Journal Club Follow-up

Dr Steven Bishop

I recently presented the DECRA study on decompressive craniectomy at Journal Club and this blog post provides a summary of some of our discussions.

The DECRA study ambitiously attempted to address whether decompressive craniectomy (DC) was beneficial and defined a small, precise patient population to enrol into the trial (unlike a number of mainly observational studies and a small RCT in children before it). Progress during the trial was slow due to the tight inclusion criteria. To address this the authors adapted the protocol, relaxed the exclusion criteria and modified the primary outcome. Other changes included the method of follow-up and a change in ICP management to recognise the use of therapeutic hypothermia. See Simon Finfer’s lecture in the background reading for more detail.

Study “protocol creep” is not uncommon in large trials and I am always unclear to how the sum of these small changes affect the outcome, if indeed they do at all. Patients recruited early in the trial are not necessarily comparable to those recruited later, although I would expect the changes to be well-balanced between the two arms. A number of DECRA patients in the standard treatment arm also underwent DC, either early as a protocol violation or late as a rescue technique. This cross-over effect was not analysed in the study, but probably had only a small impact on outcome.

Issues with the trial have been widely discussed. Instead, should we not commend the authors for answering a specific question on a relatively well-defined, homogenous patient group (admittedly a small subset of all TBI patients)? Many trials in critical care initially appear to be well-defined, e.g. all patients with septic shock, yet on closer inspection represent a heterogeneous group of patients with different underlying pathology. It is not always entirely obvious how to adopt the results of these studies into everyday clinical practice. I propose that the DECRA investigators have demonstrated something quite specific and worthwhile: that early DC in the subset of non-mass-lesion TBI patients at least does not improve neurological outcomes (if you correct for unreactive dilated pupils) and probably does result in more unfavourable outcomes. Early decompression does not have an effect on mortality – it simply shifts the balance between favourable and unfavourable outcomes, resulting in more severely disabled patients and more patients in a persistent vegetative state.

The DECRA study used a standardised surgical technique, choosing a modification of the Polin method. In Cambridge we usually perform a bifrontal decompression that extends to the coronal suture with opening of the dura with division of the falx cerebri and superior saggital sinus. I’m not entirely clear whether any of these techniques are superior. It would seem logical to adopt a technique that maximises the space available for swollen brain whilst minimising the increased morbidity of more extensive and disruptive procedures.

One of my personal bug-bears with statistics is the multiple comparison problem. I am sometimes frustrated by studies which report large numbers of secondary outcomes and post-hoc analyses. Admittedly, they do offer insight into the data and help shape future research direction, but in the wrong hands they may also mislead and result in harm. The DECRA study is commended for resisting this temptation and minimising the number of secondary analyses, although I note that the published secondary outcomes were different to those defined ‘a priori’.

We finished by asking, “What will you do the next time you have a twenty-year-old patient on ICU with diffuse axonal injury and a sizeable contusion, whom several days down the line has an uncontrolled ICP despite maximal medical therapy. Would you decompress or reach for the barbiturates and sit tight?”. I don’t personally believe that the answer to this question can be extrapolated from the DECRA trial alone as it answered a subtly different and quite specific question. In this setting perhaps decompression is a risk worth taking? Hopefully the results of RESCUEicp which is due to report next year will go someway to providing an answer.