A Systematic Review of the Effect of Delayed Appropriate Antibiotic Treatment on the Outcomes of Patients With Severe Bacterial Infections.

BACKGROUND: Patients with severe bacterial infections often experience delay in receiving appropriate treatment. Consolidated evidence of the impact of delayed appropriate treatment is needed to guide treatment and improve outcomes.

RESEARCH QUESTION: What is the impact of delayed appropriate antibacterial therapy on clinical outcomes in patients with severe bacterial infections?

STUDY DESIGN AND METHODS: Literature searches of MEDLINE and Embase, conducted on July 24, 2018, identified studies published after 2007 reporting the impact of delayed appropriate therapy on clinical outcomes for hospitalized adult patients with bacterial infections. Where appropriate, results were pooled and analyzed with delayed therapy modeled three ways: delay vs no delay in receiving appropriate therapy; duration of delay; and inappropriate vs appropriate initial therapy. This article reports meta-analyses on the effect of delay and duration of delay.

RESULTS: The eligibility criteria were met by 145 studies, of which 37 contributed data to analyses of effect of delay. Mortality was significantly lower in patients receiving appropriate therapy without delay compared with those experiencing delay (OR, 0.57; 95% CI, 0.45-0.72). Mortality was also lower in the no-delay group compared with the delay group in subgroups of studies reporting mortality at 20 to 30 days, during ICU stay, or in patients with bacteremia (OR, 0.57 [95% CI, 0.43-0.76]; OR, 0.47 [95% CI, 0.27-0.80]; and OR, 0.54 [95% CI, 0.40-0.75], respectively). No difference was found in time to appropriate therapy between those who died and those who survived (P = .09), but heterogeneity between studies was high.

INTERPRETATION: Avoiding delayed appropriate therapy is essential to reduce mortality in patients with severe bacterial infections.

CLINICAL TRIAL REGISTRATION: PROSPERO; No.: CRD42018104669; URL: www.crd.york.ac.uk/prospero/.