What place for pragmatic RCTs in intensive care?

Pragmatic trials test the effectiveness of an intervention within real-life clinical practice, prioritising applicability and generalisability. By contrast, explanatory trials evaluate the efficacy of an intervention under optimal conditions and in well-defined populations, aiming to minimise bias and confounding and to maximise the treatment effect. The majority of large-scale, multicentre academic ICU studies are overall pragmatic in nature but have repeatedly failed to demonstrate outcome differences against control populations. This relates in part to the marked heterogeneity of ICU populations, often with enrolment based on syndromic categories such as sepsis, acute respiratory distress syndrome, shock and acute kidney injury with significant inter-individual biological variability. Any beneficial treatment effects in specific, often unidentified, subsets are diluted or cancelled out by no effect or even harm in others. A second important factor underlying the lack of intervention effect is suboptimal performance of the trials, particularly poor protocol adherence, which undermines the likelihood of showing benefit, particularly when reliant on intention-to-treat analyses. In this article we discuss these issues and provide some suggestions for improvement.