Early prediction of non-survival for patients suffering cardiac arrest--a word of caution. The Belgian Cerebral Resuscitation Study Group.

A total of 6178 persons with out-of-hospital (70%) and in hospital (30%) cardiac arrests from the first of January 1982 until the end of 1989 were reviewed retrospectively with respect to 4 variables, contributing to a score for specific prediction of poor prognosis (cut-off point: greater than 3 points). These included age, initial ECG, type of respiratory arrest and bystander resuscitation. Presence of ventricular fibrillation, gasping and bystander resuscitation contributes nothing to the score, while presence of asystole or EMD (electromechanical dissociation), apnoea and absence of bystander resuscitation adds one point to it. Of patients scoring 4 or 5 points 44 were awake 14 days post CPR (Class 3). The positive predictive value of the score was 97% (95% CI 96-98%) for the out-of-hospital group and 92.2% (95% CI 88-95%) for the in-hospital group. The specificity was respectively 92.3% (95% CI 89-95%) and 94.2% (95% CI 91-96%). Although the score can weigh the likelihood of no success against that of success, we cannot recommend it for decision making as far as abandoning or continuing cardiopulmonary resuscitation efforts.