Moral Equivalency.

Dominant Western ethical opinion concerning withdraw and withhold decisions is based in a moral equivalence thesis [12]; if there is no moral difference between withdrawing and withholding, then (all else being equal) there is no instance in which it would be allowable to withhold a treatment but not to discontinue the same treatment once it is started. Subjecting hopelessly ill patients to therapy that is unlikely to be beneficial potentially violates at least two basic principles of Western medical ethics: beneficence (doing good) and nonmaleficence (avoiding harm) [13,14]. Furthermore, if physicians cannot ethically withdraw LSTs once they are started, then they may be less willing to initiate treatments that have only a small potential for benefit out of concern that the patient may become trapped into therapy that cannot be discontinued if burdens turn out later to be too great [15]. Such trial therapy is an important strategy in treating and assessing prognosis in critically ill patients. Foregoing trial treatments because of misconceptions regarding the ethics of withdrawing them later could deny patients important potential benefits.

Many physicians believe that there is an important moral difference between withdrawing and withholding, although the more experienced the physician is with dying patients, the less likely they are to report either an ethical or a psychological difference between such decisions [16]. In a survey of 1,100 US physicians, over 90 percent agreed that requests from competent patients to withdraw treatments should be honored, but more than half the physicians also reported that withdrawing LSTs was significantly more difficult psychologically than withholding them [17]. The public appears much less willing than physicians to withhold therapy (40.2 versus 82.3 percent, respectively), although a strong majority (77.7 percent in one study) is willing to withdraw therapy [18]. Misgivings are not unique to US physicians. Studies have shown reluctance with regard to withdrawing compared with withholding of LSTs among German physicians [19] and that withdraw decisions are less common than withhold decisions among Indian [20], Lebanese [21], and Greek [22] intensivists.

Some argue that if a decision to start LSTS has been justly made, the patient has prima facie claim to have therapy continued on the basis of prior acquisition, unless they waive their claim [23]. This argument is based on the principle of physician fidelity to their patients – however, it does not command the conclusion that withdrawing LSTs is unethical—rather, to do so without the patient’s consent would be unethical. The question remains as to what to do if therapy is determined to be “futile” and yet the patient refuses to have it withdrawn. Does a patient have a “right” to demand futile therapy indefinitely merely because it was begun under a later-disproven assumption that it would be beneficial? Concern over futility is at the heart of many withholding and withdrawing dilemmas, and yet it is a problematic concept, as will be discussed later.

Doing versus Allowing: Is There an Important Moral Difference?

Another way of reasoning that there is an ethical difference between withholding and withdrawing LST is to draw a moral distinction between acting (or doing) and allowing [24]. Not initiating LST allows a patient to die of their disease, whereas withdrawing an established LST involves action by the physician, who has foresight that the action will likely lead to the patient’s death. Actions, it is argued, place moral culpability on the actor, whereas inaction supposedly places responsibility for the outcomes on the patient or their health. This reasoning allows many physicians to be comfortable with withholding LSTs, but it does not support withdrawing treatment once it is begun. A problem with trying to distinguish doing from allowing is that intention is a powerful, if not determinative, moral aspect of human decisions, and if intentions and outcomes are the same, whether the physician decides to press a button (e.g., the “off” button of a ventilator in the case of withdrawing) or decides not to press a button (e.g., the “on” button of a ventilator in the case of withholding) seems morally irrelevant.

In discrediting arguments that the morality of a decision lies in a distinction between doing or merely allowing, ethicists have posed scenarios in which doing and allowing are equally morally wrong or, alternatively, in which identical actions and outcomes are in fact morally distinct. Rachels proposes that we consider Smith, who deliberately drowns his nephew in a bathtub by holding him underwater, versus Jones, who observes that his nephew is drowning in the bathtub and, though capable of rescuing him, deliberately stands by and lets him drown [25]. Both men are culpable. Brock asks us to consider a different scenario [26]. A disabled man who is ventilator dependent wishes to have the ventilator stopped; his physician turns off the ventilator, and the man dies. In another scenario, the man’s greedy nephew, anticipating a large inheritance, sneaks into the room at night and turns off the ventilator, and the man dies. If the nephew’s act is a killing, then the doctor’s act is also – both involve identical actions and outcomes. Yet most ethicists agree that these scenarios are morally distinct. This is so is because not all features of an event are morally relevant. The morally relevant features of Brock’s cases are not the actions that occur but the motives and intentions driving each actor

In most Western countries, ethicists favor the view that doing and allowing are not sufficient, and may even be immaterial, in identifying morally relevant differences of withholding and withdrawing treatments. The morally relevant features of withdrawing or withholding are the motives and intentions of reducing patient suffering and/or other burdens at end of life and of respect for patient autonomy. Since these motives are the same with both decisions, there is widespread agreement that, in most cases, withdrawing and withholding are morally equivalent [27–30]. This position has been supported by numerous professional and international organizations [2,27–30] (see Table 14.1).

