5.8 Withholding and withdrawing life-sustaining treatment (including artificial nutrition and hydration)

Although withholding medical treatment from the seriously ill is not new (Hippocrates et al., 1923), public debate over the appropriate use of life-sustaining treatment (LST) in seriously ill patients is at unprecedented levels, driven by multiple factors including advances in medical technology, an ageing population, ballooning health-care costs, the increased prevalence of life-threatening chronic disease, and more patients dying within health-care institutions.

LST is ‘any treatment that serves to prolong life without reversing the underlying medical condition and includes but is not limited to cardiopulmonary resuscitation (CPR), mechanical ventilation, hemodialysis, left ventricular assist devices, antibiotics, as well as artificial nutrition and hydration’ (Anonymous, 1992). Withholding LST is a deliberate decision not to initiate treatment aimed at prolonging life. Withdrawing LST involves removing a medical intervention without which life is not expected to continue due to the patient’s underlying health status. Withholding/withdrawing LST is therefore performed with the expectation that the patient will die due to the natural progression of their underlying illness. This is in contrast to physician-assisted dying and euthanasia (see Chapter 5.6) which are acts intended to hasten death to prevent or alleviate patient suffering.

The primary goal of this chapter is to present a practical guide for clinicians who are faced with the prospect of withholding/withdrawing LST (as outlined in Box 5.8.1).

The legality of withholding/withdrawing LST differs between countries. In many countries (e.g. United States, United Kingdom, Germany, Japan, Netherlands, Taiwan, and Australia) withholding/withdrawing LST is legal, although each country tends to have different approaches for withholding/withdrawing LST (Mendelson and Jost, 2003). In contrast, within the developing world, withholding/withdrawing LST is either of uncertain legal status (e.g. China and Brazil) or illegal (e.g. Kenya, Turkey, and India) (Blank, 2011).

Even in those countries where withholding/withdrawing LST is legal, it has been shown that poor clinician knowledge of the law is associated with higher levels of legal defensiveness and consequently more aggressive and unjustified treatment at the end of life (McCrary et al., 1992; McCrary and Swanson, 1999; Meisel et al., 2000). It is therefore important that clinicians have a basic understanding of the relevant laws. Since a thorough review of the relevant laws around the world is beyond the scope of this chapter, readers are advised to familiarize themselves with the laws governing this clinical practice in their own jurisdiction. Nevertheless, we highlight here some of the general legal principles that one might need to consider, using the US legal framework as an example.

While there is no absolute right to die (Vacco v Quill, 1997; Washington v Glucksberg, 1997), US case and statutory law generally support the right of a person to refuse life-sustaining medical treatment. Prior to 1976, however, US courts were reluctant to allow patients to refuse treatment if this would result in their death. In Re Quinlan (1976), the first reported US ‘end-of-life’ case, the New Jersey Supreme Court recognized that the existing right to determine one’s medical care included the right to refuse LST. Karen Quinlan had been in a persistent vegetative state (PVS) as a complication of drug and alcohol intoxication, and required respiratory support and artificial nutrition and hydration (ANH) to keep her alive. Quinlan’s father, her legal guardian, requested that the ventilator be withdrawn. The Court ruled that there was a right to refuse medical care based on one’s constitutional right to privacy—a right that is not expressly mentioned in the US constitution but is implied in a number of its amendments—and that this right could be exercised on behalf of a patient without decision-making capacity by her guardian.

In 1990, the US Supreme Court made its first decision on withdrawing LST in Cruzan v Director, Missouri Department of Health (1990). Nancy Cruzan was a young woman in a PVS following a car accident. The hospital applied for a court order sanctioning the family’s request to remove the feeding tube. The Missouri Supreme Court declined because of a lack of ‘clear and convincing’ evidence that this is what she would have wanted. On appeal, the US Supreme Court affirmed Missouri’s right to require ‘clear and convincing’ evidence, but otherwise left it to individual states to set the evidentiary burden. The Court confirmed that both patients with and without decision-making capacity have a constitutional right to refuse medical care under the 14th amendment’s Due Process Clause (that no person shall be deprived of life, liberty, or property, without due process of law). Shortly after the Cruzan case and its focus on patient preferences, the Patient Self Determination Act of 1990 was passed by the US Congress (Patient Self-Determination Act, 1990). This legislation requires that federally funded health-care institutions provide patients with information about their right to direct future medical care via written advance directives (ADs), which can be applied in the event of future incapacitation. By doing so, the Act forced health-care institutions to develop and outline policies regarding withholding/withdrawing of LST.

Withholding/withdrawing LST is declining treatment that is artificially sustaining a person’s life. In contrast, physician-assisted suicide (PAS) is the provision of means to end a life prematurely. In most countries where withholding/withdrawing LST is allowed, PAS is illegal. In Washington v Glucksberg (1997), the US Supreme Court held that while patients have the constitutional right to refuse medical treatment, individuals do not have a constitutional right to PAS (Washington v Glucksberg, 1997).

