S0615 Performance of Lower GI Bleeding Scores in a Large Tertiary Care Hospital

INTRODUCTION: Recent studies have proposed using endoscopic risk stratification tools such as Oakland, Strate, and NOBLADS for patients presenting in with lower GI bleeding. There is a wide range of conflicting data in the literature regarding the efficacy of these tools in predicting the severity of lower GI bleed. The primary objective of this study was to assess the performance of these bleeding scores in a large tertiary care hospital. METHODS: This is an interim analysis of retrospective data collected from a large tertiary care hospital of colonoscopies performed for GI bleeding from Jan, 2019 to Dec, 2019. The baseline characteristics of the study population are shown in Table. 1. The following risk stratification scores were calculated: Oakland, Strate, and NOBLADS. We compared the performance of these scores in predicting need for intervention (hemoclip, thermal therapy, banding, hemospray), 30-day rebleeding, and 30-day mortality using logistic regressions and receiver-operator curves (ROC). RESULTS: ROC curves are shown in Figure 1. None of the scores performed well in predicting the need for endoscopic intervention or 30-day mortality. Only the Oakland score performed moderately well in predicting 30-day rebleeding risk. Area under the curve shown in Table.2. CONCLUSION: The currently available bleeding scores for LGIB perform poorly in predicting need for intervention or mortality. Only the Oakland score shows association with 30-day rebleeding risk. There is a need for developing better risk stratification tools for LGIB.Table 1.: Baseline CharacteristicsTable 2.: Performance of Scoring ToolsFigure 1.: ROC curves for endoscopic intervention, 30-days rebleeding and 30-days mortality (L > R).