Prospective validation study of an algorithm for triage to MRCP or ERCP for investigation of suspected pancreatico-biliary disease

Abstract

Background: In patients with suspected pancreatico-biliary disease, endoscopic retrograde cholangiopancreatography (ERCP) should be reserved for those requiring therapeutic intervention. However, difficulty arises in identifying patients likely to require therapy in the early phase of diagnostic work-up. An algorithm has been developed by the authors based upon prospective assessment of ERCP patients for triage of patients to magnetic resonance cholangiopancreatography (MRCP) or ERCP with suspected pancreatico-biliary disease. We aimed to validate this algorithm in an independent group of patients using a different group of endoscopists blinded to the algorithm.

Methods: Patients were stratified into different categories by clinical, ultrasound and liver function test findings. The algorithm stratified patients by the likelihood of therapeutic intervention. The accuracy of the algorithm for a therapeutic outcome was assessed by receiver operator characteristics (ROC) curve analysis.

Results: Hundred and twenty-five consecutive patients (Oct 2005 to July 2006) were prospectively assessed by MRCP or ERCP according to the algorithm, and the outcomes recorded. Fifty-seven patients were triaged to MRCP and 63 patients were triaged to ERCP. A category was not assessable in five patients. Three patients from the MRCP group required subsequent therapeutic ERCP. Diagnostic ERCP was performed in three patients in the ERCP group. ERCP-related complications occurred in four patients. The algorithm performed well in predicting the requirement for intervention as determined by the area under the ROC curve [0.84 (95%CI 0.76-0.92)].

Conclusions: Our study confirms that an algorithm-based approach can reproducibly predict those patients requiring therapeutic biliary intervention.