The Riyadh Intensive Care Program mortality prediction algorithm assessed in 617 intensive care patients in Glasgow

The hospital mortality prediction algorithm from the Riyadh Intensive Care Program was assessed in 617 general intensive care unit patients and the results were compared with APACHE II admission scoring. Of the 119 patients predicted to die by the Riyadh Intensive Care Program, 24 (20.2%) recovered sufficiently to be discharged home. The overall false-positive rate was 5.2%, the false predictions mostly occurring in the trended component, the admission component having similar performance to admission APACHE II. For equivalent specificity the Riyadh Intensive Care Program was more sensitive than admission APACHE II risk of death figures, but the very high false-positive rate in those predicted to die precludes the algorithm's use in patient management decisions. In our view, the Riyadh Intensive Care Program does not justify the considerable extra work involved in data collection and processing over current admission scoring systems.