2025 年 42 巻 4 号 p. 472-475
Large–scale Real–World Data (RWD) has gained prominence in clinical research because it can complement randomized controlled trials due to high external validity. Large–scale RWD can be classified into three main types : patient registries, administrative claims databases, and electronic health record–based databases. Motivated by clinical experiences during surgical residency, the author turned to large–scale RWD and conducted various studies using administrative claims data, recognizing its untapped potential as a “Blue Ocean” during the doctoral course.
Several clinical epidemiological methods are essential for RWD research. Descriptive research provides foundational insights while multivariable analysis techniques―such as the missing indicator approach, restricted cubic splines, the Fine–Gray model, and generalized estimating equations―help refine analytical approaches. Additionally, propensity score analysis to compare interventions, machine learning techniques for predictive modeling, interrupted time–series analysis to evaluate nationwide policies, and sensitivity analysis to assess robustness are valuable tools.
Data pre–processing can challenge clinicians, especially in research using claims databases. Accurate linking of datasets by patient identifiers, dates, and other relevant information, is essential to form a single spreadsheet aligned with each study design. To ensure accuracy, data processing should be guided by validation studies, international standard indexes, and clinical practices.
Whereas research in fields such as cardiovascular, nephrology, and intensive care using claims databases has become a “Red Ocean”, neurological research remains a “Blue Ocean”. By applying appropriate epidemiological methods and accurate data pre–processing, researchers can solve clinical questions filling existing gaps of evidence. Japan, with its large population, relatively uniform healthcare system, and high–quality data input, offers a strong foundation for impactful RWD research across various clinical disciplines.