Predictors of favourable outcome in non-variceal upper gastrointestinal bleeding: Implications for early discharge?

Abstract

There is a lack of validated predictors on which to decide the timing of discharge in patients already hospitalized for upper nonvariceal bleeding. Identify factors that appear to protect nonvariceal bleeders from the development of negative outcome (rebleeding, surgery, death). Secondary analysis of two prospective multicenter studies. Multivariate analyses for each investigated outcome were performed; a single model was developed including all factors that were statistically significant in each sub-model. A final score was developed to predict favourable outcomes. Prognostic accuracy was tested with ROC curve analysis. Out of 2398 patients, 211 (8.8%) developed one or more adverse outcomes: 87 (3.63%) had rebleeding, 46 (1.92%) needed surgery and 107 (4.46%) died. Predictors of favourable prognosis were: ASA score 1 or 2, absence of neoplasia, outpatient bleeding, use of low-dose aspirin, no need for transfusions, clean-based ulcer, age <70 years, no haemodynamic instability successful endoscopic diagnosis/therapy, no Dieulafoy's lesion at endoscopy, no hematemesis on presentation and no need for endoscopic treatment. Overall prognostic accuracy of the model was 83%. The final score accurately identified 20-30% of patients that eventually do not develop any negative outcome. The "good luck score" may be a useful tool in deciding when to discharge a patient already hospitalized for acute non-variceal bleeding.