Spontaneous adverse event reports of Stevens-Johnson syndrome/toxic epidermal necrolysis: detecting associations with medications.

PURPOSE: Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are medically serious skin reactions that are often drug induced. The mainstay of therapy and future prevention is to discontinue and avoid the use of the suspected inducing drug. However, many cases of SJS/TEN occur in patients who are taking multiple medications, and it is often difficult to determine which drug to stop. This analysis was conducted to identify drugs that were most associated with SJS/TEN in the US Food and Drug Administration (FDA) Adverse Event Reporting System (AERS) database and to identify medications that were likely innocent bystanders.
METHODS: A Multi-item Gamma Poisson Shrinker value with an EB05 ≥ 2 was considered a disproportional increase in reporting frequency (at least two times higher than expected). The identified drugs with reporting frequency of SJS/TEN in the US FDA AERS database were then compared to the EuroSCAR (European case-control surveillance of severe cutaneous adverse reactions) study results as a reference to define signals. The EB05s were calculated as a cumulative relative reporting frequency from 1968 to 3Q2009.
RESULTS: Fifty drugs were identified as being associated with SJS/TEN. This included 12 "highly suspect" drugs and 36 "suspect" drugs. Meloxicam was the only drug that appeared on the "highly suspect" list from EuroSCAR that did not show a disproportional increase in relative reporting frequency (EB05 = 0.734). In addition, several drugs did not have an association with SJS/TEN (EB05 < 2).
CONCLUSIONS: There was good concordance between the reporting frequencies observed in the FDA AERS database and the published risk estimation of medications implicated in SJS/TEN.