Ethical Dilemmas in Postnatal Treatment of Severe Congenital Hydrocephalus

Abstract: Severe congenital hydrocephalus manifests as accumulation of a large amount of excess fluid in the brain. It is a paradigmatic example of a condition in which diagnosis is relatively straightforward and long-term survival is usually associated with severe disability. It might be thought that, should parents agree, palliative care and limitation of treatment would be clearly permissible on the basis of the best interests of the infant. However, severe congenital hydrocephalus illustrates some of the neuroethical challenges in pediatrics. The permissibility of withholding or withdrawing treatment is limited by uncertainty in prognosis and the possibility of “palliative harm.” Conversely, although there are some situations in which treatment is contrary to the interests of the child, or unreasonable on the grounds of limited resources, acute surgical treatment of hydrocephalus rarely falls into that category.


Introduction
Ethical questions arising in the care of children with congenital or acquired severe brain injury have been debated for more than 40 years. 1 As noted elsewhere in this issue, decisions about treatment for such children give rise to some of the most challenging neuroethical dilemmas in pediatrics. 2 Questions about whether or not certain types of treatment are in the best interests of the child, and whether parents are permitted to request that a particular treatment be withheld or provided, are central to the debate. 3

Severe Congenital Hydrocephalus
The human brain has a set of central fl uid reservoirs called the ventricles. These reservoirs normally contain a small amount of cerebrospinal fl uid (CSF), which circulates through tiny channels and fl ows over the outside of the brain and spinal cord. 4 Congenital enlargement of the ventricles (ventriculomegaly or hydrocephalus) is seen in approximately 5 in 10,000 live births. 5 There are a number of different causes. In some cases, severe congenital hydrocephalus (SCH) arises from blockage of the normal fl ow of CSF, for example, because one of the drainage channels is narrowed. In other cases, the ventricles are enlarged because of damage to and atrophy of the surrounding brain, for example, because of infection or infarction. 6 In the current era, 80 percent of cases of congenital hydrocephalus are diagnosed antenatally by ultrasound. 7 The author was supported in this work by a grant from the Wellcome trust, (086041/Z/08/Z). The funders had no role in the design or conduct of this study; the collection, management, or analysis of data; or the preparation, review, or approval of this manuscript. An earlier version of this article was presented at the International Society for Hydrocephalus conference in Bristol, September 2014. 85 Ventriculomegaly can be seen in isolation; alternatively, it may be identifi ed in association with abnormalities of the brain or other organs. 8 The size of ventricles before birth is often used to classify severity, with ventricles larger than 15 mm typically labeled "severe" ventriculomegaly/hydrocephalus. 9 At the most severe end of the spectrum is a condition called hydranencephaly, in which all, or almost all, of the cerebral hemispheres are replaced by fl uid. 10 The outcome for infants with mild or moderate degrees of ventriculomegaly is variable; most do not have signifi cant long-term neurological problems. 11 In contrast, infants with severe hydrocephalus have a high rate of signifi cant long-term disability. Only 5-8 percent have a normal neurodevelopmental outcome. 12 There are various surgical treatments for hydrocephalus. Principal among these is insertion of a shunt, a tube to relieve the buildup of CSF and divert it (usually into the abdomen). The shunt insertion procedure is relatively low risk. 13 However, there is a moderately high rate of complications over time. More than half of shunts block at least once over a ten-year period, and revision or reinsertion of shunts is more complicated. 14 Approximately one in fi ve children will develop an infection related to their shunt, a serious condition potentially requiring prolonged hospitalization and repeat surgery. 15 Following prenatal diagnosis of SCH, a large proportion of women choose to terminate their pregnancy. 16 However, many fetuses are diagnosed with SCH relatively late in pregnancy. 17 Although in the UK termination of pregnancy is legally available in the third trimester (on the basis of a "substantial risk of serious handicap"), 18 in other jurisdictions termination may not be an option. Where a pregnancy is continuing, in the setting of SCH obstetricians and women may elect to focus obstetric care on maternal well-being rather than fetal survival. If the fetal head is grossly enlarged because of hydrocephalus, alternatives include caesarean section (because of the diffi culty of vaginal delivery) or cephalocentesis (needle drainage of fl uid in the fetus's brain immediately before delivery, usually leading to stillbirth). 19 The ethical challenges around obstetric care following prenatal diagnosis have been described elsewhere. 20 However, there is also potential for signifi cant ethical dilemmas postnatally. Consider the following hypothetical cases. 21

