Abstract
BACKGROUND: Despite antibiotic-induced anaphylaxis being a severe allergic reaction requiring immediate care, large-scale studies examining all antibiotic subtypes remain limited. This study addresses this gap by analysing 33 antibiotic classes, along with epidemiological and regional variations. METHODS: This study utilised the world's largest pharmacovigilance database, with over 35 million individual case safety reports from 140 countries. The study employed a two-step approach: first, the antibiotics were grouped into 10 categories according to their third-level ATC codes and their frequencies were collectively analysed. These categories comprised tetracyclines (J01A), amphenicols (J01B), beta-lactam antibacterial penicillins (J01C), other beta-lactam antibacterials (J01D), sulfonamides and trimethoprim (J01E), macrolides, lincosamides and streptogramins (J01F), aminoglycoside antibacterials (J01G), quinolone antibacterials (J01M), combinations of antibacterials (J01R) and other antibacterials (J01X). Second, a more detailed analysis was performed at the fourth level of the ATC codes for the antibiotics categorised at the third level, focusing on 33 individual antibiotics. For statistical analysis, disproportionality metrics, including the information component (IC) with IC(025) and reporting odds ratio (ROR) with 95% CI, were used to classify and analyse the risk of anaphylaxis related to these drugs. RESULTS: A total of 144,820 reports were identified as antibiotic-induced anaphylaxis. All antibiotics showed significant signal detection for anaphylaxis (ROR, 20.50 [95% CI, 20.37-20.63]; IC, 3.77 [IC(025), 3.76]) across all age groups and sexes. The following three antibiotics took the most proportion of the reports: penicillins (39,696/144,820 [27.4%]; ROR, 18.82 [95% CI, 18.62-19.01]; IC, 4.04 [IC(025), 4.02]), other beta-lactam antibiotics (63,644/144,820 [43.9%]; 27.59 [27.35-27.83]; 4.48 [4.46]) and quinolones (20,303/144,820 [14.0%]; 13.40 [13.21-13.60]; 3.63 [3.61]). The median time-to-onset was 1 day (interquantile range, 1-1), with most r recovered (96.09%) and the fatality rate accounting for 1.23%. CONCLUSION: Although our findings do not permit causal inference, the analysis highlights the need for standardised grading systems, patient-specific risk factors and long-term outcome studies to improve prevention and management.