From a Western bioethical perspective, there is no moral reason to differentiate between withholding and withdrawing LST (President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, 1983; Brock, 1997; Beauchamp and Childress, 2009). Notwithstanding this, withdrawing LST is an act of commission (as opposed to withholding LST which is an act of omission), and death usually follows more quickly than for withholding LST (Sprung et al., 2003). Consequently, withdrawing LST is frequently perceived to have direct causation in a patient’s death. As a result, for many patients, family members, and clinicians withdrawal of LST carries a greater psychological burden (Solomon et al., 1993; Melltorp and Nilstun, 1997; Vincent, 1999; Dickenson, 2000). Moreover, for some orthodox religions, a distinction between withdrawal and withholding LST remains (Loike et al., 2010).

The common perception that withholding LST is morally more acceptable than withdrawing LST is further contributed to by the differential treatment the two acts are afforded in some legal jurisdictions. For instance, in New York, more onerous conditions are required for withdrawal of LST compared with withholding resuscitation (Family Health Care Decisions Act, 2010), even though there is no legal basis for doing so. At a practical level, if only withholding treatment was allowed, time-limited trials of LST would not be acceptable and patients who chose LST would need to remain on those treatments until death.

The ethical considerations around withholding/withdrawing LST have been the subject of considerable discussion. A summary of the important ethical principles that are relevant to withholding/withdrawing LST is outlined in the following sections. Since culture creates the context within which individuals comprehend the meaning of illness, suffering, and death, the moral considerations and the weight they carry are culturally determined (Vincent, 2001; Sprung et al., 2007b) and for any given patient, the most ethical option will depend on the individual circumstances of the case, and can only be determined by weighing all of the different principles.

Beneficence, nonmaleficence, autonomy, and justice are the four guiding principles of modern bioethics (Beauchamp and Childress, 2009), and any decision to withhold/withdraw LST should carefully consider each of these.

Dating back to the time of Hippocrates 2500 years ago, physicians have primarily acted to benefit their patients (beneficence) and refrained from causing harm (non-maleficence) (Will, 2011a). In modern medicine, where the ability to cure illness and extend life is widely celebrated, withholding/withdrawing LST can superficially be seen as running counter to the principle of beneficence. However, in those with serious advanced illness, clinicians need to be mindful that utilizing LST may not only be associated with significant physical harm (e.g. rib fractures frequently occur following chest compressions during resuscitation) but also be in violation of a patient’s values and goals.

Autonomy is broadly defined as the human capacity for self-determination, that is, each person has the right to choose among the best alternatives according to a self-chosen plan (Kant and Gregor, 1998) and is derived from the basic moral obligation- respect for persons. Patient autonomy has a narrower definition, being the right of patients to make medical decisions and exert control over their medical care. Patient autonomy serves as the basis not only for informed consent to treatment but also for the right to refuse treatment (President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, 1982; Will, 2011b).

When considering withdrawing or withholding LST, the desire to uphold autonomy (or more aptly, patient preferences) may come into direct conflict with other ethical principles. For instance, patients and/or their families may request interventions that the physician considers harmful. In this regard, it is generally accepted that, while patients have the right to refuse medical treatments, they do not have the right to demand harmful and non-beneficial treatments (Luce, 1995; Brett and McCullough, 2012). Moreover, some authors have noted that an overemphasis and/or uncritical acceptance of patient autonomy risks putting unrealistic responsibility on patients and their families and may in fact compromise their ability for self-determination by forcing them to make decisions for which they lack the prerequisite knowledge and experience (Quill and Brody, 1996; Billings and Krakauer, 2011).

Given the finite nature of health-care resources, justice requires these resources be fairly allocated. Despite this, many clinicians feel that they should not have to bear in mind such considerations when determining whether to withhold/withdraw LST, since their duty of care is to the individual patient and not to society at large. In reality, however, resource allocation does limit the options one considers for seriously ill patients. For example, while there are often hospital policies regarding allocation of limited intensive care unit (ICU) beds, it is clinicians who make the final determination regarding who is allocated an ICU bed. This decision is inevitably made up of judgements regarding benefits and burdens to the individual patient but also must have regard for competing demands for a bed by other patients in the hospital at any given time.

In addition to the general bioethical principles just outlined, an understanding of several other concepts is important when considering withdrawing/withholding LST.

Proportionality, a concept originating from the Catholic Church, supports the idea that LST should be forgone if it would cause more harm than benefit, and requires one to weigh the burdens and benefits of a particular treatment relative to the clinical situation and prognosis of a particular patient (Pence, 1995; Quill et al., 1997; Jansen and Sulmasy, 2002). For example, mechanical ventilation may be considered proportionate for a young, otherwise healthy adult, with a severe but reversible lung pathology/injury, but the same therapy may be disproportionate in a patient dying of metastatic cancer.

In a patient at the end of life there is a risk that LST may prolong the dying process and/or subject the patient to a quality of life that is unacceptable. In such cases where the burdens of the treatment outweigh any benefits, it has been argued that there is no ethical obligation to provide disproportionate treatment and that it is morally permissible to remove such treatments and focus exclusively on the patient’s comfort even if this results in an earlier death (Shannon and Walter, 1993).

Previously, some ethicists and theologians advocated for a distinction between ‘extraordinary’ versus ‘ordinary’ treatments as a way to decide whether to withhold/withdraw LST, with foregoing of extraordinary treatment being more ethically justified. The extraordinary/ordinary distinction is no longer considered useful given it is unclear what aspect of a treatment (usualness, complexity, artificiality, availability, cost, etc.) would make it extraordinary or ordinary, and the concept has largely been replaced by considering proportionality, that is, benefits versus burdens instead (Beauchamp and Childress, 2009).

