Hypothermia After Cardiac Arrest: A Neurological Perspective

Aabstracts & Commentary

Sources: N Engl J Med. 2002;346:549-556; Bernard SA, et al. N Engl J Med. 2002;46:557-563.

Out-of-hospital cardiac arrest is a major cause of morbidity and mortality due to the neurological sequelae of hypoxic-ischemic injury. Hypothermia may benefit this or other neurological injuries by decreasing cerebral oxygen consumption or perhaps, more importantly, by controlling downstream effects of hypoxia such as free radical production, excitatory amino acid cascades, and osmotic swelling. Hypothermia has been successfully piloted as a neuroprotective strategy in stroke but has been less successful as a potential treatment for head trauma. In the latter case, a large proportion of patients who presented to the emergency room were already hypothermic, making it more difficult to show an additional treatment effect.

Achieving therapeutic hypothermia is not simple. Hypothermia is induced to a mild (32-34ºC), moderate (30-32ºC), or more extreme (< 30ºC) temperature. This involves surface cooling blankets and may also demand more invasive measures such as urinary bladder or peritoneal irrigation. Hypothermia also requires intubation, sedation, and paralysis to assure patient comfort and blunt shivering responses. Rewarming must be done in a controlled fashion as even passive rewarming, if too rapid, may result in rebound increases in intracranial pressure with significant brain swelling. Hypothermia may result in important medical complications including infections related to impaired immunity, primarily pneumonia, and hematological problems, such as thrombocytopenia. For these reasons, successful use of therapeutic hypothermia requires a skilled ICU team able to optimally both use this modality and effectively manage complications.

Two recently published studies, one European and the other Australian, examined patients following cardiac arrest and suggest that mild hypothermia in this population might be both feasible and beneficial. In the European group, 75 of 136 patients (55%) in the hypothermia group had favorable outcomes compared with normothermia—54 of 137 patients (39%)—a statistically significant result. Favorable outcome was defined as a "good recovery" (Category 1) or "moderate disability" (Category 2). Unfavorable outcomes occurred in patients with "severe disability" (Category 3), a vegetative state (Category 4), or death (Category 5). Mortality at 6 months was also significantly reduced with hypothermia. Patients in the normothermia group were more likely to have diabetes, a previous history of coronary artery disease, or to have received bystander CPR. None of these changed the outcome in multivariate analyses. Hypothermia was initiated in an average of 105 minutes (range, 61-192 minutes) after resumption of spontaneous circulation. The goal temperature of 32-34ºC was achieved in an average of 8 hours (range, 4-16 hours) and maintained for 24 hours as per the study protocol. There was a nonsignificant trend toward increased bleeding and sepsis in the treatment group, but overall the hypothermia was safe.

In the Australian group (Bernard and colleagues), with a smaller number of patients, 21 of 43 patients in the hypothermia group had a favorable outcome compared with 9 of 34 treated with normothermia (P = 0.046). A favorable outcome was defined as discharge to home or rehabilitation. Unfavorable outcomes occurred in patients who died or in a tiny minority (n = 2 for the whole study) who were sent to a long-term nursing facility. Mortality was not significantly different between the treatment groups. Hypothermia was maintained for 12 hours. The rapidity of hypothermia induction was not reported.

The studies differed in a few important ways. The European group required that there be an estimated interval of 5-15 minutes between the patient’s collapse, with no more than 60 minutes until restoration of spontaneous circulation. This and other strict inclusion criteria resulted in a low rate of inclusion; only 275 patients were enrolled from a total of 3551 screened. The European study randomized patients 1:1 with an algorithm dictated by the coordinating center. In contrast, the Australian study split patients based on odd or even days of the month, which by chance generated 25% more patients in the active treatment group. In both studies, blinding was maintained for the physicians adjudicating outcome but not for treating physicians. As Bernard et al note, this is an unavoidable source of bias in a study such as this since cooling involves ice packs, cool air blankets, and other devices impossible to hide at the bedside. The development and approval of indwelling cooling catheters will undoubtedly facilitate therapy and might allow blinding if needed.


It is surprising that such a short duration of hypothermia (only 12-24 hours) produced such dramatic benefits. A large degree of neuronal damage due to hypoxia probably occurred in the minutes to hours before therapy was initiated and ongoing secondary mechanisms of injury likely persisted for 48-72 hours after therapy was completed. It is furthermore encouraging that even mild hypothermia, a more safe and feasible therapy than severe cooling, was so effective in these studies. A more aggressive magnitude and duration of hypothermia in the setting certainly merits further investigation.

The use of global outcome scores in these patients, which categorized outcomes in terms of level of disability as well as discharge disposition and mortality, provided these investigators the power to show an overall benefit for this therapy. With this approach, however, subtleties may be lost. A patient sent home with severe amnesia, totally dependent on family for their activities of daily living, would still be categorized as favorable based on their methodology. Because these studies do not provide any specifics regarding the neurological exam of these patients (not even a relatively blunt instrument such as the Glasgow Coma Scale), it is difficult to glean any further neurological insight into these patients’ condition.

Neurologists are often called to the bedside acutely or in the days immediately following cardiac arrest to provide prognosis. Deep sedation and paralysis for the purposes of hypothermia makes this exercise impossible. Neurologists should, nevertheless, advocate for this therapy. Given an otherwise untreatable and protean condition, cardiac arrest victims should be cooled as quickly and probably for as long as possible for optimal neuroprotection. At the very least, cardiac arrest victims who arrive already cold, should be maintained as such and any patient who develops even mild pyrexia should be aggressively defervesced at least normothermia with acetaminophen or practical cooling measures. —Alan Z. Segal, Assistant Professor, Department of Neurology, Weill-Cornell Medical College, Attending Neurologist, New York Presbyterian Hospital, Assistant Editor, Neurology Alert.