Since the 2002 SCAR study, erythema multiforme (EM), a post-infectious disease, has been distinguished from Stevens-Johnson syndrome (SJS), drug-induced. Nevertheless, EM cases are still reported in the French pharmacovigilance database (FPDB).

To describe EM reported in the FPDB and to compare the quality and the characteristics of the reports.

This retrospective observational study selected all EM cases reported in the FPDB over two periods: period 1 (P1, 2008–2009) and period 2 (P2, 2018–2019). Inclusion criteria were 1) a diagnosis of clinically typical EM and/or validated by a dermatologist; 2) a reported date of onset of the reaction; and 3) a precise chronology of drug exposure. Cases were classified confirmed EM (typical acral target lesions and/or validation by a dermatologist) and possible EM (not-otherwise-specified target lesions, isolated mucosal involvement, doubtful with SJS). We concluded possible drug-induced EM when EM was confirmed, with onset ranging from 5 to 28 days without an alternative cause.

Among 182 selected reports, 140 (77%) were analyzed. Of these, 67 (48%) presented a more likely alternative diagnosis than EM. Of the 73 reports of EM cases finally included (P1, n=41; P2, n=32), 36 (49%) had a probable non-drug cause and 28 (38%) were associated with only drugs with an onset time ≤4 days and/or ≥29 days. Possible drug-induced EM was retained in 9 cases (6% of evaluable reports). Etiological work-up was more often performed in period 2 than 1 (53.1% vs 29.3%, P=0.04), and the time to onset from 5 to 28 days was more frequent in period 2 (59.2% vs 40%, P=0.04).

This study suggests that possible drug-induced EM is rare. Many reports describe “polymorphic” rashes inappropriately concluded as EM or post-infectious EM with unsuitable drug accountability subject to protopathic bias.