In everyday practice, futility is a concept that is frequently used as a justification to withhold/withdraw LST (Wilkinson and Savulescu, 2011). Indeed, some have argued that physicians should use futility to unilaterally decide to withhold/withdraw LST if such treatment will not change the patient’s outcome (e.g. survival to discharge) (Schneiderman et al., 1990). Consistent with this notion, the Texas state legislature in 1999 passed legislation allowing health-care facilities to discontinue LST over patient and family objections, so long as the continuation of LST is deemed to be futile by the treating physicians and appropriate provisions are met (including an ethics committee review of the case) (Fine, 2009).

Despite this, we do not advocate the use of futility as a justification for withholding/withdrawing LST for the following reasons. First, there is no accepted definition of ‘futile’ treatment (Burns and Truog, 2007). Thus it is often unclear to a clinician what this actually means in practice (Solomon, 1993). For example, mechanical ventilation of a patient with multiorgan failure and end-stage cancer may enhance their respiratory function (and is therefore not futile from a physiological standpoint), but it is unlikely to have any significant impact on recovery (and is therefore futile in terms of achieving any improvement). Second, on a related note, what a clinician judges to be futile is inherently subjective and influenced by his own values (Truog et al., 1992; Ashby, 2011). For example, while a clinician may feel it is ‘futile’ to provide ongoing life support to a patient in a PVS, the family might feel quite differently, and there is no objective way to determine which of these value judgements is correct. Third, using futility as the basis for a unilateral decision to withhold/withdraw treatment enables clinicians to avoid having the difficult yet beneficial conversations that are frequently needed to bridge the gap between unrealistic family expectations and clinicians’ prognosis.

Abandonment is when a physician withdraws from a therapeutic relationship without providing notice or reasonable alternative options for care. Withdrawing of LST is not a form of abandonment since the withdrawal of LST should always include a plan to continue care and provide intensive symptom management for the remainder of the patient’s life. Especially around decisions to withhold LST, physicians should stress that the patient will not be abandoned and that every attempt will be made to ensure the patient’s comfort. We should discourage clinicians from using the phrase ‘withdrawing care’ for the withdrawal of LST, as this gives an implicit message that we are no longer caring for the patient once we discontinue LST.

The doctrine of double effect, originally formulated by Thomas Aquinas, applies to moral dilemmas in which it is impossible for a person to avoid all harmful actions and requires that (1) the nature of the act must be good or at least morally neutral; (2) the harmful effect must be foreseen but not intended; (3) the harmful effect must not be a way of producing the good effect; and (4) the good effect must outweigh the harmful effect (i.e. be proportionate). Thus, in the context of withdrawing LST, the doctrine of double effect has been used to justify the administration of high-dose opioids and sedatives for the purposes of pain and symptom relief during the withdrawal of mechanical ventilation, notwithstanding the unintended but foreseen risk of hastening the patient’s death (Sulmasy and Pellegrino, 1999). The doctrine of double effect and its applicability to withholding/withdrawing LST was confirmed in the US Supreme Court decision of Vacco v Quill (1997).

Those at the end of life who often cannot speak for themselves require additional protections to make sure that their rights are not violated. The need to focus on patient values and protect the patient’s best interests, when values cannot be discerned, is never stronger than when caring for the most vulnerable in society (children, elders, disabled, mentally incapacitated, minorities, prisoners, etc.). Members of cultural minorities with differing value structures and deep vitalist convictions also represent a group at significant risk of having their convictions dismissed by health care providers who do not ascribe the same value to survival (Burns and Truog, 2007).

For any given clinical case, one of the primary tasks of a clinician is to consider the consequences (benefits and burdens) of either utilizing or forgoing LST. At the end of this chapter, we consider specific information pertaining to several common LSTs, including cardiopulmonary resuscitation (CPR), cardiac/circulatory support devices, haemodialysis, antibiotics, and ANH. In general, however, we find the following questions helpful in determining the utility of any given LST.

Decision-making capacity is defined as the cognitive ability to participate in making medical decisions. Strictly speaking, capacity is different from ‘competence’. Competence is a legal determination made by the courts and relates to the presence or absence of global decision-making capacity. In contrast, health-care professionals determine capacity, which exists on a spectrum, or a sliding scale, rather than being strictly present or absent (President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, 1982; Drane, 1985). For example, a patient may have the capacity to make decisions about whether to have stitches for a wound, but may lack the sophisticated understanding to make decisions around withdrawing LST. In practice, the consequences that follow when a physician judges a person to lack capacity are often the same as a legal determination of incompetence. Indeed, in recent decades, competence and capacity have increasingly been used interchangeably in the legal, medical, and policy arenas.

At the end of life, cognitive failure and an associated lack of decision-making capacity is exceedingly common (due either to the disease itself, the treatments, or a combination thereof). In a recent study of almost 4000 adults who died, 42% were required to make important clinical decisions in their final days of life, but only 30% of those patients had sufficient capacity to do so (Silveira et al., 2010). Despite this, capacity assessments are frequently omitted under the mistaken assumption that sufficient capacity exists (Raymont et al., 2004; Sessums et al., 2011). And, even when an assessment is performed, clinicians repeatedly overestimate a patient’s decision-making capacity (Fitten and Waite, 1990; Etchells et al., 1997; Marson et al., 1997; Sessums et al., 2011).

