Clinical diagnosis of skin tumours: how good are we?

Abstract

Background: Few studies have addressed the accuracy of clinical diagnosis of skin malignancies. The present prospective study aims to determine the accuracy with which these lesions are diagnosed and attempts to identify the factors influencing the accuracy of clinical diagnosis.

Methods: All patients undergoing surgical excision of a skin cancer with curative intent from January 2001 to December 2002 at Peter MacCallum Cancer Institute, Melbourne, Australia were prospectively studied. Data were entered into the Filemaker Pro program and analysed using the SPSS software package.

Results: A total of 2582 lesions were surgically excised from 1223 patients. Of these 47% were basal cell carcinomas (BCC), 20% were squamous cell carcinomas (SCC), 0.9% were malignant melanomas (MM), and 32.1% were benign or premalignant. Tumours, benign and malignant, were found to be more common in men. The sensitivity for clinical diagnosis of malignancy was 97.5% while the positive predictive value (PPV) for clinical diagnosis of malignancy was 70.3%. BCC and SCC was diagnosed with a sensitivity of 89% and 56.3%, and PPV of 64.5% and 40.3%, respectively (P < 0.001). 23 MM were excised with a sensitivity of clinical diagnosis of 47.8% and PPV of 30.6%. Sensitivity and PPV were also assessed according to clinical experience of the surgeon, site of the lesion and whether surgery was performed for a primary or a recurrent lesion. Sensitivity and PPV were higher for lesions diagnosed by consultant surgeons when compared with surgical trainees (P < 0.001) - MM was diagnosed with a sensitivity of 100% by consultant surgeons.

Conclusion: The present study shows sensitivity and PPV rates comparable to published figures.