Comparison of Two Scoring Systems in Predicting Outcomes in Non-Variceal Upper Gastrointestinal Bleeding in Taiwanese Population.

Background: The role of scoring systems in detecting outcomes of non-variceal upper gastrointestinal bleeding in Taiwanese population remains uncertain.

Aims: The aim of our study was to compare Glasgow-Blatchford score with pre-endoscopic Rockall score in their utilities in predicting clinical outcomes in Taiwanese population.

Methods: We designed a prospective study to compare the performance of the Glasgow-Blatchford score and pre-endoscopic Rockall score in predicting endoscopic therapy, rebleeding and 30-day mortality in non-variceal upper gastrointestinal bleeding patients. The area under receiver operating characteristic curve was analyzed. 234 consecutive patients admitted during a 8-month period were enrolled.

Results: For prediction of therapeutic endoscopy, area under receiver operating characteristic curve was obtained for Glasgow-Blatchford score (0.629), and pre-endoscopic Rockall score (0.599). For prediction of rebleeding, area under receiver operating characteristic curve was obtained for Glasgow-Blatchford score (0.687), and pre-endoscopic Rockall score (0.581). For prediction of mortality, area under receiver operating characteristic curve was obtained for Glasgow-Blatchford score (0.505), and pre-endoscopic Rockall score (0.734).

Conclusions: In detecting low risk patients requiring endoscopy therapy, the AUC for GBS shows that it is a poor stratification tool, and the AUC for PRS reveals that it is a worthless stratification test. In detecting rebleeding, Glasgow-Blatchford score has a better performance than pre-endoscopic Rockall score. In contrast, pre-endoscopic Rockall score has a better performance in predicting 30-day mortality than Glasgow-Blatchford score.