Therapeutic decision-making in endoscopically unmanageable nonvariceal upper gastrointestinal hemorrhage.

The purpose of this study was to identify endoscopic and clinical parameters influencing the decision-making in salvage of endoscopically unmanageable, nonvariceal upper gastrointestinal hemorrhage (UGIH) and to report the outcome of selected therapy. We retrospectively retrieved all cases of surgery and arteriography for arrest of endoscopically unmanageable UGIH. Only patients with overt bleeding on endoscopy within the previous 24 h were included. Patients with preceding nonendoscopic hemostatic interventions, portal hypertension, malignancy, and transpapillar bleeding were excluded. Potential clinical and endoscopic predictors of allocation to either surgery or arteriography were tested using statistical models. Outcome and survival were regressed on the choice of rescue and clinical variables. Forty-six arteriographed and 51 operated patients met the inclusion criteria. Univariate analysis revealed a higher number of patients with a coagulation disorder in the catheterization group (41.4%, versus 20.4% in the laparotomy group; p = 0.044). With multivariate analysis, the identification of a bleeding peptic ulcer at endoscopy significantly steered decision-making toward surgical rescue (OR = 5.2; p = 0.021). Taking into account reinterventions, hemostasis was achieved in nearly 90% of cases in both groups. Overall therapy failure (no survivors), rebleeding within 3 days (OR = 3.7; p = 0.042), and corticosteroid use (OR = 5.2; p = 0.017) had a significant negative impact on survival. The odds of dying were not different for embolotherapy or surgery. In conclusion, decision-making was endoscopy-based, with bleeding peptic ulcer significantly directing the choice of rescue toward surgery. Unsuccessful hemostasis and corticosteroid use, but not the choice of rescue, negatively affected outcome.