The principle of autonomy in the case of patients with dementia

Autonomy (‘self rule’) is the capacity to think, decide and act on the basis of such thought and decision, freely and independently. For many people, respect for autonomy requires that health or other trained professionals and third party actors help individuals to come to their own decisions through the provision of valuable information and assistance to interpret their circumstances and experiences. The decisions of the patient must be respected and followed, even when the health professional may disagree with said decision. Respect for autonomy has never been the only important ethical principle in health care and public health. Beneficence (the moral importance of doing good to others), non-maleficence (we should not harm individuals) and justice are also critical, and may conflict with patient autonomy (Hope, Reference Hope2004). Nevertheless, this principle has been ‘an axiom’ in bioethics (Beauchamp and Childress, Reference Beauchamp and Childress2001; Capron, Reference Capron2010). End-of-life care controversies, for better or worse, often turn on the question of patient autonomy (Burt, Reference Burt2005).

For patients with advanced dementia, reliance on autonomy is problematic because they “may lack decision-making capacity regarding their choices and preferences at the end of life” (Parsons, Reference Parsons, Hughes, Passmore and Lapane2010). While a patient's inability to make decisions or communicate preferences creates a challenge when making decisions about starting or withdrawing life-sustaining treatment, the issue of capacity is not unique to patients with Alzheimer's disease or other forms of dementia. For example, patients who have experienced a severe stroke or suffered from a traumatic brain injury may also be unable to communicate their preferences. As our capacity to detect dementias early improves, however, dementia generates some distinct ethical dilemmas. When dementias can be detected early enough that patients still have the capacity for autonomous choices about their own future, are they capable of anticipating all of the contingencies that might ensue?

Are advance directives sufficient?

In theory, advance directives should help to address the desire of people who are diagnosed with dementia at an early stage – and the inability of patients with advanced dementia to exercise self-determination. Advance directives “extend self-determination into the future” (Vollman, Reference Vollman2001) by allowing a person, while still competent, to articulate what they want to happen and/or who they want to decide on their behalf, if and when they are unable to make these decisions for themselves. Some argue that advance directives may be particularly appropriate for patients with dementia because, in the early stages of the disease, patients are still able to make competent decisions, but unlike younger, healthy people who write advance directives, they are able to write these directives “with the advantage of life experience and on the basis of concrete symptomatology” (Vollman, Reference Vollman2001: 163).

Yet only about 25% of the population has a living will or some other form of advanced directive and their use among people with dementia is even lower. In the 1990s, only about 15% of people with dementia had advance directives. Even when patients complete these legal documents, their physicians often do not know the documents exist. Yet even when people have made use of advance directives and their physicians are aware of the documents, there are practical and ethical barriers that may limit the effectiveness of these tools. Indeed, some question whether the focus on patient autonomy is sensible at all.

One problem with advance directives is their failure to provide sufficient instructions. They are “either too vague to be clear (for example, ‘If I am close to death’) or too medically specific to be helpful in common clinical situations (for example, ‘If I am in a persistent vegetative state’)” (Hickman et al., Reference Hickman, Hammes, Moss and Tolle2005).

In 2005, a report by the President's Council on Bioethics also expressed concern about the effectiveness of advance directives. They concluded that “appointing a proxy decision maker is sensible”, but “trying to dictate the precise terms of one's future care is often misguided or ineffective” (The President's Council on Bioethics, 2005). The Council suggested that, rather than dictating what should be done, advance directives should specify the person who should make these decisions on behalf of the patient. Despite the fact that the Council emphasized the limitations of advance directives, more recent studies suggest that they, when used, may provide real value. A study by Teno et al. (Reference Teno, Mitchell, Gozalo, Dosa, Hsu, Intrator and Mor2010), for example, found that “written advance directives, do not resuscitate orders, and orders to forgo artificial hydration and nutrition are associated with lower rates of feeding tube insertions”.

