Application of agent-based simulation to the modelling and management of hospital-acquired infections

摘要

Hospital-acquired infections (HAIs) are a big threat to the well-being of patients and place a heavy burden on hospital resources. The thesis provides the first attempt to apply agent-based simulation (ABS) to describe the transmission dynamics and evaluate the intervention policies of HAIs in general and Methicillin-resistant Staphylococcus aureus (MRSA) in particular. Based on the proposed taxonomy of potential methods for modelling HAIs, the relative advantages of ABS compared to other modelling methods are investigated. The comparison provides a theoretical justification to the use of ABS. The main methodological issues, including the representation of patient agents and the modelling of the transmission process, are discussed and a framework of applying ABS on HAI modelling is proposed. Guided by the framework, a MRSA model is built and validated using observed data from an empirical study. The model is more realistic and flexible than previous MRSA models and embeds intervention policies that have not been systematically studied such as the turnaround time and frequency of screening tests and the decolonisation treatment. Various interventions and influencing factors are systematically evaluated by formal experimental design methods including the fractional factorial design and the response surface design. The experimental results indicate that the use of rapid screening tests with shorter test turnaround time is the most effective policy to reduce MRSA transmission in the hospital setting. The introduction of admission and repeat screening is another effective policy; however, the effectiveness is not linear and may depend on patients’ lengths of stay. Providing more isolation facilities is also an effective policy but its effectiveness is significantly dependent on the efficacy of isolation. To demonstrate the potential and flexibility of ABS, the MRSA model is extended to include a competitive infection, to include multiple hospital units and HCW agents, and the wider community.