To the Editor.—
Grin and colleagues1 found less improvement in the accuracy of melanoma diagnosis at New York University over a 20-year period than they had hoped. They hypothesize that this is due to the increased prevalence of thin lesions toward the end of that period, and that thin lesions are not diagnosed as accurately as thick lesions. The former may be tested in their data by examining changes over time in the prevalence of in situ lesions among all melanomas, and in the mean thickness among invasive melanomas. The second may be evaluated by examining the thickness-specific sensitivity of the clinical diagnosis of melanoma, and the sensitivity among in situ lesions. Their hypotheses would predict that (1) the prevalence of in situ lesions was greater in more recent years, (2) the mean thickness of invasive lesions was less in more recent years, (3) the sensitivity of clinical diagnosis is