Diagnostic and referral accuracy of family doctors in melanoma screening: effect of a short formal training

Abstract

Little is known about ability of family doctors in the diagnosis and management (decision as to dermatologic referral) of pigmented skin lesions. We sought to evaluate the impact of a short formal training on diagnostic and referral accuracy of family doctors in melanoma screening. A formal 4-h training session was given to a sample of 41 practising family doctors working in the Florence health district, Tuscany, Italy. Before and after the course, a diagnostic test with a series of clinical images of pigmented skin lesions including four invasive melanomas (mean thickness, 1.5 mm; range, 0.8-2.2) was performed (open intervention study). Although only 46.8% of observations yielded a correct melanoma diagnosis at baseline, 96.1% of melanoma observations were correctly associated with intention to refer the lesion to dermatologist. After training, the percentage of correct melanoma diagnosis significantly increased (76.2%, P=0.01) while no further improvement was found as to sensitivity of referral (94.8%, P=0.58). Compared to baseline, post-training evaluation showed a significant reduction of benign lesions sent to dermatologist: the percentage lowered from 52.1 to 35.8% (P=0.0014) for melanocytic nevi and from 38.6 to 17.5% (P<0.001) for benign non-melanocytic lesions (pigmented seborrheic keratoses, dermatofibromas, and vascular lesions). Grouping these two diagnostic categories, the overall specificity in dermatology referral increased from 55.0% at baseline to 73.1% after training (P<0.001). In conclusion, attendance at a 4-h formal training session was able to increase the specificity of family doctors as to dermatologist referral of suspicious lesions (less false-positive referral of benign lesions) without significant loss in sensitivity concerning melanoma.