Reliability and accuracy of dermatologists' digital image consultations

We compared the reliability and accuracy of diagnoses and management plans generated by dermatologists performing digital image review with dermatologists performing traditional clinic-based consultations. Our patient sample included referrals to a dermatology consult service from the ambulatory care clinics at a VA Medical Center. Each patient underwent a standardized history assessment and digital imaging of the referred dermatologic condition. Two dermatologists independently evaluated the patients in a traditional clinic-based setting. Three different dermatologists independently reviewed the standardized histories and digital images. Examiners were asked to provide a diagnosis (single most likely and differential) and suggest a management plan (medications, diagnostic testing, or therapeutic interventions). Simple proportion agreement among the clinic-based dermatologists for their single most likely diagnosis was 0.54 (95% CI 0.46-0.61) and was 0.92 (95% CI 0.88-0.96) when both the single most likely diagnosis and differential diagnoses were considered, A comparable level of agreement was found among the six clinic-based/digital image examiner pairs ranging from 0.41 (95% CI 0.34-0.49) to 0.55 (95% CI 0.48-0.63) and 0.79 (95% CI 0.72-0.85) to 0.95 (95% CI 0.92-0.98), respectively and among the three digital image examiner pairs ranging from 0.49 (95% CI 0.41-0.56) to 0.55 (95% CI 0.48-0.63) and 0.84 (95% CI 0.79-0.90) to 0.92 (95% CI 0.88-0.96), respectively. There was less reliability among examiner pairs for management decisions. Diagnostic accuracy, assessed among the subset of lesions that underwent definitive diagnostic testing, did not differ among the five examiners for their single most likely diagnosis (p = 0.21). When accuracy assessments included examiners' single most likely diagnosis and differential diagnoses there was a difference among examiners in the proportion of correct diagnoses (p = 0.03). One digital examiner accounted for the difference in accuracy and when removed from the analysis the remaining four examiners were comparably accurate (p = 0.64). Compared to traditional clinic-based consultations, dermatologic consultations that use digital imaging technology result in reliable diagnostic outcomes. Our study also suggests that accuracy may not differ by consult modality.