Notes on the Bankruptcy Problem: an Application of Hydraulic Rationing
Which newborn infants are too expensive to treat? Camosy and rationing in intensive care
Transcript of Which newborn infants are too expensive to treat? Camosy and rationing in intensive care
1
Title: Which newborn infants are too expensive to treat? Camosy
and rationing in intensive care.
Keywords: withdrawing treatment; resource allocation; infant,
newborn; medical ethics; neonatal intensive care
Word Count: 3684 words
2
ABSTRACT
Are there some newborn infants whose short and long-term care
costs are so great, that treatment should not be provided, and
they should be allowed to die? Public discourse, and academic
debate about the ethics of newborn intensive care has often shied
away from this question. There has been enough ink spilt over
whether or when for the infant’s sake it might be better not to provide
life-saving treatment. The further question of not saving infants
because of inadequate resources has seemed too difficult, too
controversial, or perhaps too outrageous to even consider.
However, Roman Catholic ethicist Charles Camosy has recently
challenged this, arguing that costs should be a primary
consideration in decision-making in neonatal intensive care.
In the first part of this paper I will outline and critique
Camosy’s central argument, which he calls the “Social Quality of
Life model”. Although there are some conceptual problems with the
way the argument is presented, even those who do not share
Camosy’s Catholic background have good reason to accept his key
point, that resources should be considered in intensive care
treatment decisions for all patients. In the second part of the
paper I explore the ways in which we might identify which infants
3
are too expensive to treat. I argue that both traditional personal
“quality of life” and Camosy’s “Social quality of life” should
factor into these decisions, and outline two practical proposals.
4
SHOULD NEONATAL INTENSIVE CARE BE RATIONED?
In his book, “Too expensive to treat? Finitude tragedy and the
neonatal ICU”, Charles Camosy, argues that costs should be central
in decision-making in neonatal intensive care units (NICU). His
argument is all the more striking in that it comes from a
Christian pro-life perspective. Contrary to the claims of some
religious thinkers, Camosy argues that a Christian ethic mandates
attention to the intrinsically social nature of all humans, and
consequently to the social consequences of medical treatment.i
i For this paper I will follow Camosy in addressing resource
allocation questions for newborn infants. But is it unfair to
“pick on the NICU”? Camosy defends his focus on the basis of the
very high costs of inpatient and subsequent care for at least some
critically ill newborn infants.(188-193: here and subsequently
page numbers in parentheses refer to Camosy 2010) However, he
accepts that his arguments could and should potentially relate to
expensive life-sustaining treatments for patients of any age
group.(193) Others have argued that neonatal intensive care
overall is more cost-effective than adult intensive care. If that
is the case, it may be that there is a stronger case for rationing
adult intensive care than NICU care.
5
Charles Camosy begins his book with a simple and stark
statement of facts. “1. We have virtually unlimited health care
needs. 2. We have limited health care resources”.(1) His
discussion focuses on the US health care system, and the
overwhelming injustice in its failure to provide even basic
healthcare needs for millions of Americans. However, Camosy also
refers to the situation in other developed countries and to the
global inequality in health care. He argues convincingly that the
question “should we ration” is the wrong one. We are already
rationing health care, and will always be rationing. The only
question is how we ration - whether we do so in a way that is
driven by the caprices of market and political priorities, or
whether there is a fairer alternative.(2-3)
This argument alone suggests that costs must be considered at
some level, for example, how many neonatal intensive care
nurseries are built, how many beds are funded, how many staff are
employed. Should funds be provided to NICUs or to other important
health care priorities such as antenatal care, childhood
vaccinations or support for those with disabilities? Camosy draws
on John Lantos and William Meadow’s work to point to the distorted
and distorting funding arrangements for NICUs in the United States
6
and to argue that these should be reformed. However, Camosy also
argues that resource allocation should be considered in the very
practical bedside decisions about whether individual infants are
treated or allowed to die. He refers to this as the “Social
Quality of Life” (sQOL) model of decision-making.
What does Camosy mean by the sQOL model? The term is not one
that has wide currency in the ethical or medical literature.
Camosy states that “The “social quality of life” model centers on
the newborn specifically in her family and social context”.(7)
However, the term is perhaps not ideal. “Social quality of life”
is used in some medical literature to refer to patients’
satisfaction with the social elements of their lives. It has also
been used by economists to refer to social influences on or
indicators of wellbeing (for example employment rates, expenditure
on education, economic inequality). Camosy means neither of these.
