Risk Prediction at the Emergency Department

Abstract: The severity of illness was scored in a cohort of 11751 non-surgical patients presenting at the Emergency Department (ED) during 12 consecutive months and followed for 4.7 years. The scoring system Rapid Acute Physiology score (RAPS) (including blood pressure, respiratory rate, pulse rate and Glasgow coma scale) was calculated for all arrivals at the ED. The RAPS system was also additionally developed by including the peripheral oxygen saturation and patient age, resulting in the new Rapid Emergency Medicine Score, (REMS). REMS was superior to RAPS in predicting in-hospital mortality according to ROC-curve analysis. An increase of one point in the 26 point REMS scale was associated with an Odds ratio of 1.40 for in-hospital death (95% CI 1.36-1.45, p<0.0001). Similar results were obtained in the major patient groups (chest pain, stroke, coma, dyspnea and diabetes). The association between REMS and length of stay in hospital was modest. Charlson Co-morbidity Index could add prognostic information to REMS in a long-term (4.7 years), but not in a short-term perspective (3 and 7 days). REMS was shown to be as powerful a predictor of in-hospital mortality as the more complicated APACHE II. REMS at the ED could also predict long-term mortality (4.7 years) in the total cohort (Hazard ratio 1.26, p<0.0001).REMS is a potentially useful prognostic tool for non-surgical patients at the ED, regarding both in-hospital and long-term mortality. It is less complicated to use than APACHE II and has equal predictive accuracy.