Twelve RCTs were included (n=1,327, range 22 to 396). Four trials had adequate allocation concealment. Seven trials had blinded outcome assessment. Five studies were low quality. No studies blinded patients and caregivers.
Hypothermia treatment was associated with significantly lower mortality (RR 0.73, 95% CI 0.62 to 0.85, NNT=10, 95% CI 6 to 20; 12 RCTs) and greater likelihood of a good neurological outcome (RR 1.52, 95% CI 1.28 to 1.80, NNT=5, 95% CI 3 to 8; 10 RCTs) without exceeding cut-offs for statistical heterogeneity (I2=0% and I2=34%). Subgroup analysis by length of treatment, long-term or goal-directed treatment (eight RCTs) significantly improved both mortality and neurological outcomes; short-term treatment (four RCTs) did not significantly affect either outcome.
Various adverse events were reported in the intervention groups, including bradycardia (seven RCTs), hypokalaemia (six RCTs), thrombocytopaenia (four RCTs) and hypotension (three RCTs) associated with hypothermia, and rebound in intracranial pressure associated with rewarming (three RCTs of short-term cooling).
The funnel plot was suggestive of publication bias; small negative studies were less likely to be published. Sensitivity analysis by quality did not change the overall findings.