Beyond the question of whether these legal documents are sufficient to communicate the desires expressed by the person when they were competent to do so, a more profound challenge is that advance directives may not be valid because, as the result of dementia, the person who wrote them is, in effect, a different person. If ‘psychological continuity and connection’ decrease over the life course, ‘the moral bond of obligations and documentation from a prior stage of life are weakened’. In the case of dementia, “the breaks between life before and after the onset of severe dementia can be so crass that one can speak figuratively of a ‘new’, or of ‘two different persons’” (Vollman, Reference Vollman2001: 165). If we accept the assertion that dementia leads to a loss of identity, it brings into question whether an advance directive written by a person before the onset or at an early stage of the disease are valid for the ‘new’ person.

This becomes a problem if there is some reason to believe that the wishes of the demented person might conflict with the decisions she made at an earlier period in time. For example, we can imagine a person, while still competent, might decide they want to withdraw life-sustaining equipment or refrain from treating a co-morbid condition once their dementia reaches an advanced stage because they assume that this state will involve physical or psychological anguish. This possibility will be more likely if tests that detect Alzheimer's before symptoms of dementia have developed become available (Parens and Johnston, Reference Parens and Johnston2011).

But what if, once dementia progresses, the person appears to be “satisfied enough with his or her present condition and seem[s] on the face of it to have no urge to be dead” (Callahan, Reference Callahan1995)? What if Alzheimer's results in “a new form of happiness, a release from the shackles of normal cognition?” (Parens and Johnston, Reference Parens and Johnston2011). Dresser (Reference Dresser1995) argues that, in a scenario of this sort, giving priority to an advance directive denies people, “the freedom we enjoy as competent people to change our decisions that conflict with our subsequent experiential interests”.

As Callahan (Reference Callahan1995) argues, the answer to this question depends crucially on our understanding of selfhood. Does a self continue to exist as dementia progresses? If so, it is not at all clear that advance directives should override evidence that the person may have changed her mind (Kadish, Reference Kadish1992; Callahan, Reference Callahan1995). On the other hand, are the decisions made by a person before the onset or during the early stages of dementia a reflection of an ‘authentic’ self (Hope, Reference Hope1992; Kramer, Reference Kramer1993; Tan et al., Reference Tan, Hope and Stewart2003) – or as Ronald Dworkin argues, their ‘critical’ (as opposed to experiential) interests – and therefore enjoy special legal and ethical status as the clearest statement of the person's autonomous self (Dworkin, Reference Dworkin1993)? What does it mean to speak of an authentic self? Is a person with mental illness, for example, more or less authentic when the symptoms of their condition are altered by pharmaceuticals (Parens, Reference Parens2004)? Questions about the authentic self become even more problematic when a diagnosis of Alzheimer's may lead people to make decisions about the withdrawal of treatment or, in some cases, suicide.

These are fascinating questions that, ideally, policy makers and medical professionals would revisit routinely when making decisions about end-of-life care for people with dementia. Nevertheless, even as the scientific understanding of Alzheimer's disease and other forms of dementia improves, it is important to recognize that these are not puzzles we are likely to solve. They are competing perspectives on the human condition that shape our policies and demand reflection. While ethics cannot provide the ‘right’ answer to these questions, it can and does challenge us to contemplate them and think about how to develop policies that respect the different choices that people will make.

Autonomy as fiction

The previous discussion raises the question of whether it is a mistake, in clinical practice and in policy, to place an emphasis on patient autonomy in the first place. Robert Burt argues that theoretical discussions about end-of-life care in bioethics are not connected to reality (Burt, Reference Burt2005). Concerns about the authentic self make it difficult to proceed when an advance directive exists, but most of the time it does not and we do not have evidence to confirm what the person would have wanted. Instead, court decisions that allow proxy decision makers to act on the basis of what they believe the person would have wanted means that “the autonomy framework for conceptualizing end of life decision making has a distinctly artificial cast of mind” (Burt, Reference Burt2005: S9).

Burt argues that autonomy has little practical value because most people have not made these decisions; many do not want to make these decisions; and some people are not capable of making these decisions. Rather than perpetuate the myth of patient autonomy through tortured logic, Burt proposes a three part solution that is consistent with the principle of respect for persons. Unfortunately, Burt's proposed solution merely helps to highlight the lack of consensus about how to address these questions.

