Is the Anamnesis Enough to De-Label Patients with Reported Beta-Lactam Allergy?

Background: Decision, whether to de-label patient with suspected BL hypersensitivity, is based on the risk stratification. The aim of the study was to prepare a characteristic of diagnostic risk groups and to create a model enabling the identification of the low-risk diagnostic group. Methods: We analyzed medical records of patients hospitalized due to suspected hypersensitivity to BL antibiotics. Based on the history data, patients were divided into three diagnostic risk groups, using the criteria proposed by Shenoy et al. Univariate and multivariate analysis model were used to create a diagnostic tool. Results: Among 263 patients referred for BL hypersensitivity diagnosis, 88 (33.5%) were qualified to group I, 129 (49%) to group II, and 46 (17.5%) to group III. There were significant differences between diagnostic risk groups regarding history of hypersensitivity to penicillins (p<0.001), cephalosporins (p<0.001), >1 BL (p<0.05), several episodes of BL hypersensitivity (p<0.001), medical intervention (p<0.001), documented hypersensitivity (p<0.001), time from drug intake to symptoms (p<0.001), time from hypersensitivity to diagnosis (p<0.001). 81 patients (30.8%) were de-labeled: 52 (59.8%) in group I, 27 (20.9%) in group II and 2 (4.3%) in group III. In univariate analysis, model of the low diagnostic risk group applied to the de-labeled part showed 90% specificity and 21.93% sensitivity. NPV and PPV were estimated at 72.04% and 49.53%, respectively. Multivariate model had high specificity but low sensitivity; its NPV was 76% with 68% PPV. Conclusions: The tool enabling the identification of low diagnostic risk patients based on anamnesis is not sensitive enough, to de-label patients on its basis.