An Economic Evaluation from Randomized Trial Data Comparing ERCP to MRCP in Suspected Biliary Obstruction

Abstract

Background: MRCP can prevent unnecessary ERCPs and their complications in patients with biliary obstruction (BO). Aim: To determine differences in costs between MRCP and ERCP-first strategies in patients with an intermittent probability of BO, and to identify predictors of resource use. Methods: Patients with an intermediate probability of BO based on clinical and ultrasound (US) findings were randomized to MRCP or ERCP, and followed for a year. The number of procedures, and medical direct and indirect Canadian costs (in CDN $) were analyzed per intention-to-treat. Multivariate linear regression models identified independent predictors of resource utilization. Results: 131 patients were randomized to MRCP and 126 to ERCP. Groups were similar in age (55.8 ± 17.7 yrs vs 52.3 ± 18.4 yrs), gender (60% vs 65% female), symptoms, and post-cholecystectomy status (44% vs 34%). Mean follow-ups were 10.3 ± 3.0 mos for MRCP and 10.1 ± 3.3 mos for ERCP. The proportion of additional procedures (58.8% vs 21.4%, CI on difference [0.26; 0.48]) and their total numbers (151 vs 202, CI on difference [14.2, 87.8]) were greater in the MRCP group. No clinically relevant difference was observed in days away from usual activities (268.5d, mean = 2.05d, SE [4.85d] for MRCP, and 431d, mean = 3.42d, SE [7.66d] for ERCP, CI on mean difference [-0.92; 1.55d]). Except for those related to diagnostic ERCP, all other direct medical costs were greater in the MRCP group (average (per patient) and total opportunity loss of CDN$579.55 and CDN$ 79,745.90). In contrast, indirect costs were greater in the ERCP group (average and total savings of CDN$ 36.43 and CDN$ 9,911.29). When summing direct and indirect costs, an ERCP-first strategy resulted in net total and average savings of CDN$ 69,830.37 and CDN$ 543.17 respectively. Multivariate linear regression showed that, in the ERCP group, pre-randomization US findings of CBD (beta = 1.66) and intrahepatic duct dilation (beta = 3.08) were associated with a greater number of post-ERCP procedures, while suspected CBD stone (beta = −2.10) or chronic pancreatitis (beta = −3.39) were negative predictors. In the MRCP group, an elevated serum bilirubin (beta = 0.017), pre-randomization US suspicion of a CBD stone (beta = 2.68), or of an ampullary lesion (beta = 13.6) were associated with a greater likelihood of additional tests. Conclusion: This economic analysis of a medical effectiveness randomized trial suggests that ERCP-first is a preferred strategy for the management of patients with an intermediate probability of biliary obstruction. Initial abdominal US findings provide valuable prognostic information which can improve downstream resource utilization.