Development of a clinical score system for the diagnosis of photoallergic contact dermatitis using a consensus process: item selection and reliability.

Background: Photoallergic contact dermatitis (PACD) is an uncommon condition, and there is a lack of validated criteria for its diagnosis.

Objective: To identify a set of relevant criteria to be considered when suspecting a diagnosis of PACD and to assess the reproducibility of these criteria.

Methods: This was a diagnostic item selection and reliability study performed between July 2012 and October 2012. A panel of seven recognized experts was invited to consecutive rounds of a Delphi survey and to a conclusive face-to-face meeting with the aim of obtaining an agreement on criteria for the diagnosis of PACD. The panel was also provided with a series of 16 reports of suspected PACDs to be classified according to a five-point likelihood scale. Identified criteria with the weights attributed by experts were used to develop a score system for the diagnosis of PACD. Consensus was measured by calculating the Intraclass Correlation Coefficient (ICC). The performance of the score system was evaluated in terms of overall classification accuracy.

Results: Seven criteria were identified by experts as relevant for the diagnosis of PACD. The criteria were related to the type of skin lesions, accompanying symptoms, skin area involved, general medical history, modality of exposure to the culprit substance, history of exposure to the sun or other light sources and photopatch test results. Experts reached a moderate agreement on PACD cases classification, with ICC = 0.69 (95% Confidence Interval, CI, 0.50-0.86). The score system enabled discrimination of probable and definite PACD cases from possible and unlikely or excluded ones, with a nearly perfect agreement being observed between the score system classification and judgment by experts.

Conclusion: A diagnostic score was proposed. The score should receive a comprehensive validation on a larger series of cases and with multiple evaluators.