Legal Considerations.

Legal support for withdrawing or withholding LSTs is based in the United States on the ethical principle of respect for patient autonomy and in Constitutional provisions that limit government interference and protect the right of individuals to privacy. Between 1976 and 1990, the cases of Karen Ann Quinlan [31] and Nancy Cruzan [32] culminated in the codification of a patient’s right to limit or refuse LST in the Federal Patient Self-Determination Act of 1990 [33]. This law states that competent patients have the right to refuse any medical therapy, including LST, and that hospitals must apprise patients of these rights and act accordingly and without discrimination toward them. US hospitals have been sanctioned for violations of this Act [34]. In 2005, the durability of this legal decision was repeatedly tested as the case of Terry Schiavo wound its way through state and federal courts [35]. Every single court determined that withdrawal of a feeding tube was legally permissible even though it would result in Terry’s death. This principle survived attempts to legislate mandatory continuance of Terry’s tube feedings by the Florida state legislature and the US Congress; the US Supreme Court refused requests to reconsider.

Cases have sometimes arisen in which physicians want to withdraw or withhold a patient’s LSTs over objections by the patient or surrogates, as in the case of Helga Wanglie [36,37]. In that case, the court ordered family wishes to be followed. It is important to note that such cases do not question the legality of withdrawing or withholding LST but whether it can or should be done against the patient’s or their surrogates’ will.

Sometimes physicians argue that they can unilaterally withdraw therapy if it has become futile because it no longer constitutes legitimate medical therapy. Legal decisions in the United States are not always in favor of the physician and depend on issues such as patient perceptions of futility, confusion over definitions of brain death, and other complex considerations. Western countries other than the United States tend to favor withdraw/withhold practices, but legal decisions vary elsewhere; in Japan, for example, some withdraw decisions have led to investigations of homicide [38].

Limits of Physician Authority.

Physician refusal to comply with patient and family requests regarding treatments was not a prominent issue in medical ethics until the 1980s. The mid- to late twentieth century saw an evolution of medical decision making in the United States from being physician-centric (paternalistic) to being patient-centric. This change was driven by multiple factors: seemingly limitless advances in medical technology that raised complex moral questions about the nature of medical care, a political climate in the shadow of the Vietnam War in which antiauthoritarian philosophy flourished, and the increasing prominence of personal autonomy as a cultural value [40,41].

At first, the primary limitation of physician authority arose over patients’ rights to refuse medical therapies, as in the Quinlan [31] and Cruzan [32] cases, rather than on rights of patients to demand specific treatments. Before long, however, such positive rights cases did arise, as in the case of Helga Wanglie [36,37]. Although physicians were compelled to continue to treat Mrs. Wanglie, the case raised questions about the limits of patient autonomy in demanding LST.

Medical professionalism demands that therapy be based on sound medical theory and supported by medical evidence of efficacy or at least a rational belief that the therapy has potential to work. Physicians have ethical obligations recognized as far back as Plato to refuse requested treatment that cannot restore health or meet realistic patient goals “even if they were richer than Midas” [42]. Such obligations differentiate the physician’s role from that of a mere technician who is compensated for delivering a service on demand. The principle of respect for patient autonomy must be balanced against core principles and values that define what medical practice is – and what it isn’t. Futility discussions have been important in elucidating physician concerns over “inappropriate” patient demands and in reclaiming authority regarding the value of professional education and judgment in the decision to treat.

The Cost of Patient Autonomy.

Autonomy arguments are most convincing when the harms from treatment are limited to the individual patient. However, for many medical therapies, benefits and burdens are not limited to individuals but shared among communities. Financial costs are often spread out over some risk pool, such as an insurance cohort or government payers, and thus become borne by society as a whole. Huynh et al. estimated that the cost of care that was deemed futile in a study of five ICUs was in excess of $2.6 million over one 90-day period. More concerning was the additional finding that because of futile care being provided for a single patient in the study, treatment was delayed significantly for 33 other patients who needed critical care, and at least two patients died while waiting for transfer to other ICUs [39].

