Prediction of common bile duct stones by noninvasive tests.

Abstract

To define accurate and useful predictors of common bile duct stones (CBDS). The ability to predict CBDS with noninvasive tests can avoid unnecessary, costly, or risky procedures. All patients referred for examination for CBDS by endoscopic ultrasonography (EUS) from 1993-1996 were prospectively entered in a database. In a first sample selected randomly from the whole population, predictors of CBDS were determined by univariate analysis and logistic regression. Predictors were subsequently tested in that sample and in the rest of the population. A separate analysis was done for patients planned for cholecystectomy. Eight hundred and eighty patients (328 men, 552 women), aged 57.8 +/- 17 years (range 16-94), were included. The prevalence of CBDS was 18.8%. Age, serum levels of bilirubin, aspartate aminotransferase, alanine aminotransferase, gamma-glutamyl transferase (GGT), and alkaline phosphatase, and the existence of jaundice and fever, a dilated bile duct, and a pathologic gallbladder were found to be associated with CBDS. Logistic regression was undertaken separately for patients younger than 70 years (predictors: GGT >7 x normal; pathologic gallbladder; dilated bile duct) and older than 70 years (predictors: GGT >7 x normal; fever > 38 degrees C; dilated bile duct). Odds ratios were 3 to 6.7. The model was satisfactorily applicable to the second sample; age <70 years: chi2 = 3.3 (NS); age >70 years: chi2 = 3.8 (NS). In patients younger than age 70 and planned for cholecystectomy, the combination of the level of GGT and dilated bile duct predicted CBDS accurately. A simple screening of patients at risk for CBDS can be achieved with three predictive criteria adapted for the patient's age.