Comparative Effectiveness of Rapid Versus Selective Antibiotic Initiation in Patients with Suspected Sepsis

Author(s):
Terra Hill; Lauren Kerivan; Nima Sarani; Steven Simpson; Christopher Guidry

Background:

Sepsis and septic shock are major health concerns with mortality rates over 10% and 40%, respectively. Adequate treatment involves the use of antibiotic therapy, however, literature differs in when therapy should be administered for optimal patient outcomes. This study aims to create a comparative effectiveness model using timing of antibiotic initiation to determine the optimal management of a critically ill patient with suspected sepsis.

Methods:

The model is created from the perspective of an Emergency Medicine team caring for a patient with suspected sepsis. Odds ratios and 95% confidence intervals describing the effects antibiotic timing on rates of progression to shock and rates of mortality in patients presenting with suspected sepsis were taken from the published literature. Two options of antibiotic timing were modeled within the initial 6 hours of triage: A) rapid initiation and B) selective initiation. Survival along with the most optimal management was modeled with an effectiveness of 1, while death was given an effectiveness of 0. Discounts were implemented for differing suboptimal outcomes. Rapid Initiation was defined as 3 hours or less of antibiotic administration and Selective Initiation was considered as more than 3 hours. Probabilistic sensitivity and two-way sensitivity analyses were performed to evaluate the effectiveness of each option while randomly sampling from the included distributions for odds ratios and rates of mortality of antibiotic delay and progression to shock. The model was created using TreeAge Pro Healthcare v2023 (TreeAge Software, Inc, Williamstown, MA).

Results:

After 100,000 iterations, Selective Initiation was the most effective in 64.5% of the cases. The two-way sensitivity analysis, however, demonstrated the likelihood of Selective Initiation being the most effective decreases as time progressed beyond 3 hours.

Conclusions:

Overall, the results of our comparative effectiveness model are in concordance with the current Surviving Sepsis Campaign recommendations, which allow for a short interval of selective antibiotic initiation to identify non-infectious causes of infection.