To help avoid such pitfalls, we advocate that a systematic capacity assessment be performed and documented in the medical record, whenever a decision to withhold/withdraw LST is being considered. To this end, we suggest a three-step approach. First, objective testing of cognition and reasoning should be performed (for instance, via the MacArthur Competence Assessment Tool (Dunn et al., 2006) or Aid to Capacity Evaluation (Etchells et al., 1999)). Second, clinicians should assess for the presence of depression, which will not only impair cognitive function but may also significantly bias patient preferences against LST (Ganzini et al., 1994; Hooper et al., 1996; Blank et al., 2001). If depression is suspected, discussion with family members regarding the patient’s pre-morbid values and goals should be undertaken, and relevant treatment initiated if possible. Third, the following questions should be specifically addressed (Appelbaum, 2007):

Thus, the overarching question that a clinician should seek to address is not ‘Does the patient have general decision-making capacity?’ but rather ‘Does the patient have capacity to make this particular decision at this particular time?’ (Grisso and Appelbaum, 1998).

Up to this point, the necessary groundwork for making a decision regarding LST has been laid by assessing the legal and ethical context, the likely consequences of utilizing or forgoing LST, and the decision-making capacity of the patient. To proceed, we strongly advocate the adoption of a shared decision-making model where patients or their surrogates are encouraged to participate in selecting appropriate treatment options in accordance with the patient’s values (Marshall and Bibby, 2011).

A shared decision-making model involves active dialogue and discussion between the clinician and the patient (and/or their families and surrogate), with specific roles for each (Charles et al., 1997; Frosch and Kaplan, 1999). The clinician (1) provides patients, their families and/or surrogates with information about the likely consequences of either utilizing or forgoing LST; (2) gains an understanding of the patient’s values and goals; and (3) makes a recommendation, based not only on their own medical expertise and clinical experience but also their knowledge of the patient’s values and goals and their assessment of whether utilizing LST will achieve those goals. Patients, families, and/or surrogates—with appropriate information regarding the likely consequences of LST and the clinician’s recommendation—then make educated decisions that accord with the underlying values and goals of the patient. In this way, the shared decision-making model draws upon not only the clinician’s expertise in LST but also the patient’s, family’s, and/or surrogate’s expertise regarding the patient’s values and goals (Elwyn et al., 2010; Health Foundation, 2010).

Ideally, clinicians will gain a good understanding of a patient’s values and goals in the context of a long-term therapeutic relationship. However, in many areas the delivery of health care is often fragmented, such that patients are often cared for by providers who have no prior experience with them. In such instances, questions that may be helpful in eliciting a patient’s values and health-care goals may include: What is most important to you when you think about your health? What are you hoping for? What are your most important goals? What are your biggest fears about the future? Are there specific health states you would find unacceptable (for example, being a ventilator, dependent on others)?

A shared decision-making model is a significant departure from how medicine is often practiced (Carlet et al., 2004; Cohen et al., 2005). Indeed, in some countries/cultures, it may be unacceptable to even discuss withholding or withdrawing of LST with a patient or base decisions on the patient’s preferences and/or assessment of his/her quality of life (Thompson et al., 2004; Searight and Gafford, 2005). Nevertheless, we feel that a shared decision-making model is ideally suited to end-of-life care for several reasons. First, in the context of end-of-life decisions, data have shown that physicians are not only inaccurate in predicting what treatments their patients would want if they were seriously ill (Uhlmann et al., 1988) but also consistently underestimate a patient’s quality of life (Uhlmann and Pearlman, 1991). The latter is exacerbated by the finding that ICU physicians and nurses place higher value on quality of life as opposed to quantity of life compared to patients and families (Sprung et al., 2007a). As a result, physicians are frequently biased toward withholding and withdrawing LST compared to their patients (Uhlmann and Pearlman, 1991). Second, a 2009 systematic review of 55 randomized controlled trials conducted over a 25-year period showed that patients involved in shared decision-making made clearer treatment decisions, were more likely to decline discretionary surgical intervention when compared to their doctors, and were no worse off in terms of health outcomes (O’Connor et al., 2009; Stacey et al., 2011). Third, what constitutes an acceptable quality of life for any given patient and how to weigh the benefits and burdens of treatment in this regard can only be truly known by that individual. It is therefore imperative that clinicians do not allow their own views on quality of life to unduly influence their assessment of whether it is appropriate to withhold/withdraw LST. Finally, on a related note, a decision regarding withholding/withdrawing LST will almost inevitably involve a judgement based upon values and goals and, while the physician remains the expert on the medical treatments, the patient is the only true expert on his/her values and goals. Understanding the patient’s values and goals is therefore a crucial step in making an ethically sound recommendation and enabling patient autonomy.

