Cooling after out-of-hospital cardiac arrest: The TTM2 trial

Reading Time: 6 minutes
Picture of Shyam Murali
Shyam Murali
Fellow in Trauma and Surgical Critical Care - University of Pennsylvania, Senior Editor -, Writer -, Saxophonist, EDM remixer, husband, puppy father, and new human father
Picture of Mohamed Hagahmed
Mohamed Hagahmed

Clinical Assistant Professor at the University of Pittsburgh, Department of Emergency Medicine.
Associate Program Director for the Center of Emergency Medicine of Western Pennsylvania.
EMS Medical Director.
Paramedic at heart, traveler, and multi-lingual.

The Pre-brief

The concept of “therapeutic hypothermia” was introduced in the care of post-cardiac arrest patients nearly two decades ago. Initial studies hinted at a benefit in neurological outcome for comatose patients after cardiac arrest with cooling to 32C to 34C for 24 hours. Subsequent studies which compared cooling patients to 33C vs. 36C, as well as varied durations of cooling, have yielded equivocal results and raised more questions than answers.

Enter…the TTM2 trial:

Article: 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 Jun 17;384(24):2283-2294. doi: 10.1056/NEJMoa2100591. PMID: 34133859.

We’ll cover some details about the study, but if you want to read the important parts, jump down to the Strengths/Limitations/Discussion section to catch our critical appraisal and interpretation of the study.

Thanks to Dr. Niklas Nielsen and Dr. Josef Dankiewicz for sending us supplementary materials to use for images in this post.

What did they do?

  • Open-label randomized superiority trial
  • International (included 14 countries in Australia, Europe, and North America)
  • Pre-published the design and statistical analysis
  • An independent data and safety monitoring committee reviewed the data and performed one prespecified, blinded interim analysis
  • Population
    • Inclusion
      • Screened consecutive adults after OHCA of presumed cardiac origin or unknown cause, irrespective of initial rhythm
      • Unconscious and unable to follow verbal commands
        • Score of <4 on Full Outline of UnResponsiveness scale
        • No verbal response to pain after sustained ROSC
      • Sustained ROSC (>20 minutes)
    • Exclusion
      • Interval from ROSC to screening more than 180 minutes
      • Unwitnessed cardiac arrest with asystole as the initial rhythm
      • Limitations in care
      • Temp <30C on admission
      • On ECMO prior to ROSC
      • Obvious or suspected pregnancy
      • Intracranial bleeding
      • Severe COPD with long-term home O2 therapy
  • Intervention
    • Hypothermia (33C)
  • Comparison
    • Normothermia (<37.5C)
    • If temp reached 37.8C, cooling was performed with either surface or intravascular device
  • Outcomes
    • Primary: Death from any cause at 6 months
    • Secondary:
      • Poor functional outcome at 6 months (modified Rankin scale (mRS) 4-6; determined by face-to-face or telephone interviews)
        • If structured assessment could not be completed, then a binary assessment was performed based on all available data: good (0-3) vs poor (4-6)
      • Number of days patient was alive and out of the hospital until day 180
      • Survival in time-to-death analysis
      • Health-related quality of life (VAS of EQ-5D-5L)
      • Adverse events: pneumonia, sepsis, bleeding, arrhythmia resulting in HD compromise, skin complications related to device used for targeted temperature management
    • Intention-to-treat analysis
    • Estimated sample of 1,862 patients to provide 90% power to detect a relative reduction of 15% in the risk of death in the hypothermia group


    • Enrolled 1900 patients between November 2017 and January 2020
      • Most cases enrolled from Sweden, United Kingdom, and Switzerland
    • Intention-to-treat population of 1861 after 39 exclusions for consent and randomization errors
      • 930 assigned to hypothermia
      • 931 assigned to normothermia
    • Similar characteristics between groups with regard to
      • Age
      • Gender
      • Number of shocks
      • Bystander CPR
      • Shockable vs non-shockable rhythm
      • Times from arrest to sustained ROSC and time from arrest to randomization
      • Initial temperature
      • Corneal and pupillary reflexes
      • pH, lactate, presence of shock, STEMI
    • Patients who received cooling with a device
      • Hypothermia: 95%
      • Normothermia: 46%
      • Approximately 70% surface and 30% intravascular in both groups
  • Primary Outcome – death at 6 months (RR 1.04; 95% CI 0.94 to 1.14)
      • Hypothermia: 50%
      • Normothermia: 48%
      • Results consistent across prespecified subgroups and when assessed in a time-to-event analysis
  • Secondary Outcomes
    • Poor functional outcome (mRS 4-6) assessed in 98% of patients (RR 1.00; 95% CI 0.92 to1.09)
      • Hypothermia: 55%
      • Normothermia: 55%
Created by Mark Ramzy, DO on
  • Similar EQ-5D-5L scores in patients alive at 6 months (mean between-group difference -0.8 points; 95% CI -3.6 to 2.0)
    • Hypothermia: 74
    • Normothermia: 75
Thanks to the authors of the TTM2 trial for sending us this image.
  • Adverse Events
    • Arrhythmias resulting in hemodynamic compromise (RR 1.45; 95% CI 1.21 to 1.75)
      • Hypothermia: 24%
      • Normothermia: 17%
    • No significant differences in pneumonia, sepsis, bleeding, skin complications related to device used for targeted temperature management.


