A causal analysis of the impact of early antibacterial therapy on ICU patient outcomes for sepsis

This paper addresses the global issue of sepsis, a major cause of ICU-related deaths, by focusing on the impact of early antibiotic administration on patient outcomes. The Surviving Sepsis Campaign guidelines recommend timely antimicrobial treatment, but the association with mortality remains uncertain, and other outcomes like hospital and ICU stay length are understudied. We employ the Double Machine Learning (DML) method and the eICU Collaborative Research Database to analyze the relationship between early antibiotics and patient outcomes. Although initial summary statistics imply a substantial influence of the timing of antibacterial therapy on patient outcomes, indicating earlier treatment with reduced mortality rates and shorter hospital and ICU stays, our causal model refines these results. It shows that quicker initiation of therapy primarily affects hospital and ICU length of stays, while showing no significant impact on mortality rates. Given the substantial hospital costs associated with septicemia, the findings hold implications for cost savings.