In fact, he explicitly rejects consideration of the newborn’s
future life experience in treatment decisions(12) (I will refer to
this as Personal Quality of Life (pQOL)).ii Instead, Camosy is
arguing that the social consequences of treating an infant should
ii Though, as we will later see, Camosy appears to implicitly
accept some pQOL judgements.
7
bear on whether or not they are treated.iii He contrasts two
versions of this idea:
Strong sQOL model: Broad social factors (for example
distribution of resources) should be a primary consideration in
the decision whether or not to provide treatment for a newborn
infant (8)
Weak sQOL model: Broad social factors should be a secondary
consideration in decisions whether or not to provide treatment for
a newborn infant, though they may rarely tip the balance against
providing treatment. (9)
Camosy argues in favour of the strong sQOL model of decision-
making for newborn infants, though he rejects any suggestion that
infants have reduced moral status (Figure 1). He does this on the
basis of two main arguments that have emerged from the Roman
Catholic tradition, but which Camosy claims are broadly accepted
outside Catholicism. The first of these is the long-standing
distinction between ordinary and extra-ordinary treatments, while
the second draws on Catholic social teaching, and a “relational
anthropology”.(9) The ordinary/extraordinary distinction has
considerable intuitive appeal, but it has also been the subject of
iii A better term for this might be the “Social Consequences model”,
nevertheless, I will stick with Camosy’s terminology.
8
considerable debate and criticism. I will set that argument aside
here, except to note that those who do not share Camosy’s strong
sQOL model might accept that some treatments could be
extraordinary, but hold that it is only the burden for the patient
that makes them so, not the burden on society. I focus here on
Camosy’s more distinctive argument, on the intrinsically social
nature of human persons. Camosy claims that because of this nature
it is in the best interests of any patient, including newborn
infants, to consider the impact of their treatment on others.
The ‘social nature’ argument is set out by Camosy in several
different ways. He provides a theological argument, based on
humans being made in the image of a triune God: “Relationality is
an essential element of the divine nature…and an essential aspect
of our nature as well”.(110) But he is also keen to establish that
this central part of Catholic social teaching has a wider
relevance to non-Catholics. Camosy notes that humans, whether
newborns or adults, live within groups. We grow and are nurtured
by our family and those around us, and later interact with a wider
social network.(160) Camosy draws from this two different
conclusions. He claims that newborn infants have duties to other
humans. “Certain kinds of moral duties are not predicated on
9
individualistic ideas about volitionality or consent. Rather they
are predicated on the nature of human beings as intrinsically
social”.(98) He also claims that acting in the “total best
interests” of a newborn requires us to consider the needs of
others.(120)
One worry about the first part of this argument is that it
sounds at first glance as if it may be committing the naturalistic
fallacy. We are social creatures therefore we have social duties.
Camosy does not set out explicitly the connection between human
nature and social duties, and it is possible that he could do so
while avoiding such concerns. But there might be other significant
disagreements about what is genuinely in our nature. For example,
it is arguably in our nature to be selfish and to seek treatments
that might prolong our life or especially those of our children,
even if they would come at considerable cost to others. It is also
in our nature, as Camosy notes elsewhere in the book, to respond
differently and preferentially to the needs of those who are
suffering in front of our eyes, compare with the needs of those
who are invisible and mere statistics.(132-3) Indeed, this is one
of the potential reasons why neonatal intensive care is funded
well, while other more basic treatments are neglected. But Camosy
10
does not want to derive a moral duty from this particular element
of our nature.
What about Camosy’s claim about the best interests of the
infant? Camosy does not provide a definition, but for the purposes
of what follows I take ‘best interests’ to refer to what would
overall be best for the child. There are two separate ways in
which it might be overall best for a newborn to consider the
social consequences of providing treatment. These have parallels
to the ways in which we might consider the interests of the
infant’s parents. The impacts of treatment on his family or on
wider society may influence the child’s future life and wellbeing.
