Evaluation of an Asynchronous Teleconsultation System for Diagnosis of Skin Cancer and Other Skin Diseases

Abstract

We wished to assess both diagnostic accuracy and concordance among dermatologists when evaluating digital images of skin disease presented using a low-cost asynchronous (store-and-forward) format. Each of eight board-certified dermatologists reviewed 50 clinical cases presented in digital format on a 15-inch computer monitor. For each case, the teleconsultants made a primary diagnosis and differential diagnosis and indicated whether or not a biopsy should be performed both before and after reviewing a brief history of each case. Diagnostic accuracy was calculated for each teleconsultant on the basis of biopsy, culture, or wet mount results. Concordance was determined by comparing primary and differential diagnoses made by the teleconsultants on all 50 cases with those made by the dermatologists who originally examined the patients in person. For eight skin cancers, the diagnostic accuracy for the inperson dermatologist was 88% versus 90% (range, 75-100%) for the teleconsultants. For the 25 cases (including the 8 skin cancers) confirmed by either biopsy (20), culture (1), or wet mount (4), the in-person accuracy was 84% compared to 73% (range, 65-88%) for the teleconsultants. The concordance between the in-person and teleconsultant diagnoses were in agreement 77% of the time (90% if differential diagnoses were included). After evaluating the accompanying history, teleconsultants changed their primary diagnosis in 11% of the cases (range, 2-22%). Biopsy rates were not significantly different between teleconsultants (45%) and in-person dermatologists (40%). An asynchronous software application can provide levels of diagnostic accuracy and concordance equivalent to those reported using live teleconsultation.