Diagnosis of pigmented skin lesions by epiluminescence microscopy: determinants of accuracy improvement in a nationwide training programme for practical dermatologists

Background. The poor accuracy of the clinical examination of pigmented skin lesions (PSLs) is a major limitation of secondary prevention strategies for cutaneous melanoma (CMM). In the last few years, the epiluminescence microscopy (ELM) technique has been used increasingly as an adjunct to clinical examination in the dermatology practice. Although the question of training has emerged as a priority, the diffusion, the effects, and the correlates of educational programmes in ELM have seldom been studied. Methods: Thirty ELM images of PSLs (11 CMMs, 14 melanocytic nevi (MN), and 5 nonmelanocytic lesions (NMLs) each matched with the corresponding clinical or plain photographic image were independently diagnosed before and after a one-day workshop by 83 Italian dermatologists participating in a nationwide educational programme on ELM. The original histology diagnosis was assumed as a gold standard. The overall effect of training on a set of accuracy measures by PSL type was evaluated. The association of the professional sector (public/private), number of years of general experience in dermatology (1-10/>10), average weekly number of PSLs seen (less than or equal to 10/11-20/>20), routine use of ELM (no/yes), and area of residence (northern/southern Italy) with the mean number of PSLs correctly diagnosed before and after training was evaluated with the general factorial analysis of variance. The factors associated with improvement between the two tests were evaluated with the analysis of variance for repeated measures. Results: Compared with pretraining data, the average percentage of exact diagnosis increased significantly for all PSLs (CMMs, 72% vs 55%; MN, 68% vs 64%; NMLs, 67% vs 58%; total lesions combined, 69% vs 60%). Baseline as well as final accuracy were independent from the professional sector and the years of experience but were greater among those subjects who reported >20 PSLs per week compared with the reference group (less than or equal to 10 PSLs). The routine use of ELM was associated with a slight advantage in pretraining accuracy. The area of residence was the strongest determinant of baseline as well as final accuracy. The effect of training was independent from all factors studied with the exception of the area of residence with a 13% increase in the frequency of exact diagnosis in northern Italy (from 66-79%) and 6% in southern Italy (from 55-61%). Conclusions: Though insufficient in absolute terms, a measurable increase in ELM accuracy can be achieved even with intense training sessions of short duration. Medical education to ELM in southern Italy should be a priority.