Evaluation of APACHE II, RAPS, and REMS models at the emergency department to predict hospital death in triage levels 1 and 2
Hamed Shokoohsaremi, Mohsen Ebrahimi, Shirinzadeh Feizabadi
Abstract
Background
One of the essential components for many therapeutic decisions is the initial evaluation of the patients.
Aims
The initial evaluation is usually done by the physician, which is mainly performed subjectively and is estimated more accurately if the physician is experienced. Many systems have been developed to evaluate and make these evaluations objective. The present study has used three different scoring systems.
Methods
The present study was prospective cohort on the patients admitted with triage 1 and Two levels. The study was conducted over one year. APACHE II, RAPS, REMS models were calculated for each patient. Then these models were evaluated from three main aspects of overall performance, resolution and calibration.
Results
The present study evaluated 1029 patients admitted to Edalatian Emergency Unit in Mashhad. From among them, 198 patients were at triage level 1 consistent with ESI algorithm and the rest were assigned to level two. Overall, about 29% of these patients died and 753 survived. The largest area below the curve was dedicated to APACHE II model, which shows the high discrimination of the model. AUC=0.76 (CI:0.72, 0.78). in the next ranks were REMS model with the area of 0.67 and the RAPS model with the area of 0.63. Hosmer-Lemeshow test for all models was
Full Text:
One of the essential components for many therapeutic decisions is the initial evaluation of the patients.
Aims
The initial evaluation is usually done by the physician, which is mainly performed subjectively and is estimated more accurately if the physician is experienced. Many systems have been developed to evaluate and make these evaluations objective. The present study has used three different scoring systems.
Methods
The present study was prospective cohort on the patients admitted with triage 1 and Two levels. The study was conducted over one year. APACHE II, RAPS, REMS models were calculated for each patient. Then these models were evaluated from three main aspects of overall performance, resolution and calibration.
Results
The present study evaluated 1029 patients admitted to Edalatian Emergency Unit in Mashhad. From among them, 198 patients were at triage level 1 consistent with ESI algorithm and the rest were assigned to level two. Overall, about 29% of these patients died and 753 survived. The largest area below the curve was dedicated to APACHE II model, which shows the high discrimination of the model. AUC=0.76 (CI:0.72, 0.78). in the next ranks were REMS model with the area of 0.67 and the RAPS model with the area of 0.63. Hosmer-Lemeshow test for all models was
PDF