Case 1
Andrew is a 36-week-gestation newborn infant who was diagnosed on a 28-week ultrasound with severe fetal ventriculomegaly (lateral ventricles 19 mm bilaterally). His parents had been counseled before birth that he had a high chance of long-term disability. Termination of pregnancy had been discussed but was declined on religious grounds. However, Andrew's parents have requested palliative care after birth. They do not wish to put him through surgical treatment for his hydrocephalus.

Case 2
Bianca was born prematurely at 28 weeks gestation following a severe placental abruption. She needed resuscitation and emergency blood transfusion at delivery Dominic Wilkinson 86 and was admitted to intensive care. Subsequently, Bianca showed evidence of severe brain damage from lack of oxygen. Brain scans showed a very large intraventricular hemorrhage (bleeding into the ventricles) with dilated fl uid spaces. Over the ensuing weeks Bianca remained completely dependent on the ventilator for respiratory support. She showed evidence of hydrocephalus, with increasing head circumference and a bulging fontanelle. She had a persistently abnormal conscious state, with decorticate responses to painful stimuli. Bianca's parents have rejected any suggestion of withdrawal or limitation of treatment. They have requested surgical intervention for her hydrocephalus and tracheostomy for longterm ventilation.

Case 3
Carl was born at term following a normal pregnancy. He had a cranial ultrasound performed after birth because of poor feeding and concern about some unusual movements. The ultrasound revealed the presence of hydranencephaly. His parents were counseled about this condition, and it was explained that it was likely that he would die in infancy or be profoundly disabled if he survived into childhood. Carl remained stable, and his only medical treatment was provision of nasogastric feeding. His parents researched Carl's condition on the Internet and have requested insertion of a ventriculoperitoneal shunt.

Ethics and Nontreatment of Severe Hydrocephalus
In cases like that of Andrew (Case 1), there are a series of questions that we might ask. First, what is the likely outcome for him with treatment?
This fi rst question raises an epistemic challenge. In some series, two-thirds of pregnancies are terminated, 22 leaving only a small number of live-born infants to provide data on the outcome. 23 Those infants who are born alive may not be representative of the larger group of fetuses with SCH (e.g., there is a higher rate of termination in the presence of other abnormalities). In one series from Ireland, where termination of pregnancy is not legal, out of 17 fetuses with apparently isolated SCH, 7 died (mostly in the setting of cephalocentesis). 24 Out of the 10 surviving children, 5 had severe developmental delay and/or cerebral palsy, 4 had milder degrees of impairment (e.g., mild or moderate delay in learning to walk or talk), and 1 was developing normally at 4 years of age. These data are useful; however, given the small numbers there is necessarily signifi cant uncertainty about the true rate of severe disability. 25 Furthermore, the heterogeneity of the causes and course of prenatally diagnosed SCH makes it challenging to apply published fi gures to an individual case.
The second question is clearly normatively challenging. Is the level and probability of disability in children with SCH severe enough and high enough that nontreatment is consistent with Andrew's best interests? 26 It is extremely challenging to determine where the threshold should be for permissible limitation of treatment for newborn infants. 27 One approach to answering this question appeals to normative criteria. For example, limitation of treatment would be permissible if there is clear and convincing evidence (more than a 50 percent chance) of severe disability such that the child has signifi cant risk of a life not worth living. 28 Ethical Dilemmas in Postnatal Treatment of Severe Congenital Hydrocephalus 87 Do children with SCH risk a life not worth living? There is limited data on the quality of life of children with SCH. One Canadian study provides some relevant data, albeit from a mixed group of children with different degrees and causes of hydrocephalus. 29 The study asked caregivers of children attending three neurosurgical clinics to complete a validated questionnaire assessing health-related quality of life. The majority rated their child's quality of life positively. Seventeen caregivers (5%) returned responses indicating a health utility of < 0 (i.e., potentially equivalent to a life not worth living). 30 Even if the preceding normative criteria are correct (and they may not be), it is diffi cult to know whether SCH meets them.
There is a further practical question to ask that is relevant to Andrew. If a shunt is not performed, what would his outcome be? Answering that question also poses an epistemic challenge, because in the current era almost all children with hydrocephalus receive surgical treatment. 31 There are case reports (e.g., from countries or healthcare settings with limited access to neurosurgery) of untreated children surviving infancy, who develop a massive enlargement of their head. 32 Data from an earlier period provide some insights. One series from Chicago in the 1950s described 45 infants with SCH judged to have such poor prognosis that surgery was not performed. 33 Twenty-one children died (at an average age of three years), whereas 24 survived. Nine of the surviving children (followed for an average of eight years) were found to have severe physical and cognitive impairment, whereas another nine were judged to have "near-normal" function. These striking results highlight the prognostic uncertainty noted previously-surviving infants with severe hydrocephalus do not necessarily have severe disability. However, they also highlight the possibility of what could be called "palliative harm." Infants who are untreated do not necessarily die in the newborn period. Without active treatment, infants may survive for years, or indeed into adult life. However, as a consequence of not being actively treated, those children risk surviving in a worse state-with a greater degree of physical, cognitive, or sensory disability, or with other complications of their distended ventricles and enlarged head. 34

