Accuracy of Ultrasound Imaging for Etiology of New-Onset Painless Jaundice

Abstract

Purpose Ultrasound (US) is considered a first-line study for painless jaundice. However, in our hospital system, patients with new-onset painless jaundice often have a contrast-enhanced computed tomography (CECT) or magnetic resonance cholangiopancreatography (MRCP) regardless of the sonographic findings. Thus, we investigated the accuracy of US for detection of biliary dilatation in patients with new-onset painless jaundice. Methods Our electronic medical record was searched from January 1, 2012, to January 1, 2020, for adult patients with new-onset painless jaundice. Presenting complaint/setting, laboratory values, imaging studies/findings, and final diagnoses were recorded. Patients with pain or known liver disease were excluded. A gastrointestinal physician reviewed the laboratory values/chart to classify the type of suspected obstruction. Two radiologists blindly re-reviewed the US scans, and κ between the radiologists was calculated. Fisher exact test and the 2-sample t test were used for statistical analysis. Results Three hundred sixty patients presented with jaundice (>3 mg/dL), of whom 68 met the inclusion criteria (no pain and no known liver disease). Laboratory values had an overall accuracy of 54%, but were accurate in 87.5% and 85% for obstructing stones/pancreaticobiliary cancer. Ultrasound demonstrated overall accuracy of 78%, but only 69% for pancreaticobiliary cancer and 12.5% for common bile duct stone. Seventy-five percent of the patients underwent follow-up CECT or MRCP regardless of presenting setting. In the emergency department or inpatient setting, 92% of the patients underwent CECT or MRCP regardless of US, and 81% had follow-up CECT or MRCP within 24 hours. Conclusion A US-first strategy in the setting of new-onset painless jaundice is accurate only 78% of the time. In practice, US was almost never a stand-alone imaging examination in patients presenting to the emergency department or inpatient setting with new-onset painless jaundice, no matter the suspected diagnosis based on clinical and laboratory grounds or on the US findings themselves. However, for milder elevations of unconjugated bilirubin (suspicious for Gilbert disease) in the outpatient setting, a US demonstrating lack of biliary dilatation was often a definitive study for exclusion of pathology.