  • Large sample size of nearly 2000 patients from multiple countries. This multicenter and multinational nature adds to the external validity of the study. This study is the largest conducted thus far assessing this question.
  • The study question was clear at addressing the issue at hand (hypothermia vs. maintaining normothermia).
  • Study design and statistical analyses were pre-published.
  • Baseline characteristics were very similar between groups, including the initial temperature at randomization.
  • Blinding was very robust: blinded assessment of prognosis at 96 hours, blinded participants, blinded outcome assessors, blinded authors and steering group. The only non-blinded group was the healthcare professionals caring for the patient.
  • Intention-to-treat analysis was performed.
  • More rapid cooling in TTM2 compared to HACA, possibly due to advances in technology. This figure compares the temperatures of HACA to TTM2. While HACA reached 33C, TTM2 reached approximately 33.3C, and more quickly (likely no significant difference between 33C and 33.3C). Furthermore, the 75th percentile bars (not shown here) in the HACA trial showed that about 25% of patients in the normothermia group in the HACA trial had fevers.
Adapted from materials sent to us by the authors of the TTM2 trial.
  • Incredible follow-up! 72% of participants had face-to-face follow-up with structured mRS assessment in 94% of participants. Binary assessment of functional status performed in 98% of participants. Very few missing data on mortality (<1%).
  • Variability in clinical practice was minimized by utilizing protocolized care with minimal protocol deviation related to temperature management.
  • Shivering was aggressively controlled in the study with a standardized protocol.


  • Only ~20% of the subjects enrolled in the study were female.
  • 56% of screened patients were not randomized into the trial due to various reasons
  • This study did not explore the effect of no temperature management on patient outcomes. It would have been nice to see a third group in this study that evaluated the effect of no temperature control. Based on previous studies, the working hypothesis is that avoiding fever is the name of the game with therapeutic hypothermia. In the past, the overwhelming data behind targeted temperature management, made it “unethical” to include a pure control group. Perhaps in light of this data, no temperature control versus fever prevention should be the next study conducted.
  • The external validity of this study is questionable with >90% of subjects having bystander-witnessed arrest and ~80% having bystander CPR. Nearly three-quarters of patients had an initial rhythm that was a shockable rhythm. The proportion of STEMI patients was also quite large. Most of the places where we (the authors) work, do not see this cohort of patients.
  • Surface cooling was used more frequently than intravascular devices (70% vs. 30%), and almost half of the normothermia group also received some cooling. This is also slightly questionable as more often than not, we (the authors) are inserting intravascular cooling devices to manage temperature, as opposed to using surface cooling devices.
  • In-hospital cardiac arrests were excluded from the study.
  • Unwitnessed asystole patients were also excluded from the study; however, these patients have very poor prognosis in general. Achieving ROSC in this group is a challenge in and of itself.
  • It is unclear why the authors did not randomize patients who had noncardiac causes of cardiac arrest. Since we perform targeted temperature management to preserve brain function, does the brain know why the blood isn’t flowing?
  • 130 protocol deviations in 126 participants (63 in each group); 6.7% of the study population
    • 8 randomized despite not fulfilling eligibility criteria
    • 10 participants did not have intervention started due to cerebral hemorrhage or brain death
    • 7 participants’ interventions were stopped before 6 hours and palliative care was started
    • 3 protocol deviations in temperature management
    • 29 early awakening
    • 11 missing EEGs
    • 62 participants had deviations from the protocol for neuroprognostication


Compared to the early hypothermia studies, the TTM2 trial contains a larger sample size and is generally better designed to minimize bias. Josh Farkas (@pulmcrit) aptly refers to the study as a “methodological fortress”. It showed that OHCA patients who were randomized to receive lower targeted temperature did not have better mortality and functional outcome and suffered from increased iatrogenic harm (specifically arrhythmias) compared to patients who were kept normothermic. The authors propose some possible reasons for arrhythmias causing hemodynamic instability: electrolyte disturbances, fluid status, and temperature effect on cardiac myocytes. They also note that of 53 patients that were rewarmed in the hypothermia group, 17 (32%) were due to hemodynamic compromise and 12 (23%) were due to bradycardia.

The supplementary materials in the paper also list the various unexpected serious adverse events, ranging from leg ischemia, cardiac tamponade, major stroke, tension pneumothorax, tracheal injury during intubation, and many others. In fact, patients in the hypothermia group had more than twice as many unexpected complications than patients in the normothermia group (3.7% vs. 1.4%). Is “therapeutic hypothermia” really therapeutic with all of these complications?

Based on the results of this trial, is it time to change clinical practice and completely forget about “therapeutic hypothermia”? Given the results of highly-powered studies like the TTM and TTM2 trials, hypothermia is becoming less and less useful as a means to prevent further neurologic injury. While it seems clear that targeted temperature management to 33C does not improve outcomes compared to normothermia, it is hard to ignore the skewed patient population included in this study (~90% witnessed arrest, ~80% bystander CPR, ~80% male, ~72% shockable). These demographics are also consistently found in the 2013 TTM trial. Despite this, we may be seeing a shift to temperature control at normothermic ranges and prevention of fevers.

Author’s Conclusion

“Patients with coma after out-of-hospital cardiac arrest who were treated with hypothermia did not have a lower incidence of death at 6 months than those who were treated with normothermia.”

Our Conclusion

Protocolized care that focuses on achieving hypothermia (32C-33C) is not supported by current evidence and may confer risk without clear benefit. Active temperature management targeting normothermia and avoiding fever makes more logical sense; however, further studies need to be conducted to actually determine if targeted temperature management (at normothermic temperatures) actually has any benefit.

For more FOAMed on this trial, check out:


More Posts

Related Posts

Would love your thoughts, please comment.x