Résumé
A risk stratification tool for nonsteroidal anti-inflammatory drugs (NSAIDs) hypersensitivity is currently lacking.
To develop and validate a risk stratification tool for NSAID hypersensitivity.
We conducted a retrospective study of subjects presenting between February 2001 and December 2020 at the Allergy Unit of the University Hospital of Montpellier, with a history of hypersensitivity to NSAIDs. A risk score was generated for the prediction of a positive drug provocation test (DPT) based on clinical history elements, using a data-driven methodology. We performed external validation on a cohort of 69 subjects from the University Hospital of Tours.
Of the 1,304 subjects included, 232 had at least one positive DPT and 803 had negative DPTs to reported culprit NSAIDs. The variables selected in the risk score, which will be referred to by the acronym CA4TCH, consisted of the reaction Chronology, the semiology including Anaphylaxis or respiratory symptoms, Aspirin as culprit, underlying Asthma, Atopy, Time from the reaction to the DPT, the number of culprit Chemical classes, and chronic spontaneous urticaria (Hives). CA4TCH had a sensitivity of 78.4% (95% CI, 69.8-85.3) and a specificity of 70.4% (95% CI, 64.8-79.3) with a positive predictive value of 43.3% (95% CI, 39.7-49.5), and a negative predictive value of 91.9% (95% CI, 89.7-94.2), at an outcome prevalence of 22.4%. It correctly de-labeled 565 negative subjects (70.4%) and correctly classified all positive subjects with NSAID-exacerbated respiratory and cutaneous disease. External validation yielded a similar performance.
CA4TCH enables personalized predictions at patient presentation and aligns with specialist observations and European guidelines recommendations regarding risks and symptom severity associated with previously defined phenotypes, currently used for patient triage.