First, Burt argues that “no one should be socially authorized to engage in conduct that directly, purposefully, and unambiguously inflicts death, whether on another person or on oneself”. He argues that this is important because decisions that may lead to death should be hard because it is dangerous if decisions of this kind are not. Burt does not argue it is necessarily unethical to withdraw treatment, but believes it is useful to ‘promote moral discomfort’ to avoid ‘blameworthy conduct’. The difficulty with Burt's claim is that, for many people, ‘respect for persons’ does not imply that it is wrong to take one's own life. Indeed, many would argue that respect for persons mandates such an allowance, not on grounds of autonomy but because respect for persons is different than respect for their bodies.

It is reasonable to ask: “Do some of us face fates worse than death, such that it can be rational and reasonable to request help in committing suicide? And should others of us help them to die?” (Parens and Johnston, Reference Parens and Johnston2011). The answer for many is no – and there are clear reasons to be concerned about the ethics of physician-assisted suicide – but supporting the principle of respect for persons is not sufficient to reach that conclusion.

Second, Burt argues that “decisions that indirectly lead to death should be acted upon only after a consensus is reached among many people”. The logic behind this suggestion is similar to the first. Involving more people in the decision-making process will avoid a situation in which treatment is withdrawn without careful consideration. But is it realistic, in a country with heterogeneous values, to assume that consensus is realistic?

The problems with this requirement are highlighted powerfully by the celebrated case of Terri Schiavo. Pro-life groups and their supporters in Congress were actively involved in protesting the removal of a feeding tube from Terri Schiavo, a woman who, after a heart attack, suffered extensive brain damage due to lack of oxygen and, after 2 months in a coma, was diagnosed as being in a persistent vegetative state (PVS). Schiavo's husband, Michael, who was her legal guardian, argued that his wife had told him she never wanted to be kept alive with artificial support. Her parents fought to prevent Michael Schiavo from removing their daughter's feeding tube. Despite overwhelming medical evidence that Ms Schiavo was in a minimally conscious state with no hope of recovery, the case lasted for seven years and involved several court challenges.

Along the way, right to life groups staged rallies to support Ms Schiavo's parents and several conservative politicians, including Florida Governor Jeb Bush and President George W. Bush joined their cause. Although the efforts by Schiavo's parents to prevent the removal of their daughter's feeding tube were unsuccessful, the case helped to galvanize the right to life movement. Robert Blendon et al. (Reference Blendon, Benson and Herrmann2005) found that the American public was “divided on the question of whether Schiavo's feeding tube should be removed”, but “opposition to removing Schiavo's feeding tube was associated with opposition to abortion”.

The Schiavo case invoked a powerful ‘slippery slope’ argument that equates the decision to end life-sustaining treatment with physician-assisted suicide. Nancy Berlinger rejects this claim and argues that “palliative care and withdrawing life-sustaining support are legally and morally distinct from physician-assisted suicide” (Guglielmo, Reference Guglielmo2010). Existing law clearly supports Berlinger's position, but this argument continues to be used by the pro-life movement. The Schiavo case illustrates the problem with Burt's call for consensus in end-of-life decisions. In any culture with heterogeneous values, it highly unlikely that it will be possible to reach a consensus among many people.

Burt's third solution is that end-of-life decisions should not be made “at the bedside of a specific dying person” but should be the result of “systems-wide decisions about available resources, personnel, and institutional settings”. Removing the focus away from particular dying people to system-wide decisions is ‘psychologically protective’ (Burt, Reference Burt2005: S11). If we do not know what an autonomous patient wanted, we should not pretend that any individual can or should make this decision alone. Doing so puts too much burden on that individual and invites abuse. Instead, we should recognize the need for and appropriateness of collective decisions. Along with protecting the interests of the person who is dying, Burt argues that this approach may also alleviate the burdens faced by surrogate decision makers. This third criterion is consistent with the American Medical Association's (AMA) insistence that bedside rationing is unethical in medicine more generally (AMA Code of Medical Ethics: http://www.ama-assn.org/ama/pub/physician-resources/medical-ethics/code-medical-ethics/opinion813.page). It is not at all clear, however, how it would be possible to meet Burt's first criteria if all such choices are made for social aggregates – rules that apply invariably to a population can never embody a respect for individual persons.

Are our lives our own?

