Impact of multiple medical interventions on mortality, length of hospital stay and reepithelialization time in Toxic Epidermal Necrolysis, Steven-Johnsons Syndrome, and TEN/SJS Overlap - Metanalysis and metaregression of observational studies

Abstract

Stevens-Johnson's Syndrome (SJS) and Toxic Epidermal Necrolysis are rare, life-threatening dermatologic conditions with acute onset and not clearly established treatment protocol. A plethora of observational studies are present with lack of up-to-date consensus based on evaluation of objective endpoints, among others mortality. Thorough analysis of available databases (Pubmed, EMBASE, Cinahl, Web of Science, Clinical Trials) was conducted according to PRISMA guidelines. Authors initially identified 700 papers, with 82 of them potentially eligible according to adopted criteria. A total of 42 studies were included into pooled synthesis. For continuous outcomes we analyzed the pooled means for endpoint scores using observed cases data. Categorical outcomes were analyzed by calculating the pooled event rates. We conducted subgroup and exploratory maximum likelihood random effects meta-regression analyses regarding SCORTEN of all outcomes. Using random-effects model, the overall pooled Mortality Rate was 0.191 (95%CI, 0.132-0.269). The lowest mortality rate was found to be linked with Etanercept and highest in Total Plasma Exchange (TPE) and Intravenous Immunoglobulin (IVIG). Overall reepithelization was 13.278 days (95%CI, 8.773-17.784),The highest was found in cyclosporine treatment; 14.739 whilst the lowest for steroids. Length of hospital stay in overall analysis was 19.99 days (95%CI, 16.53-23.44),the highest was linked with TPE/TPE+IvIg treatment, the lowest with steroids. Risk of bias of assessed studies was estimated to be high (for observational studies mean STROBE score 12.44). High quality TEN and SJS studies are lacking. Almost all papers report observational data without randomization and double-blind control. Therefore, the pooled analysis cannot be presented with initial bias. In our meta-analysis the most successful regimen was Etanercept treatment. It was linked with the lowest mortality. The most negative treatment outcome was observed in studies reporting TPE and IVIG. Randomized trials of high quality are needed in SJS and TEN.

Keywords: Autoinflammatory disorders; Cytokines; Drug reaction; Outcome measurement; SJS.