Background: Patients visit emergency department (ED) with varying diagnosis and severity spectrums. Simple Model Score (SMS) utilizes age and laboratory data to stratify the risk of ED patients and predict their mortality. SMS proves to perform well but its application on population with different characteristics has to be validated. Objective: To evaluate the calibration and discrimination of SMS in predicting seven day in-hospital mortality of non-surgical ED patients at Cipto Mangunkusumo Hospital (CMH), Jakarta.
Methods: A validation study with retrospective cohort designed was performed on non-surgical patients in ED of CMH during October-November 2012. The data on age, hemoglobin, platelet count, white blood count, ureum, sodium, and blood glucose level at admission to ED were used to calculate SMS. The primary outcome was seven day in-hospital mortality. Calibration was evaluated using calibration plot and Hosmer-Lemeshow test while discrimination was evaluated with area under the curve (AUC).
Results: A total of 701 patients were recruited. Mortality was observed in 92 patients (13.12%). Calibration plot of SMS showed r=0.64 and Hosmer-Lemeshow test showed p=0.75. Discrimination was shown by ROC curve with AUC
0.67(CI 95% 0.610-0.719).
Conclusion: SMS showed good calibration but less satisfying discrimination in predicting seven day in-hospital mortality of non-surgical ED patients at CMH.
Key words: Emergency, mortality, non-surgical, Simple Model Score, validation.