Special Feature

Withdrawing Life Support: DecisionMaking and Process

By Fred J. Tasota, RN, MSN, and Leslie A. Hoffman, PhD, RN

Today, 6590% of deaths in intensive care units (ICUs) occur after a decision to withdraw or withhold lifesustaining therapy.13 This high prevalence reflects a growing social and professional consensus that withdrawing or withholding treatment is ethically acceptable and clinically desirable if reversal of the disease process is no longer possible, restoration of an acceptable level of functioning is unlikely, and/or the patient is experiencing substantial pain and suffering.46 Nevertheless, health team members continue to struggle with issues surrounding such decisions and their implementation.46 This consequence is logical. Patients are transferred to an ICU so that they can receive aggressive, lifesaving care. Yet mortality rates remain high for many conditions. Further, if the patient’s condition deteriorates, guidance from the patient regarding what level of support is "unwanted" is typically not available.6 In one study of 179 patients who received a recommendation to withdraw or withhold life-sustaining therapy, only 3% were able to participate in decision-making and only 4% had a written advance directive.2 Consequently, patients continue to receive unwanted interventions after they, their families, and/or their care providers prefer to stop life-sustaining therapy.4-6

Withdrawal of mechanical ventilation, in particular, poses a difficult situation because it typically leads to the death of the patient within a short interval (24 hours or less), can be accomplished using different methods, and can be the source of patient discomfort if inappropriately managed. Consistent and compassionate communication among the health care team, family, and patient (if possible) is, thus, essential when undertaking this process.

Decision-Making Process

Two basic goals drive delivery of care in the ICU—to save patients who have a chance to live, and to help the dying do so with peace and dignity. When further care is incapable of producing a meaningful recovery—that is, futile—the latter goal moves to the forefront. The point at which goals change is difficult to determine because "futility" cannot be defined in narrow physiologic terms.4,6 Futility embraces a combination of quantitative factors, such as the probability of survival, and qualitative factors, such as quality of life.4 Therefore, each situation is unique.

It is also pertinent to consider that different members of the health team have varying beliefs and that these beliefs influence decision-making. In a study of 456 university-affiliated internists, respondents were asked to rank their preferences for withdrawing eight forms of life support (blood products, hemodialysis, IV vasopressors, total parenteral nutrition, antibiotics, tube feedings, mechanical ventilation, and IV fluids).7 Rankings were associated with gender (P = 0.039), age (P < 0.0001), and whether the physician was a general internist or specialist (P < 0.0001), but not with religion, rank, or degree of exposure to patients in an ICU. In a second study, physicians (n = 225) were asked to rank their preferences in withdrawing six life-sustaining technologies related to their special expertise: pulmonologists with mechanical ventilation, nephrologists with hemodialysis, gastroenterologists with tube feedings, hematologists with blood products, cardiologists with IV vasopressors, and infectious disease specialists with antibiotics.8 With the exception of infectious disease, specialists indicated a preference for withdrawing their "own" form of life support. Few studies have investigated nurse perceptions during withdrawal of mechanical ventilation. In one study,9 there was unanimous (100%) agreement with the decision to withdraw mechanical ventilation, but some nurses (15%) had concerns about the procedure with specific reference to patient comfort.

These studies support the statement that health team members confront this issue with their own sets of perceptions and biases. Consequently, it is essential that decisions about withdrawing or withholding life-sustaining therapy not be made unilaterally. Rather than avoiding the subject, health team members should maintain open discussions among themselves (and with the patient and/or family) regarding all medical matters, including the treatment plan, progress, and prognosis.4-6 When the patient’s condition or a request from patient or family raises doubt about the appropriateness of continuing treatment, the need for open, ongoing communication becomes paramount.

Changing the Direction of Care

The process of arriving at a decision to withdraw therapy can be arduous for all involved.10 Some family members arrive at this decision earlier than others, and it is important to air opinions openly.10 It is important that the family understand that allowing death to occur naturally is the anticipated outcome, and that the priority is relief of the patient’s suffering with medications given as needed to promote comfort. Nurses and other team members, such as respiratory therapists and social workers who have direct roles in patient care, need to be actively involved in these discussions. They can provide valuable input, uncover disagreements, and assist with mediation of disputes. It is also important that withdrawal of support, like other medical procedures, be accompanied by proper informed consent and documentation in the medical record.10 Attention to minority opinions can avoid distrust, inability to bring closure following death of a loved one, and lawsuits. Consensus is not a requirement, but it is important to try to reach agreement.10 The following guidelines are important to successful communication during the decision-making process:

• Whenever possible, initiate discussions about preferences for life-sustaining therapies before an acute event occurs making death imminent.

