Development and Validation of a Simple Model to Predict the Risk of Nonmelanoma Skin Cancer on Screening Total Body Skin Examination

Abstract

Objective. There is insufficient evidence to generate skin cancer screening guidelines at the population level, resulting in arbitrary variation in patient selection for screening skin examinations. This study was aimed at developing an easy-to-use predictive model of nonmelanoma skin cancer (NMSC) risk on screening total body skin examination (TBSE). Methods. This epidemiologic assessment utilized data from a prospective, multicenter international study from primarily academic outpatient dermatology clinics. Potential predictors of NMSC on screening TBSE were identified and used to generate a multivariable model that was converted into a point-based scoring system. The performance characteristics of the model were validated in a second data set from two healthcare institutions in the United States. Results. 8,501 patients were included. Statistically significant predictors of NMSC on screening TBSE included age, skin phototype, and history of NMSC. A multivariable model and point-based scoring system using these predictors exhibited high discrimination (AUC = 0.82). Conclusion. A simple three-variable model, abbreviated as CAP (cancer history, age, phototype) can accurately predict the risk of NMSC on screening TBSE by dermatology. This tool may be used in clinical decision making to enhance the yield of screening TBSE.