The use of local antibiogram in guiding clinical decisions is an integral part of the antimicrobial stewardship program. Conventional antibiograms are not disease-specific, ignore the distribution of microorganisms, obscure the in-vitro efficacy interrelationships, and have limited use in polymicrobial infections.
We aimed to develop an in-house empiric, disease-specific, antimicrobial prescription auxiliary for the treatment of hospitalized pediatric pneumonia patients and to present the methods which help to choose the first and the second line antimicrobial therapy, while accounting for cost and safety aspects.
A retrospective single center observational study was conducted on bronchoscopy obtained sputum culture. Analysis of probabilities, variance minimization, Boolean network modeling, and dominance analysis were applied to analyze antibiogram data. The Kirby–Bauer disk diffusion method was used to test the susceptibility of all isolates. Final optimization analysis included local drug acquisition cost (standardized to price per DDD) and safety profile.
Data of 145 pediatric patients hospitalized with pneumonia with 218 isolates over 5 years was collected. A combination of statistical methods such as probabilities of drug efficacy, variance minimization, Boolean network modeling, and dominance analysis can help to choose the optimal first-line and the second-line antimicrobial treatment and optimize patient care. This research reveals that ampicillin is the optimal choice as the first-line drug and piperacillin-tazobactam is the second-line antimicrobial drug if the first one is not effective, while accounting for cost and safety aspects.
The paper proposes a new methodology to adapt empiric antimicrobial therapy recommendations based on real world data and accout for costs and risk of adverse events.
