Management of massive haemobilia in an Indian hospital.

INTRODUCTION: Massive haemobilia carries a mortality of 25% in most reports. Although previously it was mainly due to road accidents or homicidal attempts it is now more often due to iatrogenic trauma like percutaneous liver biopsy and biliary drainage. However the management protocol is not established and there have been few reports of this serious condition from India. AIM: To review the causes of massive haemobilia and outline its management in an Indian hospital. PATIENTS AND METHODS: We retrospectively analysed the records of 20 consecutive patients with massive haemobilia (blood requirement more than 1400 ml/day) admitted to our department over six years from a prospectively maintained database. There were 10 males and 10 females who had a mean age of 43 (range 15-65) years.
RESULTS: Haemobilia accounted for 9 percent of patients admitted with upper gastrointestinal bleeding who were seen over this period. The commonest cause was iatrogenic (11) including laparoscopic cholecystectomy (6), Whipple's operation, endoscopic retrograde cholangiography (ERC), percutaneous transhepatic cholangiography (PTC), hepatic stone extraction and removal of biliary stent (1 each). The others had accidental trauma (4), visceral aneurysms (2), biliary stones (2) and chronic pancreatitis (1). The commonest clinical presentation was massive gastrointestinal bleeding. The dual phase computed tomography (CT) scan correctly identified the site of bleeding and other associated conditions in all the 11 patients in whom it was done. Conventional angiography was done in 8 patients with transarterial embolisation (TAE) being attempted in 6 and successful in 2 patients. Operations were performed in 18 patients for the following indications - failure of angiographic embolisation (6), failure of endoscopic sclerotherapy (EST) (1), duodenal erosion (2), portal biliopathy (1), haemoperitoneum (1), bile leak (1), pseudocyst (1), liver necrosis (1) and other hepatobiliary conditions (4). The surgical procedures to control bleeding were ligation of aneurysms (8), repair of the hepatic artery (4), right hepatectomy (3), lienorenal shunt, cholecystectomy and under-running of the duodenal papilla (1 each). The overall mortality was 4 patients (20 percent). There was no mortality in patients with bleeding aneurysms; the mortality being significantly higher in patients with non-aneurysmal bleeding (p=0.0049: Fishers' exact test).
CONCLUSIONS: In our experience haemobilia was usually due to an iatrogenic cause with a pseudoaneurysm following a diagnostic or therapeutic intervention(most often laparoscopic cholecystectomy) being the commonest aetiology. A dual phase CT scan accurately identified the site of bleeding. Angiographic embolisation often failed to stop bleeding and mortality was significantly higher in patients with non-aneurysmal bleeding. We should perhaps consider early surgery for haemobilia once the bleeding site has been localised by CT scan.