In reply
In their letter, Ezzedine et al suggest that VITAL, an observational study, and SU.VI.MAX, a randomized clinical trial, came to different conclusions about the relationship between supplement use and risk of melanoma mainly because of uncontrolled confounding in the observational study. We agree with Ezzedine et al that randomized trials, by design, are better protected against confounding than observational studies. However, we believe that we controlled for the factors that are associated with melanoma risk in VITAL. Moreover, we had increased precision, given that we had more cases of melanoma (n = 461) in our sample than were studied in the SU.VI.MAX trial (n = 25). Additionally, we obtained the same result of no association after examining the use of supplements and melanoma risk in several different ways, and our findings are robust in that they are supported by the results of other observational studies. For many questions of clinical importance, well-designed observational studies have constituted the best available evidence.