Performance of physiology scoring systems in prediction of in-hospital mortality of traumatic children: A prospective observational study

Abstract

BackgroundThis study is designed to compare the value of four physiologic scoring systems of rapid acute physiology score (RAPS), rapid emergency medicine score (REMS), Worthing physiology scoring system (WPSS) and revised trauma score (RTS) in predicting the in-hospital mortality of traumatic children brought to the emergency department. MethodWe used the data gathered from six healthcare centers across Iran between the April-October 2016. Included patients were all children with trauma. Patients were assessed and followed until discharge. Moreover, patients were divided to two groups of died and alive, and discriminatory power and general calibration of models in prediction of in-hospital mortality were compared. ResultsData was gathered from 814 children (average age of 11.65±5.36years, 74.32% boys). Highest measured area under the curve was for RAPS and REMS with 0.986 and 0.986, respectively. Areas under the curve of WPSS and RTS were 0.920 and 0.949, respectively (p=0.02). Sensitivity and specificity of RAPS were 100.0 and 95.05, respectively. These amounts for REMS were 100.0 and 94.04, respectively. Two models of RTS and WPSS had the same sensitivity of 84.62. Specificity of these two was 98.22 and 96.95, respectively. Three models of RAPS, REMS and RTS had proper calibrations in predicting mortality; however, it seems that WPSS overestimates the mortality in high risk patients. ConclusionAs calculations of RAPS is easier than REMS and their proper calibrations, it seems that RAPS is the best physiologic model in predicting in-hospital mortality and classifying in traumatic children based on severity of injury. However, further validation of the recommended score is essential before implementing them into routine clinical practice.