Prolonged Versus Intermittent Infusion of β-Lactam Antibiotics: A Systematic Review and Meta-Regression of Bacterial Killing in Preclinical Infection Models

Abstract

Background: Administering β-lactam antibiotics via prolonged infusions for critically ill patients is mainly based on preclinical evidence. Preclinical data on this topic have not been systematically reviewed before. Objectives: The aim of this study was to describe the pharmacokinetic/pharmacodynamic (PK/PD) indices and targets reported in preclinical models and to compare the bactericidal efficacy of intermittent and prolonged infusions of β-lactam antibiotics. Methods: The MEDLINE and EMBASE databases were searched. To compare the bactericidal action of β-lactam antibiotics across different modes of infusion, the reported PK/PD outcomes, expressed as the percentage of time (T) that free (f) β-lactam antibiotic concentrations remain above the minimal inhibitory concentration (MIC) (%fT>MIC) or trough concentration (Cmin)/MIC of individual studies, were recomputed relative to the area under the curve of free drug to MIC ratio (fAUC24/MIC). A linear mixed-effects meta-regression was performed to evaluate the impact of the β-lactam class, initial inoculum, Gram stain, in vivo or in vitro experiment and mode of infusion on the reduction of bacterial cells (in colony-forming units/mL). Results: Overall, 33 articles were included for review, 11 of which were eligible for meta-regression. For maximal bactericidal activity, intermittent experiments reported a PK/PD target of 40–70% fT>MIC, while continuous experiments reported a steady-state concentration to MIC ratio of 4–8. The adjusted effect of a prolonged as opposed to intermittent infusion on bacterial killing was small (coefficient 0.66, 95% confidence interval − 0.78 to 2.11). Conclusions: Intermittent and prolonged infusions of β-lactam antibiotics require different PK/PD targets to obtain the same level of bacterial cell kill. The additional effect of a prolonged infusion for enhancing bacterial killing could not be demonstrated.