To the Editor.
—Ms Barnes and colleagues1 give the totally false impression that, in 1981, I believed that Hirayama's British Medical Journal article,2 which suggested that environmental tobacco smoke (ETS) exposure can cause lung cancer in nonsmokers, was "correct." At that time, Mantel3 had suggested, based on a trend analysis of Hirayama's unadjusted data, that Hirayama had confused χ2 and χ, thereby considerably overstating the statistical significance of the relationship he had observed between lung cancer risk in nonsmoking women and the extent of smoking by their husbands. I had argued that Hirayama's cited trend χ value was probably based on age-adjusted data and was likely to be correct. My judgment to the effect that Hirayama was "correct" related solely to this single aspect of statistical methodology. It did not relate to his conclusion, which at that time (as I found when I visited him in
Lee PN. The Tobacco Industry and the Brown and Williamson Documents. JAMA. 1996;275(4):279. doi:10.1001/jama.1996.03530280031026