Clinical Ethics in Anesthesiology. A Case-Based Textbook

2. End-of-life issues

17. Withholding and withdrawing life support in the intensive care unit

Mark D. Siegel and Stanley H. Rosenbaum

The Case

Mrs. Jones is an 88-year-old woman with a history of diabetes, stable coronary artery disease, and hypertension. She lives independently in her own home, with weekly visits from a housekeeper. Three grown children live nearby, and one of her children (a son, age 54) lives out of state.

The housekeeper discovers Mrs. Jones unconscious in the hallway of her home after an obvious fall that is believed to have occurred sometime in the preceding several days. Mrs. Jones is intubated, and admitted to the ICU. She has a broken hip, severe dehydration, and aspiration pneumonitis. Shortly after admission, she is diagnosed with a subdural hematoma and proceeds to surgery for emergency evacuation of the subdural. She never regains consciousness.

Postoperatively, she remains ventilator dependent with significant hypoxemia requiring 100% oxygen and PEEP. She develops sepsis, with hypotension, tachycardia, and myocardial ischemia. After resuscitation from a cardiac arrest due to ventricular tachycardia, she develops acute renal failure that requires hemodialysis. Liver function tests demonstrate elevated transaminases.

Discussions with the family about withdrawing life supporting therapy begin. While the three local children are in favor of limiting treatment to comfort care, the son from out-of-state wants to continue invasive treatments.

More than one-fifth of deaths in the US follow intensive care unit (ICU) admission, usually after decisions to forgo life support.1 Each day, intensivists must decide whether to start or continue life sustaining therapies such as cardiopulmonary resuscitation, mechanical ventilation, or hemodialysis. Often, the decision to forgo treatment may protect a dying patient from ineffective care and suffering, but sometimes the decision may determine if a patient lives or dies.

End-of-life care in the ICU has many shortcomings. For example, patients with similar illnesses receive highly variable care and end-of-life decisions may be more closely related to the country, hospital, and physician in charge rather than to the patient’s wishes or disease. In many cases, treatment is exceedingly unlikely to confer benefit and resources expended could be better directed elsewhere. Dying patients often endure needless pain and suffering, while family members experience severe psychiatric symptoms. Families participating in end-of-life decisions often describe conflict with the medical team, a potential source of dissatisfaction.2

Several comprehensive reviews offer guidance to physicians and other caregivers seeking to optimize end-of-life care and are included in the “Further reading” at the end of this chapter. This chapter will consider an ethical framework to guide decision making, to explore frequently encountered obstacles, and to offer practical suggestions to enhance care for patients dying in the ICU.

Ethical framework for end-of-life decision-making

Four basic ethical principles should guide end-of-life decision-making in the ICU: (1) respect for the patient’s autonomy; (2) the duty of beneficence; (3) the duty of nonmaleficence; and (4) the obligation to ensure just distribution of care. Day-to-day dilemmas in the ICU frequently bring these principles into tension. Patients may exercise their autonomy by refusing recommended care or requesting treatments that physicians believe they should forgo. Alternatively, resource limitations may force caregivers to deny care, even if desired or potentially beneficial. Negotiating these tensions is a challenge for physicians seeking to providing ethically appropriate care.

End-of-life decisions fall broadly into two categories depending on the degree to which available choices are constrained. Constrained choices include those affected by triage or futility considerations. For

Table 17.1. An example of a triage scoring system*

Table 17.1.

* Adapted by Medical Triage Policy, Yale–New Haven Hospital.

example, when the demand for ICU beds exceeds capacity, physicians may have to deny admission to patients who might benefit.3 Similarly, physicians may not be able to provide requested care because desired outcomes cannot be achieved, which, by definition, limits options.4 Most end-of-life decisions in the ICU, however, are less constrained, leaving patients or surrogates with treatment options to consider.


