The prognosis of patients having received optimal therapy for nonvariceal upper gastrointestinal bleeding might be worse in daily practice than in randomized clinical trials.

BACKGROUND

Combination of endoscopic haemostatic and high-dose intravenous proton-pump inhibitors is considered to be the standard care for patients with acute peptic ulcer bleeding.

AIM

This study assessed predictive factors of rebleeding and death in unselected patients presented to our hospital.

METHODS

Consecutive patients with nonmalignant bleeding ulcers and stigmata of recent haemorrhage who received optimal treatment, between 22 August 2003 and 15 October 2007, were studied retrospectively.

RESULTS

Among 140 included patients, 45 (32%) rebled and 30 received another haemostatic endoscopy, which was successful in 20 cases. In multivariate analysis, the only significant predictive factor of rebleeding was duodenal site of the ulcer [adjusted odds ratio (OR): 2.75; 95% confidence interval (CI): 1.28-6.19]. In-hospital death occurred in 27 (19%) patients; with five deaths related to uncontrolled or recurrent bleeding. In multivariate analysis, predictors of in-hospital mortality were rebleeding (adjusted OR: 3.28; 95% CI: 1.17-9.16), a Rockall score higher than 6 (adjusted OR: 9.12; 95% CI: 2.57-44.29) and bleeding occurring in the intensive care unit (adjusted OR: 15.68; 95% CI: 4.41-55.82).

CONCLUSION

In unselected patients, rebleeding and mortality rates are substantially higher than those found in prospective randomized clinical trials. Intensive care unit stay is an important predictive factor of hospital mortality and should be considered in further therapeutic trials in ulcer bleeding.