Telemedicine in dermatology: a randomised controlled trial

Objectives: To compare the clinical equivalence, patient and clinician opinion of store-and-forward ( SF) teledermatology with conventional face-to-face consultation in setting a management plan for new, adult outpatient referrals. To assess the equivalence of digital photography and dermoscopy with conventional face-to-face consultation in the management of suspected cases of malignant melanoma or squamous cell carcinoma. Design: For the SF teledermatology aspect of the study, a prospective randomised controlled trial was carried out. Setting: Eight general practices and a hospital dermatology department in Sheffield, England. Participants: For the SF teledermatology part of the study, adults ( aged 16 years and over) requiring a new ( not seen by a hospital dermatologist within the past year) consultant opinion. For the digital photography element of the study, adults ( aged 16 years and over) requiring a consultant opinion due to suspicion of malignant melanoma or squamous cell carcinoma. Interventions: Patients in the telemedicine intervention group were referred to the consultant, and managed as far as possible using one or more digital still images and a structured, electronic referral and reply. The control group was managed by conventional hospital outpatient consultation. Patients referred to the 2-week wait clinic were invited to have a series of digital photographs, with and without dermoscopy, immediately before their face-to-face consultation. A second consultant viewed these and outlined a diagnosis and management plan which was compared with the actual management. Both were compared with the definitive diagnosis ( either the final clinical or histological diagnosis, where undertaken). Main outcome measure: The concordance between the consultant who had managed the case and an independent consultant who gave a second face-to-face opinion. Results: A total of 208 patients were recruited. There was also a greater loss of control cases ( 26%) than intervention cases ( 17%). A statistically significant difference in ages between the two groups completing the study ( mean age of intervention group 43.6 years, control group 49.7 years, p = 0.039) indicates that this may have introduced a bias between the two groups. A further possible source of bias is the delay ( mean difference of 54 days, p = 0.0001) between the SF opinion and the second opinion in the SF group, whereas control patients usually received their second opinion on the same day as their outpatient appointment. In 55% ( 51/92) of telemedicine cases and 78% ( 57/73) of control cases, the diagnosis concurred, with the second opinion. In 55% ( 51/ 92) of telemedicine cases and 84%( 61/ 73) of control cases, the management plan concurred with the second opinion. Of the 92 telemedicine cases, 53 were judged also to require a face-to-face consultation, mainly to establish a diagnosis and treatment plan. With the digital photography for suspected skin cancer aspect of the study, it was found that an unexpectedly high proportion ( 33%, 85/ 256) of referrals proved to have a malignancy or a severely dysplastic lesion, with almost 22% having a malignant melanoma or squamous cell carcinoma, possibly reflecting the rise in incidence of skin cancers reported elsewhere. When both standard and dermoscopic images were employed, diagnostic concordance was modest ( 68%). The approach was highly sensitive ( 98%, 95% CI: 92 to 99%), at the expense of specificity ( 43%, 95% CI: 36 to 51%). Overall, 30% of cases would not have needed to be seen face-to-face, though two squamous cell crcinomas would have been missed ( a number-needed-to-harm of 153). If the highest level of clinician confidence had been applied, no cancers would have been missed, but only 20% of patients would have avoided an outpatient appointment. Conclusions: In view of the difficulties in recruitment and the potential biases introduced by selective loss of patients and the delay in obtaining a valid second opinion in the study group, no valid conclusions can be drawn regarding the clinical performance of this model of SF telemedicine. With regard to digital photography in suspected skin cancer, it is unlikely that this approach can dramatically reduce the need for conventional clinical consultations, whilst still maintaining clinical safety. Additional research on the assessment of diagnostic and management agreement between clinicians would be valuable in this and other fields of research.