner GJ: Sleep in the intensive care unit. Chest 1995; 107:1713–1720

49. Chlan L: Effectiveness of a music therapy intervention on relaxation and anxiety for

2346 Crit Care Med 2001 Vol. 29, No. 12

patients receiving ventilatory assistance. Heart Lung 1998; 27:169 –176

50. Sulmasy DP: Is medicine a spiritual prac- tice? Acad Med 1999; 74:1002–1005

51. Waldfogel S: Spirituality in medicine. Prim Care 1997; 24:963–976

52. Shapiro BA, Warren J, Egol AB, et al: Prac- tice parameters for intravenous analgesia and sedation for adult patients in the inten- sive care unit: An executive summary. Crit Care Med 1995; 23:1596 –1600

53. Reisine T, Pasternak G: Opioid analgesics and antagonists. In: Goodman & Gilman’s The Pharmacologic Basis of Therapeutics. Hardman JG, Limbard LE (Eds). New York, McGraw Hill, 1996, pp 521–555

54. Harper MH, Hickey RF, Cromwell TH, et al: The magnitude and duration of respiratory depression produced by fentanyl and fenta- nyl plus droperidol in man. J Pharmacol Exp Ther 1976; 199:464 – 468

55. Agency for Health Care Policy and Re- search: Acute pain management: Operative or medical procedures and trauma, Part 1. Clin Pharm 1992; 11:309 –331

56. Agency for Health Care Policy and Re- search: Acute pain management: Operative or medical procedures and trauma, Part 2. Clin Pharm 1992; 11:391– 414

57. Carney MT, Meier DE: Geriatric anesthesia: Palliative care and end-of-life issues. Anesthesiol Clin North Am 2000; 18: 183–209

58. American Society of Anesthesiologists Task Force on Pain Management, Cancer Pain Section: Practice guidelines for cancer pain management. Anesthesiology 1996; 84: 1243–1257

59. Yaster M, Bean JD, Schulman SR, et al: Pain, sedation, and postoperative anesthetic management in the pediatric intensive care unit. In: Textbook of Pediatric Intensive Care. Rogers MC (Ed). Baltimore, Williams & Wilkins, 1996, pp 1547–1593

60. Hollman GA: Analgesia and sedation in pe- diatric critical care. In: Pediatric Critical Care. Fuhrman BP, Zimmerman JJ (Eds). St. Louis, Mosby, 1998, pp 1363–1379

61. Litt B, Krauss GL: Pharmacologic approach to acute seizures and antiepileptic drugs. In: The Pharmacologic Approach to the Critically Ill Patient. Chernow B (Ed). Bal- timore, Williams & Wilkins, 1994, pp 484 –506

62. Cassem EH, Lake CR, Boyer WF: Psychop- harmacology in the ICU. In: The Pharma- cologic Approach to the Critically Ill Pa- tient. Chernow B (Ed). Baltimore, Williams & Wilkins, 1994, pp 651– 665

63. Riker RR, Fraser GL, Cox PM: Continuous infusion of haloperidol controls agitation in critically ill patients. Crit Care Med 1994; 22:433– 440

64. Notterman DA: Pediatric pharmacotherapy. In: The Pharmacologic Approach to the Critically Ill Patient. Chernow B (Ed). Bal- timore, Williams & Wilkins, 1994, pp 139 –155.

65. Ostermann ME, Keenan SP, Seiferling RA, et al: Sedation in the intensive care unit: A systematic review. JAMA 2000; 283: 1451–1459

66. Truog RD, Berde CB, Mitchell C, et al: Bar- biturates in the care of the terminally ill. N Engl J Med 1992; 327:1678 –1682

67. Analgesia and sedation: In: The Harriet Lane Handbook. Barone MA (Ed). Balti- more, Mosby, 1996

68. Wilson WC, Smedira NG, Fink C, et al: Ordering and administration of sedatives and analgesics during the withholding and withdrawal of life support from critically ill patients. JAMA 1992; 267:949 –953

69. Keenan SP, Busche KD, Chen LM, et al: A retrospective review of a large cohort of patients undergoing the process of with- holding or withdrawal of life support. Crit Care Med 1997; 25:1324 –1331

70. Burns JP, Mitchell C, Outwater KM, et al: End-of-life care in the pediatric intensive care unit following the forgoing of mechan- ical ventilation. Crit Care Med, In Press

71. Bizek KS: Optimizing sedation in critically ill, mechanically ventilated patients. Crit Care Nurs Clin North Am 1995; 7:315–325

