Modeling and Simulation to Support Antibiotic Stewardship and Epidemiological Decision-Making in Healthcare Settings

点击翻译按钮获取中文摘要

PROJECT SUMMARY Our proposal assembles an exceptional group of researchers to join the Modeling Infectious Diseases in Healthcare (MIND) network. The University of Utah is the hub for our program, with major nodes at Harvard School of Public Health and Oxford University. Our proposal includes two projects: the first uses models and data to examine antibiotic selection for resistant organisms that cause HAI, including C. difficile, and the second creates modeling tools that use local data to improve implementation of infection control interventions. Thematic areas covered by our proposal include antibiotic resistance, connectedness of patients, surveillance, economic modeling, genomics, and simulations of epidemiologic studies. Our projects are highly interconnected with respect to data and methods, leveraging the broad expertise of our research teams and the availability of comprehensive data resources to support the use of models in healthcare epidemiology. The first project is intended to advance scientific understanding of the drivers of antibiotic resistance and to enhance the practical use of models to guide antibiotic stewardship policies. We will test hypotheses about which mechanisms of selection are most influential for a given organism class and type of resistance. We will categorize antibiotic treatments and organisms with respect to the magnitude of bystander selection, due to exposure of commensal organisms to antibiotics. The impact of antibiotic selection exerted by different classes of broad-spectrum antibiotics will be compared. The outputs of these analyses will support parameterization of forward simulation models, which we will use to evaluate the effects of reducing antibiotic use, particularly through decreasing treatment duration. Outcomes will be examined across drug class and organism. The second project will generate shareable tools that can be applied to local data to help healthcare epidemiologists and public health personnel make decisions regarding the implementation of infection control interventions. Our work for Aim 1 will give epidemiologists a statistical package to fit transmission models to their own carriage and infection data to estimate relevant transmission rate parameters. This will enable determination of the effectiveness of contact precautions and other infection control measures in their own institution. An extension will be to add genomic data to improve the accuracy of estimation of transmission trees. In Aim 2, we will use a “potentially prevented cases metric” to evaluate algorithms to support surveillance of pathogens that cause healthcare-associated infection. The product of this work will be a statistical package to assist epidemiologists in the decision about when it may be warranted to intervene on a possible outbreak either by launching an investigation to detect possible sources or by instituting additional control measures. In Aim 3, we will incorporate into regional models of antibiotic-resistant organisms the capacity to tailor the evaluation of alternative interventions to local patient flow and health economic data. The output will be a simulation and economic modeling tool to guide implementation of coordinated control strategies.