Publication

• Title: Duration of Therapeutic Hypothermia After Out-of-Hospital Cardiac Arrest: The ICECAP Randomized Clinical Trial
• Acronym: ICECAP — Influence of Cooling Duration on Efficacy in Cardiac Arrest Patients
• Year: 2026
• Journal published in: JAMA
• Citation: Meurer WJ, Yeatts SD, Geocadin RG, et al; SIREN Investigators. Duration of therapeutic hypothermia after out-of-hospital cardiac arrest: the ICECAP randomized clinical trial. JAMA. Published online August 5, 2026.

Context & Rationale

• Background

  • Hypoxic–ischaemic brain injury is a major determinant of death and disability after successful resuscitation from cardiac arrest, but no pharmacological neuroprotective treatment has consistently improved patient-centred outcomes.
  • Therapeutic hypothermia is highly neuroprotective in experimental models, yet translation into adult cardiac-arrest care has produced conflicting results.
  • The two influential 2002 trials reported improved neurological outcomes after cooling predominantly shockable-rhythm out-of-hospital cardiac-arrest patients to approximately 33 °C for 12–24 hours, although both trials were small and their control groups received limited temperature control.¹²
  • Subsequent evidence weakened certainty regarding both the depth and duration of cooling: TTM found no advantage of 33 °C over 36 °C; TTM48 did not demonstrate superiority of 48 over 24 hours at 33 °C; HYPERION reported a modest neurological benefit from 33 °C in patients with nonshockable rhythms; and TTM2 found no benefit of 33 °C over controlled normothermia with early fever treatment.³⁴⁵⁶
  • Most clinical protocols nevertheless continued 33 °C cooling for 12–24 hours, while preclinical work suggested that longer exposure might be required for maximal neuroprotection. The clinical duration–response relationship was therefore unresolved.
• Research Question/Hypothesis

  • Among comatose survivors of out-of-hospital cardiac arrest who had already been rapidly cooled to 33 °C, what was the shortest duration of cooling associated with the maximal 90-day neurological recovery?
  • The investigators hypothesised that neurological recovery might improve as cooling duration increased and that an upward duration–response curve would also support the biological efficacy of cooling itself.
  • Shockable- and nonshockable-rhythm cohorts were modelled separately because rhythm type was expected to identify populations with different causes, injury severity and potential treatment responses.
• Why This Matters

  • Prolonged deep temperature reduction requires continued sedation, shivering control, invasive or surface temperature-control equipment and delayed rewarming.
  • If longer treatment improved neurological recovery, routine 12–24-hour regimens might have been biologically underdosed.
  • If the duration–response curve was flat, extending exposure to 33 °C would add treatment burden without demonstrable benefit.
  • Determining duration separately from temperature target could refine post-arrest care while avoiding the false assumption that more prolonged physiological manipulation must be more effective.

