ObjectiveTo determine the efficacy and safety of using natural platelet-activating factor receptor antagonist (PAFra), BN 52021, to treat patients with severe Gram-negative bacterial sepsis.DesignA prospective, randomized, double-blind, placebo-controlled, multicenter clinical trial.SettingFifty-nine academic medical center intensive care units in Europe.PatientsSix hundred nine patients with severe sepsis, suspected to be related to Gram-negative bacterial infection, who received PAFra or placebo.InterventionsPatients were randomized to receive either a dose of PAFra (120 mg iv) every 12 hrs over a 4-day period or placebo over a 4-day period.Measurements and Main ResultsThe patients were well matched at study entry for severity of illness and for risk factors known to influence the outcome of sepsis. Among all randomized patients, the 28-day, all-cause mortality rate was 49% (152/308) in the placebo group, and 47% (140/300) in the PAFra group (p = .50). When analyzed on the basis of the previously defined target population, the 28-day, all-cause mortality rate was 50% (115/232) in the placebo group and 44% (94/212) in the PAFra group, yielding a 12% reduction in mortality rate (p = .29). In patients with documented infection involving other organisms, there was no difference between treated and placebo groups. When the outcomes of organ dysfunctions were examined in the overall population and in the documented Gram-negative bacterial infection population, the number of patients who resolved hepatic dysfunction tended to be higher in the treated group than in the placebo group (p = .06). The number of adverse events reported were not different between the two groups.ConclusionsA 4-day administration of the studied PAFra (BN 52021) failed to demonstrate a statistically significant reduction in the mortality rate of patients with severe sepsis suspected to be related to Gram-negative bacterial infection. If PAFra treatment has any therapeutic activity in severe Gram-negative bacterial sepsis, the incremental benefits are small and will be difficult to demonstrate in a patient population as defined by this clinical trial. (Crit Care Med 1998; 26:1963-1971)