Paradoxically hypothermia has been a ‘hot’ topic in emergency medicine and critical care for many years. There is good laboratory, animal and pathophysiological data to suggest that it should be neuro-protective in a broad range of conditions including cardiac arrest and brain injury. However, when this issue has been examined in pragmatic clinical trials, definitive evidence of clinical benefit for hypothermia (when compared to normothermia) in ROSC patients, status epilepticus and brain injury has not been forthcoming 1,2. Perhaps this is another bench solution that does not translate to the bedside.
We recently reported on the EuroTHERM trial3,4 that looked at rescue hypothermia in patients with severe traumatic brain injury refractory to standard tier 1 care. We felt the trial showed harm to patients in the hypothermia group, but did not answer the question as to whether hypothermia may be used as a prophylactic therapy in patients who are at the lower end of the severity spectrum, but remain at risk of secondary brain injury.
This month an international group reported the result of a randomised controlled trial in just such a group. The paper is published in JAMA5 and is (I think) currently open access. The abstract is below, but as we always say it’s vital that you read the full paper.
What kind of paper is this?
This is a superiority randomised controlled trial which is the appropriate and (usually) best way to assess the effectiveness of a therapy in a group of patients, which you believe may have advantages over standard care.
Who was studied?
The study enrolled patients who appeared to have significant brain injury at a very early stage in their illness. Patients in the ED or prehospital setting with evidence of head trauma and a GCS of less than 9 were eligible. These patients were typically intubated, or about to be intubated when randomised. This is a pragmatic group of patients that all emergency physicians will recognise. There are potentially mimics in this group (e.g. drug intoxication), but that’s OK as at the point when cooling would take place the clinician can only use whatever data is available. In general the St Emlyn’s team supports the use of pragmatic randomisation like this in clinical trials. Other trials (such as Crash 3 for example6) are increasingly taking this approach, in order to facilitate randomisation and delivery of the intervention at an early stage of the disease process.
Patients who were known or felt to be at risk of severe bleeding were not included owing to the known risks of hypothermia induced coagulopathy. The patients were mostly male and young, which reflects the epidemiology of traumatic brain injury. A significant proportion of the primary injuries were lone cerebral contusions, which again reflects what we often see in Virchester. Lastly there were a lot of exclusions, reported in the supplementary tables. Several of these were quite subjective. This is important to look at and remember, regarding generalisability.
What about the treatment?
In the intervention group hypothermia was initially induced with 2000 mL intravenous ice-cold (4°C) 0.9% saline, and surface-cooling wraps to 35C. Temperature was then taken down to 33C once severe bleeding had been excluded using the same equipment. This is important, as previous trials and many centres still use intravascular cooling devices which introduce further risks to the intervention group. Those in the control arm were kept normothermic at 37 +/- 0.5C. The treatment phase lasted for a minimum of 72 hours, and up to a week if the clinicians thought it needed. This was an interesting strategy, which implies that the trial team believed the patients were still getting some form of benefit from cooling in cases where ICP was challenging to manage. We are not sure the results of EuroTHERM support this idea.
The rest of clinical care was at the discretion of the clinicians guided by international consensus guidelines for the management of brain injury.
What about the results?
The trial enrolled over 511 patients, but after withdrawals and exclusions 500 were analysed. The main outcome (decided a-priori) was a dichotomised score of 5 or above on the GOS-E7, which the authors believed to represent a ‘favourable’ outcome. This is a well known and internationally recognised ordinal score for outcome in head injury (albeit a fairly blunt one), but the dichotomisation was the authors decision . A score of 5 equates to moderate disability. The headline figure is that there was no difference in outcome between the two arms of the trial.
Six months after injury, favourable outcomes occurred for 117 patients (48.8%) in the hypothermia group and 111 (49.1%) in the normothermia group (absolute risk difference, –0.4 per- centage points [95% CI, –9.4 to 8.7]; unadjusted relative risk with hypothermia, 0.99 [95% CI, 0.82-1.19]; P = .94)
There were no significant differences in secondary outcomes such as LOS, infections, mortality etc. These differences remained insignificant in a number of sensitivity analyses. The authors also presented some of their primary findings as distribution charts, as per the image below. These charts are ace – they make it very clear what we get for our sometimes herculean efforts. In this trial, the charts are staggeringly similar, implying we get very little. In other trials, these charts have highlighted the issues we sometimes see in neuroICU patients where delivery of an intervention perhaps saves lives, but at the cost of producing far more survivors with severe disability, and very little additional survival without. Rescue ICP was a good example of this.
In severe traumatic brain injury, early prophylactic hypothermia did not improve neurologic outcomes: The POLAR Randomized Clinical Trial | Traumatic Brain Injury | JAMA | JAMA Network https://t.co/RqwhN1u1Dy— Canberra ICU (@CanberraICU) October 25, 2018
Is hypothermia of no benefit then?
In this group of patients where it is to be used in a very early and prophylactic approach then there does not seem to be a benefit.
There are a few questions that remain though. Some neuro intensivists have argued quite rightly that this is a very specific population and that we should be mindful not to extrapolate to other groups, such as refractory raised ICP. In particular, a lot of clinicians will still reach for therapies like this when they have reached tier 3 care and there are no surgical targets or options. We are not sure this paper has provided any additional information here to help guide decision making. It’s worth following the thread below to understand this and also to once again debate the correct outcome in these trials.
Needs nuanced reply-difficult on Twitter. Briefly, POLAR is an exceptional RCT (kudos to Jamie et al), but RCTs of of hypothermia as prophylaxis (POLAR) or Tier 1 therapy (EUROTHERM) do not inform its use in refractory high ICP. Different population =different risk/benefit ratio. https://t.co/wogNHX1oRx— David Menon (@Menon_Cambridge) October 24, 2018
The endpoint of GOS-E of 5 or more is also reasonable to many, but others may argue that lower cut off points are also valid. A life with worse than moderate disability at 6 months may very well be meaningful and important to certain patients. ICU follow up clinics can really open your eyes to this. However, in this trial the primary outcome would not be affected if a cut off of 3 or 4 had been chosen. Recent neurosciences trials have been excellent at reporting GOSE in a transparent manner between groups, so that readers can understand the direct impact on outcomes and the trade off between improvements in mortality and survival with severe disability.
There is also a question about which patients may have benefited from the intervention. In our journal club it was argued that the patients most likely to benefit from hypothermia would be those with a more severe injury. In this study they used the Marshall classification8 to grade injury severity. In the most severely injured group there was roughly an 8% difference in outcomes, but normothermia was favoured in this group (Ed – and the numbers were small so did not reach statistical significance either). Either way, there is no soft signal for a benefit in more severely injured patients.
However, on the evidence available in this well conducted trial the results suggest that we should not be using early hypothermia in this group of patients for the prevention of secondary brain injury. As is so often the case, Aoife Abbey has summarised the trial beautifully following presentation of the results in Paris at the European Society of Intensive Care Medicine #LIVES2018 conference.
Although if you prefer the official JAMA version, here it is.
Congratulations to my @AlfredHealth @Monash_SPHPM & @AmbulanceVic colleagues on the excellent POLAR trial. No benefit to early prophylactic hypothermia in TBI. And, can I just say, loving the @JAMA_current infographic. https://t.co/aA0VYMmyA2 #LIVES2018 pic.twitter.com/IWVBAM8uAN— David Anderson (@expensivecare) October 25, 2018
Dan Horner @RCEMProf
Simon Carley @EMManchester
Please also read the following blogs on POLAR