In the article on cardiopulmonary resuscitation survival by Rosenberg et al,1 and commented on by McIntyre, a retrospective review of survival was done on hospitalized patients to determine characteristics that might be predictive of successful outcome of cardiopulmonary resuscitation. Survival in this study was almost 70% higher than in other recent studies. This suggests that there may have been a selection bias. This would occur if those less likely to be successfully resuscitated were eliminated from the pool of potential survivors by "do not resuscitate" orders.
If "do not resuscitate" orders were appropriately applied, those with the least chance of survival would not undergo CPR, survival rates of those resuscitated would improve, and age and disease would not be determining factors for short-term survival. If this is the case here, Rosenberg may have shown the ability of the physicians in these community hospitals to predict survival and use "do