It is important to note that we are not advocating leaving patients to make decisions on their own, which is a misunderstanding of patient autonomy and respect for persons. We have noticed that many practising physicians in the United States are less likely to make specific treatment recommendations. Instead they opt to present a range of treatment options without identifying which is best, ostensibly as a means of promoting patient choice and autonomy. However, such an approach actually withholds the expert opinion of the physician and not only risks causing harm to the patient but also places unreasonable and unwanted responsibility on patients and their families, especially in the context of end-of life decision-making (Billings and Krakauer, 2011).

If a patient has decision-making capacity, determining their preference for LST is usually straightforward, although this may require extended discussion regarding the patient’s health-care values and goals and how utilizing or forgoing LST will achieve those goals. Occasionally, even if a person has capacity, they may not wish to be involved in the decision-making process, due either to their personality, coping style, and/or cultural background (Azoulay et al., 2004; Searight and Gafford, 2005; Kelley et al., 2010). For instance, patients of some religious and cultural backgrounds will usually defer decision-making to family members or a religious authority (Kelley et al., 2010). Thus, it is important to ascertain the desire of each patient to participate in the decision-making process. Notwithstanding that, most patients still prefer to be informed that their condition is life threatening (Yun et al., 2010). Thus, even in cases where the patient does not want to play a major role in decision-making, it is essential to keep the person well informed of the basic information.

If a patient lacks capacity, determining their preference for LST is more complicated, but can usually be achieved via ADs and/or surrogate decision-makers as described next.

ADs are legal documents outlining preferences for end-of-life care and can take the form either of a document outlining one’s goals, values, and preferences regarding end-of-life treatments (including a living will, instructional directive, and values history) or of a health-care proxy form (where one appoints a proxy to make health-care decisions if capacity is lost) (Gillick, 2004).

The clinical utility of ADs has been widely debated. On the one hand, ADs promote early discussion about death and dying, help to alleviate patients’ anxiety about future decisions, guide decision-making when capacity is lost, and reduce stress, depression, and anxiety in surviving relatives (Detering et al., 2010). On the other hand, ADs have major limitations. First, individuals find it very difficult to predict what they will want when they are seriously ill (Danis et al., 1994; Loewenstein, 2005; Ubel et al., 2005). Second, only a minority of patients have completed ADs (only 29% of US adults had ADs according to a 2005 survey) (Pew Research Center for the People and the Press, 2006). Third, even when an AD exists, they are often difficult to locate or too vague to be of any use (Gillick, 2010).

Allowing for these caveats, clinicians should at the very least inquire about the existence of an AD and follow the AD if it is applicable to the situation at hand. In many US states, patients’ treatment preferences have been converted into medical orders (e.g. MOLST (Medical Orders for Life-Sustaining Treatment) and POLST (Physician Orders for Life-Sustaining Treatment)) that are portable and legally upheld throughout the state’s health-care system, whether the patient is in hospital, a nursing home, or at home.

In the event of incapacity, current bioethical thinking favours the idea that in order to protect autonomy, physicians should liaise with a surrogate decision-maker who, ideally aided by ADs, can assist in determining the patient’s goals, values, and preferences (Buchanan and Brock, 1989). At a practical level, this raises two questions.

First, who should be the surrogate? Ideally, the patient will have specified a proxy decision-maker via an AD. However, usually this is not the case, and it falls to the next of kin or else the person with the strongest genetic and/or emotional ties to the patient. Interestingly, studies of patients in the ICU have found that family members are frequently too emotionally distressed to make decisions regarding their loved ones (Pochard et al., 2001). Despite this, there is general academic agreement and support from many court decisions that close family members are in the best position to make decisions because they are most likely to know the patient’s values and goals (President’s Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research, 1983).

Second, how should a surrogate decision-maker decide whether or not to forgo LST? In principle, surrogate decision-makers are supposed to use ‘substituted judgement’, that is, what the patient would have chosen. In some cases, this will be indicated in an AD or have been discussed previously with the patient. In the remaining cases, surrogate decision-makers should base the decision to withhold/withdraw LST on the patient’s known values and goals or, if these are unknown, on what would generally be considered best for the patient (Pope, 2012). In reality however, achieving accurate substituted judgement is challenging (Emanuel and Emanuel, 1992). This was exemplified by one study which found that more than 30% of approximately 2600 surrogates incorrectly predicted patients’ end-of-life treatment preferences, and that prior discussion regarding preferences between patients and their surrogates failed to improve the accuracy rate (Shalowitz et al., 2006).

Not surprisingly, there can be significant disagreement as to whether or not to withhold/withdraw LST. In such cases, trying to better understand the nature of the conflict can be very helpful. In particular, we usually try to address two key areas.

First, what is the nature of the conflict? For example, is there an ethical dilemma (e.g. is this the right thing to do)? Is there legal uncertainty (e.g. is what I am doing prohibited by law)? Or perhaps there is still clinical uncertainty (e.g. uncertain prognosis, uncertain benefit)? Or a combination thereof? In answering the questions, clinicians will often be able to identify suitable strategies for moving forwards and obtaining resolution. For instance, where there are ethical problems, re-evaluating the relevant bioethical principles and/or consulting ethics committees can provide additional clarity. Where there are legal uncertainties, seeking proper legal counsel is important. And where there is clinical uncertainty, obtaining additional evidence from the literature and consulting with other colleagues can be helpful.

