The Claims Data Learning & Enhancing for Algorithm Refinement (CLEAR) Study: Overview of the Study Design and...

Medical claims data are increasingly used in epidemiological studies but are inadequately validated to ensure their reliability in identifying diseases. We developed the Claims data Learning & Enhancing for Algorithm Refinement (CLEAR) Study as a database platform to enable the systematic, low-cost implementation of validation studies in Japan. The CLEAR Study links routinely generated medical claims data from hospitals with diagnostic data (eg, laboratory data and diagnostic imaging reports) at the patient level. Using diagnostic data as the gold standard for disease identification, researchers can validate and refine their claims-based identification algorithms. Diagnostic data are collected as needed for each validation study, and data are linked using pseudonymized medical record numbers. Personal information is protected through the use of research identification numbers. To demonstrate the platform's feasibility, we collected data on respiratory syncytial virus (RSV) infections and intussusception cases. Eight hospitals have agreed to participate in the CLEAR Study, and three have completed claims data provision. We obtained data for 5,022 RSV infection cases with 25,920 diagnostic tests, and 1,450 intussusception cases with 561,984 diagnostic tests. We also analyzed the initial diagnostic data for 39,212 RSV infection cases and 439,088 intussusception cases, demonstrating that the database can be used to validate algorithms for these diseases. The CLEAR Study is a newly developed database platform in Japan that facilitates validation studies of claims data for various diseases. By promoting validation studies, this platform will help to improve the reliability of claims-based epidemiological studies and drug risk assessments in Japan.