Which scoring system should be used for non-variceal upper gastrointestinal bleeding? Old or new?
Hong Jae Jeon,Hee Seok Moon,In Sun Kwon,Sun Hyung Kang,Jae Kyu Sung,Hyun Yong Jeong
DOI: https://doi.org/10.1111/jgh.15555
Abstract:Background and aim: Several scoring systems for predicting outcomes in patients with non-variceal upper gastrointestinal bleeding (NVUGIB) have recently been devised, but not sufficiently validated. We compared the predictive accuracy of several scoring systems and assessed the usefulness of new scoring systems. Methods: The medical records of 1048 patients with NVUGIB were reviewed to collect demographic, clinical, laboratory, and endoscopic data. The areas under the receiver operating characteristic curve (AUROCs) were calculated for the ABC, new Japanese scoring system, Progetto Nazionale Emorrhagia Digestiva (PNED), and other scores to compare their predictive accuracy for 30-day mortality, therapeutic intervention, rebleeding, and prolonged hospital stay (≥ 10 days). Outcome predictors were identified by multivariate analysis. Results: The ABC, new Japanese scoring system, and PNED scores best predicted 30-day mortality (AUROC 0.907), need for therapeutic intervention (AUROC 0.707), and rebleeding (AUROC 0.874), respectively (all P < 0.001). The ABC and PNED scores were similarly better at predicting prolonged hospital stay (ABC AUROC: 0.765; PNED AUROC: 0.790; both P < 0.001). Thirty-day mortality was related to sex, systolic blood pressure (SBP), syncope, estimated glomerular filtration rate (eGFR), albumin, heart failure, disseminated malignancy, chronic obstructive pulmonary disease (COPD), and liver cirrhosis. Sex, age, SBP, hematemesis, blood urea nitrogen, and eGFR independently predicted the need for therapeutic intervention. Sex, SBP, pulse, albumin level, heart failure, disseminated malignancy, and COPD predicted rebleeding. Conclusion: The outcomes of patients with NVUGIB were better predicted by newly developed than by old scoring systems.