Causal inference in observational studies based on real-world data: Key points and case studies for target trial emulation
- VernacularTitle:基于真实世界数据的观察性研究因果推断—— 目标试验模拟实施要点及案例分析
- Author:
Chi YUAN
1
;
Yiling ZHOU
2
;
Yuzi CAO
3
;
Haojie ZHANG
3
;
Yiqian WANG
4
;
Sheyu LI
3
Author Information
1. 1. Department of Pediatric Surgery, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China 2. Department of Endocrinology and Metabolism, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China 3. MAGIC China Center, Chinese Evidence-Based Medicine Center, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China
2. 2. Department of Endocrinology and Metabolism, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China 3. MAGIC China Center, Chinese Evidence-Based Medicine Center, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China 4. Department of Epidemiology, University of Groningen, University Medical Center Groningen, Groningen, the Netherlands
3. 2. Department of Endocrinology and Metabolism, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China 3. MAGIC China Center, Chinese Evidence-Based Medicine Center, West China Hospital, Sichuan University, Chengdu, 610041, P. R. China
4. Department of Mathematics and Statistics, McGill University, Montreal, Quebec, Canada
- Publication Type:Journal Article
- Keywords:
Target trial emulation;
real-world data;
causal inference;
observational study
- From:
Chinese Journal of Clinical Thoracic and Cardiovascular Surgery
2024;31(12):1743-1752
- CountryChina
- Language:Chinese
-
Abstract:
Randomized controlled trial (RCT) are considered the "gold standard" for evaluating the causal effects of interventions on outcome measures. However, due to high research costs and ethical constraints, conducting RCT in clinical practice, especially in the surgical field, faces numerous challenges such as difficulties in subject recruitment, implementation of blinding, and standardization of interventions. In such cases, using real-world data to perform causal inference under the framework of target trial emulation (TTE), based on the principles of RCT design, helps to identify and reduce biases arising from design flaws in traditional observational studies, such as immortal time bias, confounding, selection bias, or collider bias. This approach can produce high-quality evidence comparable to that of RCT, thereby enhancing the clinical guidance value of real-world data studies. However, TTE has limitations, such as the inability to completely eliminate confounding, high quality requirements for source data, and the current lack of reporting standards. Therefore, researchers should be fully aware of these limitations to avoid making incorrect causal inferences. This article intends to provide an overview of the TTE framework, implementation points, application scope, application cases, and advantages and disadvantages of the framework.