• Hold regular meetings to discuss the patient’s current condition, plan of treatment, and prognosis.

• Keep discussions frank, informative, and consistent

• Document discussions in the medical record.

• Once a decision is reached that further care is futile, communicate to the family that withdrawal of support is a recommendation, not merely an option.

• Allow time to accept the recommendation.

• Establish a time for withdrawal. Incorporate a short period of delay to provide opportunity for family members to reflect on the decision, make necessary arrangements, and plan to be present, if they wish. The delay should not be longer than a day, however, because further deferral tends to increase anxiety.

Ultimately, decisions about withdrawing and withholding life-sustaining therapy should be guided by two fundamental principles, autonomy (that is, one’s duty to respect the rights of others to make decisions about treatment they wish to receive) and beneficence (that is, the duty of practitioners to do "good" for the patient). When it is not possible to reach agreement, it is advantageous to have institutional supports to facilitate decision-making. Educational programs and written policies defining acceptable circumstances for discontinuing therapies can be helpful. A multidisciplinary ethics committee may help to resolve conflicts when agreement cannot be reached.5

Implementing Withdrawal of Mechanical Ventilation

Limited research has examined how withdrawal of life support is best accomplished. In a recent study, a retrospective chart review was used to identify the process used for 419 patients admitted to three university-affiliated hospitals.3 The issue of withdrawal was first raised by the attending physician (71.9%) or primary attending service (24.9%), and rarely by the family (2.4%) or patient (0.8%). Once the issue was raised, discussions primarily involved critical care physicians (69% to 98%) and the family (83% to 100%). Nursing (16%) and social workers (14% to 19%) were infrequently involved. Once the issue was raised, a decision to withdraw life support was made at the first meeting in most cases (63%), although a minority (5%) required four or more meetings. Slightly more than half (52%) of the patients had life support withdrawn within one hour of the decision and 77% within 10 hours. Once the process began, most patients died quickly (that is within one hour [55%], within 4 hours [75%], or within 24 hours [98%]).

A subsequent study compared these findings with data obtained from six community hospitals.11 The incidence of withdrawal of life support was similar to that found in teaching hospitals, but more patients died in community hospitals as a result of withholding as opposed to discontinuing treatment. Families were more likely than physicians to be the first to raise the issue, and the time from beginning withdrawal of support to death of the patient was longer. If all discussions were in fact documented in these studies, the findings suggest that families are not actively involved in decision-making early in the process. Nevertheless, findings suggest that families are supportive of the decision, since little time ensued between making the decision and initiating treatment withdrawal.

Two distinct methods of withdrawing mechanical ventilation have been described: the endotracheal tube may be removed, or it may be left in place while ventilator rate, positive end-expiratory pressure, supplemental oxygen, and tidal volume are reduced.9,12,13 The directness of extubation represents one merit.13 However, this approach fails to protect the patient’s airway, and, thus, there is risk of stridor, air hunger, and the consequent requirement for large doses of morphine and other sedatives. The latter point can be of consequence. Although administration of morphine is accepted to relieve distress, even if it depresses respiration (principle of double effect), it is possible under these circumstances for the dosage to become so large as to call into question the issue of euthanasia.13 The best approach seems to be a combination of the two: leave the endotracheal tube in place but decrease the FIO2 and ventilatory support over a brief interval (after suctioning and medicating the patient).13 There is no reason to prolong the process of dying by prolonging reduction in support.

