Performance characteristics of an electronic medical record alert system to identify hospitalised patients with active hepatitis C virus infection: a cross-sectional analysis from two large academic centres.

Abstract

Direct-acting antivirals are highly effective in curing hepatitis C virus (HCV) infection but linkage to care remains a barrier to population-wide treatment. Hospitalisation for any cause may be an opportunity to identify candidates for treatment. Here, we present the test characteristics of an electronic medical record (EMR) alert system to identify hospitalised patients with HCV infection. We created two alerts based on HCV antibody and RNA values and HCV-specific International Classification of Diseases codes: an ever-alert designed to detect patients with any history of HCV infection (resolved or active) and an active-alert designed to only detect active infections. We performed manual chart review of alerted patients to determine positive predictive value (PPV). We reviewed consecutively hospitalised patients to determine sensitivity. Two affiliated academic tertiary care hospitals in Baltimore, Maryland, USA. PPV and sensitivity of the alerts for patients with active HCV infection. Hospitalised adults (aged ≥18 years). During 12 weeks in 2022 and 2023, the ever-alert fired for 1218 hospitalisations. The ever-alert detected 438 active infections among 1218 alerted hospitalisations (PPV 36.0% (95% CI 33.3% to 38.7%)) compared with 415 active infections among 591 alerts (PPV 70.1% (95% CI 66.4% to 73.8%)) for the active-alert. Both alerts detected 37 of 44 active cases (sensitivity 81.4% (95% CI 70.6% to 92.1%)) among 1115 consecutively hospitalised patients. The active-alert demonstrated a higher PPV than the ever-alert with a negligible decrease in case detection and similar sensitivity for active infection. The active-alert may represent an efficient method of identifying candidates for HCV linkage to care using hospital EMR data. Future directions include deploying the active-alert with a formal linkage to care team.