A risk scoring system to predict clinical events in chronic hepatitis B virus infection: A nationwide cohort study

Abstract

Many patients with chronic hepatitis B do not receive adequate follow-up. This study aimed to develop a risk score to predict clinical events in patients with chronic hepatitis B virus (HBV) infection at the population level for identifying patients at high risk to warrant regular follow-up. This study analysed population-based data from the nationwide claims database of South Korea obtained between 2005 and 2015. We identified 507,239 non-cirrhotic patients with chronic HBV infection who are not under antiviral treatment. A risk score for predicting clinical events (hepatocellular carcinoma, death or liver transplantation) was developed based on multivariable Cox proportional hazard model in a development cohort (n = 401,745) and validated in a validation cohort (n = 105,494). The cumulative incidence rates of clinical events at 5 years were 2.56% and 2.44% in the development and validation cohorts, respectively. Clinical events in asymptomatic patients with chronic HBV infection (CAP-B) score ranging from 0 to 7.5 points based on age, sex, socioeconomic status, chronic hepatitis C co-infection, diabetes mellitus, statin or antiplatelet exposure, smoking, alcohol consumption, alanine aminotransferase and gamma-glutamyltransferase had good discriminatory accuracy in both the development and validation cohorts (c-indices for 3-, 5- and 10-year risk prediction: all 0.786). The predicted and observed probabilities of clinical events were calibrated in both cohorts. A score of >3.5 points identified subjects at distinctly high risk. The CAP-B score using easily accessible variables can predict clinical events and may allow selection of patients with chronic HBV infection for priority of regular follow-up.