With increasing frequency, physicians are forced to choose among multiple patients vying for a limited number of ICU beds. Contributing factors include growth in the critically ill population, increasing recognition of better outcomes associated with ICU admission, and concerns about bioterrorism and pandemics. As a result, ICU administrators and physicians are obligated to implement effective, fair triage plans to optimize bed use.

By definition, triage implies that patients who might benefit from the ICU may be denied admission. The decision to deny admission could bring harm, including death, to the patient. Hospitals should seek to avoid the need to triage by investing sufficiently in resources to meet demand by ensuring a sufficient supply of qualified staff, beds, and equipment. In addition, steps should be taken to safely decrease ICU length of stay by promoting good practices, such as timely weaning from mechanical ventilation, making timely end-of-life decisions, and encouraging efficient throughput by transferring patients out of the ICU as soon as they are eligible. Institutions should maximize their ability to care for patients outside the unit, for example by increasing skills and staffing of non-ICU personnel. Denying ICU admission should not be equated with the decision to forgo life support, although sometimes that may be the implication.

The goal of triage is to maximize utilization of the ICU by patients most likely to benefit (Table 17.1). In choosing among patients, two key factors to consider are severity of illness and potential to benefit. Patients requiring life support technology such as intubation and mechanical ventilation, continuous renal replacement therapy, or left ventricular support devices generally have a more compelling need for ICU care than those requiring monitoring alone. At the same time, factors such as severity of the acute illness and chronic comorbidities strongly influence the likelihood of benefit. Patients who are moribund and unlikely to survive despite aggressive care are often less likely to benefit than those that are less ill. The same holds for patients with severe underlying illnesses, like advanced cancer, in whom the prognosis is likely to be poor regardless of the intervention provided.

Basic principles of fairness should govern triage. It is inappropriate to discriminate on the basis of race, gender, sexual orientation, ability to pay, or political connections. Whether to discriminate on the basis of age or functional status is controversial. Objective scoring systems should be used to maximize fairness. To minimize conflicts of interest, triage responsibility should be delegated to physicians not directly involved in patient care. Physicians responsible for triage require institutional support to support their authority while the hospital administration, attorneys, and ethics committee should be enlisted to provide oversight and guidance. Finally, when admission is denied, decisions must be made to determine if life saving efforts will be attempted outside the ICU or whether the focus should shift to palliation alone.


In the context of decision-making, futility refers to care that cannot achieve a desired end point.4 How often futile care is requested or provided is unknown; however, the frequency undoubtedly depends on the definition used. Both quantitative and qualitative criteria for futility have been proposed, although consensus is lacking. In general, agreement is more easily achieved when situations are extreme, for example, when the patient is moribund.

At least in theory, identifying futility should transform decision-making and relieve families from considering treatments that cannot alter outcome, shifting the focus to those that can. Offering futile care makes little ethical sense: it cannot help, it may cause harm, and it may waste resources and deny other patients access to care.

Surrogate decision makers will occasionally request care that physicians consider futile. Professional societies have opined that physicians are not obligated to provide care that cannot achieve treatment goals, even if requested.5 Careful, informative, empathic discussions will generally resolve most disputes. In the minority that persist, a deliberate approach that incorporates second opinions and opportunities to transfer care may ultimately lead to unilateral decisions to forgo treatment. However, not all physicians agree that unilateral decisions are appropriate and some suggest deferring to families in the small minority where impasse cannot be resolved.4

For practical reasons, the futility rationale should probably be curtailed for all but the most obvious cases. Physicians and nurses are often unable to accurately identify futility, raising the possibility of error. Concerns about self-fulfilling prophecies suggest that poor outcomes may occur simply because physicians consider the prognosis poor. Finally, survival appears to be improving for some patients with traditionally devastating illnesses, such as stem cell transplant recipients with respiratory failure, suggesting that prior beliefs about futility may be obsolete. For these reasons, we believe futility designations should be limited to obvious cases, while favoring traditional approaches that balance risk, benefit, and patient preferences when there is doubt.

