Management and outcomes of haemorrhage after pancreatogastrostomy versus pancreatojejunostomy

Abstract Background Postpancreatectomy haemorrhage (PPH) is a major cause of morbidity and mortality after pancreaticoduodenectomy (PD). It remains unclear whether performance of a pancreatogastrostomy (PG) instead of a pancreatojejunostomy (PJ) improves outcomes owing to better endoscopic accessibility. Methods A large retrospective analysis was undertaken to compare outcomes of PPH, depending on whether a PG or PJ was performed. The primary outcome was the rate of successful endoscopy. A secondary outcome was the therapeutic success after adding surgery. Results Of 944 patients who had a PD, 8·4 per cent developed PPH. Endoscopy was the primary intervention in 21 (81 per cent) of 26 patients with a PG and 34 (64 per cent) of 53 with a PJ; it identified the bleeding site in 35 and 25 per cent respectively (P = 0·347). Successful endoscopic treatment was more common in the PG group (31 versus 9 per cent; P = 0·026). Surgery was performed for PPH in 15 patients (58 per cent) with a PG and 35 (66 per cent) with a PJ (P = 0·470). The majority of haemorrhages that required surgery were non-anastomotic intra-abdominal haemorrhages (12 of 15 versus 21 of 35; P = 0·171). Endoscopic or conservative treatment for PPH was successful in 42 per cent of patients with a PG and 32 per cent with a PJ (P = 0·520). The success rate increased to 85 and 91 per cent respectively when surgery was included in the algorithm (P = 0·467). Conclusion The type of pancreatic anastomosis and its inherent effect on endoscopic accessibility had very little impact on the outcome of PPH. This was because haemorrhage frequently occurred from intra-abdominal or non-anastomotic intraluminal lesions.