COVID-19 COVID-19 Evidence Alerts from McMaster PLUS is a new service that alerts users to current best evidence for clinical care of people with threatened, suspected or confirmed COVID-19 infection.
Visit the site Sign up for alerts
Rothrock SG, Cassidy DD, Barneck M, et al. Outcome of Immediate Versus Early Antibiotics in Severe Sepsis and Septic Shock: A Systematic Review and Meta-analysis. Ann Emerg Med. 2020 Jun 24. pii: S0196-0644(20)30337-1. doi: 10.1016/j.annemergmed.2020.04.042. (Systematic review)

STUDY OBJECTIVE: Debate exists about the mortality benefit of administering antibiotics within either 1 or 3 hours of sepsis onset. We performed this meta-analysis to analyze the effect of immediate (0 to 1 hour after onset) versus early (1 to 3 hours after onset) antibiotics on mortality in patients with severe sepsis or septic shock.

METHODS: This review was consistent with the Preferred Reporting Items for Systematic Reviews and Meta-analyses guidelines. Searched databases included PubMed, EMBASE, Web of Science, and Cochrane Library, as well as gray literature. Included studies were conducted with consecutive adults with severe sepsis or septic shock who received antibiotics within each period and provided mortality data. Data were extracted by 2 independent reviewers and pooled with random effects. Two authors independently assessed quality of evidence across all studies with Cochrane's Grading of Recommendations Assessment, Development and Evaluation methodology and risk of bias within each study, using the Newcastle-Ottawa Scale.

RESULTS: Thirteen studies were included: 5 prospective longitudinal and 8 retrospective cohort ones. Three studies (23%) had a high risk of bias (Newcastle-Ottawa Scale). Overall, quality of evidence across all studies (Grading of Recommendations Assessment, Development and Evaluation) was low. Pooling of data (33,863 subjects) showed no difference in mortality between patients receiving antibiotics in immediate versus early periods (odds ratio 1.09; 95% confidence interval 0.98 to 1.21). Analysis of severe sepsis studies (8,595 subjects) found higher mortality in immediate versus early periods (odds ratio 1.29; 95% confidence interval 1.09 to 1.53).

CONCLUSION: We found no difference in mortality between immediate and early antibiotics across all patients. Although the quality of evidence across studies was low, these findings do not support a mortality benefit for immediate compared with early antibiotics across all patients with sepsis.

Discipline Area Score
Infectious Disease 7 / 7
Emergency Medicine 6 / 7
Internal Medicine 5 / 7
Intensivist/Critical Care 5 / 7
Comments from MORE raters

Emergency Medicine rater

Although this study did not show a difference in outcomes with antibiotic administration at 1 vs 3 hours, this should not be taken as license to delay antibiotic administration in septic patients.

Infectious Disease rater

This thoughtful and scholarly meta-analysis is important and overdue, viewing the low threshold to give immediate empiric antimicrobial in U.S. EDs before meaningful workup has been initiated, in fear of not being in compliance with institutional and Federal sepsis guidelines. The Surviving Sepsis campaign would be well advised to consider adoption of the findings of this study in their international guideline. They would also be well advised to put far greater emphasis on an expedited workup of at the outset, to determine whether there indeed is a high likelihood of sepsis and to identify the probable source, before commencing initial antimicrobial therapy, with the goal of minimizing unnecessary antimicrobial therapy and increasing targeted as contrasted with empiric therapy.

Intensivist/Critical Care rater

This was a meta-analysis of low GRADE quality of evidence studies. I am not a meta-analysis expert - but I am not sure that the 2011 and 2014 data should be combined with with the 18/19 data as practice may have changed in this time frame.

Internal Medicine rater

It is true that, in this clinical setting, a randomized controlled trial is unfeasible, but one will not be able to address a cause-effect relationship using a cohort study or making a metanalysis of (mostly) cohort studies.
Comments from EvidenceAlerts subscribers

No subscriber has commented on this article yet.