Design & Methods

• Research Question: To characterise the 33 °C cooling duration–response curve separately in patients with shockable and nonshockable initial rhythms, identify the shortest duration consistent with the maximal mean weighted 90-day modified Rankin Scale score, and determine whether longer cooling produced progressively better outcomes.
• Study Type: Investigator-initiated, multicentre, response-adaptive, Bayesian duration-finding randomised clinical trial with blinded central outcome assessment, conducted in emergency departments and ICUs at 71 academic and community hospitals in the United States under an FDA Investigational Device Exemption.
• Population:
  • Adults aged at least 18 years who remained comatose after resuscitation from out-of-hospital cardiac arrest.
  • Both shockable and nonshockable initial rhythms were eligible and were treated as distinct trial populations.
  • A definitive closed-loop surface or endovascular temperature-control device had to have been started before randomisation.
  • Patients were required to achieve and sustain a core temperature below 34 °C within 240 minutes of cardiac arrest.
  • Randomisation had to occur within 6 hours of starting the definitive cooling device.
  • A legally authorised representative had to consent and indicate an intention to maintain life-sustaining treatment for at least 96 hours.
  • Major exclusions were refractory haemodynamic instability, defined as systolic blood pressure below 80 mm Hg despite aggressive management; pre-existing neurological disability that confounded outcome assessment; terminal illness or anticipated death before follow-up; planned early withdrawal of life-sustaining therapy; presumed sepsis as the cause of arrest; and imprisonment.
  • Of 14,538 adults admitted alive after out-of-hospital cardiac arrest, 1,158 were randomised: 883 with nonshockable rhythms and 275 with shockable rhythms.
• Intervention:
  • All participants received therapeutic hypothermia targeted to 33 °C using a closed-loop surface or endovascular device.
  • Potential allocated durations were 6, 12, 18, 24, 30, 36, 42, 48, 60 and 72 hours.
  • The first 200 participants were randomised 1:1:1 to 12, 24 or 48 hours.
  • Subsequent allocation was response-adaptive, updated approximately every 50 enrolments and calculated separately within each rhythm cohort.
  • The algorithm allocated more patients to durations likely to be optimal while retaining allocation to durations needed to define the shape of the response curve.
  • The duration clock started when the definitive cooling device was activated, not when 33 °C was first achieved; the allocated interval therefore included induction time and any period before randomisation.
  • Controlled rewarming to 36.5 °C began after the allocated cooling duration.
  • Rewarming lasted for a period equal to the allocated cooling duration in the 6-, 12-, 18- and 24-hour groups, and for 24 hours in groups assigned to longer cooling.
  • After rewarming, active normothermia was maintained with the definitive device until liberation from mechanical ventilation or for up to 120 hours.
• Comparison:
  • There was no conventional untreated or normothermia control group.
  • Each cooling duration was compared through a common Bayesian duration–response model rather than through prespecified pairwise comparisons.
  • The shortest possible strategy was 6 hours of device-based cooling at 33 °C followed by controlled rewarming and continued active normothermia.
  • The 60- and 72-hour groups could be opened only if interim data suggested continued improvement through 48 hours; neither was opened.
• Blinding: Treating clinicians, study personnel delivering the intervention and families were aware of allocated duration. The primary 90-day modified Rankin Scale assessment was performed predominantly by a blinded central telephone assessor, and neuropsychological assessments were undertaken by blinded outcome assessors.
• Statistics: No single conventional effect-size, α or β calculation applied because this was a Bayesian response-adaptive duration-finding design. Simulation supported a maximum sample of 1,800: under the reference scenario in which benefit plateaued at 30 hours, 1,800 participants provided a 94% probability of detecting a positive duration–response and a 78% probability of correctly identifying the shortest effective duration; in a weak-effect scenario, the corresponding probabilities were 55% and 64%. The primary intention-to-treat analysis used a flexible inverted U-shaped model capable of fitting increasing, flat, decreasing or plateauing responses. The primary outcome assigned weighted scores of 10, 9, 8, 6 and 0 to modified Rankin Scale scores of 0, 1, 2, 3 and 4–6, respectively. “Optimal” was defined as the shortest duration consistent with the highest model-estimated mean weighted score.⁷
• Follow-Up Period: The primary neurological and mortality outcomes were assessed at 90 days. In-hospital outcomes, serious adverse events and withdrawal of life-sustaining therapy were also recorded.

Key Results

This trial was stopped early. Enrolment ended at a planned interim analysis after 1,158 of the maximum 1,800 participants had been randomised because the prespecified stopping rule favouring the shortest tested duration had been met. Allocations ranged from 6 to 48 hours; the 60- and 72-hour groups were never opened.

How to Interpret the Primary Outcome

• The modelled neurological outcome curve was flat to downward in both rhythm cohorts; no longer duration produced evidence of superior recovery.
• No meaningful treatment-effect heterogeneity was identified across examined subgroups, including sex, age, race or ethnicity, arrest cause, bystander CPR, cooling-device type, presenting temperature, prehospital adrenaline dose, STEMI or early head CT findings.

