The withdrawal of life support is an activity (a commission) that is ethically protected because patients' autonomy and right to refuse treatments overrides the harm (death) that comes about when physicians act to withdraw life support. This in itself is interesting because most states prohibit euthanasia (or the provision of prescriptions that enable patients to take their own lives), which is in essence a commission (as opposed to an omission) that accelerates death. I'm struggling to understand the distinction, except that the withdrawal of life support restores the patient to a "natural state" and allows nature to take its course, whereas the provision of a prescription to allow a patient to overdose is a commission that seems to interfere with nature. (Jonathan Baron has written extensively about our preference for "natural states" which often leads to worse outcomes.) That takes care of the natural versus human distinction (which of course ignores that humans are part of nature), but I still struggle to understand why the patient in Oregon has to administer his own overdose, unassisted by a healthcare professional - what's the difference between a healthcare professional assisting with the administration of an overdose and accelerating death, and his removing life support and thus accelerating death, if both acts are in deference to patient autonomy, and both are commissions, and indeed both are direct actions, as opposed to indirect ones? Maybe it's because you can act directly and cause harm in respect of autonomy as long as you restore a natural state (withdrawal of life support), but you cannot act directly to cause harm in respect of autonomy by causing an unnatural state (medication overdose). I think this stream of consciousness has led me to the distinction. Maybe. The devil is in the details.
In a related vein, we use the "Doctrine of Double Effect" (DDE) attributed to St. Thomas Aquinas to justify our actions when we administer morphine and diazepam (Valium; mother's little helper) to relieve physical and psychological suffering in dying patients. Even though the administration of these medications may accelerate death, it is not our intent to do so - the intent is to relieve suffering. The DDE in essence says that unintended harms of actions that are intended to cause good are ethically permissible. But many physicians, nurses, and hospice providers either are not aware of the DDE, choose to ignore it, or assume that because their intentions can never be known with certainty they are protected from accusations of violations of the DDE's requirements. For example, when life support is withdrawn, some providers will give a massive dose of sedatives, in effect overdosing or euthanizing the patient. (Families sometimes request this, too. They do not realize that biomedical ethics proscribes the kindness that veterinarians grant our pets at the end of life.) This behavior violates the proportionality requirement of the DDE, and indeed the dose can allow observers to infer the intent of the physician. Similarly, nurses and hospice providers (and sometimes transplant surgeons) will sometimes give additional and repeated smaller doses of medications even though there is no evidence that the patient is suffering and requires more medication. One condition of the DDE is that the good must not flow through the bad - relief of suffering must not come about through death. It must come from the provision of the medication. If medication is administered when there is no suffering the harmful effects of the medication represent a moral transgression. Thus, in my practice, I select a usual dose of morphine and valium (5-10 mg each) and allow it to be administered as needed at frequent intervals guided by symptoms of distress. This link to a NEJM article about the DDE and especially the link above are good reading for those interested in these nuances and distinctions. I will not get into terminal sedation here, which appears to be another violation of the DDE but which has been declared permissible.
Halpern and Hansen-Flaschen describe the withdrawal of oxygen in JAMA in 2006, but they are concerned primarily with doing so at patient request. More common in my experience is when a decision has been made, based on patient autonomy and surrogate decision making, to withdraw other forms of life support, and we have to decide whether to provide oxygen. Indeed, the default during a "terminal extubation" (one in which it is expected that a patient will die after withdrawal of life support) is the provision of supplemental oxygen, in my experience. Part of this is routine - respiratory therapists almost always provide at least a nominal flow of oxygen after any extubation - just for measure.
When oxygen is administered after the withdrawal of life support (or in cases where life support is to be foregone in a dying patient) two extremes can happen or anything in between. At one extreme, the patient is so desperately ill that s/he expires very quickly in spite of the oxygen. At the other extreme, say in the case of a large hemorrhagic stroke, the natural dying process can be markedly prolonged by many days with oxygen provision. (The same could be said about artificial hydration or nutrition under these circumstances, but the effects of lack of these take even longer to accrue.) This is because the natural history of death from stroke involves failure to maintain a patent airway, development of low oxygen which taxes vital organs and eventually causes death. Oxygen in these cases can significantly prolong the dying process (just as mechanical ventilation can).
So why not just withdraw oxygen? It's a good question, and I think there are several intuitive answers that explain the reluctance to do so:
- Oxygen does not seem so cumbersome as invasive life support, so removing it does not seem to remove a burden (no relief from burden means no good is done; therefore DDE does not justify it, but it could be justified on the basis of patient autonomy if it was explicitly requested)
- Removal of "heroic" measures such as life support is one thing, removal of something as basic and non-invasive as oxygen is another animal - it seems like a deprivation
- Some think that oxygen relieves shortness of breath
- Oxygen, like food and water is essential for life
- Some families hold out hope that there will be a turnaround after withdrawal of life support and that oxygen is a necessary precondition for this turnaround
- We can either infer that withdrawal of life support (or life-prolonging therapies) entails removal of oxygen as well, or explicitly address this in the discussions leading up to withdrawal
- We can stop monitoring oxygen saturation at the time of withdrawal, and argue that without monitoring there is no need for, or ability to properly titrate oxygen
- We can provide a "homeopathic" flow rate of "low dose" oxygen with or without monitoring
- We can argue that removal of oxygen does not accelerate death