Factors associated with failure of endoscopic injection haemostasis in bleeding peptic ulcers

Background: The effectiveness of a submucosal injection of adrenaline solution in endoscopic haemostasis is well documented in patients suffering from peptic ulcer bleeding. After treatment, however, a significant number of patients continue to bleed or rebleed, and require emergency surgical intervention. The aim of this study was to detine factors associated with the failure of endoscopic injection haemostatic therapy in peptic ulcer bleeding. Methods: In the period 1992 to 1998, we prospectively studied all patients suffering from peptic ulcer bleeding and identified endoscopically as being either bleeding actively or carrying a visible vessel. A total of 427 patients (343 men and 84 women: mean age 58.6 +/- 16.6 years) were all subjected to endoscopic injection with adrenaline solution on an emergency basis. Patients who eventually required surgical intervention for permanent haemostasis were considered as endoscopic haemostasis failures, whereas those who did not were considered as endoscopic treatment successes. We evaluated all clinical and endoscopic parameters that might have been related to failure of endoscopic injection therapy. Results: Endoscopic injection haemostasis was successful in 341 patients (79.9%) and a failure in 86 (20.1%) who finally underwent emergency surgical haemostasis. On analysing the examined parameters, failure was significantly related to shock on admission (OR 2.31, 95% CI 1.33, 6.97), spurt bleeding at endoscopy (OR 2.45,95% CI 1.51, 3.98), posteriorly located duodenal ulcer (OR 2.48, 95% CI 1.37, 7.01) and anastomotic ulcer (OR 3.39, 95% CI 1.37, 7.29). Endoscopic injection haemostasis therapy was less effective in patients with chronic ulcers compared to those who had acute NSAID-related ulcers. A history of peptic ulcer (OR 1.57, 95% CI 1.14, 3.05), previous peptic ulcer bleeding (OR 2.45, 95% CI 1.51, 3.98) or non-use of NSAIDs (OR 2.81, 95% CI 1.33. 4.62) were negative predictors for the outcome of endoscopic haemostasis. Conclusion: With the use of specific clinical and endoscopic characteristics it is possible to define a subgroup of high-risk patients fur continued bleeding or rebleeding despite endoscopic injection therapy. These patients may be candidates for intensive monitoring, early surgical intervention or possibly complementary endoscopic haemostatic methods.