Violeta Zaragoza‐Ninet

Background: Current relevance of positive patch‐test reactions guides management in allergic contact dermatitis (ACD), yet its clinical predictors and the use of predictive models in clinical practice remain underused. Objectives: To identify demographic and clinical factors associated with current relevance in ACD and to develop overall and allergen‐specific predictive models. Methods: A multicentric REIDAC study included data from patch‐tested patients with the Spanish baseline series. Exposure history, anatomical sites, atopic status, and occupational data were recorded. Logistic regression (LR) models were trained and internally validated to predict current relevance overall and for top frequent allergens. Model discrimination was assessed with area under the receiver‐operating characteristic curve (AUC). Results: Among 17 005 patients, 4077 (24.0%) had at least one currently relevant reaction. The final overall LR model achieved an AUC for the validation sample of 0.679. Allergen‐specific AUC parameters (LR) varied among allergens but performed best for nickel (AUC = 0.770). The independent factors associated with current relevance were female gender, specific body sites (hand, neck, head, leg, feet) and two occupations (hairdresser and construction workers). The use of other models (LASSO, gradient boosting) revealed similar results. Conclusions: Prediction modelling may moderately predict current relevance in ACD and several clinical variables are associated with current relevance in ACD.