The Hypothermia After Cardiac Arrest Study Group
Similar to the paper by Bernard and colleagues (they were published in the same 2002 NEJM edition), this paper tried to tackle the question of the benefit of therapeutic hypothermia following out of hospital cardiac arrest (OOHCA). Prior to this there had been animal study evidence and poor quality trials suggesting an improved neurological outcome if patients were cooled after cardiac arrest. They postulated that the cooling of the brain mitigated the harmful effects of persisting hypoxia and the reactive processes that resulted from this e.g. free radical generation.
Mild therapeutic hypothermia to improve the neurological outcome after cardiac arrest
Mild therapeutic hypothermia to improve the neurological outcome after cardiac arrest
What did they do?
Design: Multicentre randomised control trial with blinded outcome assessment. Computerised block randomisation
Control: Protocolised care and ‘normothermia’. Temperature measured by bladder temperature probe.
Intervention: Cooled to between 32 and 34 degrees Celsius using ‘TheraKool’ cold air device +/- ice packs. Target was to reach temperature within 4 hours of ROSC, maintain for 24 hours then allow passive rewarming over approximately 8 hours. Otherwise protocolised care as per control group.
Inclusion criteria: Witnessed cardiac arrest, ventricular fibrillation or pulseless ventricular tachycardia, presumed cardiac origin, age 18 to 75, interval of 5 to 15 mins until start of resuscitation by medical personnel.
Exclusion criteria: Temperature below 30 degrees Celsius, comatose before cardiac arrest, pregnant, response to verbal commands after ROSC, persisting hypotension after ROSC, persisting hypoxia after ROSC, known terminal illness.
Population: Patients from 9 centres in 5 different European countries.
Primary outcome: Favourable neurological outcome within 6 months – defined as Pittsburgh cerebral performance category of 1 (good recovery) or 2 (moderate disability).
Secondary outcomes: 6 month mortality, complications during first 7 days.
Control: Protocolised care and ‘normothermia’. Temperature measured by bladder temperature probe.
Intervention: Cooled to between 32 and 34 degrees Celsius using ‘TheraKool’ cold air device +/- ice packs. Target was to reach temperature within 4 hours of ROSC, maintain for 24 hours then allow passive rewarming over approximately 8 hours. Otherwise protocolised care as per control group.
Inclusion criteria: Witnessed cardiac arrest, ventricular fibrillation or pulseless ventricular tachycardia, presumed cardiac origin, age 18 to 75, interval of 5 to 15 mins until start of resuscitation by medical personnel.
Exclusion criteria: Temperature below 30 degrees Celsius, comatose before cardiac arrest, pregnant, response to verbal commands after ROSC, persisting hypotension after ROSC, persisting hypoxia after ROSC, known terminal illness.
Population: Patients from 9 centres in 5 different European countries.
Primary outcome: Favourable neurological outcome within 6 months – defined as Pittsburgh cerebral performance category of 1 (good recovery) or 2 (moderate disability).
Secondary outcomes: 6 month mortality, complications during first 7 days.
What did they find?
Numbers: 3551 patients assessed for eligibility with 275 enrolled. 137 randomised to hypothermia group, 138 to normothermia group.
Results: Similar baseline characteristics, though normothermia group more likely to receive bystander CPR and more likely to have diabetes or known coronary heart disease.
More patients in the hypothermia group had a favourable neurological outcome (75/136 - 55%) compared to the non-cooled group (54/137 – 39%).
This provided a risk ratio of 1.45 (95% confidence interval 1.08 – 1.81, p value = 0.009) with a number needed to treat of 6 to prevent 1 unfavourable neurological outcome.
The death rate at 6 months was lower in the hypothermia group (56/137 (41%)) compared to the normothermia group (76/138 (55%)).
This provided a risk ratio of 0.74 (95% confidence interval 0.58 – 0.95, p value = 0.02) providing a number needed to treat of 7 to prevent 1 death.
The number of patients who developed complication was similar in the hypothermia (98/135 (73%) compared to the normothermia group (93/132 (70%).
Similarly, the total number of complications in each group was not significantly different.
Results: Similar baseline characteristics, though normothermia group more likely to receive bystander CPR and more likely to have diabetes or known coronary heart disease.
More patients in the hypothermia group had a favourable neurological outcome (75/136 - 55%) compared to the non-cooled group (54/137 – 39%).
This provided a risk ratio of 1.45 (95% confidence interval 1.08 – 1.81, p value = 0.009) with a number needed to treat of 6 to prevent 1 unfavourable neurological outcome.
The death rate at 6 months was lower in the hypothermia group (56/137 (41%)) compared to the normothermia group (76/138 (55%)).
This provided a risk ratio of 0.74 (95% confidence interval 0.58 – 0.95, p value = 0.02) providing a number needed to treat of 7 to prevent 1 death.
The number of patients who developed complication was similar in the hypothermia (98/135 (73%) compared to the normothermia group (93/132 (70%).
Similarly, the total number of complications in each group was not significantly different.
Is it any good?
Overall: Fairly strong.
Strengths: Randomised with blinded outcome assessment.
Multicentre study across a broad population.
Weaknesses: Specific subset of cardiac arrest patients (those with VF, specific interval before resuscitation of 5 to 15 minutes) thus uncertain generalisability. The high degree of exclusion also suggests a potential for selection bias.
‘Normothermia’ group actually demonstrated a degree of pyrexia.
Treating clinician unblinded, providing another potential source of bias.
Strengths: Randomised with blinded outcome assessment.
Multicentre study across a broad population.
Weaknesses: Specific subset of cardiac arrest patients (those with VF, specific interval before resuscitation of 5 to 15 minutes) thus uncertain generalisability. The high degree of exclusion also suggests a potential for selection bias.
‘Normothermia’ group actually demonstrated a degree of pyrexia.
Treating clinician unblinded, providing another potential source of bias.
Final Thoughts
Another landmark paper that was a real game changer in the history of therapeutic hypothermia. Overall a pretty convincing study, though certain aspects need carful interpretation; most notably the presence of pyrexia in the control group.
Written: Tom Heaton
Reviewed: Not done
20th April 2016
Written: Tom Heaton
Reviewed: Not done
20th April 2016
References
- The Hypothermia After Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurological outcome after cardiac arrest. NEJM. 2002. 346: 549-556