Unconstrained decision-making

Surrogate decision-making

Most decisions in the ICU are unconstrained by triage and futility concerns, leaving physicians and surrogates with options to consider. Respect for the patient’s autonomy should dictate most decisions. Unfortunately, most critically ill patients cannot represent themselves due to cognitive impairment resulting from their acute illness, delirium, sedation, or dementia. Surrogate decisions-makers, usually family members, must speak for them.6

Ideally, surrogates may be able to articulate patients’ previously documented wishes, for example if relevant advance directive are available. If no explicit direction is available, surrogates may be able to provide substituted judgment, indicating what they believe the patient would choose, for example, on the basis of prior discussions or what is known about the patient. If no information is available to guide these judgments, surrogates may be left to choose treatment they believe to be in the patient’s best interest.

Several factors are likely to increase the burden faced by surrogates. Decision-making subjects family members to enormous stress. Many family members experience severe anxiety and depression during the ICU stay, which could interfere with their ability to be effective surrogates.7 Some may not know the patient well enough to feel comfortable making decisions. Although they are probably more accurate than physicians, family members may not reliably make the same choices the patient would.2 Even if past discussions occurred, patients’ wishes may not be static – prior conversations and advance directives may not accurately portray what patients would choose under the circumstances encountered.

Certain observations may be helpful. The willingness of individual patients to endure aggressive therapy should not be underestimated. Many patients with severe underlying diseases are willing to choose aggressive ICU care, particularly if there is a chance for functional survival. In contrast, patients tend to be less willing to undergo ICU care if the likely outcome is functional or, especially, cognitive impairment.8

Family meetings

The primary purpose of family meetings is to choose treatments best suited to meet the patient’s goals. Effective family meetings can increase family satisfaction and may mitigate long term psychiatric complications during bereavement. At its core, appropriate family meetings include all the components of any other discussion devoted to informed decision making. Physicians must ensure that surrogates are in a position to make decisions most likely to meet the patient’s goals. Surrogates must have the capacity to make informed decisions and physicians must give them the information required, including descriptions of the patient’s illness, available treatments options, and the risks, benefits, and likely outcomes of each option. Most surrogates want physicians to contribute to decision-making9 and physicians should provide input to the degree that families desire.

Unfortunately, family meetings are often poorly run.1 Many are led by physicians with little experience and tend to emphasize procedural details (e.g., intubation and CPR) without adequately discussing likely outcomes, risks, and alternatives. There is often undue focus on “code status,” rather than appropriate treatment goals, which should, in turn, guide decisions. Families are often asked to decide upon major interventions with insufficient information to guide them.10

Many families have limited understanding of their loved ones’ illnesses and prognoses. Many know relatively little about life-sustaining technologies and often overestimate the likelihood of success.10 Many express a fear, not entirely misplaced, that a do not resuscitate (DNR) decision will lead to less aggressive care overall.1 Language barriers when families do not speak English and also when imprecise terminology and jargon are used may compromise communication. An undue burden is often placed on families to make decisions independently, even though most would prefer to share responsibility with physicians.9 Physicians often miss opportunities to let families speak or to share emotional support and empathy, which, in turn, could undermine negotiation.11

Several reviews have highlighted the features of successful family meetings.1,5,11 Key among these are meeting early in the ICU stay, planning in advance, following a structured but flexible format, allowing time for family members to speak and ask questions, avoiding undue pressure to make decisions, and showing empathy. Families show greater satisfaction when they are assured that their loved one will not suffer or be abandoned, and when they receive support for their decision.11

Under the best of circumstances, physicians and family members may not be able to agree on decisions. Families may choose treatments that differ from those physicians recommend. Physicians are obliged to ensure that families understand treatment options and that coherent decisions are made in accord with the patient’s wishes, if known, or best interest, if not. Physicians may choose to gently disagree and negotiate with families. However, in most cases, they should defer to families, as long as the choices are well considered and are limited to options consistent with standard medical care.

