By Peter B. Forgacs, MD

Instructor in Neuroscience and Neurology, Feil Family Brain and Mind Research Institute and Department of Neurology, Weill Cornell Medical College; Instructor in Clinical Investigation, The Rockefeller University, New York

Dr. Forgacs reports no financial relationships relevant to this field of study.

SYNOPSIS: Current guidelines recommend the use of therapeutic hypothermia in patients with in-hospital cardiac arrest, even though its efficacy has been demonstrated only in randomized trials after out-of-hospital cardiac arrest. This non-randomized, observational cohort study based on a large national registry found that the use of therapeutic hypothermia was associated with lower likelihood of survival and less favorable neurological outcome in patients successfully resuscitated after an in-hospital cardiac arrest.

SOURCE: Chan PS, Berg RA, Tang Y, et al; American Heart Association’s Get With the Guidelines–Resuscitation Investigators. Association between therapeutic hypothermia and survival after in-hospital cardiac arrest. JAMA 2016;316:1375-1382.

Approximately 200,000 patients suffer in-hospital cardiac arrest annually in the United States. Current American Heart Association (AHA) guidelines recommend the use of a therapeutic hypothermia (or another form of a targeted temperature management [TTM] protocol, i.e., therapeutic normothermia) after both out-of-hospital and in-hospital cardiac arrest. However, its efficacy has been demonstrated only using randomized clinical trials in the specific setting of out-of-hospital cardiac arrest as a result of ventricular fibrillation. This study addresses the very important question of whether or not — similar to out-of-hospital cardiac arrest — therapeutic hypothermia is associated with better survival and neurological outcome for patients with in-hospital cardiac arrest.

This cohort study leveraged data from a large, prospective, national registry of more than 26,000 patients successfully resuscitated after an in-hospital cardiac arrest collected over a 12-year period. Primary outcome was defined as survival to hospital discharge, and secondary outcome was favorable neurological outcome determined as score 1 or 2 of the Cerebral Performance Category (CPC) scale (good recovery or moderate neurological deficits allowing independent functioning). The registry, if available, also was linked with Medicare files for patients older than 65 years of age to find associations between hypothermia treatment and one-year survival in this subset of patients. Importantly, however, only about 1,500 patients (6% of the entire cohort) received therapeutic hypothermia. These patients were matched to more than 3,700 non-hypothermia-treated patients by a model based on a propensity score that assessed detailed baseline demographic and clinical variables. After adjusting for these variables, using a multivariable logistic regression model, this study found that use of therapeutic hypothermia was associated with lower survival (27.4% vs. 29.2%) and lower likelihood of favorable neurological outcomes at hospital discharge (17.0% vs. 20.5%). At one-year follow-up, there was no difference in survival between the two groups (14.2% vs. 14.1%). These associations were similar for patients with shockable and non-shockable initial rhythms. The authors concluded that therapeutic hypothermia is not associated with better survival or neurological outcomes, and it may be even potentially harmful in the setting of an in-hospital cardiac arrest.


This non-randomized, cohort study addressed a very pressing gap in our knowledge about whether therapeutic hypothermia is an appropriate choice of treatment for all patients after cardiac arrest or if it should be restricted to patients who were successfully resuscitated after an out-of-hospital cardiac arrest. Although the study was well designed, involved a large number of patients, and was carefully analyzed, it was not randomized and, therefore, it bears several inherent limitations that warrant caution in interpreting the results and the authors’ conclusions.

First, a major limitation includes the fact that only 6% of patients with in-hospital cardiac arrest underwent therapeutic hypothermia in this cohort. The inclusion of such a small portion of the population in the main analysis raises important questions regarding generalizability of the results. The authors make a great effort to successfully match a non-hypothermia-treated population with similar demographics, comorbid conditions, likelihood of being treated with hypothermia (e.g., hospital site), circumstances of the resuscitation (such as duration, time of day, day of the week), and post-cardiac arrest interventions (medications and/or devices utilized) using a propensity score matching algorithm. Nonetheless, before propensity score matching, the non-hypothermia-treated cohort (n = 24,615) significantly differed from the hypothermia-treated cohort (n = 1,568) in several important baseline variables: e.g., non-hypothermia-treated patients were older; more likely to be in a telemetry or in an intensive care unit than in the emergency department or other procedural areas; more likely had a non-shockable initial rhythm; and differed in several preexisting comorbid conditions. The results seen after propensity matching may indicate potential ineffectiveness or harm in the small subset of inpatients who are selected to undergo hypothermia treatment under current and evolving clinical practice (distributed over the last 12 years); however, it is unclear at this point if these findings can be applied directly to all patients who suffer an in-hospital cardiac arrest.

Additional limitations include lack of detailed data on hypothermia protocols and efficacy of treatments in the registry that may have affected the outcomes; lack of specific information about level of consciousness (i.e., if all patients were comatose at the time of enrollment, however, only patients on mechanical ventilation were included to mitigate this concern); and variability in assessing neurological outcome status at different sites. In addition, there is still a possibility of residual bias in patient selection despite a lot of effort taken to reduce biases, e.g., the indication bias, including a sensitivity analysis that excluded patients who died within the first 24 hours after cardiac arrest that yielded similar results as the main analysis. Most importantly, as this study was not randomized, there may be many other unknown confounders that may have affected the outcomes; therefore, these results cannot be translated readily to clinical practice.

Importantly, this study also highlights our lack of knowledge regarding neuroprotective mechanism of hypothermia after severe brain injuries that could provide some explanation for these results. Indeed, a few studies involving other injury mechanisms, such as severe trauma or bacterial meningitis, showed no benefit or even potential harm of using therapeutic hypothermia. It is plausible that patients who suffer in-hospital vs. out-of-hospital cardiac arrest have different baseline variables and comorbid conditions that may shift neuronal vulnerability or alter the ability to maintain optimal conditions during hypothermia protocols. In addition, the different circumstances of resuscitation efforts in hospital or out-of-hospital settings (e.g., time of initiation of chest compressions, shocks or medications after the arrest) also may influence the possible effects of cooling on various neuronal rescue mechanisms (such as reduction of free-radical formation or halting apoptosis).

In summary, this study addresses a very important question and raises attention to the gap in our knowledge of appropriate utilization of therapeutic hypothermia after cardiac arrest, as well as in our lack of precise understanding of the neuroprotective effects of cooling. Although the results of this study raise some concerns, given the inherent limitations of a non-randomized study design, they should not alter current clinical practice and recommendations to use therapeutic hypothermia after in-hospital cardiac arrest. Nevertheless, these findings strongly emphasize the need for a randomized, prospective clinical trial to establish efficacy of therapeutic hypothermia in patients with in-hospital cardiac arrest.