Internal Validity

• Randomisation and Allocation: Assignment was centrally generated through a web-based system. The first 200 allocations were fixed at 1:1:1, after which response-adaptive probabilities were updated approximately every 50 enrolments. Treating investigators were kept unaware of the evolving allocation probabilities, reducing the opportunity to anticipate assignment.
• Dropout or Exclusions: All 1,158 randomised participants were included in the intention-to-treat analysis. Primary outcome data were unavailable because of withdrawal or loss to follow-up in only 10 patients; these outcomes were imputed. Attrition was therefore unlikely to materially bias the principal analysis.
• Performance and Detection Bias: Clinical blinding was impossible because cooling duration, rewarming and device use were visible. This allowed duration to influence sedation, neuromuscular blockade, ventilation, investigation and prognostic impressions. Blinded central mRS assessment substantially reduced detection bias for the primary outcome but could not remove performance bias arising during ICU care.
• Protocol Adherence: Temperature trajectories demonstrated excellent separation by allocated duration. Nevertheless, only 92% of participants actually had a documented temperature below 34 °C within 4 hours, despite this being an eligibility criterion.
• Baseline Characteristics: Characteristics were broadly comparable across duration groups, but response-adaptive allocation produced markedly unequal group sizes. The nonshockable 12-hour group contained 288 patients compared with 42 in the 36-hour group; corresponding shockable groups contained 90 and 11 patients. Chance imbalances were consequently plausible in the smallest groups and explain some extreme unadjusted event rates.
• Illness Severity: The population was severely ill. Mean weighted mRS across durations was only 0.8 in the nonshockable cohort, with 81.3% mortality, compared with 3.5 and 52.0%, respectively, in the shockable cohort. The nonshockable population may therefore have contained many patients beyond the point at which modest neuroprotection could alter outcome.
• Heterogeneity: Modelling rhythm cohorts separately was appropriate, but substantial residual heterogeneity remained. Nonshockable arrests included respiratory causes, overdose, primary cardiac disease, pulmonary embolism and neurological events, each with different mechanisms and reversibility.
• Timing: Median time from cardiac arrest to definitive device initiation was 151 minutes: 152.5 minutes in nonshockable and 146 minutes in shockable arrests. Randomisation occurred on average 2.76 hours after device initiation. The intervention was therefore delivered early relative to many clinical series, but only after cooling had already been selected and commenced as standard care.
• Dose: The randomised “dose” was duration from device activation rather than duration actually maintained at 33 °C. Induction time was included, so a patient assigned to 6 hours could have received substantially less than 6 hours at the target temperature.
• Separation of the Variable of Interest: The 6-hour group began rewarming after 6 hours and was rewarmed over a further 6 hours, whereas the 48-hour group remained targeted to 33 °C until hour 48 and was then rewarmed over 24 hours. Despite this 42-hour difference in assigned hypothermia, all groups subsequently received active normothermia for up to 120 hours.
• Key Delivery Aspects: Closed-loop surface devices were used in 55% and endovascular devices in 45%. Blood pressure, oxygenation, ventilation, shivering, rebound fever, infection management and prognostication were addressed by clinical-standardisation guidance, but sedation and shivering treatment remained governed largely by local practice.
• Withdrawal of Life-Sustaining Therapy: Neurological prognostication leading to withdrawal was intended to be deferred for 96 hours, yet withdrawal occurred within 96 hours of device initiation in 140/881 nonshockable patients (15.9%) and 31/275 shockable patients (11.3%). During the admission, life support was withdrawn in 59.6% and 37.8%, respectively; poor neurological prognosis was recorded as the reason in 48.2% and 27.6%. Early prognostic decisions therefore remained a potential self-fulfilling source of outcome bias.
• Outcome Assessment: The 90-day mRS was clinically meaningful and centrally assessed, but the selected weighting assigned the same value of zero to mRS 4, mRS 5 and death. It therefore preserved distinctions among favourable outcomes while discarding distinctions between moderately severe disability, severe disability and death.
• Statistical Rigor: The Bayesian model, adaptive allocation and stopping rules were prespecified and well matched to a multidose duration-finding question. However, stopping at 1,158 patients reduced information in the shockable cohort and at longer durations, while the primary result was expressed as posterior target probabilities rather than familiar comparative effect sizes with confidence intervals.

Conclusion on Internal Validity: Internal validity is moderate to strong for the conclusion that extending 33 °C cooling beyond a short duration did not improve outcomes in rapidly cooled, haemodynamically stable out-of-hospital cardiac-arrest patients. It is more limited for declaring precisely 6 hours optimal, particularly in the shockable cohort, because of early stopping, boundary effects, sparse longer-duration groups, early withdrawal of life support and the absence of an untreated comparator.

