Neurosurgical trainee, Wessex
Traumatic brain injury (TBI) is among the most common conditions leading to admission to a neurosciences ICU, where the cornerstone of management is intracranial pressure monitoring as a guide to interventions that might prevent secondary neurological injury.
At the Wessex Neuro-ICU, as in most centres around the country, we follow a step-wise treatment protocol for the management of intracranial hypertension. The first step here is simply to adjust basic physiological parameters (ABCs) to augment cerebral blood flow and oxygenation. Where this becomes insufficient we then escalate our management using ever more effective but also risky interventions as becomes necessary.
In 1901 Kocher noted that it may be helpful to “decompress the brain by widely opening the skull to decrease the pressure” (Holder 1901; 262-6). Over a century later, the role of decompressive craniectomy remains one of the most hotly-debated topics in the management of severe TBI. Observation and experience alone tell us that opening up a rigid compartment dramatically improves compliance and reduces the requirement for potentially dangerous interventions such as prolonged therapeutic hypothermia and tightly regulated invasive ventilation. This immediately begs the question why we do not decompress all patients with intracranial hypertension from TBI. Indeed in some situations decompression is the only conceivable option, such as after evacuation of a large extra-axial haematoma where the underlying brain immediately herniates out through the defect. Where equipoise remains, however, the question is whether decompressive craniectomy is a safe and effective option to treat raised ICP due to severe generalised cerebral swelling.
The main reason for our trepidation lies in the significant morbidity associated with decompressive craniectomy itself both in the short and longer terms. Quite aside from the cosmetic defect resulting from a sunken flap, a large craniectomy is thought to hinder rehabilitation and in most cases eventually necessitates re-admission for cranioplasty. The frequency and severity of complications from craniectomies performed indiscrimately would unsurprisingly negate any potential survival benefit from performing the procedure in the first place.
The key then must be patient selection and timing of surgery. If we can reserve decompressive craniectomy for a select group of patients whose ICP management has escaped control from basic interventions and rigorous medical management, we could potentially prevent death without causing unnecessary complications to patients who may have otherwise had a favourable outcome without an operation. Enter the vociferous debate over the two major randomised controlled trials examining the effect of decompression in diffuse TBI.
The first to conclude and be published is DECRA (NEJM 2011; 364:1493-1502), led by the Australian and New Zealand Intensive Care Society Clinical Trials Group. Adults with severe diffuse TBI with raised ICP refractory to ‘first-tier’ therapies (n = 155) were randomised to receive either early bifrontotemporoparietal decompressive craniectomy or standard care. Their definition of ‘medically refractory’ was ICP greater than 20mmHg sustained for more than 15 minutes. Defying general expectations, although surgery resulted in lower ICPs and shorter duration of ICU care it also resulted in a greater risk of unfavourable outcome at 6 months on the Extended Glasgow Outcome Scale.
This called into question whether it was even ethical to continue recruiting patients to the second major trial. RESCUEicp is an ongoing RCT led by NIHR and the University of Cambridge looking to recruit 400 adults with severe TBI. Here the threshold for randomisation was set at ICP greater than 25mmHg sustained for more than 1-12 hours at any time after injury. It therefore follows that decompression is likely to be reserved for more severe injuries with more sustained intracranial hypertension than those randomised in DECRA. Most neurosurgeons and intensivists will probably not decompress patients on the basis of an ICP in the low 20s for under an hour, choosing the less aggressive option of persisting with medical management in the first instance and reserving craniectomy as a last resort for cases where all else fails.
There are a number of other important points to consider when comparing these two trials:
- DECRA screened 3478 patients and recruited only 155, suggesting that the results are only applicable to the small proportion of patients who meet their inclusion criteria;
- DECRA excluded all patients who had received previous neurosurgical intervention for evacuation of mass lesions, in whom contemporaneous craniectomy is a common practice (Neurosurgery 2005; 57:1183-92);
- RESCUEicp gives clinicians the choice of an FTP decompression for unilateral swelling or a bifrontal decompression for bilateral diffuse hemisphere swelling, whereas DECRA specified the bifrontal decompression for patients randomised to the surgical arm. This may be associated with higher morbidity as it is often a longer procedure and may carry a higher infection risk by breaching the frontal air sinus.
Where we stand
If we were to rely on current evidence alone then we should probably avoid decompressive craniectomy in TBI at all costs. However, as Smith et al. (BMJ 2003; 327:1459) eloquently express in their systematic review on parachute use to prevent death from gravitational challenge, the value of clinicians’ judgement and consensus opinion should not be underestimated.
As a unit we generally feel that decompressive craniectomy should be used selectively and reserved only for patients in whom all measures short of barbiturate coma have failed to control ICP. In recent years our preferred surgical intervention in diffuse TBI has been to insert EVDs earlier on, including in patients whose ventricles would not be considered abnormally enlarged under normal circumstances. EVDs allow removal of small volumes of CSF that - as a result of where patients lie on their compliance curves - probably causes a disproportionate ICP-lowering effect. We try to do this early in order to prevent the additional risk of insertion in hypothermic and coagulopathic patients whose parenchymal swelling may be so severe as to virtually obliterate their ventricles. Our impression is that proactive early EVD insertion is a relatively safe intervention that may reduce the number of subsequent interventions required and the overall duration of care.
While awaiting the results of RESCUEicp we will continue to perform decompressive craniectomy in patients with life-threatening intracranial hypertension that fails to respond to medical management. Further, if RESCUEicp were to show equivocal outcomes we feel that a health economics analysis must be considered to analyse the socio-economic impact of judicious decompressive craniectomy.