Dominic Wilkinson, a neonatal physician and medical ethicist, has written a searching, moving, and philosophically sophisticated book about the ethics of life and death decision making in the neonatal intensive care unit. Although I will devote much of this review to criticism, I want to say at the outset that Death or Disability represents interdisciplinary work at its very best. Wilkinson’s exposition is both rich in detail and uncompromising in its ethical analysis. He spares the reader none of the clinical, psychological, and moral complexities and uncertainties of the decisions made daily in the neonatal intensive care unit (NICU).
I especially recommend this book to moral philosophers and practical ethicists worried that their debates on issues like time-relative interests and non-identity are ingrown and academic. Wilkinson explains the relevance of these and other seemingly arcane issues to the real, wrenching decisions that parents, physicians, and the larger society must make about the treatment of newborns at risk of very severe impairment.
After briefly outlining the structure of the book and some of its complex arguments, I will focus on three of the few areas where Wilkinson’s thorough, painstaking analysis strikes me as incomplete: the role of impersonal considerations in decisions about continuing or ending the lives of impaired neonates; the implications of the time-relative interest account Wilkinson adopts for balancing potentially conflicting interests in the neonate’s survival; and the role of physicians in helping to make neonatal treatment decisions under the threshold framework Wilkinson proposes.
The book begins with two prologues about imagined visits to learn the fate of a sickly neonate: one to the Roman Temple of Carmentis in 30 AD, the other to The Carmentis Machine in 2030. These prologues are vivid and engaging, but a bit disorienting. They give the reader the impression that Wilkinson will be focused on the implications of prognostic technologies that enable physicians and parents to reliably predict a neonate’s development. But the book is not, or is much more than, an exercise in speculative bioethics. The second, longer half is devoted to decision making under the prognostic uncertainty that we will continue to face in the near future. Moreover, the first half focuses on (relative) certainty only about significant neonatal impairments – certainty that current physicians and parents can sometimes achieve. Wilkinson does not discuss the issues that would be raised by information about minor health conditions and nonmedical characteristics, which the Carmentis Machine would reveal. Those issues are interesting in their own right, but they would distract from his intense focus on the dilemmas posed by severely impaired neonates.
Wilkinson begins his analysis by explaining how the contemporary NICU forces painful decisions on parents and physicians: it keeps alive 23-28 week old neonates who would have previously died, but who face extremely severe physical and cognitive impairments once they leave the unit. He illustrates the difficulties with detailed accounts of real and composite cases, to which he refers back throughout the book. In chapter 2, he considers the prevailing standard for making those decisions: with reference to “the best interests of the child.” He questions how, or even whether, neonatal decision makers can assess those interests, or judge that a neonate will have a life “not worth living.” He argues that they make those judgments, albeit with an uncomfortable degree of speculation, by balancing the expected benefits and harms of the neonate’s continued life. But he departs from the best-interests standard in arguing that the interests of parents, siblings, and others should also be taken into account.
Chapter 3 argues that impersonal considerations, concerning the effect of a decision about neonatal treatment on aggregate well-being, do not have a significant role to play in making those decisions. This chapter raises a number of philosophically interesting claims that I discuss below. Chapter 4 concludes Part I with a richly detailed review of the potentially conflicting personal interests that are most directly implicated in neonatal treatment decisions: those of parents, siblings, and the child itself. Wilkinson defends the controversial claim that the neonate’s own interests should be discounted by its limited psychological continuity with its future self – a claim whose implications I discuss below.
Part II, starting with Chapter 5, introduces the manifold uncertainties that physicians and patients confront in the NICU, not all of which can be resolved by improved predictive technologies. I found this part of the book the most difficult, because it presents complex normative arguments that are resolved in part by the critical examination of empirical research. Chapter 5 reviews the evidence we have on the prognosis for severely impaired neonates. Wilkinson enumerates the factors that limit the inferences we can draw from existing data. For example, most studies count the deaths of infants for whom treatment was withdrawn, raising the acute problem of self-fulfilling prophecies. But counting only those neonates for whom treatment was extended may overrepresent the relatively healthy; those for whom treatment would have seemed more promising. Yet the sickest neonates also may be overrepresented in the leading studies, which come from specialized care centers. Some readers may find this methodological rigor daunting; there were times I struggled to see the forest for the trees. Perhaps the most accessible and important section of the chapter deals with the difficulties of quality-of-life assessment. These assessment blur medical prognosis with the normative evaluation of medically-challenged lives. Health professionals consistently give lower quality of life estimates than families (and much lower than those given by disabled individuals able to judge themselves); this may reflect their professional emphasis on health more than their superior knowledge.
