The mention of differential diagnosis is inversely associated with diagnostic concordance. The graph shows the distribution of all cases with concordant and discordant diagnosis divided into those with and without mention of differential diagnosis. Data are plotted for both online and physical consultations together for clarity’s sake, with both kinds of consultations showing a clear association (online differentials, χ2P = .001; consultation differentials, χ2P < .001).
The type of previous physician contact is associated with discordance of diagnosis between online and physical consultation. Previous physician contacts were categorized as no physician contact, contact with a general practitioner only, a dermatologist only, both a general practitioner and a dermatologist, more than 1 dermatologist, and with another specialist. Data are shown for 80 patients; data were missing for 8 patients (χ2P = .02).
Grünig E, Schmidt-Weitmann S, Brockes-Bracht C, Hofbauer GFL. The Accuracy of Diagnosis of an Online Consultation Service Compared With Physical Consultation With a Dermatologist. JAMA Dermatol. 2015;151(12):1375-1376. doi:10.1001/jamadermatol.2015.2537
The division of Clinical Telemedicine at University Hospital Zürich in Switzerland has run an email-based online consultation service for laymen since 1999, offering a physical consultation when deemed appropriate.1 We compared the adequacy of online diagnosis with the benchmark of an in-person dermatological consultation.
A retrospective review was performed for inquiries from June 2007 through September 2013. Paired online and physical consultations with a dermatologist after referral were included after institutional review board approval (NCT01969422). The institutional review board requested that patients be informed about a right to veto, which none exercised. Diagnoses were classified according to the dermatological diagnostic index developed by the German Dermatological Society DDG expanding the classifications of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision.2 Concordance was defined as a matching of digits of the index code, resulting in 6 groups. IBM SPSS 22 software was used for descriptive statistics, χ2, and Kendall τ. P < .05 was considered significant.
One hundred two submissions (32 with photographs) were referred for dermatological consultation. A total of 88 patients presented, and 82 patients (median [SD] age, 38  years, 41 [50%] male) were included. Diagnostic concordance was evenly distributed from low to high (χ2P = .16). A lack of differentials online (χ2P = .001), as well as on physical consultation (χ2P < .001), was associated with diagnostic concordance (Figure 1), but neither photograph submission, prediagnosed conditions, nor personal or family history information was significant (Kendall τ P = .80, P = .04, P = .30, P = .16, respectively). The kind of physician contact was associated with diagnostic concordance (χ2P = .02) (Figure 2). In 37 cases (42%), a second diagnosis was made at physical consultation.
The online and physical consultation diagnoses were concordant in 52% (n = 43, missing data n = 6) and enlarged concordant in 66% (n = 54, missing data n = 6). The patients’ dermatological problems were complex (53 [60%] had already visited a physician). Other studies show concordance or enlarged concordance rates from 47% to 89%, while a concordance rate greater than 60% is deemed reliable in teledermatology.3- 6
Previous physician contact was not associated with diagnostic concordance except in cases in which more than 1 dermatologist or another specialist was visited. None of the information provided by users was associated with diagnostic concordance. Other studies did not look at this association of information provided and diagnostic concordance, but reported that a thorough history as such increases diagnostic concordance for clinical photographs from 57% to 70%.4- 6 We noted absence of differential diagnosis as a factor in diagnostic confidence in both online (χ2P = .001) and physical consultation (P < .001).4,5 In 37 cases (42%), a second diagnosis was made during the physical consultation, with 1 basal cell and 1 squamous epithelial carcinoma diagnosed at consultation. The limitations of our study include its retrospective design, lack of randomization, and the limited patient numbers.
Accepted for Publication: June 19, 2015.
Corresponding Author: Günther F. L. Hofbauer, MD, Dermatologische Klinik, Universitätsspital Zürich, Gloriastrasse 31, 8091 Zürich, Switzerland (firstname.lastname@example.org).
Published Online: September 16, 2015. doi:10.1001/jamadermatol.2015.2537.
Author Contributions: Ms Grünig and Dr Hofbauer had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Ms Grünig and Dr Schmidt-Weitmann served as co–first authors, each with equal contribution to the manuscript. Drs Brockes-Bracht and Hofbauer served as co–last authors, each with equal contribution to the manuscript.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Grünig, Brockes-Bracht, Hofbauer.
Drafting of the manuscript: Grünig, Brockes-Bracht, Hofbauer.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Grünig.
Obtained funding: Brockes-Bracht, Hofbauer.
Administrative, technical, or material support: Brockes-Bracht, Hofbauer.
Study supervision: Schmidt-Weitmann, Brockes-Bracht, Hofbauer.
Conflict of Interest Disclosures: None reported.