The treatment might, for example, be so expensive or burdensome
that it leads to the break-up of his family, or it may consume all
his allocated health insurance, leaving him with no support for
later health care. This is the narrow sense in which social costs
might be relevant, and would be relevant to what Camosy calls the
weak sQOL model.(107) Alternatively, it may be thought that even
if he is never aware of it the newborn has an interest in the
wellbeing of his family and of others in society, is benefited by
their lives going well, and harmed when their lives go badly. It
is this latter interest that forms the basis for Camosy’s strong
11
sQOL model. But does it make sense to say that a newborn is
benefited by considering the health care needs of others in
society? It depends on what counts as a benefit for the child.
This clearly would make no sense to those who hold some theories
of wellbeing, for example hedonism, or some forms of preference-
satisfaction, since the newborn does not, and may not ever be
aware of these things, nor will he gain pleasure from them.
Consideration of the costs to others might be included in an
‘objective list’ account of wellbeing, though it has not
traditionally been included in such lists. But even on this basis
it is not clear if this claim is coherent.
The problem is that an assessment of whether a treatment would
benefit the child, and whether we can afford to provide it are two
different questions. We may have treatments that we can afford
that are of no benefit to the child, or treatments that would be
of benefit that we cannot afford. But Camosy’s ‘total best
interests’ collapses these distinctions. It makes it conceptually
difficult for us to claim that there is a need for more funding
for a particular treatment. We would want to claim that the
treatment is beneficial and therefore it should be funded. But on
12
Camosy’s account this treatment is not of benefit overall (albeit
because we cannot afford it).
Camosy’s attempt to include the costs of treatment within a
“total best interests” assessment is understandable. It is
understandable because there is considerable ethical pressure for
physicians to focus on what would be best for individual patients.
It would be, in a sense, a great consolation for doctors to know
that they do not have to choose which patient to benefit. They can
always act in patients‘ best interests (even if some patients die
for lack of treatment). It is also understandable because
clinicians, too, sometimes slip between justification for limiting
treatment based on the best interests of the patient, and
justification based on limited resources. It is much easier to
convince families to allow their loved child or elderly relative
to die if the families are told that treatment can no longer help
the patient (even if that is not strictly true). And sometimes the
two will coincide, as where treatment is both beneficial and
affordable, or neither beneficial nor affordable. Nevertheless it
is, I would argue, a mistake to conflate benefit to individual
patients with benefits to others. It is important to keep the two
distinct for the sake of clear thinking and honest communication.
13
It may also be that our response to disagreements will be
different depending on the reason why we are proposing to limit
treatment.
But even if the ‘social nature’ argument is unlikely to
convince all readers, there are nevertheless alternative, simple
and widely applicable arguments in favour of Camosy’s sQOL model.
Given that public healthcare is a limited resource, it is
intuitively plausible that we should seek to use that resource to
secure the greatest benefit for the greatest number of those in
society. A fair distribution of those resources ought to reflect
the needs of patients and the potential benefits of treatment, and
will necessarily incorporate an assessment of the costs. This is
accepted as one of the fundamental principles of medical
professionalism. Rationing at some level has long been argued to
be widespread, inevitable, justified, and, ultimately acceptable
to the general public. A full discussion of the arguments for and
against rationing is beyond the scope of this paper, but it is
difficult to avoid concluding that the costs of treatment should
be factored into decisions about whether to provide it. To
reiterate, the question is not whether we ration, it is how we
ration, and who we choose to deny treatment.
14
WHICH NEWBORN INFANTS ARE TOO EXPENSIVE TO TREAT?
Although Camosy’s arguments and other arguments converge in
suggesting that the social consequences of treatment should be
factored into decisions, they do not necessarily distinguish
between strong and weak versions of the sQOL model. Should social
consequences be of primary importance in treatment decisions? Should
doctors be thinking first about the potential costs whenever they are
about to start resuscitating an infant, child or adult? Camosy
does not set out clearly what he means by primary importance, but
in a related context David Archard has distinguished between
different descriptions of the importance of the best interests of
the child, involving a spectrum from “the paramount”, through “a
paramount” or “the primary”, to “a primary” consideration.
According to Archard, “A consideration that is ‘primary’ is a
leading consideration, one that is first in rank among several.”
Using this definition we might still think that doctors should
start with an assessment of whether treatment would be in the
child’s best interests, but then consider whether social factors
outweigh that benefit. The strong sQOL model suggests that social
consequences could sometimes justify not providing treatment even
when it would clearly be beneficial to the child. Tragically, this
15
is often the case even for fairly basic treatments in low-resource
countries. It is easy to decline intensive care admission when
there is no intensive care unit, or there are no vacant beds. In
well-resourced countries, though, the challenge comes in
determining when available, beneficial treatment may be withheld
or withdrawn on the grounds of cost.