Chapter 6 discusses ways of managing this uncertainty, focusing on the strategy of obtaining greater knowledge by prolonging treatment. This strategy raises a concern about the moral risk of extending treatment beyond the so-called “window of opportunity,” when the neonate’s survival is dependent on medical technology. Wilkinson argues for greater parental discretion on both ends: early withdrawal will often be justified given the evidence parents already have and the limited prognostic gain likely from continued treatment; parents may decide to let the child die past the “window” if the prognosis remains grim, by removing nutrition and hydration – a withdrawal that Wilkinson finds no more problematic in principle than removing a ventilator. Chapter 7 revisits the interests at stake in neonatal treatment decisions in light of these uncertainties. Its highlight is an original and highly sensitive account of the complex relationship between the interests of the neonate and its parents.
The book concludes with a careful proposal in chapter 8 for decision making under this prognostic and moral uncertainty. Wilkinson offers a two-threshold framework that carves out a space for parental discretion, between cases where the neonate’s dire prospects make continued life-support unacceptable and cases where its decent prospects make withdrawal unacceptable. He gives examples of cases that arguably fall within this “grey zone,” while acknowledging the daunting conceptual and practical problems in setting the thresholds. He finds grounds for optimism in the apparent consensus among an array of professional groups on guidelines. Less encouraging is his report of wide variability among individual physicians on specific cases, a finding that suggests that the consensus among groups may be largely verbal. In setting the thresholds, Wilkinson says very little about the impersonal considerations he discussed earlier, as well as the (ultimately personal) interests of the larger society in the cost of neonatal care. Although he has been criticized for his neglect of social costs (Foster 2013), that omission will be welcomed by those, like me, who deny that physicians, let alone parents, should engage in bedside rationing.
Wilkinson is particularly good at anticipating questions and objections; most of mine were thoughtfully addressed soon after they arose. There are, however, a few significant issues that could have benefited from further discussion:
1. In his analysis of the moral reasons for “replacing” a severely disabled neonate with a healthy child, Wilkinson does not deny that impersonal considerations have moral force; he merely argues that they provide only weak moral reasons for replacement, which can easily be outweighed even by fairly minor parental interests. His argument is, roughly, that 1) in most cases, the impersonal gain from replacing a less with a more healthy child will be smaller than the gain from having a more healthy child rather than none; 2) The gain in impersonal good from the latter is not enough to give parents a strong moral reason to have a healthy child; so 3) the moral reason parents have to replace a less with a more health child will be even weaker.
Although I doubt that impersonal considerations have any moral force, a proponent of replacement could respond by challenging a key assumption in Wilkinson’s argument: that if there is a moral reason to have a more rather than less healthy child, there is a moral reason to have a healthy child rather than no child. Johann Frick (2014) has recently argued that selection among possible children is a conditional duty, analogous to promise-keeping. One has no duty to make promises, only a duty to fulfill them if one has made them. Making a promise subjects the agent to a standard that would not otherwise apply to her. Similarly, one has no duty to have children. But, Frick contends, if one decides to have children, one has a duty to have the child with the happiest possible life; not only to make one’s child happy, but to select the child with the greatest potential for happiness. One needn’t accept this as the applicable standard for prospective parents – I emphatically don’t – to accept Frick’s more general point that whatever the correct standard is, it applies only conditionally.
In the context of procreation, Frick’s analysis challenges Wilkinson’s claim that it is inconsistent to acknowledge a moral reason to pick the happier or healthier child but not a moral reason to select either over no child. Even if this inference is blocked, however, Wilkinson could plausibly argue that the applicable standard is different for replacement than conception. It would be implausible to claim that parents had a strong moral reason to continue replacing neonates until they obtained the best one they could reasonably expect. But Wilkinson could not set a less demanding standard based on a comparison with the impersonal gains expected from conception – that comparison would simply be irrelevant.
More broadly, Wilkinson does not offer a decisive verdict on the scope, weight, and role of impersonal considerations. He shares the widely-held intuition that a world containing multitudes of people with lives just barely worth living would be worse than one with fewer, better-off people, even if the former had greater aggregate well-being. Moreover, he denies this this “repugnant conclusion” can be avoided by limiting impersonal considerations to “same number” choices, in which either choice will result in the same number of people ever existing. Wilkinson finds this limitation not only ad hoc but unhelpful, since replacement results in a different number of people ever existing. Yet at the same time, he recognizes the intuitive force of impersonal considerations in “substitution” cases, where prospective parents must choose between having a worse-off child now or waiting to have a better-off child later. Wilkinson, then, is unwilling to deny impersonal considerations some role in the NICU, even if he concludes that this role will usually be insubstantial. They remain as strangers at the neonates’ bedside, along with the personal interests of the neonate, parents, siblings, and others that Wilkinson so fully acknowledges.
