Going down the right path: Using causal diagrams and multiple comparison procedures to steer clear of the Table 2 Fallacy and multiplicity

The Table 2 Fallacy is an interpretation error commonly encountered in medical research. This error occurs when coefficient estimates in multivariable regression models, apart from that of the primary exposure, are interpreted as valid total effects on the outcome. Causal diagrams, notably directed acyclic graphics (DAGs), can be used to avoid this error. DAGs can be used to identify sets of covariates that, when adjusted for, allow for unbiased estimation and correct interpretation of the total effect for each primary exposure. However, the risk of a false positive finding, or the type I error rate, increases when inference is conducted for multiple associations of interest. While multiple comparison procedures are recognized as a critical feature of randomized controlled trial (RCT) design and analysis, opinions remain divided on their use within observational studies. Target trial emulation is an increasingly used framework that aligns the design and analysis of causal observational studies with that of RCTs. This article demonstrates that when multiple causal relationships are of primary interest to an observational study, causal diagrams can be used in conjunction with multiple comparison procedures to simultaneously avoid the Table 2 Fallacy and control the risk of spurious findings. The proposed framework should further align these two study types and assist in ensuring that results are reproducible. Methods are demonstrated by simulating a hypothetical causal mechanism using the dagR library.