Second, where is the conflict arising? Is it within the treating team? Between the team and the patient/family? Or among the patient and their family? For conflicts within the treating team, an independent mediator can be helpful to draw out the relevant clinical issues and re-direct focus to achieving an outcome in best accordance with the patient’s values and goals.

For conflict between the team and the patient/family, the role of good communication cannot be overstated. For example, studies in the ICU suggest that many families lack even a basic understanding of a patient’s diagnosis, prognosis, and treatment options (Azoulay et al., 2000), and that better and more frequent communication is associated with greater consensus between families and medical teams regarding ICU treatment (Prendergast, 1997; Garros et al., 2003). Along similar lines, remembering that the emotional distress of families in ICU impairs their ability to make decisions on a patient’s behalf (Pochard et al., 2001), addressing this distress either directly or with the support of social workers, psychologists, and/or nursing staff, can assist the therapeutic relationship and may also aid in resolving or avoiding conflict (McDonagh et al., 2004; Stapleton et al., 2006) as well as improve the mental health of surviving relatives (Lautrette et al., 2007). Adoption of mediation techniques such as the ‘principled negotiation’ approach (Fisher et al., 2011) can help to focus attention on common interests rather than individuals and emotions and may be a useful tool for resolving end-of-life conflicts (Burns and Truog, 2007). In addition, hospital ethics committees—if available—can be an invaluable resource to not only provide an opinion on valid ethical options but also help with conflict resolution (Schneiderman et al., 2003).

There sometimes remain cases that cannot be resolved despite good communication. For instance, some patients and families will want to ‘do everything’ to maintain life regardless of the quality of life or the chances of ‘success’, leaving clinicians in moral distress about providing inappropriate LST. If clinicians truly believe surrogates are making decisions that are harmful to a patient, an application to appoint a different surrogate decision-maker can be made to the courts. In the rare cases where there is in fact no real harm to the patient, one option that has been suggested is that care be continued and that the focus should be on finding additional ways to support the medical staff to minimize the damage to the morale of the treating team (Burns and Truog, 2007).

Finally, for conflicts between the patient and their families, it is important to maintain focus on the patient’s goals and, again, support from colleagues in social work, psychology, and chaplaincy can be of tremendous assistance.

Once a decision to withhold/withdraw LST is made, it is important to make a clear treatment plan to manage the likelihood that the patient will become more symptomatic following the withdrawal/withholding of LST (Truog et al., 2008). The presence of interdisciplinary palliative care teams may help alleviate patient, family, and staff distress around the time of withholding or withdrawing LST. It is also important to carefully document the decision-making process, the consent of families and patients, the withdrawal process (if applicable), and the treatment of symptoms until death.

Before LST is withdrawn, clinicians should inform families of the expected signs of the dying process, including Cheyne–Stokes respiration, skin mottling, and increased secretions. Families should be given an approximation of how long the patient is expected to live but also advised of the possibility that death may not be as soon as expected. In the event there is prolonged time between withdrawal and death, opioids, sedatives and all medications should be administered in doses necessary for the patient’s comfort. Attempts should also be made to avoid unnecessary interventions during this time, including monitoring devices, alarms, blood tests, intravenous fluids, and so on (Truog et al., 2008).

At the same time, it is important to provide psychosocial support not only to patients and their families but also clinicians. The decision to withhold or withdraw LST is emotionally charged for all involved, and existential and spiritual issues may become more prominent as death nears. Referrals to psychology, social work, and chaplaincy can be helpful where available and appropriate. Families should also be allowed adequate time with the patient to say their goodbyes. Ongoing support via bereavement services is valuable in aiding families after death and a forum or other professional services for clinicians to debrief may help to decrease burnout and distress amongst professional staff (Felton, 1998; Bar-Sela et al., 2012).

Here, we consider specific information pertaining to several common LSTs.

In 1960, Kouwenhoven and colleagues first described CPR as a medical intervention for cardiac arrest (Kouwenhoven et al., 1960). The success of this simple technique, originally used in the perioperative setting and in newly established coronary care units, was subsequently applied to anyone who suffered a cardiopulmonary arrest both in and outside the hospital. In 1973, the National Conference on Standards for CPR and Emergency Cardiac Care in the United States recommended universal use of CPR for sudden death. Today, presumed consent for CPR is still the norm in US hospitals and CPR is generally attempted whenever the heart stops.

The outcomes of CPR are generally quite poor. Unlike on TV where 67% of TV hospital patients who undergo CPR survive to discharge (Diem et al., 1996), in reality the success rate is considerably lower (15–18%) (Saklayen et al., 1995; Ehlenbach et al., 2009). Moreover, among high-risk patients, the prognosis is even more discouraging with survival to discharge rates of only 6–8% in those with hepatic insufficiency, haematological malignancy, metastatic cancer, or septicaemia (Larkin et al., 2010). Similarly, both advanced age and low pre-existing functioning are predictors of poor survival following CPR. For example, cancer patients with poor performance status at the time of admission (e.g. spending > 50% of time in bed) had only a 2.3% chance of surviving to discharge after receiving CPR (Vitelli et al., 1991). Indeed, the only cancer patients likely to survive to hospital discharge after CPR are those with good functional status who suffer an acute insult such as medication-induced cardiac toxicity (Faber-Langendoen, 1991). Thus, when cardiac arrest results directly or indirectly from advanced life-threatening illness, CPR is very unlikely to enable a patient to be discharged home.

