• Nine centres in five European countries
• March 1996 to January 2001
• Stopped before full recruitment due to lower than expected enrollment rate and funding shortage

• Inclusion: 18 – 75 year olds; witnessed cardiac arrest; pulseless VT or VF as initial rhythm.
• Exclusion: longer than 15 minutes before advanced life support; longer than 60 minutes before ROSC; responsive after ROSC; prolonged hypotension after ROSC (30 minutes at MAP < 60 mmHg); hypoxaemia for more than 15 minutes after ROSC; cardiac arrest after medical support already present
• 3551 assessed of which 275 randomised

• 137 patients
• Induced hypothermia targeting 32–34°C using surface cooling (cold air blanket or ice packs) for 24 hours, followed by passive re-warming
• Sedation and shiver-management was protocolised: midazolam, fentanyl and pancuronium
  • Median time from ROSC to initiation of cooling was 105 minutes [IQR 61–192]
  • Median time from ROSC to target temperature was 8 hours [IQR 4–16]
  • 12% failed to reach target temperature

• 128 patients
• Standard hospital bed with no temperature management
• Sedation protocol as per intervention arm
  • Diagrammatic data suggests median temperature was 37.5°C or higher from ~12 hours onward

• Primary outcome: more patients in the hypothermia group were left with good neurological outcomes at six months
  • Good neurological outcome assessed as Pittsburgh Cerebral Performance Categories 1 [no disability] or 2 [mild disability]
  • Bad neurological outcome assessed as P-CPC 3 [severe disability], 4 [persistent vegetative state] or 5 [death]
  • 75 (55%) versus 54 (39%) had favourable neurological outcomes, favouring hypothermia
  • Risk ratio for death in normothermia group was 1.40 (95% CI 1.08–1.81).
  • Number-needed-to-treat was 6 (95% CI 4–25)
  • Secondary outcome: a difference in mortality favoured the hypothermia group

• Among patients in whom spontaneous circulation has been restored after cardiac arrest due to ventricular fibrillation, systemic cooling to a bladder temperature of 32–34°C for 24 hours increases the chance of a favourable neurological outcome, as compared to standard normothermic life support

• Randomised trial with blinded assessment.
• First clinical trial testing theory drawn from cohort and animal studies.

• Narrow inclusion criteria limit generalisability
• Minor baseline differences between groups, but no change in risk ratio after statistical correction
• Attending physician not blinded, which may bias toward a positive outcome
• Blinding of assessing physician not verified, but relatively objective outcome assessment should limit impact of any accidental un-blinding
• Only 8% of screened patients were included; selection bias cannot be excluded
• Control group became hyperthermic, which may be harmful. Difference observed could be due to harm of conventional therapy rather than benefit of hypothermic therapy

The Bottom Line

Links

Full text pdf / abstract / doi: 10.1056/NEJMoa012689

Editorial, Commentaries or Blogs

Metadata

Summary author: @DuncanChambler
Summary date: May 2014
Peer-review editor: @stevemathieu75