Terminal Withdrawal of Life-Sustaining Supplemental Oxygen

An influential report released in 1983 defined life-sustaining therapies as “all health care interventions that have the effect of increasing the life span of the patient.”¹ This definition is highly inclusive: aspirin for stable coronary artery disease, intravenous antibiotics for osteomyelitis, and mechanical ventilation for respiratory failure all qualify. However, when considering withholding or withdrawing life-sustaining interventions, clinicians commonly refer to a more discrete group of therapies intended to forestall impending death by augmenting or replacing a vital bodily function. A hallmark of life-sustaining therapies, therefore, is that withholding or withdrawing them leads to physiologic decompensation foreseeably to cardiac arrest.

Supplemental oxygen has not commonly been considered a life-sustaining therapy. Yet it clearly serves this purpose for spontaneously breathing patients in whom pulmonary gas exchange is so impaired that the needs of vital organs cannot be met with ambient air alone. Supplemental oxygen may be lifesaving, as in the acute treatment of severe pneumonia or pulmonary embolism, or life-sustaining, as in the subacute or long-term management of patients with advanced pulmonary fibrosis, extensive intrathoracic cancer, or cardiovascular conditions causing right-to-left shunting of venous blood.

As cardiopulmonary diseases associated with hypoxemia increase in incidence, and as new technologies are available to provide high-flow oxygen to patients living at home, physicians are more commonly caring for patients whose lives are sustained by supplemental oxygen. Although improvements in oxygen delivery systems have led to improved functional capacity for some patients, the quality of life associated with long-term oxygen dependence may remain unacceptable. As a result, some patients have asked their physicians for assistance with or acquiescence to their plans to withdraw supplemental oxygen.

Informed patients with decision-making capacity have well-established rights to forgo any and all forms of life-sustaining therapy.^{2 - 3} Although these rights clearly extend to supplemental oxygen, requests to remove this form of life-sustaining therapy raise difficult questions. Should physicians help patients remove such a minimally invasive and often comforting intervention when death is the expected result? If so, should oxygen be replaced with palliative sedation? Should such sedation be administered before oxygen is removed in anticipation of distress, or afterward in response to manifest discomfort? These and other concerns may prevent some physicians from heeding requests for the withdrawal of life-sustaining oxygen as readily as they may heed requests that other therapies be withdrawn. As the Association of Palliative Medicine noted, “once oxygen has been given to a patient it is often difficult to stop its use.”⁴ In this commentary, we discuss these concerns and suggest strategies for physicians to overcome them.

When discussing the possibility of terminal withdrawal of life-sustaining interventions with patients or family members, physicians often point to the discomfort, invasiveness, and functional impairment associated with continuation of a therapy that is no longer serving an intended purpose. Patient comfort and dignity are improved by removing (or not inserting) indwelling catheters and by eliminating life-supporting machines that limit patient movement and interfere with family access to the bedside. Compared with discontinuation of invasive life-sustaining treatments, such as mechanical ventilation or hemodialysis, the benefits of discontinuing supplemental oxygen may seem less clear.

In the past, dependence on high-flow oxygen imposed pragmatic constraints on patients, such as effectively prohibiting transfer from the hospital to a preferred location for terminal care (eg, home or a skilled nursing facility). Dependence on tightly fitting reservoir oxygen masks also limited movement and inhibited patients' abilities to eat and speak clearly. However, newer oxygen concentrators that generate flow rates up to 10 L/min effectively overcome some of these burdens. One or 2 such concentrators can reliably supply adequate oxygen in any location with electrical wall outlets. Furthermore, high-flow, humidified nasal cannulae⁵ and transtracheal catheters⁶ can deliver oxygen at flow rates that are sufficient to replace a reservoir mask in many instances, thereby facilitating eating and speaking.

Thus, although some supplemental oxygen devices are inconvenient, oxygen is less burdensome than many other forms of life support. In addition, it may be beneficial by alleviating dyspnea.^{4 ,7} Although explanations for this benefit remain unclear and include the possibility that airflow rather than oxygen per se contributes importantly to palliation,^{7 - 8} most physicians believe that oxygen improves dyspnea.⁹ More clearly, supplemental oxygen sustains cognitive function for patients with severe hypoxemia at rest,¹⁰ and thereby preserves their ability to interact meaningfully with others. Physicians and other caregivers observe these benefits whenever unintended displacement or interruption of supplemental oxygen results in patient confusion, air hunger, or panic.

