Making cost-effectiveness analyses clinically relevant: the effect of provider expertise and biliary disease prevalence on the economic comparison of alternative diagnostic strategies.

RATIONALE AND OBJECTIVES This study was performed to assess the incremental cost-effectiveness of initial magnetic resonance cholangiopancreatography (MRCP) and initial endoscopic ultrasonography (EUS) compared with initial endoscopic retrograde cholangiopancreatography (ERCP) and to evaluate the effect of MRCP provider expertise on the relative cost-effectiveness of the three methods. MATERIALS AND METHODS Thirty patients with suspected biliary disease and referred for ERCP were prospectively evaluated with EUS, MRCP, or ERCP within 24 hours of referral, according to institutional review board-approved protocol. Performance characteristics were measured for EUS and MRCP, with ERCP as the reference standard. A decision analysis compared the clinical and economic effects of three diagnostic strategies (ERCP, EUS followed by ERCP [EUS-ERCP], and MRCP followed by ERCP [MRCP-ERCP]) using prospective EUS and MRCP test characteristics and Medicare reimbursements. The added costs per additional correct diagnosis and per additional false-positive finding averted and the rates and costs of ERCP-related complications were calculated for EUS-ERCP and MRCP-ERCP. Two additional MRCP readers reviewed MRCP data to evaluate interobserver variability and estimate provider expertise. Additional economic analyses incorporated these estimates. RESULTS Compared with initial ERCP, EUS-ERCP demonstrated 72% of biliary abnormalities and reduced ERCP-related complications by 60%; the corresponding percentages for MRCP-ERCP were 48% and 40%. Initial EUS and initial MRCP decreased the number of ERCP procedures performed by 69% and 49%, respectively. Each correct diagnosis made with ERCP that would not have been made with initial EUS or initial MRCP cost an additional 4,875 dollars or 2,580 dollars, respectively. Each false-positive diagnosis averted with initial ERCP that would have been made with EUS-ERCP or MRCP-ERCP cost an additional 9,750 dollars or 1,548 dollars, respectively. The decision model was most sensitive to disease prevalence. As provider expertise increased, the additional cost of an additional correct diagnosis increased for ERCP compared with MRCP-ERCP, with disease prevalence accentuating provider effects. CONCLUSION Initial EUS and initial MRCP are less costly than initial ERCP, but provider expertise, biliary disease prevalence, and procedural costs influence incremental cost-effectiveness.