Feasibility study of using clinical trial individual-level data sample bank as external control to support drug and device development:taking transcatheter aortic valve replacement device as an example
10.3969/j.issn.1004-8812.2025.08.006
- VernacularTitle:临床试验个体水平数据样本库作为外部对照支持药械研发的可行性研究——以经导管主动脉瓣膜置换器械为例
- Author:
Xiao-ying LIN
1
;
Chi-lie DANZENG
1
;
Duo-er WANG
1
;
Ying-xuan ZHU
1
;
Ye LU
1
;
Fan GAO
1
;
Yuan-xin LI
1
;
Meng-zhu SU
1
;
Zi-long ZHANG
1
;
Min CHEN
1
;
Qi-ze LI
1
;
Ru JIANG
1
;
Yan-yan ZHAO
1
;
Yang WANG
1
Author Information
1. 国家心血管病中心 北京协和医学院 中国医学科学院阜外医院医学统计部,北京 102308
- Publication Type:Journal Article
- Keywords:
Clinical trials;
Individual level data;
Sample banks;
External controls
- From:
Chinese Journal of Interventional Cardiology
2025;33(8):459-466
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the feasibility and corresponding implementation methods of constructing a sample resource bank based on individual-level data of completed clinical trials and using it to construct external controls for drug/device clinical trials.Methods Taking the pre-marketing clinical trial of transcatheter active valve replacement(TAVR)for the treatment of aortic valve stenosis as an example,the individual-level databases of multiple trials were standardized to form a sample bank.The original data of any trial in the sample bank were selected as the experimental group,and the remaining samples were selected as the control group.The potential confounding was handled by using the propensity score matching and stratification methods to clarify the process of constructing external controls based on the sample bank of individual-level data of clinical trials.Results This study included individual-level data of single-group trials of 4 TAVR devices,with a total of 569 subjects(59.2%male).The number of subjects in Trials 1 to 4 was 120,120,163,and 166,respectively.Propensity score matching enabled the matching of 113,117,125,and 147 subjects with comparable or similar characteristics from individual-level data from other trials,respectively,demonstrating a high matching success rate.The PS score distribution plot after stratification showed that the proportions of subjects in the experimental and control groups in strata 1 to 5 in scheme 1 were 4/103,11/103,22/92,32/87,and 51/64,respectively.For all constructed external controlled trials,a certain number of control samples with similar baseline characteristics to the experimental groups were distributed within each propensity score stratum.The results of the simulation test also reflected the potential differences between different devices in the 12-month all-cause mortality rate.Conclusions The sample bank constructed with individual-level data from clinical trials,as a high-quality data source,can serve as a source of external control for single-arm trials in the same field,and as a useful supplement to the external control scenario of real-world evidence to support drug and device development.At the same time,targeted research on research methods and bias control measures in related fields is also needed.
