Stop Randomizing All Cardiac Arrests.

Myron L. Weisfeldt, MD The field of resuscitation science is characterized nearly uniformly by failed clinical trials: be it sodium bicarbonate, epinephrine at low or high dose, vasopressin, continuous or interrupted chest compressions, temperature management, antiarrhythmic drug use, or devices to augment perfusion, none has been shown convincingly to be of value. Nearly all of these large expensive trials included victims of cardiac arrest who met entrance criteria and received efforts to resuscitate to the point in time of the intervention. In many of these trials, a subgroup appears to have greater survival in 1 of the arms of the study. Those subgroups are often patients with likely better survival because the arrest is witnessed and a higher frequency of initial shockable rhythm is present. We are then prone to believe the correctness of the benefit in the subgroup and approach guidelines for treatment with these subgroups in mind. An extensive review and cautionary note on the risks of subgroup analysis when the overall result is null, and the risk of believing borderline statistically significant overall results, has been published recently by Pocock and Stone.1 Even recognizing these concerns, I predict, recent clinical trial subgroup differences will haunt cardiopulmonary resuscitation guideline committees going forward. One example is the recently published study2 of amiodarone, lidocaine, and placebo in refractory ventricular tachycardia or ventricular fibrillation. The overall result showed no statistically significant difference between the individual drugs and placebo, but for the more optimistic subgroup of witnessed cardiac arrest, both drugs were significantly better than placebo. Will the guideline be Class IIb (may or might be reasonable), Class IIa (can be useful or is reasonable), or Class I (is recommended)? A similar plea to mine, focused on not randomizing patients with an initial rhythm of asystole, was published by Kreutziger and Wenzel3 in 2009. As they point out, rarely is an asystolic arrest going to respond to treatment. Sinus arrest and heart block are rare and respond frequently to chest compression. Since the time of that publication, the issue of nonshockable rhythms has become even more important. The frequency of initial pulseless electric activity (PEA)4 arrests has increased remarkably. Recent epidemiological studies have identified initial shockable arrests (pulseless ventricular tachycardia or ventricular fibrillation) as occurring in nearly 25% of all arrest victims.4 Appropriately, clinical trials in resuscitation have focused on factors that appear to show benefit in experimental large animal models. The vast majority of these models are acute shockable arrests with <10 minutes of untreated ventricular fibrillation. Even in excellent emergency medical service systems, PEA arrest survival is far below that with initial shockable rhythm. Rarely has PEA arrest been modeled in large animals. We just do not know what to try. Major ambiguity surrounds the etiology of PEA arrest. Pathology studies have been limited, but some suggest that pulmonary embolism4 is frequent. If correct, management would likely be quite specific to that etiology. Stop Randomizing All Cardiac Arrests