Families will occasionally request treatment that arouses misgivings among team members. In some cases, physicians may feel that the requested care is likely to cause excessive suffering and little good; alternatively, physicians may be concerned that families are refusing potentially effective care and putting patients at risk. Ultimately, physicians are obligated by basic ethical principles to avoid causing pain and suffering if there is no reasonable associated treatment benefit.

Most perceived impasses and conflicts between family members and caregivers are actually due to identifiable problems that can be readily addressed (Table 17.2). Different perceptions regarding prognosis may respond to open, patient discussion. Many perceived sources of conflict such as anger, distrust, and unrealistic expectations may simply be the manifestations of grief associated with the trauma of hearing bad news and may dissipate if caregivers show empathy. Various cultural barriers may impose obstacles, particularly when physicians and families are from different racial, religious, ethnic, or cultural backgrounds or speak different languages.12 Families from non-Western backgrounds may not share standard beliefs regarding patient autonomy, delivering bad news, or the family’s appropriate role in decision-making. Families from traditionally oppressed groups may be less inclined to trust that physicians have the patient’s best interest at heart. Families from some religious backgrounds may believe in miracles and may not be inclined to consider the physician’s prognostic estimates relevant. Sensitivity to these potential sources of conflict and input from allied personnel such as professional interpreters, chaplains, and social workers may help identify and overcome barriers. Finally, it is important to recognize that most families experience anxiety and depression when faced with losing a loved one. Patience, empathy, and willingness to share in the burden of decision making may prove valuable. In the setting of a trusting relationship, physicians should rarely feel the need to override families’ choices. Negotiation, education, and good communication about prognosis and treatment options should eliminate or minimize barriers to consensus.

Withholding and withdrawing life support

If the decision is made to withhold or withdraw life support, it is generally appropriate to proceed expeditiously, while allowing time for family and other visitors to gather. When appropriate, local organ banks should be notified so that donation options are pursued – often local policy may require notification, particularly in the setting of brain death.

The patient’s physical comfort must be assured prior to withdrawing life support, particularly when intubation and mechanical ventilation are discontinued. Families may differ regarding their preference

Table 17.2. Sources of conflict in end-of-life decision-making and potential solutions

Table 17.2.

* Suggested solutions are not meant to imply that conflicts raised have easy or predictable solutions, nor that the cause is always easily identified. The nature of the dispute cited by the family or identified by caregivers may mask deeper conflicts requiring further exploration.

for removing the endotracheal tube and their wishes should generally be respected as long as the patient’s comfort is paramount. Withdrawal of life support is best approached as a procedure with specific, necessary components. Medications should be targeted to ensure comfort. To achieve symptom control, the dosing of some medication, particularly opiates, may lead to respiratory depression or even quicken death in some cases. From an ethical perspective, such dosing is acceptable (i.e., the principle of “double effect” – see Chapter 15) as long as the primary purpose is to alleviate discomfort and not primarily to suppress respiration or hasten death. In most cases, however, opiates do not shorten time to death and sometimes symptom management may lead to a period of temporary stabilization.

Families require substantial support during and after withdrawal of life support. Physicians may offer reassurance by visiting the patient’s room intermittently, answering questions about the dying process, reinforcing their support for the decision, and expressing empathy and support. It may be helpful to explain the role of medications to treat distressing symptoms. Similarly, it may be helpful toexplain the dying process, particularly agonal breathing, and reassure families that symptoms will continue to be effectively managed. Recognizing that family members, particularly spouses, are at significant risk for developing psychiatric disorders during bereavement, plans for follow-up care and appropriate referrals might be considered.

If the bed is available and death is imminent, patients should be kept in the ICU and families should be allowed to spend time with the body afterwards if they wish. However, if process is likely to be prolonged, transfer to a hospice unit or a floor able to provide continue expert care may be necessary, particularly if ICU beds are in short supply.

