Predicting neurologically intact survival after in-hospital cardiac arrest-external validation of the Good Outcome Following Attempted Resuscitation score

Background: A do-not-attempt-resuscitation order is issued when it is against the wishes of the patient that cardiopulmonary resuscitation is performed, or when the chance of good quality survival is minimal. Therefore it is essential for physicians to make an objective prearrest prediction of the outcome after an in-hospital cardiac arrest (IHCA). Our aim was external validation of the Good Outcome Following Attempted Resuscitation (GO-FAR) score in a population based setting. Methods: The study was based on a retrospective cohort of adult IHCAs in Stockholm County 2013–2014 identified through the Swedish Cardiopulmonary Resuscitation Registry. This registry provided patient and event characteristics and neurological outcome at discharge. Neurologically intact survival is defined as Cerebral Performance Category score (CPC) 1 at discharge. Data for the GO-FAR variables was obtained from manual review of electronic patient records. Model performance was evaluated by measure of discrimination with the area under the receiver operating curve (AUROC) and calibration with assessment of the calibration plot. Results: The cohort included 717 patients with neurologically intact survival at discharge of 22%. In complete case analysis (523 cases) AUROC was 0.82 (95% CI 0.78–0.86) indicating good discrimination. The calibration plot showed that the GO-FAR score systematically underestimates the probability of neurologically intact survival. Conclusion: The GO-FAR score has satisfactory discrimination, but assessment of the calibration shows that neurologically intact survival is systematically underestimated. Therefore, only with caution should it without model update be taken into clinical practice in settings similar to ours.