Interobserver Reliability of Store-and-Forward Teledermatology in a Clinical Practice Setting

Introduction and objectives: Although many studies have evaluated the diagnostic reliability of store-and-forward (SF) teledermatology, the reliability of the technique for the diagnosis of general skin conditions in a clinical practice setting has never been demonstrated. We evaluated the reliability of SF teledermatology in clinical practice by analyzing the diagnostic agreement achieved in a subgroup of patients from the DERMATEL-2 study. Material and methods: Patients referred from primary care settings were randomized to 3 groups: SF, a combination of videoconferencing and SF technology (VC-SF), and a control group. This article focuses on the SF group. Clinical data were recorded and photographs taken by primary care physicians, who forwarded the data electronically. Each SF consultation package was assessed by 3 dermatologists (D1,D2,D3). Subsequently all the patients were assessed by a single dermatologist (D1) in a face-to-face consultation. Finally, 2 other dermatologists (D4,D5) assessed the agreement between the diagnoses obtained by SF and FF. Results: In total, 457 patients (200 males and 257 females) aged between 2 months and 86 years were randomized (192 to SF, 176 to VC-SF, and 89 to the control group). The diagnostic categories were as follows: tumors (49.4%), inflammatory (25.7%), adnexal (11%), infectious (9.4%) and other processes (4.4%) Since 170 patients had consultations deemed valid for analysis, the study included a total of 510 SF assessments. Most of the images and clinical records were of high quality (71.2% and 91.2% respectively), and diagnostic confidence was high in 81.4% of the cases studied. In 58.4% of cases the condition was managed exclusively by teledermatology. Levels of complete and aggregate interobserver agreement between SF and face-to-face evaluators were 0,72 and 0.90, respectively, for diagnosis and 0.61 and 0.80 for treatment. Diagnostic agreement correlated with the image quality (P < .001), diagnostic confidence (P<.001), felt need for conventional consultation (P<.001), and the quality of the clinical record (P=.013). Conclusion: The interobserver reliability of SF diagnosis in clinical practice is good. Dermatologists are able to predict errors in diagnosis by analyzing their own diagnostic confidence and evaluating the quality of the images. (C) 2013 Elsevier Espana, S.L. and AEDV. All rights reserved.