BACKGROUND Although injection treatments for ulcer haemostasis seem to be effective, recurrent bleeding remains a serious problem. Large randomised clinical trials are required to show differences between treatment modalities for gastrointestinal bleeding. The aim of this study was to compare the safety and efficacy of repeated endoscopic injection of fibrin glue (FG) with that of single endoscopic injection of polidocanol
in the prevention of recurrent bleeding. METHODS 854 patients with active gastroduodenal bleeding (spurting, oozing), or ulcers with a visible non-bleeding vessel, were randomly assigned one of three endoscopic treatments: single application of polidocanol 1%, single application of FG, or daily repeated application of FG until the visible vessel had disappeared. All patients were pretreated with local injection of epinephrine (1/10,000), and had daily repeat endoscopies until the vessel observed at initial endoscopy was no longer visible. FINDINGS Recurrent bleeding rates among the 790 patients in whom the rates could be assessed were 58 (22.8%) of 254 in the polidocranol group, 51 (19.2%) of 266 in the FG-single group, and 41 (15.2%) of 270 in the FG-repeated group. The difference between FG-repeated treatment and polidocanol was significant (p = 0.036). Treatment failed, making other treatments (including surgery) necessary, in 34 (13.0%) of 261 in the polidocanol group, 34 (12.4%) of 274 in the FG-single group, and 21 (7.7%) of 274 in the FG-repeated group. The difference between FG-repeated treatment and polidocanol was significant (p = 0.046). The 30-day-mortality rates were low in all three treatment groups (polidocanol 4.7%; FG-single treatment 5.3%, FG-repeated treatment 4.3%). The safety profiles of the three treatment strategies were similar. INTERPRETATION Repeated injection with FG glue is significantly more effective than injection with polidocanol 1% in the treatment of bleeding from gastroduodenal ulcers.