Outcome of acute nonvariceal gastrointestinal haemorrhage after nontherapeutic arteriography compared with embolization

In acute nonvariceal gastrointestinal (GI) bleeding, immediate arteriographic haemostasis is presently assumed to be a therapeutic advantage. This study assesses whether the risk of a delayed haemostasis, caused by arteriographic findings precluding embolization, might influence patient outcome. We performed a 5.5-year retrospective database search to find all patients referred for arteriography to arrest acute nonvariceal GI bleeding with embolization. The embolized and nonembolized patients were compared for differences in baseline characteristics and bleeding parameters. In both groups the outcome of all endoscopic or surgical interventions after catheterization was included in the follow-up. Clinical success (at 30 days, after all therapy) and in-hospital mortality in the embolized and nonembolized group were compared. We retrieved 63 nonembolized bleedings in 58 patients and 49 embolized bleedings in 49 patients. In the nonembolized group, transfusion need and haemodynamic instability were significantly less severe. Forty-two of 63 (66%) nonembolized bleedings persisted requiring haemostasis by surgery (n=23), endoscopy (n=13) or supportive transfusions. Thirteen of 49 (27%) embolized bleedings recurred and were managed by surgery (n=7), endoscopy (n=3) or transfusion. Overall clinical success rate was 88.9% (56 of 63) in the nonembolized and 87.8% (43 of 49) in the embolized group. Mortality rate was 17.2% (10 of 58) in the nonembolized vs 30.6% (15 of 49) in the embolized patients (P=0.115). Whether or not arteriographic findings afforded the opportunity to embolize, outcome of acute nonvariceal GI bleeding did not differ significantly; however, patients undergoing embolization were more critically bleeding and ill.