Clinical scenario

‘Sarah is a baby girl born by an emergency caesarean section following a period of observation for non-reassuring cardiotocographic recordings. She was initially ‘flat’ and received positive pressure ventilation for 3 min before establishing spontaneous breathing. Her Apgar scores were 1, 6 and 8 at 1, 5 and 10 min, respectively; cord pH was 7.08 and standard base excess (sBE) was −12.1. Sarah stayed with her mother as she was breathing normally and centrally pink despite being mildly hypotonic with minimal activity. At 10 hours of age, she started to develop recurrent seizures. Cerebral MRI showed extensive diffusion restriction patterns compatible with acute hypoxic–ischaemic insult.’

Sarah is a composite case, developed to include real events that we and others have observed. Unfortunately, many neonatal units receive similar cases every year and they often end up not offering therapeutic hypothermia, the only available treatment with proven safety and efficacy to this condition.1 The current guidelines are not inclusive and do not consider borderline cases.2 3

The simple question clinicians should ask themselves, is it unreasonable to treat a newborn with perinatal asphyxia and moderate encephalopathy? Babies, in a situation like Sarah, may lose the opportunity to be treated with therapeutic hypothermia because they miss a single criterion from the current cooling guidelines. The selection criteria in the initial randomised controlled trials of hypothermia were developed to identify the highest risk newborns who had been exposed to hypoxia–ischaemia. Newborns who had lower levels of risk were pragmatically excluded. Now that the evidence for benefit is well established,1 4 we propose that those entry points should be used as a guideline rather than a limiting factor for intervention. Collectively we should expand and ideally simplify those criteria based on our clinical judgement to maximise the potential benefits.

Apgar score can be one of the limiting factors to offer treatment. Apgar score, despite being an objective measure of the physiological changes of newborns at birth, is subjective to the person who is scoring. It has methodological weaknesses and the interobserver reliability is poor.5 6 In one randomised controlled trial of selective head cooling, some newborns with Apgar scores ≥6 developed moderate to severe encephalopathy and were enrolled.7 Similarly, pH is weakly associated with risk of encephalopathy and neonatal encephalopathy can occur at substantially higher pH values than those used in the large clinical trials of therapeutic hypothermia. Vesoulis et al found that universal cord blood gas screening with a pH threshold ≤7.10 and mandatory encephalopathy examination detect improved timely initiation of therapeutic hypothermia.8 Similarly, Yeh et al found that the absolute risk of an adverse neurological outcome was significantly increased with cord pH below 7.10.9 Alternatively, lactate may be a more consistent index than pH or sBE, with most cases of hypoxic–ischaemic encephalopathy found to have lactate above 8 mmol/L.10

Hypoxia–ischaemia is a dynamic continuum of clinical and pathological phenomena. There are no distinct boundaries between mild, moderate and severe cases. The precise timing of the insult and the type of injury are unclear in most cases. Perinatal hypoxia can be intermittent or total or a combination.11 Moreover, the stage of encephalopathy and the severity of electroencephalographic changes evolve over time.7 12 Biselele et al reported both improvement and deterioration in the Thompson score during the first 6 hours of life of newborns with perinatal encephalopathy.13

Clinical neurological assessment is intrinsically subjective and needs rigorous training and cross-validation with video-recordings to be a valid tool. Electroencephalography (EEG) is a much more objective measure but needs significant training. Interestingly, in a long-term follow-up study of an uncooled cohort of newborns with mild and moderate hypoxic–ischaemic encephalopathy graded clinically or by early EEG, survivors of mild encephalopathy had higher rate of disability than their peers and had cognitive outcomes similar to those of children with moderate encephalopathy.14 This emphasises the difficulty of early prognostication and that the severity of the injury may not be reliably quantified by clinical assessment or EEG alone. Moreover, there is some evidence that the neuroprotective effect of hypothermia may vary according to the underlying pathology.15 Given these confounding factors, it is currently challenging for clinicians to decide beforehand who will or will not benefit from therapeutic hypothermia. It is currently ambiguous how borderline cases of hypoxic encephalopathy will evolve without intervention. What we know is that therapeutic hypothermia improves the stage of encephalopathy if the intervention is started within 6 hours of birth and improves the chance of normal survival.16

The decision to offer hypothermia for neuroprotection may engender subsequent litigation because treatment presupposes an acute intrapartum injury. Conversely, failing to offer cooling may be interpreted as a violation of the standard of care.17 Understandably, the balance between costs, risks and benefits of treatment is not clear in many situations. This conundrum should be mitigated by the potential benefits from treatment. The proven safety of therapeutic hypothermia in an intensive care environment should ameliorate the fear of overtreating. Conversely, the window to initiate the intervention is limited. Therefore, in many cases the decision to treat should be based on the clinical context, and the guidelines for therapeutic hypothermia should be used as an ancillary rather than a restraining dynamic to treat.

The current guidelines for therapeutic hypothermia have been a limiting step in offering treatment for a group of vulnerable newborns who suffer perinatal hypoxia–ischaemia and could benefit from this intervention. We propose that when there is a perinatal history of potential ischaemia, for example, cord prolapse, placental abruption or as simple as abnormalities in the fetal heart monitoring, we should use our clinical judgement with frequent, careful and thorough neurological examination for the presence of encephalopathy as the cardinal criterion for intervention. The potential benefits in these situations outweigh any potential harms. Simply, we have nothing else to offer.

Quality improvement projects with standardised screening tools including formal clinical and EEG assessment to monitor the outcomes of borderline cases who are considered but not offered therapeutic hypothermia are essential to establish reliable criteria for treatment. Such projects could incorporate early risk factors such as perinatal events in addition to the current criteria for cooling and evaluate the evolution of the newborns’ condition using frequent clinical assessment.

The objective of this letter is not to propose new indications for therapeutic hypothermia. Rather we suggest that it is reasonable to look for ways to optimise existing criteria for treatment. Although only a small number of newborns fall in the margins for consideration, the individual benefits can be significant. It is important to acknowledge that there are potential harms associated with treatment, for example, admission to neonatal intensive care, central lines, use of sedative drugs and parenteral nutrition, separation from mother and risk of subcutaneous fat necrosis. It will be necessary to discuss with the parents the trade-off between potential neuroprotective benefits of therapeutic hypothermia versus potential harms in cases that would not qualify according to the current guidelines. The irreversible nature of potential brain damage and the reversible and treatable nature of other side effects should be emphasised in those discussion with proper documentations and open explanations.

In conclusion, pending further research into better ways of assessing newborns for treatment, especially in borderline cases, we suggest that in doubt, it is appropriate to initiate therapeutic hypothermia. As part of this, we suggest that any newborn who has been exposed to perinatal hypoxic events and required active resuscitation should be observed for at least 6 hours and examined routinely against the criteria for therapeutic hypothermia particularly clinical or electroencephalographic signs of encephalopathy. Ultimately, our collective goal should be to refine the criteria for benefit with therapeutic hypothermia.