72. Mancini I, Body JJ: Assessment of dyspnea in advanced cancer patients. Support Care Cancer 1999; 7:229 –232

73. Wilson RC, Jones PW: A comparison of the visual analogue scale and modified Borg scale for the measurement of dyspnoea dur- ing exercise. Clin Sci 1989; 76:277–282

74. Bruera E, de Stoutz N, Velasco-Leiva A, et al: Effects of oxygen on dyspnoea in hy- poxaemic terminal-cancer patients. Lancet 1993; 342:13–14

75. Booth S, Kelly MJ, Cox NP, et al: Does oxygen help dyspnea in patients with can- cer? Am J Respir Crit Care Med 1996; 153: 1515–1518

76. McCann RM, Hall WJ, Groth-Juncker A: Comfort care for terminally ill patients: The appropriate use of nutrition and hydration. JAMA 1994; 272:1263–1266

77. Printz LA: Terminal dehydration, a compas- sionate treatment. Arch Intern Med 1992; 152:697–700

78. Gillick MR: Rethinking the role of tube feeding in patients with advanced dementia. N Engl J Med 2000; 342:206 –210

79. Viola RA, Wells GA, Peterson J: The effects of fluid status and fluid therapy on the dy- ing: A systematic review. J Palliat Care 1997; 13:41–52

80. Steinbrook R, Lo B: Artificial feeding— Solid ground, not a slippery slope. N Engl J Med 1988; 318:286 –290

81. Cook DJ, Guyatt GH, Jaeschke R, et al: Determinants in Canadian health care workers of the decision to withdraw life support from the critically ill. Canadian Critical Care Trials Group. JAMA 1995; 273: 703–708

82. Christakis NA, Asch DA: Biases in how phy- sicians choose to withdraw life support. Lancet 1993; 342:642– 646

83. Asch DA, Christakis NA: Why do physicians prefer to withdraw some forms of life sup- port over others? Intrinsic attributes of life- sustaining treatments are associated with physicians’ preferences. Med Care 1996; 34: 103–111

84. Asch DA, Faber-Langendoen K, Shea JA, et al: The sequence of withdrawing life- sustaining treatment from patients. Am J Med 1999; 107:153–156

85. Christakis NA, Asch DA: Medical specialists prefer to withdraw familiar technologies when discontinuing life-support. J Gen Intern Med 1995; 10:491– 494

86. Wall SN, Partridge JC: Death in the inten- sive care nursery: Physician practice of withdrawing and withholding life support. Pediatrics 1997; 99:64 –70

87. Collins JJ, Grier HE, Kinney HC, et al: Con- trol of severe pain in children with terminal malignancy. J Pediatr 1995; 126:653– 657

88. Grenvik A: “Terminal weaning”: Discon- tinuance of life-support therapy in the ter- minally ill patient. Crit Care Med 1983; 11: 394 –395

89. Gilligan T, Raffin TA: Withdrawing life sup- port: Extubation and prolonged terminal weans are inappropriate. Crit Care Med 1996; 24:352–353

90. Gilligan T, Raffin TA: Rapid withdrawal of support. Chest 1995; 108:1407–1408

91. Krishna G, Raffin TA: Terminal weaning from mechanical ventilation. Crit Care Med 1999; 27:9 –10

92. Gianakos D: Terminal weaning. Chest 1995; 108:1405–1406

93. Campbell ML, Carlson RW: Terminal wean- ing from mechanical ventilation: Ethical and practical considerations in patient management. Am J Crit Care 1992; 1:52–56

94. Levetown M: Palliative care in the intensive care unit. New Horizons 1998; 6:383–397

95. Smedira NG, Evans BH, Grais LS, et al: Withholding and withdrawal of life support from the critically ill. N Engl J Med 1990; 322:309 –315

96. Heytens L, Verlooy J, Gheuens J, et al: Laza- rus sign and extensor posturing in a brain- dead patient. Case report. J Neurosurg 1989; 71:449 – 451

97. Ropper AH: Unusual spontaneous move- ments in brain-dead patients. Neurology 1984; 34:1089 –1092

98. Campbell ML: Breaking bad news to pa- tients. JAMA 1994; 271:1052

99. Krahn GL, Hallum A, Kime C: Are there good ways to give “bad news”? Pediatrics 1993; 91:578 –582

100. Ptacek JT, Eberhardt TL: Breaking bad news. JAMA 1996; 276:496 –502

101. Quill TE, Townsend P: Bad news: Delivery, dialogue, and dilemmas. Arch Intern Med 1991; 151:463– 468

102. Sharp MC, Strauss RP, Lorch SC: Commu- nicating medical bad news: Parents’ experi- ences and preferences. J Pediatr 1992; 121: 539 –546

