To the Editor.
—The article by Clark et al1 has very important implications. The results suggest that increased selenium intake results in a substantial reduction in cancer incidence in only 5 years. This is especially remarkable with regard to lung cancer, since the effects of smoking cessation usually require a much longer follow-up.2 These results can be due to (1) a true beneficial effect of selenium; (2) random variation, ie, type I error; and (3) an unexplained higher rate in the control group; in other words, the differences are not due to the benefit of selenium, but due to a much higher rate in the control group than would be expected. Unfortunately, the authors do not provide information on age- and sex-specific cancer rates in either group or the age distribution of cancer cases. They report a mean age of 63 years at entry, 75% men, and approximately 8000 person-years of observation.