22
The split in this literature is between designs, not between studies. Combined with the observation that a patient on ECMO has a temperature set by the circuit whether anyone chooses or not, it means the observational comparison is frequently between two hypothermic targets rather than between treatment and no treatment.
Normothermia or fever prevention alone
DOI 10.3389/fmed.2026.1882223; PROSPERO CRD42023435353
Therapeutic hypothermia
Only three randomised trials; observational arm crude and at high risk of bias; certainty rated low to very low by the authors.
Pooled randomised evidence shows no benefit: survival RR 1.30 (0.70 to 2.40), favourable neurological outcome RR 1.80 (0.86 to 3.77); no benefit at 1, 3 or 6 months. Pooled observational evidence shows benefit: survival RR 1.41 (1.14 to 1.74), neurological RR 1.61 (1.21 to 2.14), crude and unadjusted. Any association is confined to 32 to 34 degrees; nothing appears at 34 to 36. No increase in major complications.
31 studies, 6,184 adult ECPR patients, of which only 3 were randomised