Development of a clinical diagnostic matrix for characterizing inherited epidermolysis bullosa

Background Accurately diagnosing the subtype of epidermolysis bullosa (EB) is critical for management and genetic counselling. Modern laboratory techniques are largely inaccessible in developing countries, where the diagnosis remains clinical and often inaccurate. Objectives To develop a simple clinical diagnostic tool to aid in the diagnosis and subtyping of EB. Methods We developed a matrix indicating presence or absence of a set of distinctive clinical features (as rows) for the nine most prevalent EB subtypes (as columns). To test an individual patient, presence or absence of these features was compared with the findings expected in each of the nine subtypes to see which corresponded best. If two or more diagnoses scored equally, the diagnosis with the greatest number of specific features was selected. The matrix was tested using findings from 74 genetically characterized patients with EB aged > 6 months by an investigator blinded to molecular diagnosis. For concordance, matrix diagnoses were compared with molecular diagnoses. Results Overall, concordance between the matrix and molecular diagnoses for the four major types of EB was 91.9%, with a kappa coefficient of 0.88 [95% confidence interval (CI) 0.81-0.95; P < 0.001]. The matrix achieved a 75.7% agreement in classifying EB into its nine subtypes, with a kappa coefficient of 0.73 (95% CI 0.69-0.77; P < 0.001). Conclusions The matrix appears to be simple, valid and useful in predicting the type and subtype of EB. An electronic version will facilitate further testing.