As clinicians and the public have become more aware of the poor outcomes associated with CPR—not to mention the likelihood of secondary chest pain and rib fractures—decisions not to attempt resuscitation using DNR (do not resuscitate) or AND (allow natural death) orders have become commonplace. Discussion of a patient’s so-called ‘code status’ (i.e. whether or not to attempt CPR in the event of cardiac arrest) is therefore now quite common in places such as the United States, although most physicians lack adequate training in how to conduct such conversations (Anderson et al., 2011).

Like any LST, CPR is a medical intervention that requires a clinician to make a recommendation based on its likely outcomes and the patient’s values and goals. We would therefore advise adopting the general approach we described earlier using a shared decision-making model to determine whether or not to utilize CPR. In actively dying patients or patients with significant co-morbid disease (e.g. multiorgan failure, metastatic cancer, advanced dementia), CPR is unlikely to prolong survival and should not be recommended (Blinderman et al., 2012).

In some cases, suspension of DNR/DNI orders is appropriate. Ultimately, a decision will be made in light of a patient’s values and health-care goals. For example, if a patient is planning to undergo a palliative procedure that would deliver good symptom relief, and expects post-operatively to have a reasonable quality of life and whose goal is prolongation of life and comfort, a suspension of the DNR/DNI order may be appropriate. In other cases, the patient or proxy decision-maker may prefer that no attempts to prolong life be made, even if the patient were to likely recover from an intraoperative cardiopulmonary arrest. In any case, the risks and benefits of the procedure should be made explicit with the patient or proxy decision-maker (American College of Surgeons, 1994) and a clear plan should be made with the responsible physician and the consulting interventionalist for the perioperative period.

Stopping mechanical ventilation is one of the most commonly performed withdrawals of LST. This issue is addressed in detail in Chapter 10.7.

Both continuous and intermittent inotrope therapy can provide symptomatic relief in patients at the end of life and may form part of a patient’s hospice care plan (Stevenson, 2003; Lopez-Candales et al., 2004). Similar to other LSTs, the discontinuation or limitation of inotrope therapy should be considered when the therapy is no longer consistent with the values and goals of the patient, especially when continuation of inotropes seems only to be prolonging the dying process.

The deactivation or removal of cardiac and circulatory support devices, such as pacemakers, ventricular assist devices (VADs) and extracorporeal membrane oxygenation (ECMO), should be considered when such devices are no longer meeting the patient’s goals, prolonging the dying process against the wishes of the patient, or when the benefits are outweighed by the burdens. The withdrawal of these devices is emotionally charged, since patients typically die shortly after the device is discontinued. Indeed, some describe these acts of withdrawal as ‘justified killing’ (Brock, 1992).

One scenario that is becoming increasingly common is the alert patient with a VAD or ECMO that has become a ‘bridge to nowhere’. In such cases, the patient will usually have had a VAD or ECMO placed, as a bridge to future transplantation or to enable more time for clinicians to assess the patient’s condition and prognosis and suitable treatment options, but subsequently become unsuitable candidates for either long-term device therapy (e.g. CentriMag^®) or transplantation. Such patients thereby become confined to the critical care environment for the remainder of their lives and—although death will inevitably result from comorbid illness, an acquired nosocomial infection, or complications related to the device—discomfort about withdrawal of the device is inevitable as some of these patients appear ‘well’ and are not actively dying. Ultimately, the decision to discontinue the device rests with the patient and family, and withdrawal against patient or family objections should not be done (Abrams, et al., 2014). Ethical dilemmas encountered with the use of extracorporeal membrane oxygenation in adults. Chest. 145(4), 876–882). However, a disaster setting or serious epidemic that requires the availability of LST such as ECMO may change the ethical calculus in patients maintained on such therapies, but are not believed to survive to hospital discharge (Powell et al., 2008).

Interestingly, the withholding and withdrawing of haemodialysis frequently appears more acceptable to many people, since the patient is perceived to die more ‘naturally’. Contrary to popular belief, death from uraemia is not symptom free but is often associated with symptoms such as pain, agitation, myoclonus/muscle twitching, dyspnoea, pruritis, and nausea, particularly in the last 24 hours of life (Cohen et al., 2000; Murtagh et al., 2007a, 2007b; Kane et al., 2013). Appropriate symptom assessment and management must therefore be incorporated into the care plan in patients who decline or discontinue haemodialysis (Moss, 2000), and collaboration with renal specialists to ensure appropriate selection and dosing of medications is desirable. If severe agitation or delirium develops, neuroleptics and sedatives may be necessary. Discussions with the patients and their families regarding expected survival times post withdrawal of dialysis should occur: in patients with no renal function, average survival is 8–12 days (Sekkarie and Moss, 1998; Murtagh et al., 2007b).

Patients with advanced disease (including dementia) often lose interest in food and drink, which frequently prompts the use of ANH. However, although effective in providing short-term support for acutely ill patients, ANH is unlikely to increase life expectancy in patients with advanced disease (Borum et al., 2000). In fact, ANH may actually add to suffering or even decrease life expectancy due to various complications such as line sepsis, aspiration pneumonia, diarrhoea, hypervolemia, as well as pressure ulcers, pain and local infection at the feeding tube site (Quill, 1989; Finucane et al., 1999; McClave and Chang, 2003; Casarett et al., 2005).

