Prediction of choledocholithiasis using a pocket microcomputer.

A computerized method, using a small pocket computer, has been used to predict the presence of choledocholithiasis in a prospective series of 239 patients undergoing cholecystectomy. From an initial data base of 424 patients 36 factors were evaluated and the most important 2 of these were determined by multivariate analysis for use in the prospective analysis. Satisfactory operative cholangiograms were a prerequisite to evaluation of the statistical method and were obtained in 90.4 per cent of cases. Using the computerized method a common bile duct stone would have been overlooked in only 1 patient but 17 unnecessary explorations would have been carried out. The overall accuracy of the computerized method was 92.5 per cent. When the method was applied to a further study of 97 patients from a separate centre the overall accuracy was 85.6 per cent. If the method was used to aid selective use of operative cholangiography, cholangiograms would be performed in 20 per cent and stones would be overlooked in less than 1 per cent.