LAUSR

Background The effects of intravenous corticosteroids in patients with sepsis remain controversial due to mixed results from randomized trials. Moreover, updated definitions of sepsis, Sepsis-3, were proposed in 2016, and findings related to the effects of corticosteroids in patients defined by the Sepsis-3 criteria are scarce. Objective To investigate the effectiveness of corticosteroids in patients with sepsis or septic shock using real-world data to complement the findings of randomized controlled trials, and to determine whether the treatment effects differ by sepsis definitions. Methods We conducted this study by utilizing a large, multi-center healthcare database, eICU, in which we identified patients with sepsis admitted to 208 intensive care units across the US from 2014 to 2015 based on two different definitions: prior explicit definitions (i.e., based on diagnosis codes) and the Sepsis-3 definitions (i.e., based on SOFA score). The association between intravenous corticosteroids and in-hospital survival up to 50 days in patients with sepsis was retrospectively analyzed. A parametric hazard model with stabilized inverse probability of treatment weight adjustment was used to control for baseline confounders. Results Of the 7,158 patients identified based on the explicit definition, 562 (7.9%) received corticosteroids; of the 5,009 patients identified based on the Sepsis-3 definition, 465 (9.3%) received corticosteroids. In the explicit cohort, adjusted in-hospital survival at day 50 was 0.62 in the treated vs 0.57 in the non-treated, with a survival difference of 0.05 (95%CI: -0.11, 0.17). Similar results were seen in the Sepsis-3 cohort (0.58 vs 0.56 in treated and non-treated, respectively), with a 50-day survival difference of 0.02 (95%CI: -0.19, 0.17). Conclusions In patients with sepsis or septic shock, intravenous corticosteroids were not associated with a higher in-hospital survival up to 50 days regardless of the sepsis definitions. Further research may be necessary to definitively confirm effectiveness in real-world practice.