For the millions of cardiac arrest survivors who reach a hospital unconscious, the question of how long to keep the brain cooled has never had a definitive answer — until now. The ICECAP trial is the largest and most methodologically sophisticated attempt to resolve a debate that has quietly shaped intensive care practice for two decades, and its findings carry direct implications for how neurological rescue is managed in the critical hours after the heart restarts.
The trial enrolled adults who suffered out-of-hospital cardiac arrest, remained comatose afterward, and achieved target cooling below 34 °C within four hours of arrest. Conducted across 71 U.S. hospitals between June 2020 and June 2025, ICECAP used an innovative response-adaptive randomization design that tested ten distinct cooling durations — ranging from 6 to 72 hours at 33 °C — rather than the conventional two-arm comparison. The adaptive algorithm continuously reallocated enrollment toward durations showing the greatest neurological benefit, assessed via a weighted modified Rankin Scale at 90 days analyzed with a Bayesian duration-response model. Critically, the algorithm operated separately within each presenting cardiac rhythm, acknowledging that shockable and non-shockable arrest may have fundamentally different cooling requirements.
This design is itself a landmark contribution to critical care methodology. Previous trials — TTM, TTM2, and HYPERION among them — either compared hypothermia to normothermia or fixed cooling at 24 hours, leaving the dose-response relationship largely unmapped. ICECAP's adaptive architecture is better suited to identify a true therapeutic optimum rather than simply confirming or refuting a single pre-specified duration. The Bayesian framework also allows probabilistic conclusions about where benefit peaks, information that binary hypothesis testing cannot provide. Limitations include the inherent complexity of adaptive trial interpretation and the requirement for rapid cooling within four hours, which may not reflect real-world prehospital timelines. Whether the optimal duration differs meaningfully by rhythm type, arrest-to-cooling interval, or initial neurological status remains a critical open question that secondary analyses will need to address.