Balancing benefits against burdens for high-flow supplemental oxygen is further complicated by the potential for pulmonary oxygen toxicity. Continuous inhalation of 100% oxygen at atmospheric pressure causes injury to airways and consequent chest pain in healthy volunteers after as little as 16 hours.¹¹ Pulmonary edema typically follows after several days. Considerably less is known about the tolerance of patients with severe lung disease to inhalation of variable concentrations of oxygen in clinical settings.¹² Thus, the burden of high-flow supplemental oxygen may change over time and in an unpredictable fashion for patients who are dependent on this therapy.

Another difficulty in removing supplemental oxygen is that oxygen is often perceived as meeting a basic bodily need, and so its use may comfort family and friends by suggesting that caregivers are actively attending to the dying patient. Perhaps for this reason, some physicians regularly provide oxygen via face mask or nasal cannula after terminally withdrawing mechanical ventilation, even for those patients who are deeply sedated and unresponsive. Thus, unlike the withdrawal of most other forms of life support (although similar, for some, to artificial feeding), the unexplained withholding or withdrawal of oxygen can appear neglectful or callous.

Many illnesses for which oxygen is used are accompanied by acute impairments of cognitive function due to sleeplessness, hypercarbia, hypoxemia, or the administration or accumulation of sedating drugs. Therefore, physicians may wonder whether patients who wish to end their lives by removing oxygen are, in fact, competent to make such choices. The challenge of testing decision-making capacity in this circumstance may be compounded if patients' abilities to communicate are limited by rapid breathing or exertional dyspnea.

Some physicians will also be concerned about withdrawing a life-sustaining and potentially palliative intervention, only to replace it with analgesic or sedative medications. Doses of opioid and benzodiazepine medications that are sufficient to alter consciousness can also suppress ventilation, particularly when used in combination.¹³ Although administration of opioids and benzodiazepines does not seem to influence the time until death after palliative withdrawal of mechanical ventilation in an intensive care unit,¹⁴ such drugs could easily hasten death among spontaneously breathing patients who are only marginally able to sustain adequate ventilation.

The religious and ethical doctrine of double effect has frequently been invoked to address such situations. The doctrine holds that outcomes that would be morally unacceptable if caused intentionally might be acceptable if caused as a byproduct of some other intended effect. Although this doctrine has clear philosophical limitations,¹⁵ it has been explicitly accepted by the Supreme Court¹⁶ and remains an integral concept in end-of-life decision making.¹⁷ Accordingly, if competent patients wish to discontinue supplemental oxygen therapy and dyspnea results, it becomes acceptable and humane to relieve this dyspnea pharmacologically.

Even physicians who accept this justification, however, may be unsure about how to administer sedatives and analgesics in this situation. It is unclear what physiological changes occur after abrupt replacement of high-flow oxygen with ambient air, how patients experience these changes, and at what pace they occur. In light of such uncertainty, should clinicians sedate fully conscious patients prior to oxygen withdrawal? Will clinicians be able to titrate medications effectively and responsibly if distress develops suddenly after oxygen is discontinued? As difficult as these questions may be for physicians faced with requests for oxygen withdrawal among hospitalized patients, they become even more challenging when patients request that oxygen be withdrawn at home.

If high-flow supplemental oxygen is minimally invasive, widely available, and palliative to both patient and family, why would rational patients ask that it be discontinued? The answer to this question will be unsettling for some clinicians: these patients may be seeking relief from the burden of life itself.

Some patients never complain about the discomfort or expense associated with administration of supplemental oxygen, and yet wish to have it removed. For many terminally ill patients, progressive loss of functional capacity and sense of self lead to substantially reduced opportunities for happiness and satisfaction.¹⁸

Such limited benefits to ongoing life may be dominated by steadily increasing burdens. In addition to the dyspnea that may result from such commonplace activities as eating, bathing, or defecating, living near the threshold of dependency on mechanical ventilation may cause anxiety, despondency, and other forms of emotional distress. For some patients, fear of suffocation is ever present and episodes of panic arise whenever breathing is compromised, such as with mucous plugging. Patients are also burdened by their increasing dependence on others for transportation and hygiene.

