Appendix A. Editorial comment

Abstract

Dr Kshivets: Sometimes the traditional method, biometrics matches the system, although you didn't find any predictors, but using neural networks computing, and at last simply knowledge engineering and expert system technology you may find these predictors.Dr Athanasiou: Yes. Dr Kshivets: And these predictors you can verify on learning samples. Dr Athanasiou: I have an answer. You asked me if we can use neural networks in our series. It has been published already by Tim Coats, who is Professor of Emergency Medicine in the Royal London and it was found that didn't have any better ROC curve performance in comparison to multivariate models. The multivariate models have a big disadvantage in that they can identify only patients at very high risk, with high incidence. So basically there is not an efficient way to identify the patients who will survive after emergency thoracotomy. We are concentrating on other types of models to analyze these data.Dr F. Vyhnanek (Prague, Czech Republic): We know it is a problem, emergency thoracotomy in blunt injury. Do you have some experience with emergency thoracotomy in serious blunt injury of the abdomen?Dr Athanasiou: We know that blunt trauma has a very bad outcome when we perform emergency thoracotomy, and what we have found as well is that between the years 1991 to 1993 we performed many thoracotomies and the majority of them were from blunt injury, and the results showed that we didn't have any survivors. So we introduced a new algorithm for penetrating injury which has been published in the Journal of Trauma.Emergency thoracotomy (ET) as a life-saving measure either in a prehospital or hospital setting remains a controversial procedure. Its role in traumatic cardiac arrest is defined as resuscitative allowing the evacuation of pericardial tamponade, direct control of intrathoracic hemorrhage, open cardiac massage and cross-clamping of the descending aorta.In this issue of the European Journal of Cardiothoracic Surgery the authors Athanasiou et al. have reviewed their experience with ET in a prehospital or hospital setting (53 patients in 9 years) with a high survival rate of 18.8% and suggest that ET might be a significant procedure in a well-selected group.Although cardiopulmonary arrest after trauma is associated with a poor prognosis, there are also other reports on the use of resuscitative thoracotomy in the emergency department that seem to hold some promise for injured patients with cardiopulmonary arrest [1 -6]. Generally, the ET literature consists of retrospective series suffering from methodological limitations. Analysis of this literature turns to be really complicated. Survival rates vary from 2 to 31% [7] but have not always been correlated with the mechanism and location of injury, the field and transport times, the duration of arrest and the physiologic status of the patient. Often survival data do not present neurologic outcomes in survivors.The ATLS guidelines regarding ET make recommendations solely on the basis of mechanism of trauma and do not take into ac...