Can statistical adjustment guided by causal inference improve the accuracy of effect estimation? A simulation and empirical research based on meta-analyses of case–control studies

摘要

Directed acyclic graph in the target population. A, exposure; Y, outcome; L, confounder; R, risk factor; M, mediator; C, collider. The causation from A to Y can be reached in four ways: (1) the direct path A → Y; (2) the indirect path through mediators A → M → Y; the backdoor path through common causes A ← L → Y or A ← L → R → Y; and (3) the front-door path by conditioning on common effects