Defining Futility.

Labeling a treatment as futile is powerful; many physicians assume that it ostensibly allows them to unilaterally withhold or withdraw LSTs, since it is tantamount to saying that to provide it is not compatible with the ethical practice of medicine [43]. The reality, however, is not so simple. One problem (if not the problem) in preventing or discontinuing futile treatments lies in defining exactly what futile therapy is. As Tonelli states, “despite the apparent acceptance of the existence of medically futile interventions, physicians simply cannot agree on when they have a futile case in front of them” [44].

Clinical outcomes of therapy can only be tallied retrospectively. Defining futility prospectively (e.g., should we withhold this treatment because it will be futile) rests in predicting that treatment is unlikely to deliver desired results. But how improbable must a positive outcome be to be considered futile? A commonly employed definition of futility is that a physician must conclude that it will succeed in fewer than 1 of 100 cases [45]. This number appears to have been extrapolated from assertions that a 1 percent difference in outcomes in clinical research (p < 0.01) is usually considered insignificant [46]. However, in one study, when physicians were asked to propose a probability of success below which a treatment should be considered futile, the answers ranged from 0 to 60 percent. One of five physicians chose a cutoff of 20 percent or higher [46]. Even after they were educated that futility is a probability of success of less than 0.01, one-third of medical residents defined futility as a probability of success of 5 percent [47].

What outcome should be determinative in defining futility? Even using a “straightforward” outcome such as mortality presents problems because mortality is notoriously difficult for physicians to predict accurately in hospitalized patients. Several scoring systems (APACHE IV, SAPS 3, MPM III) have been shown to discriminate groups of patients with elevated risks of mortality, but all of them overestimate final mortality rates [48]. Physicians also overpredict mortality when using clinical judgment. In one study, the positive predictive value of physicians’ 2-year mortality estimates was only 57.4 percent – although, interestingly, this was comparable to the APACHE II and PREDICT risk-scoring systems [49]. Predicting future outcomes is furthermore operator dependent; the less experienced the physician, the less accurate is the prediction [50]. Neville et al. found that ICU fellows determined LST was futile on average 2 days earlier than attendings, weren’t able to cite as many reasons for their determination and were less accurate than attendings in identifying patients who would die in hospital. Of note, overall accuracy of all physicians in the study was poor, irrespective of experience; a surprisingly high number of patients adjudged to be receiving futile treatment (38 and 15 percent for fellows and attendings, respectively) ended up surviving for more than 6 months [50]. Imbus and Zawaki reported their experience in the Burn Center at UCLA, underscoring this point: “No burn is certainly fatal until the patient dies … unable to prophesy and unwilling to strip the patient of any hope he may cherish, we prefer to diagnose burns as ‘fatal’ or ‘hopeless’ only in retrospect” [51]. This is not to say that they shrank from telling severely burned patients the truth. They told their most severely injured patients that while they could not predict the future, to their knowledge, “no one in the past of your age and with your size of burn has ever survived this injury, either with or without maximal treatment.”

Is a therapy automatically futile if it fails to deliver the desired medical result? Or is it only futile if it fails to meet the social and health goals of the patient and their family? Patients and physicians often disagree about what makes a treatment futile and what benefits are worth pursuing, even if survival is unlikely. A treatment that is unlikely to extend life by more than a couple of days may be futile in the physician’s eyes, whereas the patient may believe that even a few hours of extended time with family is a valuable goal [52]. In such cases, the physician and patient are defining benefits and futility with different measuring sticks – the physician is defining futility quantitatively, while the patient’s surrogates are using qualitative definitions. Jecker describes the differences between quantitative and qualitative futility as follows: quantitative futility is the overwhelming probability that a treatment will fail to produce a desired physiologic effect, whereas qualitative futility describes a situation in which, whether or not the physiologic effect can be achieved, it does not offer a significant benefit to the patient [37]. In the former formulation of futility, evidence and physician experience predominate. In the latter formulation of futility, patient perception and values prevail.

Disagreements between physicians and patients regarding futility are based on differences in values between the patient and physician [53]. Disentangling values disagreements requires discussion, mutual respect, and negotiation [54], and persistent disagreements sometimes reach the courts. Some legislative efforts, such as the Texas Advance Directive Act (TADA), try to strike a balance between patient and physician interests by codifying a physician’s authority to withhold or withdraw treatments they deem futile, despite disagreement with the patient or their surrogates [55,56].