EUROTHERM Header Image

Hypothermia for Intracranial Hypertension after Traumatic Brain Injury

Andrews. NEJM Published first on line Oct 2015. DOI: 10.1056/NEJMoa1507581

Clinical Question

  • In patients with traumatic brain injury (TBI), does hypothermia (32-35°C) and standard care compared to standard care alone reduce death and major disability at 6 months after injury?


  • Multi-centre, multi-country RCT with a weighted recruitment to increase UK population
  • Centralised randomisation with minimisation procedure to balance centre, age, GCS motor score, time from injury and pupillary response
  • Appropriate concealment of allocation until inclusion
  • Open-label – patients, families and treating clinicians aware
  • Outcome assessor blinded
  • Pilot trial period with subsequent adjustment of inclusion criteria and target sample size
  • Powered at 80% to detect a 9% absolute reduction in the rate of unfavourable outcomes from an expected incidence of 60% in the standard care group with 0.05 two-sided alpha significance level, if 600 patients were included
    • This was initially 1800, but this was adjusted after the pilot trial period identified a higher incidence of pathology with raised intracranial pressure in the study population and a change in the statistical method (ordinal analysis of outcome measure rather than binomial)


  • 55 centres (although only 47 actually recruited) in 18 countries
    • 53% patients recruited in UK
  • Nov 2009 – Oct 2014


  • Inclusion: Adults with a primary closed TBI and intracranial pressure of more than 20mmHg for at least 5 minutes after stage 1 treatments with no obvious reversible cause
      • stage 1 treatment
        • mandatory: mechanical ventilation; sedation; analgesia; head of bed elevated to 30 degrees; intravenous fluids with or without inotropes to maintain mean arterial pressure > 80mmHg
        • optional: ventriculostomy with or without CSF drainage; surgical removal of space occupying lesion
    • an initial head injury that occurred no more than 10 days earlier
    • availability of a cooling device or technique for > 48 hours
    • core temperature of at least 36°C at time of randomisation
    • abnormal CT brain
  • Exclusion: patients already receiving therapeutic hypothermia or were unlikely to survive for the next 24 hours; barbiturates infused prior to randomisation; temperature <34°C at hospital admission, and pregnancy
  • 2498 patients were screened, 387 were randomised, 386 received intended treatment, 376 were evaluated in an intention to treat analysis


  • Therapeutic hypothermia (32-35°C) for at least 48 hours and continued until ICP controlled + standard care
    • Cooling achieved by bolus of 20-30mls/kg of refrigerated, intravenous 0.9% sodium chloride and then maintained using individuals departments usual practice
    • Core temperature was reduced by the minimum required to maintain ICP of 20mmHg or less
    • Stage 2 treatments were added if hypothermia failed to control ICP
      • Included: mannitol; hypertonic saline; inotropes to maintain cerebral perfusion pressure of >60mmHg
    • Stage 3 treatments were added if ICP was still inadequately controlled
      • Included: barbiturate therapy with processed EEG monitoring; decompressive craniectomy further surgical intervention if required
    • Rewarming strategy was controlled at 0.25°C per hour once ICP was controlled


  • Standard care
    • Stage 1 treatment and escalation to stage 2 and then 3 if ICP was not controlled


  • Primary outcome:
    • Ordinal analysis of the Extended Glasgow Outcome Score (GOS-E) at 6 months after injury, with adjustment for covariates, demonstrate a shift to poorer outcomes in the hypothermia group
    • Binomial analysis of GOS-E (good outcome vs bad outcome) also demonstrated a statistically significant difference favouring standard care
      • Good outcome in hypothermia group 25.7% vs standard care group 36.5%
      • Absolute risk increase 10.85% (95% CI 1.61% to 20.10%); NNH = 10; p=0.03
      • Adjusted common odds ratio 1.53 (95% CI 1.02 to 2.30; p-value 0.04)
      • Fragility Index is 3
        • If 3 patients in the experimental group (hypothermia) had a good outcome instead of a bad outcome, then it would no longer be considered statistically significant using Fishers Exact test – however it is worth noting that this binomial analysis was not the main statistical method the authors used; the ordinal analysis above provides greater statistical strength
  • Secondary outcome:
    • 6 month mortality was statistically significantly increased in the hypothermia group
    • Failure of all stage 2 therapies to control ICP occurred less often in the hypothermia group
      • More barbiturates were used in the standard care group, but there was no difference in the incidence of decompressive craniectomy
    • Incidence of pneumonia at days 1-7 was no different between the groups
    • Length of ICU stay was not different between the groups
    • Modified Oxford Handicap Scale (MOHS) grade at day 28 or discharge from an acute-care hospital was not statistically significantly different between the groups
    • Serious adverse events were more common in the hypothermia group
      • 33 events vs. 10 events

Authors’ Conclusions

  • In patients with intracranial pressure of more than 20mmHg after TBI, therapeutic hypothermia plus standard care to reduce intracranial pressure did not result in better outcomes compared with standard care alone


  • Trial protocol was published a priori
  • Participating centres all had evidence of expertise with intracranial-pressure monitoring and therapeutic cooling
  • Rewarming strategy was controlled at 0.25°C per hour
  • Use of higher strength statistical methods based on ordinal outcome scale
  • Patient centred outcomes rather than physiological measure (e.g. ICP)
  • Combined death, vegetative state and severe disability into one GOS-E category so that conclusion would not favour severely disabled survivors (however, this modification of the previously validated scale could be considered a weakness)
  • Pragmatic and based on existing guidelines (Stage 1, 2 and 3 therapy for raised ICP)


  • The inclusion criteria was changed during the study to remove an upper age limit (previously 65 years) and to increase the time from injury from 72 hours to 10 days
  • Early termination risks bias (tends to over estimate effect magnitude) however remaining results collected after termination did not show regression toward the mean suggesting this is not the case in this trial
  • Loss of outcome data for 10 cases (given fragility index was 3, these 10 could have changed the result significantly)
  • Lack of data on additional Stage 2 therapies limits opportunity to draw new hypotheses – it could be that some other Stage 2 therapy caused the harm / benefit effect seen in this trial and that hypothermia was not the causal therapy

The Bottom Line

  • In patients with elevated intracranial pressure after TBI, therapeutic hypothermia, in addition to standard treatment, results in a greater risk of death and worse neurological outcomes in survivors compared to standard measures

External Links


Summary author: @stevemathieu75
Summary date: 11th December 2015
Peer-review editor: @DuncanChambler


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.