A comparable situation confronts some patients who are dependent on hemodialysis. Voluntary withdrawal of dialysis is a common prelude to death for patients with end-stage renal disease, contributing to approximately 1 of every 4 deaths among patients with end-stage renal disease in the United States.¹⁹ The burden of dialysis does not fully explain this phenomenon. An unacceptable overall quality of life and the desire not to remain a burden to others commonly contribute to patients' decisions to discontinue dialysis.²⁰

Physicians may be concerned that by condoning a patient's plan to end his or her own life and providing sedative medication to make the withdrawal of oxygen or dialysis more comfortable, the physician may be criticized for assisting in the patient's suicide. However, when a competent patient requests that a physician discontinue life support, this is not equivalent to a request for assisted suicide.^{3 ,21} The distinction between assisting in a patient's death and removing a life-sustaining treatment holds even if the life-sustaining treatment is unobtrusive and potentially palliative. Importantly, the fact that the primary purpose of such requests may be to escape the burden of life itself, rather than the burden of therapy, does not absolve physicians of their duty to heed patients' requests for therapy withdrawal.

Although it may be difficult for physicians to overcome all of these concerns, the following 4-step approach may be helpful.

First, physicians should assure themselves, other health care professionals involved in the patient's care, and the patient's family members or close friends that supplemental oxygen is a form of life-sustaining medical treatment. As such, requests to discontinue oxygen should be honored with the same judiciousness as requests to withdraw other forms of life support. The physician should further recognize that requests to discontinue this intervention are not inherently irrational. Even if the practical burdens of oxygen are minimal and the palliative benefits real, the patient's dissatisfaction with his or her current quality of life may justifiably underlie such requests.

Second, physicians should ensure that patients requesting the terminal withdrawal of oxygen are free from undue influences, including family members' wishes, economic considerations, or treatable depression. Physicians should search for sources of distress or unfulfilled needs that can be addressed in other ways. For example, problems related to oxygen supplementation, such as an ill-fitting mask or overly restrictive apparatus, may be resolved by consulting physicians who more commonly use new technologies for supplying and conditioning high-flow oxygen.

Third, physicians should ensure that the patient has the capacity to make medical decisions. Patients should be required to show consistency by conveying the same preference on at least 2 separate occasions; to show understanding by acknowledging the alternatives to and consequences of their decision; and to show rationality by explaining why their decision serves their goals.

And fourth, physicians should ensure that patients and their family members understand the difficulty of predicting patients' experiences after oxygen withdrawal. Although hypoxia may often produce unconsciousness without discomfort, increased dyspnea, anxiety, and agitation are also possible results. Physicians should further explain that pharmacotherapy is indicated if dyspnea or anxiety do occur, but that such therapies may be difficult to titrate; undersedation may enable distress and oversedation may limit the quality of patients' final interactions with loved ones.

In an acute care facility, patients who choose to discontinue oxygen may be administered opioids and benzodiazepines to alleviate dyspnea and anxiety. In some instances, a recent temporary interruption of oxygen therapy will allow prediction of the patient's subjective response after palliative withdrawal. If such an experience predicts rapid onset of patient distress, sedation can be provided in anticipation of oxygen withdrawal, as is recommended in anticipation of withdrawing mechanical ventilation.²² When the patient's response is less certain, sedating drugs should be reserved but readily available for administration as needed after withdrawal. Dosing will be inexact, but the presence of a clinician who is experienced in the palliative titration of intravenous or oral opioids and benzodiazepines will help to optimize this difficult process.

Some patients will request physicians' assistance in withdrawing supplemental oxygen at home. In certain cases, patients provided with prescriptions for sedatives and analgesics can manage their own end-of-life care with the assistance of family members. However, a hospice nurse or a visiting physician may provide more skilled assessment, titrated dosing of medications, comfort to others who are present, and appropriate documentation of this legitimate medical service.²³

Regardless of whether supplemental oxygen is withdrawn in a hospital or at home, physicians should explain that the patient's major recourse in the event of uncontrolled distress is to reapply their oxygen. The plan can then be reconsidered in light of the experience gained. False starts and ambivalence have been described in the context of withdrawing dialysis²⁴ and also may occur after withdrawal of supplemental oxygen.