Upper gastrointestinal bleeding: predictors of risk in a mixed patient group including variceal and nonvariceal haemorrhage

Abstract

Background and study aims: Effective management of upper gastrointestinal bleeding (UGIB) relies on the application of clinical risk scores. The validation of risk scores has to date focused mainly on nonvariceal UGIB groups. We aimed to evaluate our clinical and endoscopic management of UGIB, and to validate existing risk scores for a mixed patient population with a high percentage of variceal bleeds.

Study design and methods: Analysis included UGIB patients presenting consecutively to a tertiary referral university hospital. All patients had been admitted by our emergency department and had undergone upper gastrointestinal endoscopy. Clinical, biochemical and endoscopic data were recorded. Clinical and complete Rockall and Blatchford risk scores were calculated for all patients and statistical analysis was carried out by a multiple logistical regression model.

Results: A total of 21% of patients had variceal bleeds. There was considerable heterogeneity between groups with the variceal group having more comorbidities (P=0.003), lower haemoglobin (P=0.003) and lower systolic blood pressure (P=0.013) at presentation. This translated to higher risk scores (P<0.0001) and worse clinical outcomes (rebleeding P=0.004). Only complete Rockall score was predictive of outcome (rebleeding P=0.004, AUC 0.8). Blatchford score did not predict bleeding or mortality. However, no patient with a Blatchford score of 0 had an adverse clinical outcome.

Conclusion: Postendoscopic Rockall score can be used as a predictor of outcome for mixed UGIB groups. Although Blatchford score did not predict outcome in our study, at a 0 level it does appear to be a safe triage tool for pre-endoscopic identification of patients with variceal bleeds, even where there is no known history of liver disease.