When should newborns be declined intensive care, even if there
are beds available? Camosy provides some tentative suggestions. He
argues that it should be illegal to use resources to treat
imperiled newborns “that cannot possibly benefit from the
treatment”, listing as examples anencephaly, trisomy 18,
holoprosencephaly, large encephalocoeles, pulmonary vein atresia,
multicystic/dysplastic kidneys, congenital severe hydrocephalus
with absent or minimal brain growth.(176) He also proposes a form
of triage categorisation for newborns needing intensive care, with
some patients being given a “must not treat” classification on the
basis of their prognosis.(197) The basis for the triage assessment
would be on the basis of a formula such as that given in box 1.
Chance of Survival x Length of survival Short and long-term costs of treatment
16
Box 1 Camosy’s sQOL formula (Camosy does not express it in this
mathematical way, but this appears to capture the view that he
describes)
Camosy’s suggestion of refusing to provide treatment for
newborns with so-called “lethal” or “terminal” malformations is
intuitively highly plausible, however he fails to acknowledge
several problems. First, such an attempt parallels the long-
standing attempt by clinicians to identify treatment that is
‘futile’ and may, therefore be withheld or withdrawn even if the
patient or their surrogates request it. However, attempts to
codify this in the form of futility statues have been largely
unsuccessful. Second, at least in paediatric intensive care units,
published studies from the US and UK suggest that futile treatment
constitutes only a very small proportion of ICU bed days,.
Similarly, only a very small proportion of patients in newborn
intensive care have the malformations that Camosy identifies, and
the average costs per patient are small. The vast majority of NICU
patients (and the bulk of the costs of newborn intensive care) are
either premature babies born between 26 and 34 weeks of gestation
whose outcomes are very good, or babies with correctable
congenital anomalies. This does not mean that the savings from
17
avoiding inappropriate treatment are not worth pursuing - simply
that they are unlikely to yield large benefits for society. Third,
the designation of conditions as lethal or terminal is far less
clear cut than Camosy imagines. For one thing, 6 of the 16
conditions that Camosy lists (drawing on a paper by Steven
Leuthner), have severe brain abnormalities, and low survival rates
are heavily influenced by non-treatment decisions. Prolonged
survival has been described in all but one of the abnormalities
that Camosy lists. It is simply incorrect to state that a child
with trisomy 18 or holoprosencephaly “cannot possibly benefit”
from treatment. They are highly likely to die if they receive
palliative treatment. However, the reason that they are usually
treated this way is because of the severity of impairment that
they will experience if they survive. It may well be that the
limited benefit to the child is outweighed by the burdens of
treatment for the child, and for others. But central to this is an
assessment of the child’s personal quality of life.
Camosy’s equation for triage classification in intensive care
is essentially a version of the simple utilitarian principle of
maximising life-years saved. This principle has intuitive appeal,
and has been used in disaster triage and allocation of scarce
18
antibiotics. It does, however, arguably neglect other important
relevant considerations.iv For example, Camosy explicitly excludes
consideration of pQOL. Is this plausible? He rejects the use of
Quality-adjusted-life years (QALY) in health care allocation
because of concerns about the subjectivity of assessment, problems
with quantifying life quality, incommensurability of different
benefits, and the narrow nature of QALY models.(185-187) He also
rejects models of decision include pQOL like that of H. Tristram
Engelhardt, whose treatment formula is otherwise very similar to
Camosy’s own.(Box 2) Camosy’s main reason for rejecting these
models is that they potentially imply a reduced moral status for a
subgroup of newborns with poor quality of life.
Box
2.
Engelhardt’s sQOL model
But one problem with Camosy’s sQOL model is that it appears to
yield some counterintuitive conclusions. Imagine that we were
contemplating long-term mechanical ventilation support for two
children with identical chances of survival and length of life.
One of these children has anencephaly, while the other has a
iv For a detailed description of alternative allocation principles
and their merits see Persad et al.