Regardless of the method chosen, health team members must frequently assess and aggressively manage patient discomfort with rapid titration of appropriate medications. Morphine is the most useful choice for analgesia due to its potency, its ability to relieve dyspnea, and its wide therapeutic range. In patients prone to bronchospasm, fentanyl may be substituted. Dose requirements vary significantly based on level of pain, tolerance, and factors affecting systemic distribution. More important than specific dosages are the principles that discomfort should be anticipated and medication should be immediately available. Intravenous infusions facilitate a consistent effect and should be combined with intermittent boluses, administered when ventilator support is decreased or when patient discomfort is increased. In order to optimally reduce discomfort, the physician must either be at the bedside at all times or provide nurses with wide-dosing latitude. Benzodiazapines are indicated for sedation, anxiety, restlessness, and delirium. Optimal management is best achieved using a combination of opiates and benzodiazapines with lower doses of both drugs. Depending on the situation, opiates and/or benzodiazapines may not be indicated or necessary. Likewise, circumstances may warrant the use of other agents (e.g., anticholinergics to decrease secretions, bronchodilators to ease breathing, and antiemetics for nausea).

In the previously cited study with findings from university hospitals, morphine was used in all instances in which medication was considered appropriate, and sedation was used in 61% of cases.3 Morphine was most often (44%) given by bolus injections, supported by a background infusion. However, in 36% of cases, a morphine infusion alone was used. The median hourly dose was 14.4 mg, with a range of 0.7-350 mg/h (mean 21 ± 33 [SD]). Benzodiazepines were most commonly given as boluses alone (80%). In 20% of cases, a background infusion was used either with (10%) or without (10%) bolus injections. The median hourly dose was 5.1 mg, with a range of 0.2-80 mg/h (mean 8.6 ± 11).

Neuromuscular blocking agents are contraindicated, since they mask signs and symptoms that alert the health team members of the need to provide additional analgesia and/or sedation. If necessary, reversal agents (neostigmine) may be used. If paralysis cannot be reversed in 2-3 hours, most authors recommend proceeding while administering high doses of opiates and sedatives.9,12

While maintaining the patient as the primary focus of the process, family support remains vital throughout in order to allay anxiety experienced by loved ones. Comfort measures directed toward the patient in an unhurried, compassionate manner demonstrate continued caring by the health team. In addition, the following are important to promote patient and family well-being:

• Maintain interaction with the patient and family (not talking is equated with not caring).

• Explain that death is the anticipated outcome, the priority is relief of suffering.

• Explain the dying process and what to expect during withdrawal.

• Reinforce that the use of medications is not to hasten death, but to promote comfort.

Although families often ask questions about how long the process will take, it is best to refrain from providing specific estimates, as time may vary considerably from that expected. Although the majority of patients succumb within minutes to a few hours, there are instances when patients survive for longer periods and may be transferred to a ward.5


If these management strategies are understood and applied, the difficult process of withdrawal of life support can become more humane. Attention to patient comfort, to clear, open and ongoing communication, and to promoting death with dignity are vital for all concerned. (Tasota is Project Director, Transtracheal Assist Grant, School of Nursing, University of Pittsburgh.)


1. Faber-Langendoen K. Arch Intern Med 1996;156: 2130-2136.

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

3. Keenan SP, et al. Crit Care Med 1997;25(8):1324-1331.

4. Luce JM. Am J Respir Crit Care Med 1997;156: 1715-1718.

5. Tasota FJ, Hoffman LA. Critical Care Nursing Quarterly 1996;19(3):36-51.

6. Karlawish JHT, Hall JB. Am J Respir Crit Care Med 1997;155:1-2.

7. Asch DA, Christakis NA. Med Care 1996;34(2): 103-111.

8. Christakis NA, Asch DA. J Gen Intern Med 1995; 10:491-494.

9. Daly BJ, Thomas D, Dyer MA. Am J Crit Care 1996; 5(5):331-338.

10. Rubenfeld GD, Crawford SW. Crit Care Alert 1995; 2(12):92-96.

11. Keenan SP, et al. Crit Care Med 1998;26(2):245-251.

12. Brody H, et al. N Engl J Med 1997;336(9):652-657.

13. Gilligan T, Raffin TA. Crit Care Med 1996;24:352-354.


In the article "Music Therapy Reduces Anxiety in Mechanically Ventilated Patients" (Crit Care Alert 1998;6[7]:49-50), the source was incorrectly cited. The correct citation should read: Chlan L. Effectiveness of a music therapy intervention on relaxation and anxiety for patients receiving ventilatory assistance. Heart Lung 1998;27:169-176. We regret any confusion we may have caused.