103. Viswanathan R, Clark JJ, Viswanathan K:

2347Crit Care Med 2001 Vol. 29, No. 12

Physician’s and the public’s attitudes on communication about death. Arch Intern Med 1986; 146:2029 –2033

104. Rosenbaum GE, Burns J, Johnson J, et al: Autopsy consent practice at US teaching hospitals—Results of a national survey. Arch Intern Med 2000; 160:374 –380

105. Zarbo RJ, Baker PB, Howanitz PJ: The au- topsy as a performance measurement tool—Diagnostic discrepancies and unre- solved clinical questions: A College of American Pathologists Q-Probes study of 2479 autopsies from 248 institutions. Arch Pathol Lab Med 1999; 123:191–198

106. Federal Register Final Rule: Hospital Con- ditions for Participation for Organ Donation (42 CFR Part 482), 2000

107. Gortmaker SL, Beasley CL, Sheehy E, et al: Improving the request process to increase family consent for organ donation. J Transpl Coord 1998; 8:210 –217

108. Institute of Medicine: Non-Heart-Beating Organ Donation: Medical and Ethical Issues in Procurement. Washington, DC, National Academy Press, 1997

109. DeVita MA, Snyder JV: Development of the University of Pittsburgh Medical Center Policy for the care of terminally ill patients who may become organ donors after death following the removal of life support. Kennedy Inst Ethics J 1993; 3:131–143

110. Anderson AH, Bateman LH, Ingallinera KL, et al: Our caring continues: A bereavement follow-up program. Focus Crit Care 1991; 18:523–526

111. McClelland ML: Our unit has a bereavement program. Am J Nurs 1993; 93:62– 68

112. Tobias JD: Propofol sedation for terminal care in a pediatric patient. Clin Pediatr 1997; 36:291–293

113. Quill TE, Lo B, Brock DW: Palliative options of last resort—A comparison of voluntarily stopping eating and drinking, terminal se- dation, physician-assisted suicide, and vol- untary active euthanasia. JAMA 1997; 278: 2099 –2104

114. Quill TE, Brody RV: “You promised me I wouldn’t die like this!” A bad death as a medical emergency. Arch Intern Med 1995; 155:1250 –1254

115. Sulmasy DP, Pellegrino ED: The rule of double effect: Clearing up the double talk. Arch Intern Med 1999; 159:545–550

116. Enck RE: Drug-induced terminal sedation for symptom control. Am J Hosp Palliat Care 1991; 8:3–5

117. Fainsinger R, Miller MJ, Bruera E, et al: Symptom control during the last week of life on a palliative care unit. J Palliat Care 1991; 7:5–11

118. Ventafridda V, Ripamonti C, De Conno F, et al: Symptom prevalence and control during

cancer patients’ last days of life. J Palliat Care 1990; 6:7–11

119. Mount B: A final crescendo of pain? J Palliat Care 1990; 6:5– 6

120. Roy DJ: Need they sleep before they die? J Palliat Care 1990; 6:3– 4

121. Hatherill M, Tibby SM, Sykes K, et al: Dilemmas exist in withdrawing ventila- tion from dying children. BMJ 1998; 317:80

122. Hoyt JW: Persistent paralysis in critically ill patients after the use of neuromuscular blocking agents. New Horiz 1994; 2:48 –55

123. Rushton C, Terry PB: Neuromuscular blockade and ventilator withdrawal: Ethical controversies. Am J Crit Care 1995; 4:112–115

124. Truog RD, Burns JP: To breathe or not to breathe. J Clin Ethics 1994; 5:39 – 41

125. Truog RD, Burns JP, Mitchell C, et al: Phar- macologic paralysis and withdrawal of me- chanical ventilation at the end of life. N Engl J Med 2000; 342:508 –511

126. Schweitzer AS: On the Edge of the Primeval Forest. New York, Macmillan, 1931

127. Balestrieri F, Fisher S: Analgesics. In: The Pharmacologic Approach to the Critically Ill Patient. Chernow B (Ed). Baltimore, Wil- liams & Wilkins, 1994, pp 640 – 650

2348 Crit Care Med 2001 Vol. 29, No. 12