Résumé
Objectives: Delabeling algorithms for unexplored beta-lactam (BL) drug hypersensitivity reaction (DHR) suspicions in patients with ur-gent need of BL administration have been published in recent years. The aim of this study was to evaluate the safety (i.e. the risks as-sociated with BL exposure) of two algorithms that were applied on a retrospective cohort of patients with positive and negative drug allergy work-up to penicillins.Method: Data of patients who underwent a drug allergy workup for the identification of a penicillin DHR between January 2014 and December 2021 at the Allergy Unit of the University Hospital Montpellier were extracted from the Drug Allergy and Hypersensitivity Database (DAHD). According to their clinical history, these patients were independently stratified into risk groups, following two al-gorithms, proposed by Blumenthal K. et al 2015 and Romano A. et al 2020.Results: In total, 875 patients were included in a stepwise manner. About 44.2% had a positive drug allergy work-up, either by means of skin testing or drug provocation test. The most frequent index clini-cal history was maculopapular exanthema (321, 36.9%), followed by anaphylaxis with (106, 12.1%) or without shock (145, 16.5%) and urticaria/angioedema (170, 19.5%). Each algorithm comprised three risk classes of immediate/delayed reactions with different sever-ity. For the Blumenthal algorithm, 56% of patients were classified as high-risk, IgE-mediated reactions, while 37% as low-risk, delayed reactions. For the Romano algorithm, 50.4% were classified in the high-risk immediate reactions and 42.5% in the low-risk immediate/delayed reactions. Both algorithms classified the rest of 6.9% of pa-tients as high-risk delayed reactions. The algorithms failed in terms of safety for 10.5% of the patients in the Blumenthal algorithm and 0.5% in the Romano algorithm.Conclusions: The algorithm proposed by Romano et al was safer, fail-ing in only 0.5% of the patients, in a population tested in Southern France for penicillin allergy suspicion.