Inter‐ and intra‐observer variability in the selection of therapy for infantile hemangiomas among pediatric dermatologists in Spain

Abstract

Guidelines and expert recommendations on infantile hemangiomas (IH) are aimed at increasing homogeneity in clinical decisions based on the risk of sequelae. The objective was to analyze the inter- and intra-observer agreement among pediatric dermatologists in the choice of treatment for IH. We performed a cross-sectional inter-rater and intra-rater agreement study within the Spanish infantile hemangioma registry. Twenty-seven pediatric dermatologists were invited to participate in a survey with 50 clinical vignettes randomly selected within the registry. Each vignette contained a picture of an infantile hemangioma with a clinical description. Raters chose therapy among observation, topical timolol, or oral propranolol. The same survey reordered was completed 1 month later to assess intra-rater agreement. Vignettes were stratified into hemangioma risk categories following the Spanish consensus on IH. The agreement was measured using kappa statistics appropriate for the type of data (Gwet's AC1 coefficient and Gwet's paired t test). Twenty-four dermatologists completed the survey. Vignettes represented 7.8% of the Spanish hemangioma registry. The inter-rater agreement on the treatment decision was fair (AC1 =0.39, 95% confidence interval [CI]: 0.30-0.47). When stratified by risk category, good agreement was reached for high-risk hemangiomas (AC1 =0.77, 95% CI: 0.51-1.00), whereas for intermediate- and low-risk categories, the agreement was only fair (AC1 0.31, 95% CI: 0.16-0.46 and AC1 =0.38, 95% CI: 0.27-0.48, respectively). Propranolol was the main option for high-risk hemangiomas (86.4%), timolol for intermediate-risk (36.8%), and observation for low-risk ones (55.9%). The intra-rater agreement was good. The inter-rater agreement between pediatric dermatologists on the treatment of IH is only fair. Variability was most significant with intermediate- and low-risk hemangiomas.