2. The relevance and weight of impersonal considerations are left uncertain not only in replacement, but in treatment decisions where replacement is not at issue. Even if impersonal considerations lack great weight, they do not need much to serve as a thumb on the scales in close cases. This poses an especially significant challenge for Wilkinson, since he gives comparatively little weight to the interest of neonates in continued existence. According to Jeff McMahan’s Time-Relative Interest Account (TRIA) (2002), which Wilkinson adopts, a being’s interest in its future depends on the psychological connections it has with that future, connections that are very tenuous for a neonate. No matter how rich and rewarding its life is expected to be, a neonate’s time-relative interest in continuing it will be fairly slight. Wilkinson denies that this discounting will license treatment withdrawal in most cases, but his denial relies to an uncomfortable extent on the interests most parents have in the neonate’s survival, especially if its impairments are only moderate or mild. Not only are those interests highly contingent and variable, they may be outweighed in close cases by impersonal considerations – a prospect Wilkinson fails to address in his discussion of competing interests.
Even if parents are reluctant to withdraw support from an impaired neonate, an impersonal assessment of their interests may favor withdrawal. They may be significantly better-off in the long run without the expense and stress of raising a severely impaired child; they may be fully aware of this despite opposing withdrawal. (I agree with Wilkinson both that the challenges of raising a severely impaired child are often exaggerated, and that they can be oppressive or overwhelming for some parents.) Their willingness to sacrifice their own well-being arguably takes their personal interests off the scale, but it may have little or no relevance for the impersonal calculus. Even taking account of grief and guilt, their net well-being might be substantially increased if treatment is withdrawn. That gain in well-being counts impersonally, and it could easily outweigh the neonate’s highly discounted interest in survival. Allowing impersonal considerations any role in the NICU complicates, and arguably distorts, the formidable task of balancing the interests of actual persons– the child, parents, siblings, and the larger society.
3. I want to conclude by highlighting one of the difficulties in implementing the threshold framework that Wilkinson develops: the challenge to the physician’s traditional patient-centered role. Under that framework, it is not clear whether the physician can or should remain an advocate solely for the child’s interests. That role would require physicians to advocate withdrawal or continuation in the “grey zone,” depending on what they saw as in the neonate’s interest, against parents who had come to the opposite conclusion. Given the well-documented pessimism of physicians about prognosis and quality of life, their advocacy would likely reflect and reinforce the asymmetry between let-die and keep-alive decisions defended by Wilkinson, favoring withdrawal in close cases.
But there is a price for the clarity about potentially-conflicting interests promoted by such physician advocacy. Parents faced with agonizing decisions about whether to let-die or keep alive a very severely impaired neonate may often conflate their own interests with those of the child, a conflation facilitated by the extensive overlap in their interests. Clearly distinguishing the interests that do not overlap might improve the quality of parental decision making in some ways, but it might also have a substantial cost for parents who choose to withdraw support based in part on their non-overlapping interests. In cases where the physician believes it is in the child’s interest to continue living despite the heavy burdens for the parents, her well-intentioned advocacy to that effect will throw their “selfish” interests into sharp relief. I suspect it is much easier for parents to rationalize withdrawal decisions as made exclusively in the child’s interests than to acknowledge that their own interests tipped the balance. This may be self-deception, but it would be problematic for physicians to deny them this small mercy.
To conclude, this is an impressively serious, wise, and humane book. If there is a consequentialist aspect to Wilkinson’s analysis, it is not found in appeals to aggregate welfare or the impersonal good to resolve difficult questions. Rather, it is reflected in a deep, informed concern for all of the interests implicated in decisions about severely impaired neonates. Wilkinson’s perspective is encompassing, not impersonal; he comes close to the ideal of an empathic but unsentimental Impartial Observer, displaying a rare capacity to appreciate the multiplicity of interests and values that must be taken into account in making life-and-death decisions about severely impaired neonates.
Department of Bioethics
National Institutes of Health
Bethesda, MD, USA
Disclaimer: The views expressed are the author’s own. They do not represent the position or policy of the National Institutes of Health, U.S. Public Health Service, or the Department of Health and Human Services.
Foster, Charles. 2013. “Review of Death or Disability: The Carmentis Machine and Decision-Making for Critically Ill Children.” European Journal of Health Law 20: 532-539.
Frick, Johann David. 2014. “‘Making People Happy, Not Making Happy People’: A Defense of the Asymmetry Intuition in Population Ethics.” PhD diss., Harvard.
McMahan, Jeff. 2002. The Ethics of Killing: Problems at the Margins of Life. New York: Oxford University Press.