Premium
Current clinical algorithms for predicting common bile duct stones have only moderate accuracy
Author(s) -
Ebrahim Mohamed,
Sorensen Lars T.,
Jorgensen Lars N.,
Kalaitzakis Evangelos
Publication year - 2018
Publication title -
digestive endoscopy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.5
H-Index - 56
eISSN - 1443-1661
pISSN - 0915-5635
DOI - 10.1111/den.12994
Subject(s) - medicine , endoscopic retrograde cholangiopancreatography , algorithm , confidence interval , common bile duct , gold standard (test) , diagnostic accuracy , gallstones , gastroenterology , pancreatitis , computer science
Background and Aim A risk‐stratified approach for selecting patients likely to harbor common bile duct ( CBD ) stones to proceed directly to endoscopic or surgical stone clearance, rather than undergo less invasive testing, has been proposed. We assessed the performance of three clinical algorithms used to predict CBD stones. Methods All patients undergoing first‐time endoscopic retrograde cholangiopancreatography ( ERCP ) in 2011–2012 as a result of suspected CBD stones were enrolled prospectively in a clinical database. Data such as imaging test findings and liver function tests (LFTs) were collected 48 h prior to and on the day of ERCP . Patients were stratified into different risk groups for harboring CBD stones according to three clinical algorithms using imaging and laboratory data. Findings on ERCP were used as gold standard. Performance characteristics of each algorithm were separately calculated for each time point of LFT assessment. Results Overall, 186 patients were analyzed, 75% of whom presented CBD stones on ERCP . Proportion of patients categorized as high‐risk for harboring CBD stones varied among the three algorithms (67% vs 73% vs 56%). Also, the algorithms showed only moderate, albeit comparable, accuracy for predicting the presence of CBD stones (0.65, 95% confidence interval [ CI ] 0.62–0.68 vs 0.68, 95% CI 0.63–0.67 vs 0.59, 95% CI 0.57–0.61). Similar results were obtained when performance characteristics were recalculated using LFT from 48 h prior to ERCP (data not shown). Conclusion Three diagnostic algorithms commonly used for predicting CBD stones have comparable but only moderate accuracy. Further research is warranted to improve risk stratification of patients with suspected CBD stones.