The strength of using the First National Health and Nutrition Examination Survey database to address a clinically relevant question (that would be very expensive and premature to answer using a prospective, randomized, placebo-controlled design) should not be underestimated. Examining existing databases to test new hypotheses is generally regarded as a cost-effective and appropriate epidemiological strategy.1
The comparability of our study to prior studies that have measured thickness of non—sun-exposed skin is not relevant. The end point, the appearance of the skin, is intended to be different than skin thickness for 2 reasons: first, patients are more likely to care about how their exposed skin looks than how thick their nonexposed skin is, and second, the effect of estrogen on the clinical outcome of skin appearance can be seen as corroborating the intermediate end point of estrogen's effect on skin thickness. Taken together, these findings suggest a mechanism and a