Futility and neurotrauma: can we make an objective assessment?

In the often fraught setting of acute neurotrauma, clinicians and patients’ surrogate decisionmakers face enormous psychological pressure when deciding whether to proceed with what may be considered life-saving but potentially non-restorative surgical treatment. As this case has illustrated, considerable conflict can also arise between the various stakeholders. On one hand, there is the powerful human proclivity to take all measures to save the life of a single identified individual regardless of cost — this is described as the rule of rescue.6 In this clinical scenario, all medical measures failed to control the ICP, and it is extremely difficult to be seen to “give up” on a young patient when there may be at least a small chance of a reasonable clinical recovery. On the other hand, there is the risk that while surgical intervention may be life saving, it may leave the patient in a neurological condition that he would deem unacceptable. The ethical tension arises from the uncertainty regarding the long-term outcome and this has been the subject of ongoing research in Western Australia.7,8

Perth is unique in its geographic isolation and, because there are only two neurotrauma hospitals served by a single statewide neurosurgical service, it is possible to obtain accurate, long-term, population-based outcome data. By using the recently established CRASH (Corticosteroid Randomisation After Significant Head Injury) Trial collaborators’ outcome prediction model9 to stratify patients with traumatic brain injury according to injury severity, it is possible to compare predicted outcome following decompressive craniectomy with observed long-term outcome7 (Box 1). The web-based outcome prediction model provides a percentage risk of an unfavourable outcome at 6 months (defined by the Glasgow outcome scale [GOS] categories of severely disabled, persistently vegetative and dead), and this is derived from clinical and radiological features on initial presentation.

For the clinical scenario we have presented, the clinical variables required for the model are shown in Box 2. This patient has a predicted risk of an unfavourable outcome at 6 months of 92.5%. Data from WA show that most patients with a similar predicted risk of an unfavourable outcome (or index of injury severity) remain severely disabled or in a vegetative state if they survive (Box 1). So the obvious question is whether the patient (were it possible) or his parents (as surrogate decisionmakers) would have provided consent if they had been provided with these data.

A large study has examined the influence that this type of objective data would have on the opinions of anaesthetists in WA.10 The participants were shown a particular clinical scenario in which there was intractable intracranial hypertension despite maximal medical management and were asked if they would consent to surgical decompression if they themselves were the injured patient. They were then shown the CRASH Trial collaborators’ model and the objective long-term outcome data. The responses to a predicted risk of an unfavourable outcome of 94.4% suggested that anaesthetists were strongly influenced by an objective assessment of the possible outcomes and felt strongly that, for themselves, survival with severe disability would be unacceptable. Whether these views are shared by the general population is yet to be examined, but the anaesthetists’ responses suggest that careful consideration is required when discussing long-term outcomes with surrogate decisionmakers. Before performing what is presumed to be life-saving decompressive surgery, outcome cannot be dichotomised into life or death. When the prediction of an unfavourable outcome is greater than 80%, the most likely outcome if the patient survives is that of severe disability, and if that is unacceptable to that individual then the intervention should be deemed futile. Conversely, substantial benefit describes an outcome that now or in the future the patient would regard as worthwhile, and it may be that over time patients can adapt to a level of disability that they would previously have regarded as intolerable. Relatively little is known about the inner world of patients who survive with severe disability and we must accept that there are significant limitations to making abstract judgements such as “I would rather be dead than alive with severe disability”. Finally, even when there is a very high chance of survival with severe disability, life is sacrosanct in certain cultures and religions and these views should be respected.

In conclusion, we believe that the concepts of substantial benefit and risk of unacceptable badness provide a useful framework for discussing medical futility. The CRASH Trial collaborators’ prediction model provides a useful surrogate index of injury severity. We acknowledge the limitations of applying a mathematical model (based on what must be, by definition, historical data) to individual patients and we agree with the CRASH Trial collaborators that the model should only be used to support, and not replace, clinical judgement. However, we feel that comparing predicted outcomes with observed outcomes may provide useful information when discussing realistic outcome expectations with surrogate decisionmakers.