Introduction The Glasgow Blatchford Score (GBS) is a pre-endoscopic risk assessment tool for patients presenting with upper gastrointestinal haemorrhage (UGIH). It can predict need for intervention or death and identifies low risk patients suitable for out-patient management.1 There are no published data assessing its use in variceal haemorrhage. Our aim was to compare the GBS with both admission and full Rockall scores in assessment of patients with variceal bleeding.

Methods Data on consecutive patients presenting to four UK hospitals (Glasgow Royal Infirmary, Royal Cornwall Hospital Truro, University Hospital of North Tees and Ninewells Hospital Dundee) were collected. Admission history, clinical and laboratory data, endoscopic findings, intervention and follow-up were recorded. We compared the ability of GBS and both Rockall scores to predict intervention and death in those patients with a final diagnosis of variceal bleeding.

Results 1556 patients presented with UGIH to the four hospitals during the study period. 78 had a final diagnosis of variceal bleeding. The mortality of these patients was higher than the non-variceal patients (18% vs 4%; p < 0.0005). On presentation, no variceal bleeding patient had a GBS <3; however, six had an admission Rockall score of zero. The median(range) GBS, admission Rockall and full Rockall scores for the variceal bleeding group were 10(2–18), 3(0–7) and 5(1–10), respectively. The comparable figures for all other patients were 3(0–19), 1(0–7) and 3(0–9), respectively (all p < 0.00005 vs varices). When comparing variceal bleeding patients with those who required intervention or died from another bleeding source, there was no difference using any of the three scores. In predicting need for intervention in the variceal bleeding group, AUC (95% CI) for GBS, admission Rockall and full Rockall scores were: 0.72 (0.56–0.89), 0.46 (0.30–0.62) and 0.66 (0.51–0.83), respectively. For predicting death, the figures were: 0.58 (0.41–0.75), 0.68 (0.54–0.82) and 0.72 (0.58–0.86), respectively.

Conclusion At presentation, the GBS correctly identifies patients with variceal bleeding as being at high risk for requiring intervention and appears superior to the admission Rockall score for this. However, it is a poor predictor of mortality in this patient group.