(Last Updated On: September 17, 2018)

Association Between Therapeutic Hypothermia and Survival After In-Hospital Cardiac Arrest.

Chan, Paul  et al. JAMA. Oct 2016. [paper]

Why I chose this study

I recently rotated through the ICU and experienced multiple in-hospital cardiac arrests. There is research that suggests therapeutic hypothermia improves outcomes for out-of-hospital cardiac arrest. We were cooling patients with in-hospital cardiac arrest and I was wondering about the data behind this guideline. I ran into this article sometime last year, and am interested in its possible implications.


Previous research for therapeutic hypothermia, or targeted temperature management, has shown that it improves both overall survival and rates of favorable neurological survival for comatose patients who have received ROSC after out-of-hospital cardiac arrest due to ventricular fibrillation¹. This technique is thought to reduce oxygen demand by decreasing the metabolic rate and thereby decreasing oxidative stress from circulating free-radicals and reperfusion injuries which can result after prolonged tissue hypoxia.

In-hospital cardiac arrest affects approximately 200,000 individuals annually in the US. Despite there being no previous randomized trials conducted in the in-hospital setting, therapeutic hypothermia is recommended for comatose patients following both out-of-hospital and in-hospital cardiac arrest, regardless of initial rhythm.

Research Question

Is therapeutic hypothermia associated with better survival outcomes for patients with in-hospital cardiac arrest?

Study Design

Multi-center, retrospective, propensity-matched cohort study performed at 355 US hospitals in cooperation with the national Get With the Guidelines-Resuscitation (GWTG-Resuscitation) registry.

26,183 patients successfully resuscitated from an in-hospital cardiac arrest between March 1, 2002, and December 31, 2014, and either treated or not treated with hypothermia. Follow-up ended February 4, 2015.

  • 1568 patients were treated with therapeutic hypothermia
  • 1524 patients treated with hypothermia were successfully propensity matched to 3714 patients not treated with hypothermia

Inclusion Criteria

  • Patients aged 18 years and older enrolled in GWTG-Resuscitation between March 1, 2002, and December 31, 2014.
  • Only patients with ROSC after an index in-hospital cardiac arrest.
  • For those aged 65 years and older, patients who were linked to Medicare inpatient files were included to enable examination of post-discharge survival.
  • Only cases occurring after the first documented use of therapeutic hypothermia for in-hospital cardiac arrest at each hospital.
  • The cohort was restricted to patients on mechanical ventilation at the time of cardiac arrest or after cardiac arrest (to approximate for comatose patients).

Exclusion Criteria

  • No return of ROSC.
  • Those aged 65 years and older not linked to Medicare inpatient files (no unique match or enrolled after 2012).
  • Patients from hospitals with no cases of therapeutic hypothermia.
  • Patients prior to the first documented use of therapeutic hypothermia for in-hospital cardiac arrest at each hospital.
  • Patients with missing information on survival to discharge and comorbidities for model adjustment.
  • Patients with an initial out-of-hospital cardiac arrest followed by an in-hospital cardiac arrest.
  • Death within 24 hours of index cardiac arrest.


  • Primary Outcome: In-hospital survival (i.e., to hospital discharge)
  • Secondary Outcomes: Favorable neurological survival and cumulative survival among patients aged 65 years and older.


Propensity score analyses were conducted to evaluate the association between therapeutic hypothermia treatment and survival outcomes. A multivariable logistic regression model was constructed to estimate the patient’s likelihood of being treated with therapeutic hypothermia after in-hospital cardiac arrest, including hospital site, age, sex, self-identified race, initial cardiac arrest rhythm, location of cardiac arrest, comorbid conditions, medical conditions present within 24 hours of cardiac arrest, and interventions in place at the time of cardiac arrest.

The model also adjusted for duration of acute CPR, the time of day, and day of the week of the cardiac arrest. Besides matching by propensity score, hypothermia-treated and non-hypothermia-treated patients were additionally matched on 3 other criteria: cardiac arrest within 365 days of the hypothermia-treated patient’s cardiac arrest, initial cardiac arrest rhythm, and duration of acute CPR.


  • Initial inclusion: 117,005 patients with in-hospital cardiac arrest with ROSC from 674 hospitals
  • 90,822 patients were excluded.
  • Final cohort: 26,183 patients from 355 hospitals.
  • 1568 patients (6.0%) were treated with therapeutic hypothermia.
    • Patients treated with hypothermia were younger, less likely to have a cardiac arrest in the intensive care unit, and more likely to have an initial cardiac arrest rhythm of ventricular fibrillation.
    • Patients treated with hypothermia were less likely to have hypotension, respiratory insufficiency, pneumonia, acute stroke, and a metastatic or hematologic malignant neoplasm at the time of the cardiac arrest.
  • The propensity score for the overall cohort had good discrimination and led to the successful matching of 1524 patients treated with hypothermia to 3714 patients not treated with hypothermia.
  • The median lowest temp for hypothermia treated patients was 33.1°C with 20.9% falling below the recommended nadir of 32°C.  The median lowest temp was 36.3°C for the non-treated patients.

