Interpretation of results of pooled analysis of individual patient data – Authors' reply

We thank Nick Freemantle and Domenico Pagano, and Peter Jüni and colleagues for their comments on our Article on mortality after coronary artery bypass grafting (CABG) versus percutaneous coronary intervention (PCI), in which we identified important subgroup differences that might affect the choice of revascularisation modality.
Freemantle and Pagano challenge our conclusions because we performed no tests for interaction between left main and multivessel disease and did not correct for multiplicity. However, a subgroup analysis according to left main versus multivessel disease would be unsound; many patients in the subgroup with left main disease also had multivessel disease and therefore groups were not mutually exclusive. Hence, testing for a subgroup effect might risk falsely rejecting the null hypothesis.
[...]