Performance of the I3A Scoring System for Risk Stratification in a Retrospective Cohort of Children With Ibuprofen-Induced NSAID Hypersensitivity.

Abstract

Two clinical history-based risk-stratification tools for nonsteroidal anti-inflammatory drug (NSAID) hypersensitivity have recently been developed: the I3A score, a pediatric ibuprofen-specific tool based on age and index reaction characteristics, and the CA₄TCH score, a broader multivariable tool applicable across NSAIDs and age groups. To externally validate the I3A score, and compare its performance with the CA₄TCH score. We retrospectively analyzed data of consecutive pediatric subjects (<18 years) presenting between February 2001 and December 2020 at the Allergy Unit of the University Hospital of Montpellier, with a history of hypersensitivity to ibuprofen, who underwent an ibuprofen challenge. The I3A and CA₄TCH scores were used to predict challenge outcomes, and their performance was evaluated by comparing predictions to the real challenge outcomes. Of the 103 subjects included, 16 (15.5%) had a positive challenge to ibuprofen. The I3A score had a sensitivity of 62.5% (95% CI, 37.5%-87.5%), a specificity of 88.5% (95% CI, 81.6%-95.4%), a negative predictive value of 92.8% (95% CI, 88.5%-97.4%), and a positive predictive value of 50.0% (95% CI, 33.3%-70.0%). It misclassified 6 positive subjects, among which were 1 subject with angioedema and 1 with anaphylaxis during the challenge. In their clinical history, these subjects had reported urticaria and anaphylaxis, respectively. In comparison, the CA₄TCH score had a sensitivity of 68.8% (95% CI, 43.8%-87.5%), a specificity of 89.7% (95% CI, 82.8%-95.4%), a negative predictive value of 94.0% (95% CI, 89.8%-97.6%), and a positive predictive value of 55.0% (95% CI, 38.9%-75.0%); it misclassified 5 positive subjects, all of whom experienced urticaria when challenged. The validation of the I3A score yielded similar specificity and negative predictive value as in the original derivation cohort, but substantially lower sensitivity and positive predictive value. Although I3A and CA₄TCH had similar performances in terms of quantitative measures, for the I3A score to generate a high-risk prediction, it is necessary (although not sufficient) that angioedema is reported in the clinical history, which may lead to undertriage when this symptom is absent. This highlights the need for larger derivation cohorts, which would enable construction of more stable, albeit more complex risk-stratification scores, incorporating more variables, various combinations of which may yield a positive prediction.