Towards Understanding the Effective Use of Antibiotics for Sepsis

BACKGROUND: The benefits of early antibiotics for sepsis have recently been questioned. Evidence for this mainly comes from observational studies. The only randomized trial on this subject, the Prehospital Antibiotics Against Sepsis (PHANTASi) trial, did not find significant mortality benefits from early antibiotics. That subgroups of patients benefit from this practice is still plausible, given the heterogeneous nature of sepsis. RESEARCH QUESTION: Do subgroups of sepsis patients experience 28-day mortality benefits from early administration of antibiotics in a prehospital setting? And what key traits drive these benefits? STUDY DESIGN AND METHODS: We used machine learning to conduct exploratory partitioning cluster analysis to identify possible subgroups of sepsis patients who may benefit from early antibiotics. We further tested the influence of several traits within these subgroups, using a logistic regression model. RESULTS: We found a significant interaction between age and benefits of early antibiotics (P = .03). When we adjusted for this interaction and several other confounders, there was a significant benefit of early antibiotic treatment (OR, 0.07; 95% CI, 0.01-0.79; P = .03). INTERPRETATION: An interaction between age and benefits of early antibiotics for sepsis has not been reported before. When validated, it can have major implications for clinical practice. This new insight into benefits of early antibiotic treatment for younger sepsis patients may enable more effective care. CHEST 2021; 160(4):1211-1221

Automated summary

The study investigated whether subgroups of sepsis patients benefit from early administration of antibiotics in a prehospital setting and which key traits drive these benefits. An interaction between age and benefits of early antibiotics for sepsis was found, which has not been reported before. This new insight may have major implications for clinical practice in providing more effective care for younger sepsis patients.