Accuracy of a pharmacovigilance algorithm in diagnosing drug hypersensitivity reactions.

BACKGROUND This study was performed to evaluate the diagnostic accuracy of a pharmacovigilance algorithm in patients with 1 or more histories suggestive of drug hypersensitivity. METHODS We performed a retrospective analysis of a clinic case series. We analyzed patients with suspected clinical reactions of drug hypersensitivity. Patients with severe skin reactions were excluded. Patients with history of drug allergy were subjected to additional testing to validate this history. Following a detailed clinical history, skin tests were performed. If skin tests were not available or validated, drug provocation tests were conducted. Assessment of causality was established by an investigator unaware of drug testing results using a pharmacovigilance algorithm that was then compared with the final diagnosis. RESULTS A total of 677 consecutive patients with 1001 reactions were analyzed. No score could be given because of the absence of 1 of the criteria required for 204 reactions (20.4%). For 720 reactions (71.9%), a dubious causality assessment score was given. Drug hypersensitivity was confirmed by drug testing in 175 reactions (17.5%) and eliminated in 826 reactions (82.5%). Sensitivity of the algorithm was 10.3% and specificity was 76.9%. Although there were 1.7% false-positive scores, there were no false-negative scores. The logistic regression that was performed to look for independent clinical risk factors linked to the drug hypersensitivity diagnosis found 3 parameters: likely causality assessment score, drug reintroduction in clinical history, and delay between reaction and last drug intake of less than 1 hour. CONCLUSION A pharmacovigilance algorithm is not accurate for the diagnosis of drug hypersensitivity reactions and cannot replace drug allergy testing.