A Critical Analysis of the Literature on Time-to-Antibiotics in Suspected Sepsis

Abstract

Abstract The Surviving Sepsis Campaign recommends immediate antibiotics for all patients with suspected sepsis and septic shock, ideally within 1 hour of recognition. Immediate antibiotic treatment is lifesaving for some patients, but a substantial fraction of patients initially diagnosed with sepsis have noninfectious conditions. Aggressive time-to-antibiotic targets risk promoting antibiotic overuse and antibiotic-associated harms for this subset of the population. An accurate understanding of the precise relationship between time-to-antibiotics and mortality for patients with possible sepsis is therefore critical to finding the best balance between assuring immediate antibiotics for those patients who truly need them versus allowing clinicians some time for rapid investigation to minimize the risk of overtreatment and antibiotic-associated harms for patients who are not infected. More than 30 papers have been published assessing the relationship between time-to-antibiotics and outcomes, almost all of which are observational cohort studies. Most report significant associations but all have important limitations. Key limitations include focusing just on the sickest subset of patients (only patients requiring intensive care and/or patients with septic shock), blending together mortality estimates from patients with very long intervals until antibiotics with patients with shorter intervals and reporting a single blended (and thus inflated) estimate for the average increase in mortality associated with each hour until antibiotics, and failure to control for large potential confounders including patients’ presenting signs and symptoms and granular measures of comorbidities and severity of illness. In this study, we elaborate on these potential sources of bias and try to distill a better understanding of what the true relationship between time-to-antibiotics and mortality may be for patients with suspected sepsis or septic shock.