Ethics and Treatment of Severe Hydrocephalus
I have described serious potential ethical challenges for nontreatment in some cases of SCH. In contrast, Cases 2 and 3 represent situations in which clinicians might feel ethical qualms about provision of treatment. Bianca's clinical picturesevere hypoxic brain injury and hydrocephalus-would make severe cognitive and motor disability virtually inevitable if she survives. 35 Surviving children with hydranencephaly all have profound cognitive impairment. 36 Some clinicians might regard treatment as futile in either or both of these cases and could refuse to provide surgery. 37 Yet there are signifi cant challenges in determining whether treatment for Bianca or Carl is futile. The concept of medical futility has come under sustained criticism. 38 There are multiple different competing defi nitions of futility, all of which contain value judgements-about the appropriate goal of treatment and about the point at which the level or chance of benefi t is suffi ciently low that treatment should not be provided.
Fundamentally, there are only two ethical justifi cations for declining to provide treatment in Cases 2 or 3. Treatment should not be provided if it would be harmful to the child, or if it would be harmful to others (e.g., because of use of limited resources). 39 Would neurosurgical treatment be harmful to Bianca or Carl? In Bianca's case, the clinical team might be concerned about the possibility of medical treatment causing her to suffer. A recent UK court case concerning an infant referred to as "ZT" involved a very similar situation: an infant with hypoxic brain injury and hydrocephalus. 40 Expert medical professionals gave evidence of ZT's response to tracheal suction, required multiple times a day, and expressed concern that he was experiencing pain from this procedure. However, there are three arguments that undermine this concern about harm. First, those infants who are most severely affected may have little or no apparent response to pain. In ZT's case, other medical testimony pointed to his lack of responsiveness to pain or deep stimulation as evidence of the severity of his brain damage. 41 For Carl, with hydranencephaly, the absence of any cerebral cortex might be thought to be incompatible with pain perception. 42 Paradoxically then, the greater the severity of hydrocephalus, the more tolerable treatment may be, and the less we may have reason to fear harm to the child. 43 Second, a concern about suffering for the child is, or should be, something that intensive care consultants, anesthetists, or pain specialists should be able to address. Discomfort from tracheal suction is a very familiar problem in intensive care, one that neonatologists and other medical professionals have considerable experience in managing. Third, it might be the case that medical treatments cause ongoing symptoms of pain or discomfort for Bianca. Importantly, though, those symptoms are not likely to be directly related to treatment for her hydrocephalus. Shunt surgery is a relatively small surgical procedure and not usually particularly painful. Indeed, surgical treatment of her hydrocephalus might well reduce symptoms related to raised intracranial pressure. Concern about harm from medical treatment might give reasons not to continue invasive and unpleasant treatments such as prolonged mechanical ventilation for Bianca. It wouldn't necessarily provide a clear reason to decline surgical treatment of hydrocephalus in either Bianca's or Carl's case. 44 Would treatment for Bianca or Carl harm other people? In public health systems with limited resources, provision of highly costly treatment may mean that other patients are unable to access treatment. Even if treatment were not harmful, this concern might provide a strong ethical rationale for not providing treatment. 45 Yet it is unclear whether neurosurgery in such cases would be prohibitively expensive. The costs of neurosurgery for hydrocephalus range from US$10,000 for simple shunt insertion/revision to $70,000 for hospital admission associated with shunt infection. 46 Based on published rates of shunt infection and mortality in hydrocephalus, we can roughly estimate surgical treatment to cost an average of $9,400 per year of life saved across childhood. 47 Whether or not this is judged to be prohibitive will depend on the resources available and therefore the costeffectiveness threshold used, whether quality of life is included in the assessment of benefi t, and the value placed on survival in a state of severe disability. By way of illustration, if survival were judged to have a utility of 0.1 for Bianca or Carl, 48 treatment would be regarded as cost-effective at a threshold of $94,000 per qualityadjusted life year (QALY). This level of cost might be judged reasonable. 49 Furthermore, in Carl's case, the acute costs of shunt insertion are likely to be relatively modest, and he does not appear to require other expensive or limited medical resources. The most signifi cant cost for Bianca is her ongoing need for mechanical ventilation and intensive care. At a daily cost of more than $3,500, and with no apparent prospect for weaning from respiratory support, Bianca's Ethical Dilemmas in Postnatal Treatment of Severe Congenital Hydrocephalus 89 medical care over a year might well exceed $1,000,000. 50 It may be that some healthcare systems are suffi ciently well resourced that they can meet these costs as well as the healthcare needs of all other patients. However, that is not the case in most public healthcare systems.

