Store-and-forward teledermatology based on photographs taken with a smartphone by a general practitioner in case of suspicion of melanoma: evaluation of inter- and intra-rater concordance (Preprint)

Abstract

The incidence of melanoma is increasing worldwide. The effectiveness of treatment is associated with diagnostic and therapeutic delays. In this context, teledermatology (TD), especially store-and-forward TD, is a promising technology, as it can accelerate the care process. However, several studies have indicated that the efficiency and reliability of this practice in cases of pigmented skin lesions is not proven and have called for further work.

The main objective of this study was to evaluate inter- and intra-rater management concordance among teledermatologists concerning the appropriate consultation delay when reviewing photographs of suspicious pigmented skin lesions taken with a smartphone by a general practitioner (GP).

We first analyzed management concordance among different teledermatologists with experience in TD in 2020. The teledermatologists determined the consultation delay to propose to 110 patients with suspicious pigmented skin lesions viewed online based on 241 photographs taken by GPs in a real-life context. The practitioners had to decide among 4 possibilities for care management: 1. urgent need for consultation (15 days), 2. non-urgent need for consultation, 3. no need for consultation ("simple letter to the attending physician"), and 4. impossible to assess: request for additional information or photographs from the GP. "Major" discordance was defined as follows: one teledermatologist decided that the patient needed a consultation within 15 days, while another teledermatologist decided that it was not necessary to see the patient. We performed the same analysis two years after to evaluate the intra-rater concordance.

This study revealed poor agreement among dermatologists when they assessed photographs taken by a GP using a smartphone. The interrater agreement was poor regardless of the analytical modality used. The 2-to-2 interdermatologist management concordance revealed a low Cohen's kappa (between 0.11 and 0.43). Overall agreement was low (Fleiss’ kappa 0.24 in 2020 et 0.27 in 2022). We observed major discordance regarding care management for 38% of the patients in 2020 and 25% of the patients in 2022. Those results of interrater agreement were confirmed by the second analysis in 2022. The intra-rater concordance was poor also with low Cohen's kappa indices (between 0.37 and 0.46).

The poor inter- and intra-rater agreement calls into question the relevance of the store and forward dermatology, without the use of dermoscopy, for suspicious pigmented lesions. This study is in line with recent studies calling for postponement of the development of store-and-forward TD for the management of suspicious pigmented skin lesions until additional studies have been carried out.