Objective: To assess the ability of the Rockall score to predict outcome in patients who undergo endoscopic therapy for peptic ulcer haemorrhage.
Design: Retrospective data analysis.
Setting: Patients admitted to the three major acute hospitals in Lothian, UK.
Participants: Details of 211 patients involved in two randomized trials of endoscopic therapy between 1995 and 1999 were accessed, and Rockall scores calculated. All patients had ulcers with active bleeding or non-bleeding visible vessels requiring endoscopic therapy. The patients were followed up for 6 months and end points included rebleeding and death.
Main outcome measures: A comparison of mean Rockall scores for those patients who did not rebleed, those who re-bled and those who died. Identification of those patients at greatest risk of rebleeding or death after endoscopic therapy.
Results: One hundred and seventy-six patients did not rebleed, with mean score 6.17 (SD = 1.61). Rebleeding occurred in 35 patients whose mean score was 6.97 (SD = 1.52). There were 29 deaths with mean score 7.34 (SD = 1.40). The differences between the three groups were significant by one-way ANOVA (P < 0.001). Fifty-six patients had a Rockall score of 8 or over and, of these, 16 (29%) re-bled and 14 (25%) died. Of the 155 patients with scores of 7 or less, 19 (12%) re-bled and 15 (10%) died. The difference between these groups was significant with chi2 = 7.912 (P = 0.005) for rebleeding and chi2 = 8.147 (P = 0.004) for death.
Conclusions: The Rockall score can be used to predict poor outcome in patients who undergo therapeutic endoscopy for major peptic ulcer bleeding.