Objectives: Accurate preoperative prediction of choledocholithiasis is essential in order to minimize patient risk and curtail health care expenditures. This study was designed to identify independent risk factors for choledocholithiasis in patients who had undergone cholecystectomy for symptomatic cholelithiasis and to develop a predictive model based on those factors.
Methods: The charts of 1264 consecutive patients who had undergone cholecystectomy at one of three North Carolina hospitals between January 1, 1989 and December 31, 1991 were reviewed; 465 of these patients had confirmed presence or absence of choledocholithiasis by cholangiography and/or common bile duct exploration and were eligible for analysis. Candidate predictor variables included age and maximum preoperative values for each of the following: temperature, alkaline phosphatase, bilirubin, AST, amylase, white blood cell count, and common bile duct diameter. Model development and validation were conducted using standard data-splitting (60% "training," 40% "test") and logistic regression techniques.
Results: Choledocholithiasis was confirmed in 115 (25%) of the 465 eligible patients. Univariate analysis identified bilirubin, common bile duct diameter, AST, temperature, alkaline phosphatase, and age as predictors. Multivariable analysis subsequently identified bilirubin, common bile duct diameter, AST, alkaline phosphatase, and age as independent predictors of choledocholithiasis. A final model containing these variables (except age, whose contribution to the model was small) accurately predicted choledocholithiasis (c-index = 0.76).
Conclusions: Accurate estimates of choledocholithiasis risk can be made using maximum preoperative bilirubin, common bile duct diameter, AST, and alkaline phosphatase values. Use of the model may help physicians select those patients with symptomatic cholelithiasis who would most likely benefit from further investigation to exclude choledocholithiasis.