Primary Outcome

  • 417 patients treated with therapeutic hypothermia (27.4%) survived to hospital discharge, as compared to 1084 non-hypothermia treated patients (29.2%).
  • This association was similar (p = 0.74) for non-shockable cardiac arrest rhythms (22.2% vs 24.5%) and shockable cardiac arrest rhythms (42.3% vs 44.1%).
  • A lower proportion of patients in the hypothermia-treated group died during the first day than in the non-hypothermia treated group (29.1% vs 45.0%,  p < 0.001).

Secondary Outcomes

  • Favorable Neurological Survival: Therapeutic hypothermia was associated with a lower likelihood of favorable neurological survival for all rhythms (17.0% vs 20.5%; RR, 0.79; p < 0.001), with similar patterns in patients with non-shockable and shockable rhythms (p = 0.88).
  • One-year Survival: Cumulative survival throughout the first year was similar between the 2 groups (mean, 2.21 months vs 2.20 months, p = 0.92).


This is a large national registry study with huge numbers compared to previous studies. Even with the large numbers the authors pointed out a lot of limitations. The main one being that it was a retrospective propensity matched cohort study and not a randomized controlled trial. Depending on the variables picked for the matching and how well the cohorts matched there may still be uncontrolled differences between the two groups. Another big limitation was missing data on the hypothermia protocols used and even the temperature of the patients studied. Looking at the mean lowest temp the patients receiving hypothermia were on aggressive protocols and many were cooled below the recommended temp of 32°C. Current protocols typically recommend less aggressive cooling with target temps between 34-36°C. The temperatures may account for the lack of benefit and also the potential for harm. Even with the limitations, these findings do not support use of therapeutic hypothermia for patients with in-hospital cardiac arrest.

Evidence does favor therapeutic hypothermia protocols for comatose patients after out-of-hospital cardiac arrest after achieving ROSC. The difference may be because therapeutic hypothermia is not necessarily beneficial for all brain injury patterns. For instance, clinical trials have found that therapeutic hypothermia leads to worse survival outcomes for other conditions, such as traumatic brain injury and bacterial meningitis. If we consider in-hospital cardiac arrest as a different entity and disease process than out-of-hospital cardiac arrest, we can begin to see why similar treatments might not lead to similar results, and might even be contraindicated or lead to harm.

As is apparent upon consideration, with in-hospital cardiac arrest, the response times, comorbidity burden, and cardiac arrest etiology often differ markedly from the out-of-hospital setting. Notably, more than 80% of in-hospital cardiac arrests have initial non-shockable rhythms of asystole or PEA. For these rhythms, the benefit of therapeutic hypothermia has not been shown. In-hospital cardiac arrest has median response times of <1 minute to CPR, 1 minute to first defibrillation, and 3 minutes to first epinephrine dose. This may limit the theorized benefit of therapeutic hypothermia to reduce free-radical mediated reperfusion injury from anoxic brain injury.

The conclusions are rather surprising, considering there may actually be harm after utilizing therapeutic hypothermia for in-hospital cardiac arrest, regardless of initial rhythm. As mentioned before this may be related to aggressive cooling and new protocols may not have the same risk of harm. For the time being, I will continue to follow the guidelines and initiate therapeutic hypothermia in truly comatose patients after in-hospital cardiac arrest. I would be more likely to cool the patients who have longer arrests, although my actions may differ on a case-by-case basis. I will also likely use 36°C as my target temperature based on other research, given the possibility of harm from hypothermia, although again, this research was done for out-of-hospital cardiac arrest².

¹Bernard, et al. “Treatment of Comatose Survivors of Out-of-hospital Cardiac Arrest with Induced Hypothermia. NEJM. 2002; 346(8): 555-63.

²Nielsen, et al. “Targeted temperature management at 33°C versus 36°C after cardiac arrest.”  NEJM. 2013; 369(23): 2197-206.


Will Brumley is an EM/IM Resident, class of 2021



EM/IM Sessions are reviewed in journal club style by the current attendings and residents, as well as alumni of the UIC IM/EM program prior to publication. This post was specifically reviewed by Adam Rodos, MD, Assistant Program Director for the EM/IM Residency. Elspeth Pearce, MD Editor.