Improving the diagnosis of hypersensitivity reactions associated with sulfonamides

The differential diagnosis of idiosyncratic adverse drug reactions (ADRs) is difficult because symptoms are seldom pathognomonic. The authors compared the performance of two new diagnostic aids in the differential diagnosis of 27 cases of skin reactions associated with sulfonamide therapy. One test, the Bayesian Adverse Reaction Diagnostic Instrument (BARDI), calculates the posterior probability (PsP) of a drug being the etiology of an idiosyncratic ADR. The other, the lymphocyte toxicity assay (LTA), is a biochemical test that determines the percent of cell death because of toxic metabolites of a drug. Cases were assessed independently and blindly by BARDI and then the LTA. Skin reactions consisted of 19 exanthematous eruptions, 2 vasculitis, 1 erythema multiforme, and 5 urticarial rashes. BARDI indicated that the drug was very likely associated with 20 adverse events (PsP = 0.59 to 0.99); and unlikely to have caused 7 (PsP = 0.02 to 0.38). The LTA showed that 19 patients (70%) were test positive, and 8 were test negative. Comparing BARDI to LTA showed that BARDI had 79% sensitivity and 67% accuracy, but only 38% specificity. However, when the results of the LTA were incorporated into the calculation of the PsP based on this comparison, BARDI indicated that the drug was very likely associated with 18 reactions (PsP = 0.61 to 0.99) and unlikely to have caused 9 reactions (PsP = 0.003 to 0.47). The agreement increased from 67% to 96%. These findings indicate that the accuracy of diagnosis of idiosyncratic ADRs improved when the results of BARDI and LTA were combined. These findings also crossvalidate the results obtained by using BARDI.