External Validity

• Population Representativeness: The 71-centre network included both academic and community hospitals, and the cohort was more demographically and aetiologically diverse than many previous temperature-control trials.
• Rhythm and Cause: Nonshockable rhythms accounted for 76.3% of enrolment. In that cohort, only 22.5% of arrests were classified as cardiac, while 29.4% were respiratory and 28.1% were related to overdose or poisoning. This improves relevance to contemporary North American practice but differs substantially from European trials dominated by witnessed, shockable, presumed-cardiac arrests.
• Screened Versus Enrolled: Only 1,158 of 14,538 admitted patients were randomised. Many exclusions were appropriate because the screened population included awake patients and those with refractory shock, but the trial ultimately represented approximately 8% of admitted survivors.
• Demographic Comparability: Median age was 61 years in both enrolled and non-enrolled patients. Women comprised 39.6% of the trial versus 35.4% of those not enrolled. Black patients comprised 38.7% of enrolled participants compared with 26.0% of non-enrolled patients. Survival to discharge was 28.0% among enrolled and 30.6% among non-enrolled patients.
• Important Exclusions: Results do not directly apply to in-hospital cardiac arrest, children, awake survivors, arrests attributed to sepsis, patients with profound refractory haemodynamic instability, those with major pre-existing disability, or patients whose families would not support at least 96 hours of continued treatment.
• Temperature-Practice Selection: Participants had already been selected for rapid cooling to 33 °C before randomisation. The findings therefore apply most directly to centres that choose deep hypothermia and can achieve a temperature below 34 °C within 4 hours.
• Resource Requirements: Both intervention delivery and subsequent normothermia required a closed-loop surface or endovascular device. Applicability is lower in resource-limited environments relying on intermittent or non-feedback temperature-control methods.
• Different Temperature Strategies: The trial does not determine duration in patients managed from the outset with normothermia or reactive fever treatment, and it does not compare 33 °C with alternative targets between 34 and 37.5 °C.

Conclusion on External Validity: Generalisability is good for contemporary US adults with out-of-hospital cardiac arrest who remain comatose, are sufficiently stable for intensive treatment and are rapidly cooled with a definitive device. Applicability is limited outside this selected population and should not be extrapolated to patients managed without deep hypothermia, without active fever prevention or after in-hospital arrest.

Strengths & Limitations

• Strengths:
  • Large, investigator-initiated randomised trial across 71 hospitals.
  • Innovative response-adaptive design capable of examining multiple clinically plausible durations without requiring equally large fixed groups.
  • Separate modelling of shockable and nonshockable rhythms.
  • Broad inclusion of respiratory, toxicological and noncardiac arrest causes.
  • Central concealed allocation and blinded primary outcome assessment.
  • Excellent separation of temperature exposure across allocated durations.
  • Only 10 participants lacked observed primary outcome data.
  • Prespecified adaptive stopping rather than post hoc termination.
  • Detailed temperature curves, mRS distributions, safety reporting and comparison with the screened population.
  • Standardisation of major post-arrest care and an explicit attempt to delay neurological prognostication.
• Limitations:
  • No normothermia, fever-prevention-only or zero-duration control group.
  • The shortest permitted duration was 6 hours; the design cannot determine whether an even shorter exposure or no 33 °C exposure would have performed similarly.
  • The trial stopped at 1,158 rather than the maximum 1,800 participants.
  • The 60- and 72-hour groups were never opened, so the trial provides no direct clinical evidence regarding these durations.
  • Only 275 participants had shockable rhythms, with as few as 11–36 patients in several individual duration groups.
  • Very high mortality and poor neurological outcome created a floor effect, particularly in nonshockable arrest.
  • The rapid-cooling criterion may have inadvertently selected patients who arrived spontaneously hypothermic or cooled easily because severe brain injury had impaired thermoregulation.
  • Treating teams were unblinded, and sedation, shivering treatment and other co-interventions were not fully standardised.
  • Substantial early withdrawal of life-sustaining therapy persisted despite protocol safeguards.
  • The weighted primary outcome assigned identical value to mRS 4, mRS 5 and death.
  • Neuropsychological analyses were based on small numbers of survivors and had little power to detect clinically important cognitive differences.
  • The protocol-specified patient-reported quality-of-life outcome was not presented in the primary publication.
  • The Bayesian model did not provide the conventional pairwise risk ratios, confidence intervals or noninferiority margins that many clinicians would use to judge whether 6 and 12 hours were acceptably equivalent.

