|
BACKGROUND/RATIONALE:
Teledermatology has the potential to radically change the delivery of dermatological care for veterans. Pilot studies have demonstrated the clinical usefulness of teledermatology when used for a wide variety of skin disorders but preliminary evidence suggests that this system may be less optimal for skin neoplasms, which comprise approximately 50% of all dermatological visits in the VA.
OBJECTIVE(S):
The primary objective of this study was to determine the diagnostic accuracy of store-forward teledermatology for skin neoplasms. Secondary objectives included determination of: 1) diagnostic agreement and 2) appropriateness of management plans of teledermatologists as compared to clinic dermatologists.
METHODS:
This study was designed and executed as a repeated-measures equivalence trial to compare the diagnostic accuracy of store-forward teledermatology (asynchronous, still images of skin viewed by remote dermatologists) and traditional, in-person encounters by clinic dermatologists. Eligible patients included either those referred to the Minneapolis VA dermatology clinic for evaluation of a skin neoplasm or evaluated in the dermatology clinic for a skin neoplasm. After informed consent, each lesion was examined first by a clinic dermatologist and then, at a later time, by a teledermatologist; both generated a primary diagnosis, up to two differential diagnoses, and a management plan. Histopathology served as the gold standard. The primary statistical analyses used two-sided equivalence testing to determine whether the diagnostic accuracy of teledermatology is equivalent to standard, in-person, clinic-based exams (i.e., whether a 95% CI for the difference between accuracy rates lies inside the interval -10% to +10%). A sample size of 2080 was calculated to provide a power of >80% to assess both primary and secondary outcomes for both pigmented and non-pigmented lesions.
FINDINGS/RESULTS:
2,152 patients were enrolled; the majority were male (2082, 96.8%) and Caucasian (2098, 97.5%). Mean age was 68 years (range 19-94). 45.1% (967/2145) of the target lesions were located on the face or ears. A quarter of the 1270 biopsied target lesions were basal cell carcinomas (303, 24.3%) followed by benign keratoses (173, 13.6%) and squamous cell carcinomas (165, 13.0%). 2.8% (36) were melanomas.
For the primary outcome, partial diagnostic accuracy, teledermatologists and clinic dermatologists were not equivalent. The primary diagnostic accuracy of teledermatologists and clinical dermatologists was neither equivalent nor not equivalent. The appropriateness of the respective management plans for clinicians and teledermatologists were equivalent. The results for all target lesions as well as for the two pigmented and non-pigmented subtypes of target lesions were consistent with the overall group. Excluding cases rated as moderate picture quality, prespecified equivalence requirements were reached for primary diagnostic accuracy for all target lesions with high quality pictures as well as for both high and low quality for management plans.
Despite the indeterminate results concerning equivalence, the accuracy rates of clinical dermatologists were significantly better than teledermatologists. This held for both pigmented and non-pigmented lesions and for both primary diagnostic accuracy as well as partial diagnostic accuracy. The accuracy of standard, in person, clinical dermatology was also statistically significantly superior to teledermatology for the categories of basal cell carcinoma, squamous cell carcinoma, and melanoma. For melanoma, the average accuracy rate for clinic dermatologists was 17-25% better than the average accuracy rate for clinic dermatologists.
There was a statistically significantly higher rate of severity for teledermatologists’ management plans as compared with the clinic dermatologists. Of all lesions (target and non-target lesions) which were rated as “severe” significance by the expert panel for the clinic dermatologists, two lesions were melanoma as compared with nine lesions for teledermatologists.
When all target lesions, as well as just non-biopsied lesions or just biopsied lesions, were considered, the rates of partial diagnostic agreement for teledermatologists and clinic dermatologists were higher than just the primary diagnostic agreement alone (84-86% vs. 48-84%). The percent agreement for management plans ranged from 72-77%. The primary diagnostic agreement rates were higher for non-biopsied target lesions as compared to biopsied target lesions, regardless of pigment status.
IMPACT:
This study is the first to focus on skin neoplasms specifically in veterans. The results will have direct impact on the formation of teledermatology programs within the VA as well as in the private sector.
PUBLICATIONS:
None at this time.
DRA:
Health Services and Systems
DRE:
Diagnosis and Prognosis, Technology Development and Assessment
Keywords:
Cancer, Telemedicine
MeSH Terms:
Diagnosis, Computer-Assisted, Telemedicine, Dermatology
|
