International Case-Control Study of Toxic Epidermal Necrolysis (TEN) and Stevens-Johnson Syndrome (SJS)

Objectives

The aim of the study was to provide quantitative estimates of the risk of TEN and SJS among the users of various drugs. The conditions are rare, but frequently caused by drugs. Numerous drugs have been implicated, based mostly on case reports, and there are no satisfactory risk estimates.

Methods

The study was conducted in France, Germany, Italy, and Portugal, with Slone acting as coordinating center. Cases were identified by study personnel from hospitals in the regions; all patients diagnosed with TEN, SJS, or erythema multiforme major were potentially eligible. All cases were reviewed without knowledge of drug use by an independent committee of dermatologists, meeting at regular intervals, to confirm the diagnoses. Cases were classified on a continuum between EM and TEN, in the following categories: EM, EM/SJS, SJS, overlap SJS/TEN, TEN with “spots,” pure plaque TEN, and “other.” Controls were other hospital patients with acute conditions (e.g., pneumonia, trauma, appendicitis) or nonacute conditions (e.g., hernia, hemorrhoids) not associated with drug use. A minimum of two controls were enrolled for each case, matched according to age, sex, and region. A total of 800 cases of TEN, SJS, and EM and 1,950 hospital controls were included in the final dataset.

All subjects were interviewed in hospital by trained study personnel,mainly physicians trained in dermatology. The information included demographic data, details of the clinical course of the patient’s illness up to the time of hospital admission, relevant past medical history, habits such as alcohol consumption, and a detailed history of drug use in the month before hospital admission. The data were analyzed using standard case-control methods.

Results

Quantitative estimates of risk have been provided for numerous drugs. Although relative risks were large, ranging up to 156 for sulfonamides, excess risks were low because SJS/TEN is a rare occurrence. The risk for drugs that are taken longterm, particularly anticonvulsants, was mostly confined to the first few weeks of therapy. The study has also made an important contribution to the nosology of severe cutaneous reactions.