Possibility of Real-Time Risk Assessment of Acute Upper Gastrointestinal Bleeding by Using TRI (TIMI Risk Index) Model

Purpose: The purpose of this study is to evaluate the predictive value of TRI(Thrombolysis in myocardial infarction risk index) in predicting UGIB(upper gastrointestinal bleeding), by comparing it to Rockall scores. Methods: We retrospectively studied 583 patients who admitted with the diagnosis of UGIB. Based on the recorded data, we compared AUCs(area under curves) of TRI and Rockall scores for rebleeding and death. Rebleeding rates and mortalities of all risk groups of TRI were obtained to check up the difference significances between neighboring groups. Results: AUC of TRI for rebleeding was 0.599, which was lower than AUCs of clinical and completed Rockall scores-0.655, 0.696. AUC of TRI for death was 0.741, which was also lower than AUCs of clinical and completed Rockall scores-0.838, 0.846. Rebleeding rates did not increase correlatively with the step-up of the risk grades of TRI. Mortality rates were 2.9%, 6.3%, 17.2%, 46.7%, 57.1%, respectively in 5 risk grades of TRI with increasing tendency by grade step-up, however, there were significant differences between rebleeding rates of low(10-60-80) risk groups(P=0.00017, 0.01763 respectively), but not between lowest(80) risk groups(P=0.41, not applicable respectively). Conclusion: Predictive value of TRI to forecast the outcomes of acute upper gastrointestinal bleeding is lower than Rockall scores. However, TRI is useful to predict UGIB mortality quickly, since this score shows significant risk stratification for mortality rate when divided into 3 risk groups-low(60).