Conclusions
How, then, should clinicians resolve dilemmas like the ones described previously for children with SCH? A palliative approach to care may be ethically permissible for some infants with SCH, particularly in situations in which the hydrocephalus is accompanied by other major abnormalities. However, it is important that prognosis-including the degree of uncertainty about the outcome and the quality of life likely to result-be assessed carefully, on the basis of the best available data, and communicated clearly to parents. Acute neurosurgical intervention for raised intracranial pressure may be a component of good palliative care for the child, both providing symptom relief and reducing the possibility of palliative harm. It should not usually be withheld, except in the context of a dying child.
Conversely, a palliative approach to treatment might be ethically mandatory in some children with the most severe forms of congenital hydrocephalus. Concerns about harm to the child, or about unreasonable use of limited resources, could justifi ably lead to treatment being withheld, even if it were desired by parents. Yet, I have argued that the focus in such cases should usually be on the harmful nature and cost of intensive care and mechanical ventilation (or other similar interventions) rather than on neurosurgery per se. This would not apply to children who do not require expensive and burdensome treatments, such as Carl.
How will advances in neuroscience impact on decisionmaking in children with SCH? Developments in neuroimaging-for example, using functional imaging or tractography-might make it possible to predict the degree and nature of longterm impairment with a greater degree of certainty. Alternatively, functional imaging or related technologies might make it possible to detect patterns of neural activity that would substantiate concerns about a child suffering. 51 Although these possibilities are intriguing, the technical challenges are considerable, and it seems unlikely that they will be available any time soon. 52 Sadly, however, such advances will not in any way resolve the fundamental normative challenges that are at the heart of these neuroethical dilemmas.  27, Wilkinson 2013, at 67-9. However, it should be noted that such infants may also have substantial reductions in terms of wider goods that they appreciate-that is, in the benefi ts that should be balanced against suffering. Accordingly, it is possible that even if infants suffer relatively little, the overall balance of benefi ts and burdens is tipped against treatment.