Chance of Treatment Success x Probable Quality of Life x Length of Life
19
severe neuromuscular disorder but normal cognition. Camosy’s model
appears to imply that we have no reason to prefer to give this
treatment to one child rather than the other. To be fair to
Camosy, he appears to be sympathetic to the idea that life-
sustaining treatment would be ‘extraordinary’ and hence, optional,
for a patient in a persistent vegetative state,(71-2) and perhaps
he would make a similar judgment about anencephaly. But the reason
for treatment being extraordinary for this patient must relate
ultimately to some qualitative assessment of the nature of their
future existence. It must be related to the patient’s pQOL.
Second, it is not necessarily the case that taking into account
a patient’s future level of impairment (particularly cognitive
impairment) necessarily imputes that they have reduced moral
status. There are a number of ways that severe cognitive
impairment might affect the future interests of the child, and
hence influence the balance of benefits and burdens of treatment.
Finally, a number of Camosy’s concerns about incorporating pQOL
into treatment decisions (difficulties in predicting QOL,
uncertainties and subjectivities in measurement, problems in
trading off quality against length of life, and a slippery slope
leading to euthanasia), are also objections that have been or
20
could be raised to the sQOL model, and in particular to the
incorporation of long-term health care costs into decisions.
Camosy responds to each of these objections against the sQOL
model, but his own responses to these arguments provide a parallel
defense for including pQOL in decisions.
How would the sQOL model be applied in practice? There are
strong reasons for thinking that where resource allocation is
impacting upon decisions that the process for doing so is
transparent, and accountable. Here are two suggestions. The first
is the idea of time-limited trials of intensive care treatment.
Despite the arguments of ethicists that withholding and
withdrawing treatment are equivalent, medical professionals
continue to mark a difference between decisions not to start, and
decisions to withdraw treatment. This means that in practice
treatment is sometimes withheld because of resource limitations,
but rarely (at least explicitly) withdrawn for the same resource
limitations. One way to avoid this phenomenon, and perhaps to
provide a structure for resource-based treatment withdrawal would
be to offer some patients (those with a high risk of death despite
intensive care) a trial of treatment for a defined time period.
For example, newborns (and their families) could be offered a 48
21
or 72-hour period of intensive care with an expectation that at
the end of that time treatment would be withdrawn unless the
patient had shown a definite response to treatment. This policy
would enable more prognostic information to be collected prior to
treatment limitation. It would allow information to be collected
about the potential long-term costs of care. But what sort of
prognosis at the end of this period would justify cessation of
treatment?
The second suggestion would be to develop an explicit
prognostic threshold for treatment withdrawal based on existing
consensus. So, for example, international guidelines for
resuscitation of extremely preterm infants suggest that intensive
care should usually not be offered for infants born at 22 weeks’
gestation. Outcome studies of infants born at this gestation
indicate that if they are provided with intensive care these
infants have approximately a 10% chance of survival without
profound impairment. There are a number of serious criticisms of
the practice of basing resuscitation decisions solely on
gestational age, but an alternative would be to use the concept
of “gestational age equivalent prognosis”. On this basis, it would
be reasonable to withhold or withdraw intensive care from infants
22
with a chance of survival without profound impairment that is
equivalent to an average 22 week gestation infant ie <= 10%. For
example, this would include some severe forms of congenital
diaphragmatic hernia, some extremely small infants, and perhaps
some of the conditions in Camosy’s ‘terminal’ list.
CONCLUSIONS
Charles Camosy’s first book is a thought-provoking and well-
written attempt to grapple seriously with the challenge of
resource allocation in neonatal intensive care. He argues
persuasively that the social consequences of care should be
factored into treatment decisions for newborn infants, and indeed
for all patients. While Camosy’s argument draws heavily on
Catholic Social Teaching and has some problems, its basic
foundations are likely to be plausible to those of other religions
and no religion. Camosy is right to assert that “there are no interests
of our own that should not always and directly be evaluated in
light of the claims of others”.(120) Some newborn infants are
indeed too expensive to treat. Yet translation of this insight
into practical decision-making remains a huge challenge, both for
policy-makers and for clinicians.
23
Figure 1. Should the costs of treatment be included in
decisions about life-sustaining treatment for a newborn. A
comparison of Camosy’s view with alternative models including a
strict best interests model (only the child’s interests should be
included in decisions eg Ramsay ), the weak sQOL model (Costs may
be included where they impact on the interests of the child) eg
Shaw ), a and views that are based on reduced moral status (eg
Kuhse and Singer ).