The impact of an institutional sepsis guideline on selecting appropriate empirical treatment in patients with carbapenem-resistant gram-negative bacilli bacteremia.

Objective: We aimed to investigate the impact of our institutional sepsis protocol on the empirical treatment of carbapenem-resistant Gram-negative bacteria in a setting where infectious disease consultation (ID) is available 7 days / 24 h and broad-spectrum antibiotic use requires ID approval.

Methods: A total of 612 patients (168 patients pre-guideline, 444 patients post-guideline) who received empirical antibiotics for suspicion of sepsis before documentation of antibacterial susceptibility were included. Demographic, clinical and microbiological data were collected from the hospital's electronic medical record system, retrospectively. Compliance with institutional guidelines and the rate of appropriate antibiotic use prior to the availability of antibiograms were assessed.

Results: There was a statistically significant increase in the utilization rate of empirical antibacterial treatment based on pre-defined risk factors of multidrug resistance [OR (95% CI) 1.73 (1.21-2.48), p = 0.003]. Furthermore, appropriateness of the initial antibacterial treatment according to the antibiogram results increased significantly in the post-guideline period [OR (95% CI) = 3.25 (2.09-5.06), p < 0.001]. The rate of compliance with guideline recommendations (p = 0.004) and the rate of appropriate empirical antibiotic treatment (p < 0.001) by each year were significant when compared with the pre-guideline period. Also, practices that improve drug pharmacokinetics such as loading dose, prolonged infusion of meropenem and adjusting antibiotic doses according to renal function increased statistically after the release of guideline.

Conclusion: An institutional sepsis protocol based on risk factors for multidrug resistance and local epidemiology increased the rate of appropriate empirical antibiotic treatment even in a setting where ID consultation is readily available.