Ethics of therapeutic hypothermia

Wilkinson in this issue cautiously addresses the transition from human experiments to ‘standard of care’ for hypothermia therapy in neonates with post-intrapartum asphyxial hypoxic-ischemic encephalopathy (HIE) (1). Under the term ‘standard of care’, he appears to hesitate yet his ‘opinion’ is unequivocal: ‘given the current evidence about hypothermia, … failing to at least discuss with parents the option of cooling (including transfer if necessary) would be unethical’. Other authors, including many co-investigators in the hypothermia trials, too numerous to cite here, are similarly reticent about declaring hypothermia a ‘standard of care’. The meaning of this term in the medical context is unclear (although legally defined, the term is difficult to apply clinically even when medical societies take a stand, for example, electronic fetal heart rate monitoring). Wilkinson's ethical stance on offering hypothermia, written in the plain language cited above, is unambiguous. Bravo! Like Wilkinson, we considered it ‘our duty … to offer this treatment to every eligible patient with moderate or severe neonatal HIE’ (at experienced centres) (2). We avoided mention of ‘standards’. ‘Standard’ or not, we consider that hypothermia is here to stay. Even if another 1500 research subjects were to show no differences in outcome between hypothermia and controls, meta-analysis would still favour therapy (3). In addition, it is likely that the evaluation of the efficacy of hypothermia is conservative for two reasons: firstly, only moderate and severe adverse outcomes were accounted for in the published clinical trials of hypothermia. Recent studies indicate that intrapartum asphyxia causes important learning disabilities and behavioural disorders in the absence of cerebral palsy (4,5). It will take time to determine the effect of hypothermia on these outcomes. Secondly, it is likely that the hypothermia trials included a few subjects with antepartum brain injuries, for which hypothermia therapy would not be efficacious. Another line of evidence that supports the clinical acceptance of an experimental therapy is the premature termination of a clinical trial, as occurred with hypothermia for HIE. Research ethics committees routinely reconsider their original decision as new knowledge is brought to their attention. As a result the committee may mandate a change in the research protocol, varying from an amended consent form (with implications for recruitment of research subjects) to premature termination. The research ethics process has parallels to clinical guideline and standards formulation such as consideration of scientific evidence, and accounting for patient perceptions (represented by lay committee members); moreover, decisions are made by consensus. Thus a well-implemented research ethics process helps define the boundaries between human experiment and clinical practice, and the grey area between. The outcome of the relatively well-designed research ethics process can thus preempt the formulation of clinical standards. A second set of ethical issues raised by Wilkinson are related to patients with most severe HIE and the worst prognosis, in whose best interests withdrawal of life-support may be offered. Hypothermia therapy, especially when started early, may obfuscate prognosis and constrain withdrawal of life-support decisions. Withdrawal of life-support involves information-provision, informed parental (or legal guardian) consent and parental autonomy. Although hypothermia reduces both death and severe long-term morbidity rates, some individuals who may have died if untreated, may survive with severe disability. Timing is of great pragmatic importance for withdrawal of support decisions. As Wilkinson pointed out, withdrawal may not achieve its goal if implementation is delayed. We deal briefly with the practical implications of administering hypothermia below: The initiation of hypothermia is facilitated by dispensing with the need for consent; the earlier onset of treatment is a desirable goal, which may be associated with better outcomes than those achieved in the controlled trials. Hypothermia treatment can be started before the prognostic information is sufficient to decide whether to offer the withdrawal option to parents or not. Severity of illness can be evaluated pre-hypothermia, within minutes of birth by a fairly accurate prediction rule such as ours consisting of chest compressions, age of onset of breathing and base deficit at birth (6). More evidence of prognosis, albeit not absolutely accurate, can be obtained on an ongoing basis from other indicators such as age of onset of seizures, early imaging and electrophysiological findings as data become available. In general, in any clinical context, it may be appropriate to discontinue life-support, provided this is done within customary ethical boundaries, with informed patient/parent consent. Although prognostic evaluation is inevitably confounded by hypothermia therapy, withdrawal decisions are not unconditionally precluded in treated infants. Hypothermia therapy may also be discontinued if there are insufficient grounds for its continuation, for example, when further information reveals a disorder of insufficient severity to warrant continued therapy. Now, we address questions pertinent to the above considerations: Is it obligatory to finish a course that has been started (the 72-h period is in any event somewhat arbitrary) before making major decisions on the provision of care? We see no reason to delay withdrawal of life-support decisions until the ‘full course’ of hypothermia has been administered, if appropriate. What information should be provided to parents and when? It is our opinion that in the interests of parents’ autonomy, they should receive all relevant information (including some idea of the ‘confidence limits’ of that information) at the time of the first informational encounter. This should include the severity of the infant's illness, the prognosis, proposed interventions and the potential risks and benefits including survival with severe disability (vs. death), the concept of the child's best interests and customary institutional practice of withdrawal and its criteria (which may depend on new information as it becomes available from the evolution of the illness and laboratory testing). In the current context in which parental consent is no longer required for hypothermia therapy (as we see it), it is immaterial whether the initial provision of all or any of the above information occurs before or after the therapy has been started. The third domain of issues raised by Wilkinson relates to broadening eligibility criteria for hypothermia, that is, identifying candidates for ‘compassionate’ therapy. Before addressing this question, it should be pointed out that knowledge in these areas could be expanded by secondary analyses of existing data. In our opinion, data mining (requiring collaboration between the primary investigators to conduct individual patient data meta-analyses (6)) would provide insight into the eligibility criteria below, as well as other issues. Infants with borderline prematurity (e.g. 34–36 gestation): infants of 36 and 37 weeks gestational age could be compared with the more mature trial infants. Delayed initiation of hypothermia beyond 6 h: infants who started hypothermia therapy late in the 0–6 h window of time could be compared with those who started earlier (the categorical cut-off age for analysis, as late as possible within this window, to be determined by sample size considerations). Postnatal onset of severe asphyxial insult: little is known about this population. Owing to its rarity a randomized controlled trial is probably not feasible. If no important rationale contraindicates hypothermia therapy for these infants, we see no reason why it cannot be used as a ‘compassionate treatment’. If hypothermia is used, we strongly suggest a systematic approach with prospective data collection. Hypothermia therapy is not available shortly after birth: the effect of ‘passive hypothermia’, serendipitous or improvised, could also be clarified by post-hoc analyses of trial patients. Based on the literature, it can be stated with reasonable confidence that infant and environmental temperatures should be monitored vigilantly, pyrexia should be treated and environmental hyperthermia should be strictly avoided. In conclusion, we agree with Wilkinson that today, failing to ‘at least discuss with parents the option of cooling (including transfer if necessary) would be unethical’. Regarding the impact of hypothermia on withdrawal of life-support decisions and expanding eligibility criteria, we are more cautious. However, we suggest ways in which new evidence can be mustered quickly, to broaden the basis on which clinicians can make decisions. Within a few months of the time of writing, the data of more than 1200 research subjects will be available from trials. Sample size and quality of data permitting, individual patient data meta-analyses (7) could be performed and the results used to make more nuanced decisions for individual patients. Aggregated data could also be used to create contemporary prognostic models for hypothermia-treated and untreated patients, at various postnatal ages. We encourage primary investigators to collaborate in performing secondary analyses. As stated in another domain, ‘it is time to collaborate’ (8), to which we append a question: is it ethical to neglect to do so?

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.