When biology tells usexactly where to look —listening changes everything.

A phase 3 Bayesian RCT enrolling adults with septic shock, multiorgan dysfunction, and endotoxin activity (EAA) between 0.60 and 0.89 units — the "high but not maximal" endotoxin phenotype identified in the EUPHRATES trial. Patients were randomized 2:1 to two sessions of polymyxin B haemoadsorption plus standard care versus standard care alone.

The study used a prespecified Bayesian analysis that formally pooled evidence from both EUPHRATES and TIGRIS, making this the first prospective validation of the enriched endotoxin endotype in septic shock research.

Parallel-group, open-label RCT across 20 NHS emergency departments. The intervention arm used a procalcitonin-guided algorithm combined with standard NEWS2 scoring to guide early sepsis identification and antibiotic initiation decisions. The control arm followed NEWS2-based usual care alone.

Three pre-specified interim analyses with group-sequential stopping rules for both effectiveness and futility. Co-primary endpoints: antibiotic initiation rate and 28-day all-cause mortality.

Multicenter double-blind placebo-controlled RCT in 9 ICUs enrolling non-intubated critically ill adults with hyperactive delirium. Patients received continuous IV dexmedetomidine or placebo for a minimum of 36 hours.

The primary endpoint was a joint composite model incorporating agitation duration, delirium duration, and need for intubation with deep sedation — capturing the full clinical burden of the hyperactive delirium phenotype in the non-intubated patient, a population historically excluded from most sedation trials.