Accuracy of SCORTEN to predict the prognosis of Stevens-Johnson syndrome/toxic epidermal necrolysis: a systematic review and meta-analysis

Background The SCORTEN score is a specific predictor of the probability of death for patients diagnosed with Stevens-Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN). Objectives To evaluate the overall accuracy of SCORTEN and the influence of several moderators such as age, sex, geographical region and age of the study. Methods A systematic search was performed on MEDLINE, The Cochrane Library, EMBASE, SCOPUS and Web of Knowledge, with no restriction on language (last update 5 February 2019 for all databases). Original studies on the use of SCORTEN were eligible. The standardized mortality ratio (SMR), defined as the quotient between the number of deaths observed and the number expected following SCORTEN, was taken as the measurement of analysis. Results Sixty-four papers were part of the main meta-analysis carried out in the study. A pooled log(SMR) of -0.0889 (95% CI: -0.2023 to 0.0245) was obtained, suggesting a reasonable behaviour of SCORTEN as a predictor of mortality. The possible influence of several factors in the accuracy of SCORTEN was studied by means of meta-regression models. Multivariate meta-regression allowed finding that the mean age of the patients and the ending year of the study are the only factors that significantly influence SCORTEN predictions. The mean age of the group of patients was associated with a significant increase in the observed/expected ratio, whereas a progressive reduction in the observed/expected ratio can be appreciated over the years. Finally, an underestimation of mortality was found for SCORTEN values of 3 or less and the opposite for those above 3 (SCORTEN range: 0-7). Conclusions The rarity of the disease and the heterogeneity of the studies included are major limitations. Despite the overall remarkable accuracy displayed by SCORTEN, the influence of several factors, as comorbidities (e.g. renal impairment), involved body surface area and patient's age, seem of enough relevance to consider a redefinition of the scale.