LAUSR

Corticosteroids have been used for more than 60 years to treat various forms of severe infection. Yet, corticotherapy remains one of the most controversial treatments for sepsis. The past 2 years, guidelines for clinical practices about corticosteroids for sepsis have been released by at least five entities (1–5). All but one (2) of these guidelines recommended the use of corticosteroids only in patients with septic shock who are poorly responsive to fluid replacement and vasopressor therapy. One group suggested that corticosteroids should be given in patients with sepsis (weak recommendation) with or without shock (2). In 2018, five systematic reviews addressing the effects of corticosteroids in sepsis were published (6–10). The number of included trials differed between systematic reviews and ranged from 14 to 42. The relative risk (RR) of death in the short-term varied from 0.91 to 0.96 and the upper limit of its 95% CI varied from 0.98 to 1.03. Then, the magnitude and direction of the pooled RR of dying in the short-term favoring corticosteroids were consistent across these five meta-analyses which differed mainly by the presence of some imprecision in the point estimate. Among the trials of corticosteroids for sepsis, five trials deserve specific attention for three main reasons (11–15). First, they account for more than 60% of all participants in trials of corticosteroids for sepsis. Second, they were at moderate to low risk of bias in the five domains of allocation, blinding, incomplete data, selective reporting, and other potential sources of bias. Third, they were conducted after the first consensus definition for sepsis (16) and their target population was septic shock. Two of these trials found survival benefits from corticosteroids (11, 15) and three did not (12–14), and will be referred to as “positive” and “negative” trials, respectively. This article explores potential explanations for discrepancies in trials’ findings in four domains, that is, study population, experimental interventions, outcomes, and risk of bias.