Several immunomodulatory agents are used in the treatment of epidermal necrolysis, but evidence of their efficacy is limited. The Autonomous Community of Madrid has two reference burn units to which all patients with epidermal necrolysis are referred. One burn unit has mostly used cyclosporine (CsA), and the other has used non-CsA therapies (mainly high-dose intravenous immunoglobulin). The allocation of patients to one or the other burn unit was mainly based on proximity, resembling a random assignment. Thus, we took advantage of this "natural experiment" to estimate the mortality risk ratio (MRR) of CsA (n = 26) compared with non-CsA (n = 16) treatment using hospital as an instrumental variable over the period from 2001 to 2015. We also computed the observed versus expected (O/E) MRR in a case series of 49 CsA-treated patients (including 23 patients from other regions treated in Madrid), and using the Score for Toxic Epidermal Necrolysis (i.e., SCORTEN) scale to estimate the expected values. The instrumental variable-based MRR of CsA versus non-CsA was 0.09 (95% confidence interval = 0.00-0.49). The O/E analysis also showed a reduction in mortality risk (MRROE = 0.42; 95% confidence interval = 0.14-0.99). We identified five other case series of CsA-treated patients providing MRROE and meta-analyzed their results. The pooled MRROE (including from this study) was 0.41 (95% confidence interval = 0.21-0.80). All three approaches consistently show that CsA reduces the mortality in epidermal necrolysis patients.
Copyright © 2017 The Authors. Published by Elsevier Inc. All rights reserved.