Background: Computer-assisted diagnostic systems are not substantially more accurate than the clinician in the differential diagnosis of jaundice but may help in optimal selection and sequencing of tests. The present study aimed to assess with an electronic diagnostic tool the pattern of ordering tests and the diagnostic contribution and related financial cost of each test in jaundiced patients with pancreatobiliary carcinoma, in an effort to make the clinician's diagnostic behaviour more efficient and economical.
Methods: Clinical and diagnostic test data were prospectively gathered from 356 jaundiced patients with pancreatobiliary carcinoma and entered in a Bayesian diagnostic programme. The test results were added to the existing diagnostic evidence, and the programme calculated the diagnostic contribution of each test.
Results: A total of 1804 diagnostic tests were ordered. Quantitative assessment of the diagnostic contribution of each test showed that percutaneous transhepatic cholangiography and computed tomography were associated with the highest provision of information. The most cost-effective tests were ultrasonography and liver function tests.
Conclusions: It is possible to determine objectively the diagnostic contribution of each test in establishing the diagnosis of pancreatobiliary carcinoma. The observed physician behaviour in ordering the various diagnostic tests might be improved with regard to its efficacy and its cost-effectiveness profile.