Interpretation & Why It Matters

• What ICECAP Resolves

  In patients already selected for rapid therapeutic hypothermia at 33 °C, extending deep cooling from 6 hours through 48 hours did not improve neurological recovery, mortality, cognition or safety outcomes.
• What “6 Hours” Means

  The 6-hour strategy was not cessation of temperature management after 6 hours. It was 6 hours from device activation, followed by 6 hours of controlled rewarming and then active normothermia for up to 120 hours.
• Practical Consequence

  For centres choosing 33 °C, routine prolongation of deep hypothermia beyond a short exposure is unsupported. The trial favours minimising unnecessary time at 33 °C while retaining controlled rewarming and active fever prevention.
• What It Does Not Prove

  ICECAP does not establish that 33 °C is superior to controlled normothermia, that 6 hours is superior to 12 hours, or that temperature-control devices can be stopped after 6 hours.
• Implication for Biological Reasoning

  The trial directly challenges the extrapolation that more prolonged hypothermia must be more neuroprotective because it is more effective in some experimental models. A compelling physiological mechanism did not translate into a clinically detectable human duration–response.
• Remaining Question

  The central unresolved issue is no longer simply “how long should everyone be cooled?”, but whether particular biological injury phenotypes benefit from a specific temperature, onset time or duration.

Controversies & Other Evidence

• Early termination was planned, not a trial failure: The accompanying editorial places ICECAP within modern adaptive trial methodology, in which prespecified stopping can indicate that a trial has answered its design question efficiently rather than that recruitment has failed.⁸
• The stopping rule was not equally demonstrated in both rhythm cohorts: The protocol allowed stopping within a rhythm cohort once more than 50 patients had received 6-hour cooling and the posterior probability that 6 hours was the target duration exceeded 0.50. At publication, the nonshockable cohort clearly fulfilled these numerical conditions with 89 patients and a posterior probability of 0.507. The shockable cohort had only 45 patients assigned to 6 hours and a posterior probability of 0.490. The primary report’s statement that the trial met its prespecified stopping rule therefore compresses two different evidentiary states, and the reason for terminating further shockable-rhythm enrolment is less transparent.
• “Optimal” is a model definition, not proof of clinical superiority: The target duration was defined as the shortest duration statistically consistent with the modelled maximum. When a fitted curve is flat or declining, probability naturally accumulates at the shortest permitted boundary. The resulting 0.49–0.51 probability for 6 hours is not equivalent to a 95% probability that 6 hours is truly best.
• The lower boundary was 6 hours: Because neither zero hours nor durations below 6 hours were tested, the trial cannot establish whether 6 hours represents a biological minimum. It identifies the left boundary of the tested range as sufficient relative to longer exposure.
• No absolute efficacy comparator: A positive duration–response might have supported the causal efficacy of cooling despite the absence of normothermia. The observed flat-to-decreasing response cannot do so. All groups may have benefited similarly from brief cooling, none may have benefited, or the subsequent common fever-prevention strategy may have dominated any duration effect.
• The randomised exposure was device time rather than target-temperature time: Duration began at device activation and included induction. The 6-hour group therefore did not necessarily receive 6 complete hours at 33 °C. ICECAP evaluates a pragmatic policy of device-based cooling from initiation, not a pure biological dose measured from attainment of target temperature.
• Outcome weighting embeds a value judgement: Assigning zero to mRS 4, mRS 5 and death avoids treating transition from death to severe dependency as a trial success, but it also discards potentially important distinctions between survival with severe disability and death. Separate mortality analyses were therefore essential.
• Withdrawal of life support remained a competing intervention: More than one in nine participants underwent withdrawal within 96 hours despite planned deferral of neurological prognostication. If early withdrawal was influenced by unblinded treatment duration or institutional nihilism, it could have reduced the opportunity for delayed awakening and reinforced the very poor outcome floor.
• The historical literature remains internally inconsistent: The 2002 trials suggested benefit from 12–24 hours of 33 °C compared with poorly controlled normothermia, whereas TTM and TTM2 found no advantage over structured higher-temperature strategies. TTM48 did not show that 48 hours was superior to 24 hours, and HYPERION’s favourable neurological outcome was 10.2% with 33 °C versus 5.7% with normothermia. An individual-patient meta-analysis combining nonshockable-rhythm participants from HYPERION and TTM2 subsequently found no significant survival or functional benefit from 33 °C.^1234569
• Current guidelines predate ICECAP: The 2025 American Heart Association guideline considers it reasonable to maintain temperature control between 32 and 37.5 °C for at least 36 hours in adults who remain unresponsive.¹⁰ The 2025 ERC–ESICM guideline emphasises active fever prevention, generally for 36–72 hours, rather than mandating prolonged 33 °C hypothermia.¹¹ ICECAP is compatible with continued temperature control because even its 6-hour group received controlled rewarming and subsequent active normothermia; it should not be interpreted as evidence for abandoning fever prevention.

