Evaluating Treatment Effectiveness from Nationwide Population-Based Clinical Databases

• Activity Number: 583 - Statistical Applications in Observational Studies
• Type: Contributed
• Date/Time: Wednesday, August 2, 2017 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Chen-Hsin Chen* and Y.-C. Tsay and C.-W. Ou-Young and M.-S. Lai
• Companies: Academia Sinica (Institute of Statistical Science) and Academia Sinica (Institute of Statistical Science) and Academia Sinica (Institute of Statistical Science) and National Taiwan University (Institute of Epidemiology and Prevention Medicine)
• Keywords: immortal time ; left truncation ; propensity score ; right censoring ; time-dependent model ; viral hepatitis therapy

Abstract

A randomized trial is ethically infeasible for studying the long-term efficacy of antiviral therapies, so we designed a retrospective cohort study. The National Viral Hepatitis Therapy Program was launched in Taiwan to treat patients with chronic hepatitis B and/or C. Our study assembled National Health Insurance related databases to examine whether antiviral therapy can improve clinical outcomes of these patients. Due to concern of potential selection bias in the observational study, we adjusted baseline characteristics using inverse probability of treatment weighting by the propensity score. Considering age as the time-scale and dissimilar entry ages of patients in the study, we encountered left-truncated and right-censored data. To account for the immortal time rendered by treatment, we analyzed the data using the treatment status based on initiation age with time-varying covariates of comorbidities and liver diseases. The outcomes included all-cause mortality and risk of hepatocellular carcinoma based on the proportional hazards model and the mixture cure model, respectively. Our approach provides an appropriate evaluation for drug effectiveness in pharmacoepidemiology.