Burt is not alone in questioning whether patient autonomy should be the dominant principle in end-of-life planning, but others who reject this principle do not necessarily endorse Burt's process. For example, some may prefer “to delegate their autonomy to a family member, religious leader, or others, and defer discussions about prognosis and treatments for cultural or other reasons” (Jennings et al., Reference Jennings, Kaebnick and Murray2005). Family involvement in end of life and other care decisions may help to preserve a patient's identity, and reflect the priority that they place on these relationships. The recognition that family involvement in decision making may reflect patient preference rejects a narrow conception of autonomy as unrealistic and undesirable, but is consistent with the principle of ‘respect for persons’. This is why Ho argues that, while we should respect the desire for family involvement in end-of-life decisions, it is important to guard against the possibility that “patriarchal and other oppressive relationships often disguise manipulation, exploitation, control and abuse as love and familial bond” (Ho, Reference Ho2008).

A more radical move away from autonomy – whether it is exercised by the individual patient or in consultation with other in that person's social network – is offered by the pro-life movement. Many pro-life groups reject the idea of the patient autonomy framework because they do not believe people have a right to choose when and under what circumstances to die. These decisions are made by God. As Lutherans for Life explain, “The Bible clearly tells us that there is a time to die (Ecclesiastes 3:2). The Bible is also clear in reminding us that our times are in God's hands (Psalm 31:15). The Lord is the One who gives and takes life (Job 1:21). This is not our right” (Bartlett and Rehder, Reference Bartlett and Rehder2010).

Rejecting completely the idea of an autonomous patient (as opposed to suggesting that it may, under some circumstances, be impractical to rely on this principle) is inconsistent with mainstream bioethics and most existing law, but it is critical to be aware of this perspective because, in the United States, the Christian Right has a powerful impact on health and social policy. Along with their objection to physician-assisted death, pro-life groups have frequently argued against the removal of ventilators and feeding tubes and the use of palliative sedation because, in their view, removing technology that keeps people alive – or providing palliative sedation, which is the monitored use of medications to relieve pain and distress for patients with a terminal diagnosis – are the equivalent of murder. The majority of scholars in bioethics, as well as US courts, reject the argument that palliative sedation and physician-assisted death are ethically or legally equivalent, but this perspective does influence the politics of end-of-life care and perpetuates myths among health care professionals that may alter practice.

Although members of the pro-life movement express support for the idea that futility is not ethical medical practice, the circumstances under which members of pro-life groups are willing to accept the conclusion that care is futile are quite limited. Indeed, there is a religious divide in the United States about whether health policy should have as its central purpose the relief of suffering. Evans found that, although the majority of Americans view the relief of suffering as a legitimate goal, many conservative Protestants (evangelicals, fundamentalists and Pentecostals), do not always support using reproductive genetic technologies to relieve suffering (Evans, Reference Evans2006: 1069).

Do we agree on the meaning of futility?

There is broad consensus in bioethics that medical professionals should avoid futility. Interventions that generate burden without benefit and which prolong, rather than alleviate suffering, are not appropriate. If an intervention harms more than it helps, it is not ethically sound (Saucier et al., Reference Saucier, Berlinger, Thomson, Gusmano and Wolfe2010). But as the discussion above suggests, there are huge disagreements about the meaning of futility. Its power as an abstract principle seems to evaporate once we try to apply it.

For example, Carol Levine argues that the recommendations of the 2005 President's Council on Bioethics described above reflect, not just doubts about whether advance directives are sufficiently precise, but “a sustained critique of individual autonomy as a guiding principle” and an implicit claim that care is never futile. According to Levine:

Although few are willing to defend the proposition that there is always something you can do, no one would deny that it is difficult for families and health professionals to acknowledge they have reached a point of futility. While it is understandable and – as Burt argues, desirable – for end-of-life decisions to be difficult, US policies and politics clearly tip the balance in the direction of treatment, even in the face of strong evidence that such care does more harm than good.

Financial incentives

The use of palliative care in the United States has increased during the past decade, but most advocates argue that it is still underused and is particularly difficult to arrange for people with dementia (Meier et al., Reference Meier, Lim and Carlson2009; Quill, Reference Quill2008). Because the majority of people who die of dementia-related causes die in nursing homes, it is important to understand the barriers to palliative care in these institutions.

