Research Papers: Skin CancerIdentifying individuals at high risk of melanoma: a simple toolFortes, Cristinaa; Mastroeni, Simonaa; Bakos, Luciod; Antonelli, Gianlucaa; Alessandroni, Liviaa; Pilla, Maria Antoniettaa; Alotto, Massimoa; Zappalà, Albab; Manoorannparampill, Thomasa; Bonamigo, Renane; Pasquini, Paoloa; Melchi, FrancocAuthor Information aClinical Epidemiology Unit bOncology Unit cVIII Dermatology Unit, IDI-IRCCS, Rome, Italy dDepartment of Dermatology, Hospital de Clínicas de Porto Alegre, Federal University of Rio Grande do Sul, Porto Alegre (UFRGS) eDepartment of Dermatology, Fundação Faculdade Federal de Ciências Médicas, Porto Alegre, Brazil Correspondence to Dr Cristina Fortes, PhD, Clinical Epidemiology Unit, Istituto Dermopatico dell'Immacolata, IDI, Via dei Monti di Creta, 104, 00167 Roma, Italy Tel: +39 6 6646 4305; fax: +39 6 6646 4307; e-mail: [email protected] Received 16 February 2010 Accepted 22 April 2010 European Journal of Cancer Prevention: September 2010 - Volume 19 - Issue 5 - p 393-400 doi: 10.1097/CEJ.0b013e32833b492f Buy Metrics Abstract Simple and reliable tools for identifying patients at high risk for melanoma with preventive measures have important public health implications. An individual risk score for cutaneous melanoma was constructed and externally validated. With the summary coefficients of the risk factors for cutaneous melanoma, derived from a meta-analysis, a melanoma risk score was tested in an Italian population and externally validated in a Brazilian population. Common nevi, skin and hair color, freckles, and sunburns in childhood were the variables included in the final predictive model. The discriminatory ability of the models was assessed by the receiver operating characteristic (ROC) curve. The performance of the model was also evaluated by conducting an external validation. The area under the curve (AUC) of the candidate model was 0.79 (95% confidence interval: 0.75–0.82). The same model, when applied in the Brazilian population, presented an AUC of 0.79 (95% confidence interval: 0.70–0.86). At the cut-off level of 3 and more, 89 and 80% of the melanoma cases were correctly classified as ‘at risk for melanoma' in the Italian and in the Brazilian populations, respectively. The risk model is a simple tool that identifies patients for preventive measures and may be used with reasonable confidence in different populations. The risk model may help family doctors in referring patients to dermatological clinics and thus improve early diagnosis. © 2010 Lippincott Williams & Wilkins, Inc.