tions of acute biliary pancreatitis nearly in half (OR, 0.56; 95% CI, 0.38 – 0.83). This improvement in outcome is entirely accounted for by the group with predicted severe pancreatitis (OR, 0.27; 95% CI, 0.14 – 0.53); there was no reduction in complications in the group with predicted mild pancreatitis. The same analysis showed no reduction in overall mortality for early ERCP, although another meta-analysis did calculate a reduction in mortality in the subgroup with predicted severe pancreatitis.130 The fourth and most recent randomized trial128 has not been included in these meta-analyses. In this most recent trial, 103 patients with acute biliary pancreatitis who also had a dilated bile duct ( 8 mm) on initial ultrasonography and a bilirubin level 1.2 mg/dL were randomized to early ERCP within 72 hours of admission. Patients with cholangitis were excluded. Although bile duct stones were seen and removed in 72% of the group randomized to early ERCP, there was no difference in the primary outcomes of organ failure, mean CT severity index, local complications, overall morbidity, or mortality.

Taken together, these data and clinical experience provide important guidelines for ERCP in the management of acute biliary pancreatitis. ERCP should be urgently performed when acute cholangitis has complicated acute biliary pancreatitis (about 10% of patients). ERCP should also be performed when clinical or radiographic features suggest a persistent common bile duct stone (a dilated common bile duct or visible common bile duct stone, or jaundice or persistently abnormal liver chemistry values).

In some centers, EUS is used to identify patients with acute biliary pancreatitis who have persistent bile duct stones and thus select patients for early ERCP.

Early ERCP may also be considered in the absence of these situations, when biliary pancreatitis is severe or is predicted to be severe (based on APACHE II, Ranson’s criteria, or modified Glasgow criteria). Early ERCP in this situation (for severe or predicted severe pancreatitis in the absence of concomitant cholangitis or a high suspicion of a persistent common bile duct stone) is more controversial, and the data from randomized trials are not uniform in support of this practice. If early ERCP is performed, it should be undertaken within 48 –72 hours of the onset of illness. In these randomized trials, sphincterotomy was not performed in the absence of stones in the common bile duct; it is not known if this strategy is justified. The decision whether to perform sphincterotomy if no stones are visualized in the common bile duct at the time of ERCP is individualized and may be influenced by the size of the cystic duct, the size of stones remaining within the gallbladder, the size of the common bile duct, and the expected wait until cholecystectomy.

Irrespective of these issues, cholecystectomy is indicated as soon as possible and in no case beyond 2– 4 weeks after discharge to prevent relapses of acute pancreatitis. In patients who are not fit for surgery, endoscopic sphincterotomy alone provides acceptable protection

from subsequent attacks of acute biliary pancreatitis.132 In 8 case series comprising 320 patients with gallstone pancreatitis or choledocholithiasis and gallbladder in situ managed by ERCP and sphincterotomy alone, only 3 (1%) developed recurrent biliary pancreatitis but 56 (17%) developed other biliary symptoms or complications (such as acute cholecystitis or biliary colic).132–139 This rate of biliary symptoms and complications is high enough to warrant laparoscopic cholecystectomy if the patient is fit for surgery.

In patients with mild or resolved acute biliary pancreatitis who are scheduled for cholecystectomy, there is usually little need for preoperative ERCP because the risk of persistent common bile duct stones is low. There is no evidence that routine preoperative ERCP reduces complications, cost, or length of stay.140 A randomized trial of routine preoperative ERCP compared with selective use of postoperative ERCP based on the results of intraoperative cholangiography noted shorter hospital stays and lower cost in the postoperative ERCP group.141 This trial excluded patients with associated cholangitis; urgent ERCP is obviously required in these patients. In patients with a high likelihood of persistent common bile duct stones, preoperative ERCP is appropriate. In one analysis, preoperative ERCP was the most cost-effective approach when the prevalence of common bile duct stones reached

80%.142 In situations in which the prevalence of common bile duct stones was 80%, laparoscopic common bile duct exploration or, if unavailable, postoperative ERCP were most cost effective. In patients in whom a preoperative question exists as to the presence of persistent common bile duct stones, preoperative EUS or

MRCP is appropriate rather than proceeding directly to

ERCP.143

Prophylactic Antibiotics

Infection of pancreatic necrosis is the major cause of morbidity and mortality in acute pancreatitis after the

first week of illness. The prevention of infection in patients with pancreatic necrosis has therefore been a sought-after clinical goal. Early trials of antibiotic prophylaxis used antibiotics that were later shown to have inadequate penetration into pancreatic necrosis. Several recent randomized trials have assessed the efficacy of antibiotic prophylaxis using agents with better tissue penetration.144 –151 These trials have used different patient selection criteria, different antibiotics, different outcome measures, and different durations of treatment. Only 2 of these studies are double blinded.150,151 Several systematic reviews and meta-analyses have been performed on these studies, but the heterogeneity of the studies reduces the reliability of such analyses. The most recent Cochrane Database review152 combined data from 4 studies (not including the 2 most recent studies that are double blind) and concluded that prophylactic antibiotics reduced mortality (OR, 0.32; P .02) and pan-

AGA INSTITUTE