因果关系模型在中医药观察性研究中应用的探讨

Causal inferences in experimental studies and observational studies have common empirical philosophical foundations. Both of them obtain counterfactuals by adjusting confounders. Randomization and non-randomization are two traditional schools in causal inference quantificationally. Randomized controlled trials obtain counter factuals by randomization. Causal association is discussed and tested with casual inference in non-randomization. Observational studies could be used in intervention studies of traditional Chinese medicine as randomized controlled trials are not always applicable. Instrumental variable, regression discontinuity design and propensity score methods were introduced in this article.