General practitioner screening for melanoma: sensitivity, specificity, and effect of training

Abstract

Objective To measure the performance of trained and untrained general practitioners (GPs) in screening men and women aged 50 or more for melanomas. Methods GPs trained in melanoma diagnosis, untrained GPs, and skin cancer specialists examined groups of volunteers, each of which included a small number of subjects with prediagnosed suspicious pigmented lesions (SPLs) that were subsequently excised for histopathological examination. Results Trained and untrained GPs achieved mean sensitivities of 0.73 and 0.71, and mean predictive values of 0.40 and 0.37, respectively, for the detection of prediagnosed SPLs. When the SPLs had been excised and examined histopathologically, reanalysis showed mean sensitivities of 0.98 and 0.95, mean specificities of 0.52 and 0.49, and mean positive predictive values of 0.24 and 0.22 for the detection of subjects with melanomas by trained and untrained GPs respectively. Trained GPs were significantly better than untrained GPs at diagnosing as melanomas SPLs that subsequently proved to be melanomas (p=0.04). Conclusions GPs in this study achieved high sensitivities in screening older Australian men and women for melanomas, but at the cost of low specificities and positive predictive values. Training in melanoma diagnosis had no significant effect on sensitivity, specificity, and positive predictive value for screening. Data from the study were tested in a model of population screening for melanomas, and costs per life year saved for men aged 50–70 ranged from $A11 852 to $A40 259 depending upon the screening interval and whether the GPs excised the SPLs diagnosed, or referred all patients to skin cancer specialists; this would be as cost effective as cervical cancer screening.