Summary

• ICECAP randomised 1,158 comatose out-of-hospital cardiac-arrest patients at 71 US hospitals to adaptive durations of 33 °C cooling ranging from 6 to 48 hours.
• The trial stopped at a planned interim analysis after the duration–response curve failed to show better neurological recovery with longer cooling.
• The posterior probability that 6 hours was the shortest duration consistent with maximal outcome was 0.507 in nonshockable and 0.490 in shockable rhythms; 12 hours retained probabilities of 0.305 and 0.301.
• Longer cooling did not improve 90-day mortality, dichotomised neurological outcome, neuropsychological performance or adverse events.
• The result supports avoiding prolonged exposure to 33 °C, but does not prove that 6 hours is uniquely optimal, establish efficacy versus normothermia or justify stopping active temperature control after 6 hours.

Overall Takeaway

ICECAP is an important and methodologically innovative duration-finding trial that provides strong evidence against routinely prolonging therapeutic hypothermia at 33 °C in rapidly cooled comatose survivors of out-of-hospital cardiac arrest. Its central lesson is narrower than “6 hours is best”: longer deep hypothermia did not improve outcome, while controlled rewarming and subsequent fever prevention remained integral parts of every tested strategy.

Bibliography

• 1.Bernard SA, Gray TW, Buist MD, et al. Treatment of comatose survivors of out-of-hospital cardiac arrest with induced hypothermia. N Engl J Med. 2002;346(8):557-563.
• 2.Hypothermia After Cardiac Arrest Study Group. Mild therapeutic hypothermia to improve the neurologic outcome after cardiac arrest. N Engl J Med. 2002;346(8):549-556.
• 3.Nielsen N, Wetterslev J, Cronberg T, et al; TTM Trial Investigators. Targeted temperature management at 33°C versus 36°C after cardiac arrest. N Engl J Med. 2013;369(23):2197-2206.
• 4.Kirkegaard H, Søreide E, de Haas I, et al. Targeted temperature management for 48 vs 24 hours and neurologic outcome after out-of-hospital cardiac arrest: a randomized clinical trial. JAMA. 2017;318(4):341-350.
• 5.Lascarrou JB, Merdji H, Le Gouge A, et al; CRICS-TRIGGERSEP Group. Targeted temperature management for cardiac arrest with nonshockable rhythm. N Engl J Med. 2019;381(24):2327-2337.
• 6.Dankiewicz J, Cronberg T, Lilja G, et al; TTM2 Trial Investigators. Hypothermia versus normothermia after out-of-hospital cardiac arrest. N Engl J Med. 2021;384(24):2283-2294.
• 7.Meurer WJ, Schmitzberger FF, Yeatts S, et al; ICECAP Trial Investigators. Influence of cooling duration on efficacy in cardiac arrest patients (ICECAP): study protocol for a multicenter, randomized, adaptive allocation clinical trial to identify the optimal duration of induced hypothermia for neuroprotection in comatose, adult survivors of after out-of-hospital cardiac arrest. Trials. 2024;25(1):502.
• 8.Huang AJ, Lewis RJ. Trials terminated early: when is enough, enough? JAMA. Published online August 5, 2026.
• 9.Taccone FS, Dankiewicz J, Cariou A, et al. Hypothermia vs normothermia in patients with cardiac arrest and nonshockable rhythm: a meta-analysis. JAMA Neurol. 2024;81(2):126-133.
• 10.Hirsch KG, Amorim E, Coppler PJ, et al. Part 11: post-cardiac arrest care: 2025 American Heart Association guidelines for cardiopulmonary resuscitation and emergency cardiovascular care. Circulation. 2025;152(16 Suppl 2):S673-S718.
• 11.Nolan JP, Sandroni C, Cariou A, et al. European Resuscitation Council and European Society of Intensive Care Medicine guidelines 2025: post-resuscitation care. Intensive Care Med. 2025;51(12):2213-2288.