Background: Detection of common bile duct (CBD) stones in patients with acute biliary pancreatitis (ABP) proves challenging. We hypothesized that grouping clinically significant predictors would increase reliability of detection.
Methods: A retrospective review was performed of 144 consecutive patients who presented to a single tertiary care institution from 2002 to 2007 with ABP.
Results: Of the 144 patients, 32 had a persistent CBD stone. Following multivariate analysis, admission CBD size on ultrasound, gamma glutamyl transferase (GGT), alkaline phosphatase (AP), total bilirubin (TB), and direct bilirubin (DB) significantly correlated with persistent CBD stone. Receiver operator curve analysis and linear regression were applied to obtain optimal and equitable predictive values, and variables combined. Optimal values were: CBD >or= 9 mm; AP >or= 250 U/l; GGT >or= 350 U/l; TB >or= 3 mg/dl; and DB >or= 2 mg/dl. Presence of five variables had an associated odds ratio (OR) of 53.1 (p < 0.001) and four variables an OR of 8.97 (p = 0.004) for presence of persistent CBD stone. Zero variables conferred a significantly decreased probability of CBD stone, OR 0.15 (p < 0.001). Presence of one to three variables did not predict presence of CBD stone.
Conclusion: Presence of four or five variables significantly correlated with persistent CBD stone. Biliary evaluation by endoscopic retrograde cholangiopancreatography is suggested, as initial magnetic resonance cholangiopancreatography (MRCP) may only increase cost and delay time to intervention. In the absence of any variable, biliary evaluation by intraoperative cholangiogram may be sufficient. Decisions regarding patients with one to three variables should occur on a case-to-case basis. Initial biliary evaluation by MRCP is likely preferable, however, as no increased probability of CBD stone was identified, thus not warranting risks associated with intervention.