Gastrointestinal bleeding after aortic surgery: The role of laparotomy to rule out aortoenteric fistula

The frequency and causes of gastrointestinal (GI) bleeding occurring after aortic surgery were evaluated retrospectively to determine the incidence of aortoenteric fistula (AEF) in relation to other causes and to place in perspective the role of laparotomy for the diagnosis of AEF. Two hundred fifty-three patients in whom aortic prostheses have been inserted were observed for a mean of 46 months. Seventy-four bleeding episodes occurred in 21% of patients between 1 and 108 months after surgery (mean 29 months). Only one AEF appeared that was associated with GI bleeding, for an incidence of 1.4% of bleeding episodes and 0.4% of grafts inserted. No diagnostic workup for GI bleeding was performed for 20 of the 74 episodes and no AEFs were noted in this group during a mean follow-up of 28 months. Diagnostic evaluation was done for the remaining 54 episodes. No cause for bleeding was identified in 16 patients and no AEF developed in this group during a mean follow-up of 26 months. A potential bleeding site was identified in 38 patients, of whom 30 had intrinsic GI lesions and no subsequent evidence of AEF during a mean follow-up of 28 months. Laparotomy for a suspected AEF was recommended for the remaining eight episodes in eight patients and was performed in six patients. An AEF was seen in one patient with abnormality found on preoperative CT scanning; an intrinsic GI lesion was identified in three patients; and no pathologic condition was found in the remaining two (negative laparotomy rate, 33%). Two patients, who refused surgery, subsequently died and no AEFs or other lesions amenable to surgical intervention were found at autopsy. The projected total negative laparotomy rate would have been 50% (four of eight patients). We conclude that GI bleeding is common whereas AEFs are rare after aortic surgery. Laparotomy performed to exclude an AEF appears to be unwarranted when appropriate diagnostic efforts to demonstrate the lesion do not reveal any abnormality and there is no evidence of occult sepsis.