In reply
We agree that observational studies are vulnerable to confounding. However, one can hardly expect a randomized clinical trial on secondhand smoke. Rigorous observational studies are the only design that can be used. Regarding the article by Ioannidis,1 the editors of PloS Med,2 in their editorial about that article, state that “He stresses that reliable evidence generally comes from several studies and from several teams of researchers, and that what matters is the totality of the evidence.”