The causal diagnosis of obstructive jaundice. A Bayesian approach

To compare the contribution of clinical, laboratory and imaging data for the diagnostic accuracy in establishing the cause of obstructive jaundice, the records of 333 patients operated on for a presumed obstruction of the biliary ducts were analysed. The final diagnoses, after surgery, were divided into six groups: stones of the gallbladder and biliary tract, pancreatic cancer, biliary tract neoplasms, ampullary carcinoma, postoperative stenosis of the bile ducts, intrahepatic cholestasis (biliary cirrhosis, chronic cholangitis) and other causes of jaundice (liver tumours, cysts, extended cancer of the upper abdomen). Twenty-three parameters (9 clinical symptoms and signs, 10 biochemical alterations and 4 imaging methods) were examined. Not every case had all the investigations performed but the number of each group of data was large enough to allow a statistically significant conclusion. The contribution of each of the 23 parameters in increasing the probability of correct diagnosis was determined using a computer program based on Bayes' theorem. This analysis showed that for patients which presented suggestive clinical signs and symptoms for stones of the gallbladder and biliary tract and for pancreatic cancer, the diagnosis can be predicted with a probability of 90% only on the basis of clinical findings. The probability of a correct preoperative diagnosis was increased to 99% by imaging methods. On the contrary, for patients with a less clear diagnostic probability (with ampullary carcinoma, intrahepatic cholestasis and other causes) only ultrasonography and computed tomography could increase the probability of correct diagnosis.(ABSTRACT TRUNCATED AT 250 WORDS)