Generalization of treatment effect using similarity of populations

2022 Fiscal Year Final Research Report

• Project/Area Number: 19K03627
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 12040:Applied mathematics and statistics-related
• Research Institution: National Hospital Organization Nagoya Medical Center
• Principal Investigator: Kada Akiko 独立行政法人国立病院機構(名古屋医療センター臨床研究センター), その他部局等, 室長 (70399608)
• Co-Investigator(Kenkyū-buntansha): 橋本 大哉 名古屋市立大学, 医薬学総合研究院(医学), 准教授 (50775715)
• Project Period (FY): 2019-04-01 – 2023-03-31
• Keywords: 対象集団 / 疾患登録 / 外部対照 / 一般化

Outline of Final Research Achievements

Focusing on small clinical trials, we examined methods that use population similarity to generalize or transport treatment effects to a broader population. Simulations confirmed the performance of inverse probability of sampling weighting, g-formula, calibration weighting, and augmented method with doubly-robust estimator. In situations where the number of subjects in a clinical trial is small, the advantage of the method using a doubly-robust estimator was confirmed. Confidence intervals for the restricted mean survival time for small sample sizes were examined by simulation in situations of a single-arm trial. The performance of methods with four different types of transformations was evaluated.

Free Research Field

医療統計

Academic Significance and Societal Importance of the Research Achievements

稀少疾患など対象者数が少ない状況で介入試験を行う場合がある。そのような場合に疾患登録が利用できる状況であれば、対象集団の違いを特定し、解析方法を工夫することで、より広い集団での解釈につながる可能性がある。