Despite the lack of strong evidence to support the use of ANH at the end of life, religious and cultural beliefs frequently support its use and oppose its withdrawal (Geppert et al., 2010). For example, according to Catholic, orthodox Jewish, and Islamic authorities, ANH has a special status and may not simply be removed in the same way. For example, ANH is considered basic humane care rather than medical treatment for devout Catholics (Pope John Paul II, 2004). That being said, one may be allowed to withhold or withdraw ANH if it is adjudged likely to cause more harm than benefit (Alsolamy, 2014). Similarly, for many cultures, food is believed to be necessary for the comfort of the patient, eliciting concerns that patients may be ‘starving’. This issue needs to be approached with great care and sensitivity, and families can be reassured that once the patient no longer has the desire to eat the mechanism by which they feel hunger is no longer present and that the alleviation of thirst can be achieved by stopping anticholinergic medications, and providing good mouth care, ice chips, and sips of water (Burge, 1993; McCann et al., 1994).

Legally, US appellate courts have consistently found not only that ANH is a medical procedure that can be forgone in the same manner as other treatments but also that the cause of death is the disease that causes the inability to eat rather than the lack of nutrition itself (Cruzan v. Director, Missouri Department of Health, 1990). For example, in the highly publicized case of ‘Terri’ Schiavo, a woman who had been in a PVS for 8 years, her husband (and guardian) requested that ANH be withdrawn against the objections of her parents, and the Florida Supreme Court subsequently affirmed that her husband could exercise her right to discontinue treatments. After successive court appeals, the US Supreme Court refused to hear the case but reaffirmed the right to refuse medical treatments including ANH. Similarly, federal regulations and judiciary precedence in the US recognize that residents in long-term care facilities have the right to refuse all medical treatments including ANH (Bouvia v Superior Court, 1986; Department of Health and Human Services, 1989).

Patients have a right to discontinue any and all medical treatments, including those that may be effective, like antibiotics for specific infections. The continuation or discontinuation of antibiotics is a highly contextualized medical decision at the end of life and antibiotics should not be withheld as a matter of policy from patients on hospice or those who wish to focus exclusively on comfort. In some cases, when the goal of care is to focus exclusively on comfort, starting or continuing antibiotics may assist in symptom relief (e.g. reduction of fevers from a bacterial infection) (White et al., 2003). On the other hand, discontinuation of antibiotics may be reasonable if the goal is no longer life prolongation, as antibiotics may prolong the dying process in patients with a terminal condition and acute infection.

Patients with severe neurological injury may present a diagnostic dilemma that must be resolved if the withdrawal/withholding of LST is being considered. Specifically, clinicians should discriminate between patients in a PVS and those in a minimally conscious state (MCS), since management of the two is quite different. A patient in a PVS by definition lacks consciousness and therefore does not experience pain or dyspnoea and has no awareness of suffering. A MCS in contrast shows intermittent evidence of consciousness and offers the chance of further recovery. Despite these important differences, up to 40% of patients have been wrongly diagnosed as being in a PVS when in fact they have been in a MCS (Schnakers et al., 2009), highlighting the need to more accurately differentiate the two conditions (see Chapter 5.3).

In managing a patient in a PVS, the lack of consciousness is an important clinical factor that influences management. For example, if families perceive some behaviours as being signs of distress (e.g. reflexive coughing, posturing, secretions, tremors, respiratory patterns, and other physical signs), clinicians can gently and compassionately provide reassurance that the patient does not perceive these signs as distressful. Some have argued that for patients in a PVS, the focus of the clinician should turn to serving the best interests of the family rather than the patient who no longer has any true interests (Arras, 1991). In such cases, clinicians may therefore choose to treat upsetting physical signs not to alleviate suffering in the patient but to contain the family’s distress.

In managing a patient in a MCS, it is important to optimize the conditions for neurological recovery and to attempt to communicate with them given the potential for awareness. Clinicians should also familiarize themselves with any special rules governing those in MCS. For example, in England and Wales, all decisions relating to the withdrawal or withholding of ANH in patients in MCS must be referred to the Court for a judicial ruling (Airedale Hospital Trustees v Bland, 1993; Mental Capacity Act 2005, 2005).

In patients where there is uncertainty about the neurological status of the patient, or the patient is found to have MCS, the same approach for non-neurologically injured patients should be employed when mechanical ventilation is withdrawn.

The appropriate use of LSTs in seriously ill patients is a controversial topic that continues to generate intense debate amongst private citizens as well as health-care professionals, bioethicists, lawyers, government, and religious institutions. Decisions to withhold/withdraw LST are difficult and will remain so. Ultimately, a clinician’s approach to the issue of withdrawing and withholding LST will be influenced by many factors that shape the relationship between the patient and the physician/health-care system, and include the individual culture of the hospital/clinic in which a clinician works, the broader health-care system of a country and the historical developmental of that system, the availability of health-care resources, the ethical values and cultural norms dominant in that society, religious practices/beliefs, the legal system, as well as government policies.