NOBLADS - External Validation of a Risk Scoring System for Severe Acute Lower Gastrointestinal Bleeding.

Abstract

BACKGROUND\nAcute lower gastrointestinal bleeding is a common cause of hospital admission. NOBLADS is a lower gastrointestinal bleeding clinical risk score.\n\n\nOBJECTIVE\nThis study aimed to externally validate NOBLADS in predicting severe acute lower gastrointestinal bleeding and clinical outcome.\n\n\nDESIGN\nObservational retrospective study.\n\n\nSETTING\nperformed in a single large tertiary hospital.\n\n\nPATIENTS\nincluded patients admitted with acute lower gastrointestinal bleeding during a 15-month period that underwent endoscopic evaluation. Excluded chronic lower gastrointestinal bleeding or acute lower gastrointestinal bleeding in patients admitted for other conditions.\n\n\nMAIN OUTCOME MEASURES\nPrimary outcome was acute severe lower gastrointestinal bleeding (classified as severe if requires >2 red blood cells transfusion and/or produces a >20% hematocrit fall). Secondary outcomes: red blood cells requirement, therapeutic intervention and hospital stay duration. NOBLADS score was applied to all patients. Its accuracy to predict acute severe bleeding and secondary outcomes were study using ROC analysis.\n\n\nRESULTS\nA total of 173 patients (50.3% males, age 69 ± 17 years) were included. The most common lower gastrointestinal bleeding etiologies were diverticular bleeding (18.5%) and ischemic colitis (15.6%). 50 patients (28.9%) presented criteria for severe bleeding and 33 patients required instrumental intervention: endoscopic (n=28), surgical (n=4), radiologic (n=1) therapy. NOBLADS score was significantly different according with acute lower gastrointestinal bleeding causes (p < 0.001), being accurate to predict severe bleeding (AUROC 0.923 ± 0.018 (p < 0.001). NOBLADS ≥2 detected acute severe lower gastrointestinal bleeding with 100% sensitivity/62.4% specificity, identifying the need for higher red blood cells requirement (3.6 vs 0.08, p < 0.001), therapeutic intervention (38% vs 13%, p < 0.001) and longer hospital stay (12.8 vs 3 days, p < 0.001).\n\n\nLIMITATIONS\nUnicenter retrospective study; number of unprepared sigmoidoscopies/left side colonoscopies; comparison between NOBLADS and other scoring systems was not performed.\n\n\nCONCLUSION\nNOBLADS is a practical, highly accurate tool and predicts the need of in-hospital management, early colonoscopy, red blood cells transfusion and longer hospital stay in patients admitted with acute lower gastrointestinal bleeding. See Video Abstract at http://links.lww.com/DCR/B748.