Palliative care is not the same as hospice, but they are often confused in the United States because the Medicare hospice program is the primary mechanism available for providing palliative care services to patients. Unfortunately, the rules governing the Medicare hospice program undermine the goal of providing early palliative care to patients with dementia. First, to qualify for the Medicare hospice benefit, a patient must receive a diagnosis that they have a life expectancy of 6 months or less. Because patients diagnosed with Alzheimer's often have life expectancies that exceed 6 months and may need a level of geriatric-psychiatric care difficult to obtain in hospice or deliver in a home setting, physicians caring for these patients often recommend hospice too late (Mitchell et al., Reference Mitchell, Kiely, Miller, Connor, Spence and Teno2007). Second, they must agree to forgo all curative care and accept hospice benefits only. Third, the hospice program must coordinate its efforts with the health professionals in the institution – most often the nursing home – where the patient with dementia has been receiving their care. To do so, the “nursing home staff must invite an outside team into their facility” (Meier et al., Reference Meier, Lim and Carlson2009). Fourth, nursing homes are penalized financially if they enroll a patient in the Medicare hospice program rather than transferring them to an acute hospital. Finally, the quality measures used to evaluate nursing homes also encourage them to transfer dying patients to hospitals because they assume that outcomes like weight loss and functional decline, both of which are associated with late-stage dementia patients, are markers of poor quality care (Meier et al., Reference Meier, Lim and Carlson2009).

Financial and regulatory incentives also play a role in encouraging the use of feeding tubes among patients with late-stage dementia. The development of the percutaneous endoscopic gastrostomy (PEG) feeding tube in the late 1970s reduced the risk of pressure sores and aspiration associated with earlier technologies (Gauderer, Reference Gauderer1991). First used on children, the success of the new technology encouraged its use for a broader array of patients, including patients with dementia (Hartsell and Williams, Reference Hartsell and Williams2010). In the later stages of the diseases, people with Alzheimer's and other forms of dementia often experience difficulty eating. “Common complications of late-stage dementia include anorexia, dysphagia, aspiration, and apraxia, and these often result in malnutrition and weight loss” (Hartsell and Williams, Reference Hartsell and Williams2010: 55). Recent studies suggest that use of PEG tubes in patients may not be necessary and provide little clinical benefit for patients with dementia, but they are still used with great frequency and removing them is controversial (Gillick, Reference Gillick2000; Murphy and Lipman, Reference Murphy and Lipman2003). Some argue that because assisted oral feeding, which is often a preferable alternative, is time consuming, nursing homes face a financial incentive to rely on PEG (Li, Reference Li2002). This example captures the complexity of what good care entails for patients with late stage dementia and the challenge of providing it in understaffed nursing homes. Perhaps if the Medicaid payment rates to nursing homes were higher this might be less of an issue? But this would require addressing the deeper issue of our collective unwillingness to pay for decent long-term care, particularly for those who cannot speak up for themselves when they are being ill-treated.

In addition to the financial challenges associated with providing appropriate alternatives, “the desire to avoid regulatory sanctions, bad publicity, and liability exposure creates a further incentive for nursing homes to provide tube feeding” (Finucane et al., Reference Finucane, Christmas and Leff2007). This highlights another way in which existing policy may undermine good end-of-life care for people with dementia. While there is little evidence that medical malpractice lawsuits explain the high rate of health care spending in the United States, the threat of lawsuits – and what Alan Meisel et al. (Reference Meisel, Snyder and Quill2000) “legal myths” – do influence practice. Although relatively few hospitals, nursing homes or physicians are sued for administering palliative sedation, failing to provide life-sustaining treatment, or withdrawing life-sustaining treatment – the fear that this might happen is often sufficient to prevent health care providers from acting in the best interests of their patients.

There is also evidence that limited understanding among many physicians contributes to the unnecessary use of PEG. In a study of primary care physicians, Shega et al. (Reference Shega, Hougham, Stocking, Cox-Hayley and Sachs2003) found that “a majority of physicians believe speech therapists, nurses, and nutritional support teams recommend feeding tubes” and that many physicians recommended the use of a PEG because they had received a request from a nursing home.