3375 Where do eus and mrcp fit in the management of the patient with acute biliary pancreatitis: a decision analytic model.

Abstract

Background: In the absence of biliary sepsis, the role of endoscopic retrograde cholangiopancreatography (ERCP)in acute biliary pancreatitis (ABP) is controversial. ERCP has not reliably decreased morbidity or mortality, and procedure-related complications may occur. Magnetic resonance cholangiopancreatography (MRCP) and endoscopic ultrasonography (EUS) are new modalities for CBD visualization that could lower costs and complications related to ERCP.We performed cost-effectiveness and Bayesian analyses of these modalities to define their potential role in ABP. Methods: A decision analytic model of the natural history of ABP was constructed (DATA 3.5, TreeAge Inc., Williamstown, MA). The transition probabilities between health states and test performance characteristics (sensitivity and specificity)were derived from the medical literature using Medline™. All costs(including the diagnostic procedures evaluated) were obtained from Medicare reimbursement rates and supplemented by the medical literature. The strategies evaluated were: 1)ERCP at 48 hours, 2)MRCP at 48 hours followed by ERCP if positive for CBD stone, 3)EUS at 48 hours followed by ERCP if positive for CBD stone and 4) Observation and endoscopic intervention only if biliary sepsis ensued. We compared costs and performed cost-effectiveness analysis between strategies. The outcome measures utilized were total costs, costs per ABP death prevented, costs per ERCP death prevented and costs per stone detected. Results: The total costs per patient with ABP were: Observation $8,821; EUS $9,828; ERCP $9,970; MRCP $10,253. The incremental cost-effectiveness ratio (ICER) per ABP death prevented was equal for ERCP, EUS and MRCP ($900K). The ICER per ERCP-related death prevented was $70K for Observation, $92K for EUS and $95K for MRCP. At a pre-test probability of CBD stones between 0-40%, EUS was the most cost-effective strategy in terms of dollars per stone detected, whereas ERCP was preferable at probabilities >40%. Conclusions: In terms of preventing deaths due to ABP and ERCP, observation is the most cost-effective strategy. From a practical standpoint, if concern exists about retained stones, EUS is most cost-effective, at a pretest probability of stones between 0-40%. If the probability of retained CBD stones is greater than 40%, ERCP is the dominant strategy. Development and validation of clinical predictors of retained CBD stones will facilate cost-effective care in these patients.