Gram negative bloodstream infections (GN BSI) are a leading cause of mortality worldwide, and antibiotic treatment approaches remain understudied. BALANCE+ is a perpetual Bayesian adaptive platform trial to test multiple treatment questions for hospitalised patients with GN BSI. The vanguard phase objective was to test the feasibility of the main trial.

Adaptive platform trial with five initial domains of investigation, each with open label 1:1 randomisation.

Ten hospitals across four Canadian provinces.

Individuals admitted to hospital with blood cultures yielding Gram negative bacteria.

The five initial domains of investigation included: antibiotic de-escalation versus no de-escalation; oral transition to beta-lactam versus non-beta-lactam treatment; routine versus no routine follow-up blood cultures (FUBCs); central vascular catheter replacement versus retention; and, ceftriaxone versus carbapenem treatment for low risk AmpC organisms.

Domain-specific recruitment rates and protocol adherence.

During the vanguard phase, 719 patients were screened, of whom 563 (78.3%) were eligible, with 179 (31.8%) enrolled into the platform. The platform recruitment rate was 1.37 patients/site-week. Recruitment varied by domain: routine versus no FUBC domain 1.23 patients/site-week; oral beta-lactam versus non-beta-lactam domain 0.48; de-escalation versus no de-escalation domain 0.28; low risk AmpC domain 0.02; catheter replacement versus retention domain 0.01. Domain specific protocol adherence rates were 145/158 (91.8%) for routine versus no routine FUBC, 53/60 (88.3%) for oral beta-lactam versus non-beta-lactam, 26/33 (78.8%) for de-escalation versus no de-escalation, 3/3 (100%) for low risk AmpC, and 0/1 (0%) for line replacement versus retention. There was complete ascertainment of all study outcomes in hospital 170/170 (100%) and near complete ascertainment at 90 days 162/170 (95.3%).

The vanguard phase demonstrated overall trial feasibility by recruitment rate and protocol adherence, with differences across interventions, leading to a transition to the main BALANCE+ platform trial with minimal protocol modifications.

NCT05893147.