Endoscopic retrograde cholangiopancreatography for suspected choledocholithiasis: Testing the current guidelines

Background: Current guidelines include an algorithm for predicting choledocholithiasis. Presence of any very strong predictor or both strong predictors confers a high (>50%) probability of choledocholithiasis. Absence of predictors confers low risk (<10%) of choledocholithiasis. Other combinations have an intermediate risk of choledocholithiasis. Aim: Determine accuracy of the proposed algorithm in predicting choledocholithiasis. Methods: Retrospective analysis of all endoscopic retrograde cholangiopancreatographies performed for suspected choledocholithiasis in 3 years in a Tertiary care hospital and a community hospital serviced by The University of Texas Health Science Center at Houston Division of Gastroenterology. Application of the guidelines, and comparing results to endoscopic retrograde cholangiopancreatography findings. Results: A total of 1080 endoscopic retrograde cholangiopancreatographies were performed; 521 for choledocholithiasis. Most patients were Hispanic and female. Univariate analysis: presence of any very strong predictor and both strong predictors had an OR for choledocholithiasis of 3.30 and 2.36 respectively. Multivariate analysis: odds of choledocholithiasis with any very strong predictor was 2.87, and both strong predictors 3.24. Choledocholithiasis was present in 71.5%, and 41% of patients with high, and intermediate risk respectively. Conclusion: This study confirms the utility of clinical predictors for the diagnosis of choledocholithiasis. All of the very strong predictors and one of the strong predictors increased the odds of choledocholithiasis. Patients with high risk for choledocholithiasis had a probability of 79% of choledocholithiasis. Sensitivity and specificity of current predictors are too low to obviate the possible need of non-invasive tests to confirm or exclude choledocholithiasis in all risk groups.