Between January 2005 and December 2009, consecutive patients presenting with acute UGIB were considered for this study. These patients were referred to the Emergency Unit of the Department of Internal Medicine at the “Sestre milosrdnice” University Hospital, Zagreb, Croatia and then if necessary hospitalized at the Interventional Gastroenterology Unit at the same hospital.
UGIB was suspected if hematemesis, melena or hematochezia were seen and/or bloody nasogastric aspirate was observed. In all those patients upper gastrointestinal endoscopy was performed within 6 h of hospital admission. Patients were included only if emergency endoscopy disclosed a gastric or duodenal bleeding ulcer with major stigmata of hemorrhage (“coffee ground” material or blood in the stomach and/or duodenum) and nonbleeding visible vessel in an ulcer bed (Forrest IIa)[8
]. Exclusion criteria were as follows: major comorbid or terminal illness that made endoscopy hazardous; inability or unwillingness to consent to endoscopy and endoscopic therapy; gastric malignancy; minor stigmata of hemorrhage at endoscopy such as oozing from ulcer borders without a visible vessel, flat-pigmented spots, or clean ulcer base. Patients with gastric and duodenal ulcer with either an actively bleeding vessel (spurting or oozing; Forrest I), or adherent clot (Forrest IIb) were also excluded.
Endoscopy was performed with standard upper endoscopes (GIF Q140 and GIF Q160, Olympus Optical Co., Japan). Endotherapy was carried out by the well-trained endoscopists, each with at least five years experience in the treatment of patients with GI bleeding. Endoscopic characteristics, including ulcer localization, ulcer size, and type of stigmata, were recorded (Endobase, Olympus, Japan).
Possible complications of endoscopic treatment and complete study protocol were discussed with patients and their relatives, and written informed consent was obtained before endoscopy and entry into the trial. The ethics committee of our hospital approved the treatment protocol. Randomization of eligible patients was carried out at the time of endoscopy by an individual uninvolved with the procedure who opened sealed numbered envelopes containing treatment assignments generated with a computer randomization program. The treatment group allocation was then communicated to the endoscopist in the endoscopy suite. Patients were randomized to a small-volume epinephrine group (15 to 25 mL injection group; Group 1), a large-volume epinephrine group (30 to 40 mL injection group; Group 2) and a hemoclip group (Group 3). In the small-volume epinephrine group (Group 1) 15 to 25 mL of a 1:10 000 solution of epinephrine was injected around the visible vessel (2-4 mL/injection at 2-3 mm from the visible vessel). In the large-volume epinephrine group (Group 2), 30-40 mL of a 1:10 000 solution of epinephrine was injected around the visible vessel at the ulcer bed as in the small-volume epinephrine group. Mechanical hemostasis was performed with stainless steel hemoclips (Olympus, Japan) as has been previously described[9,10
]. During endoscopy and endotherapy, electrocardiographic monitoring was used to detect arrhythmias.
Once hemostasis was achieved the bleeding site was observed for at least 10 min and it was assessed by water irrigation at maximal pressure. Failure of the initial hemostasis has been defined if any hemorrhage occurred immediately (within 10 min) after initial endoscopic hemostasis. In these patients crossing over to the other treatment group was not allowed. In all patients two biopsy specimens were taken from the gastric antrum and body, and the presence of Helicobacter pylori (H. pylori) infection was assessed by histopathological examination of the specimens. In patients with gastric ulcer in whom recurrent bleeding was not observed, control endoscopy was performed 4 d to 5 d after initial hemostasis and biopsy specimens were obtained from the margins and base of gastric ulcers to exclude malignancy.
After initial endoscopic hemostasis, patients were hospitalized and cared for by a physician who was blinded to the endoscopic treatment that had been delivered. Vital signs were monitored hourly whereas blood counts were observed every 6 h for the first 48 h and every 12 h to 24 h thereafter. All patients were given acid suppressive therapy: pantoprazole 80 mg iv, (bolus) and then 40 mg iv, every 8 h for at least 48 h, followed by 40 mg daily by mouth, or esomeprazole 80mg iv, (bolus) and then 40 mg iv, every 8 h for at least 48 h, followed by 20 mg once a day by mouth. Shock was defined as a systolic blood pressure of less than 90mmHg with symptoms or signs of organ hypoperfusion.
Recurrent bleeding was defined as one or more signs of ongoing bleeding, including fresh hematemesis or melena, hematochezia, aspiration of fresh blood via nasogastric tube, instability of vital signs, and a reduction of Hb by more than 2 g/dL over a 24 h period (early recurrence) or over a 7 d period (late recurrence) after initial stabilization of puls, blood pressure and Hb concentration. If recurrent bleeding was suspected, endoscopy was performed immediately. If “coffee ground” material or blood in the stomach and/or duodenum has been found together with active bleeding or a fresh blood clot in the ulcer base were found, recurrent bleeding was considered confirmed. For ethical reasons, additional endoscopic methods for treatment of recurrent bleeding were discussed with patients and their relatives and therapeutic option in all patients with recurrent bleeding was hemoclip application. Patients in whom endoscopic treatment or retreatment was unsuccessful underwent emergency surgery.
The rate of recurrent bleeding, as the primary outcome, was compared between the groups of patients included in the study. Secondary outcomes compared between the groups were primary hemostasis rate (defined as the absence of hemorrhage occurred immediately after initial endoscopic hemostasis), permanent hemostasis (defined as the absence of recurrent bleeding within the 30 d period after initial or secondary endoscopic hemostasis), need for emergency surgery, 30 d mortality and bleeding-related deaths, length of hospital stay, and transfusion requirements.
Base on assumption that injection of a large-volume epinephrine decreased the expected rate of recurrent bleeding from 17.1% after injection of small-volume epinephrine solution to zero, 39 patients would have been needed in each group for a power of 80% and a significance level of 0.05[6
Continuous data were summarized as mean [95% confidence interval (CI)]. The Student t test was used to compare the mean values of continuous variables. The Pearson chi-square test and the Fisher exact test were used when appropriate for the comparison of categorical variables. All analyses were performed with a statistical package (SPSS for Windows, United States). A P values less than 0.05 were regarded as statistically significant.