Case resolution

Mrs. Jones failed to improve. Her physicians saw death as inevitable and tried to persuade the family to withdraw life support. Three children embraced these recommendations, but the son from out of town did not. Over the next 2 days, the medical team met frequently with the family, accompanied by ICU nurses, a social worker, and a chaplain. During these meetings, the son acknowledged a strained relationship with his siblings and expressed regret that he had not visited his mother more often. He ultimately acknowledged that his mother was not improving and came to support his siblings’ view that she would not want to remain on the ventilator given the poor prognosis. Consensus was soon reached to switch the focus to palliation. A morphine infusion was started for pain and dyspnea and the ventilator was disconnected. Mrs. Jones died peacefully several minutes later, surrounded by her family and the ICU team.

Key points

• End-of-life care, particularly making decisions to withhold or withdraw life support, is a fundamental component of critical care practice.

• Appropriate end-of-life decision-making hinges on the intensivist’s understanding of key ethical principles. In some cases, decisions are constrained by limits in available ICU resources, particularly beds; in other cases, decisions are constrained by limitations imposed by the patient’s disease.

• Careful, deliberate, and empathic explanations and negotiation usually lead to choices acceptable to both family members and the medical team. Impasses may occasionally arise, although most are likely to resolve with patience and identification of the barriers preventing agreement.

• When life support is discontinued, it is crucial to providing ongoing care both to patients and their newly bereaved families.


1* Siegel, M.D. (2009). End-of-life decision making in the ICU. Clin Chest Med1, 181–94.

2* Way, J.Back, A.L., and Curtis, J.R. (2002). Withdrawing life support and resolution of conflict with families. BMJ325, 1342–5.

3* Guidelines for intensive care unit admission, discharge, and triage. (1999). Task Force of the American College of Critical Care Medicine, Society of Critical Care Medicine. Crit Care Med27(3),633–8.

4* Burns, J.P. and Truog, R.D. (2007). Futility: a concept in evolution. Chest132(6), 1987–93.

5* Truog, R.D.Campbell, M.L.Curtis, J.R.Haas, C.E. et al. (2008). Recommendations for end-of-life care in the intensive care unit: a consensus statement by the American College [corrected] ofCritical Care Medicine. [erratum appears in Crit Care Med. 2008 36(5), 1699]. Crit Care Med, 36(3), 953–63.

6* Berger, J.T.DeRenzo, E.G., and Schwartz, J. (2008). Surrogate decision making: reconciling ethical theory and clinical practice. Ann Intern Med149(1), 48–53.

7* Pochard, F.Azoulay, E.Chevret, al. (2001). Symptoms of anxiety and depression in family members of intensive care unit patients: ethical hypothesis regarding decision-making capacity. Crit Care Med29(10), 1893–7.

8* Fried, T.R.Bradley, E.H.Towle, V.R., and Allore, H. (2002). Understanding the treatment preferences of seriously ill patients.[see comment]. N Engl J Med346(14), 1061–6.

9* Heyland, D.K.Cook, D.J.Rocker, al. (2003). Decision-making in the ICU: perspectives of the substitute decision-maker. Intensive Care Med29(1), 75–82.

10* Heyland, D.K.Frank, C.Groll, al. (2006). Understanding cardiopulmonary resuscitation decision making: perspectives of seriously Ill hospitalized patients and family members. Chest130(2),419–28.

11* Curtis, J.R. and White, D.B. (2008). Practical Guidance for Evidence-Based ICU Family Conferences. Chest134(4), 835–43.

12* Kagawa-Singer, M. and Blackhall, L.J. (2001). Negotiating cross-cultural issues at the end of life: “you got to go where he lives.” JAMA286(23), 2993–3001.

Further reading

Beauchamp, T.L. and Childress, J.F. (2009). Principles of Biomedical Ethics. 6th edn. New York: Oxford University Press.

Medical futility in end-of-life care (1999). Report of the council on ethical and judicial affairs. JAMA281, 937–41.

Ratnapalan, M.Cooper, A.B.Scales, D.C., and Pinto, R. (2010). Documentation of best interest by intensivists: a retrospective study in an Ontario critical care unit. BMC Med Ethics11, 1.