The transmission dynamics of methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci in hospital wards
This thesis presents a study of the transmission dynamics of nosocomial pathogens in hospital wards, with particular reference to methicillinresistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE). The work makes use of mathematical models, and observational and epidemiological studies. Transmission dynamics of a potential pathogen are first explored using a stochastic host-vector epidemic model, where health-care workers' hands are assumed to be the vectors. Consequences of changes in patient management are presented, and stochastic effects are shown to be essential to an understanding of ward dynamics. Observations of carer handwashing behaviour and carer-patient contact patterns are described, and the factors associated with hand washing compliance and contact rates explored using statistical models. Patient-carer mixing patterns are investigated. Refinements to the host-vector model are used to show how different aspects of observed contact patterns may both increase and decrease the spread of nosocomial pathogens. For contact patterns typical of intensive care units, the model predicts that infection rates will increase as the staff-to-patient ratio decreases, so understaffing may result in more cross-infection even if handwashing levels do not change. The effect of antibiotic treatment on the spread of resistant strains is studied using a two-strain model. Changing patterns of antibiotic use are shown to be capable of causing large and rapid changes in ward prevalences of resistant strains. To investigate possible fitness costs, growth kinetics of methicillin-sensitive and methicillin-resistant S. aureus strains are compared. No evidence for differences in growth rates is found, though there is a suggestion that MRSA stains may have longer lag periods. Finally, a Markov chain Monte Carlo (MCMC) approach is developed to enable model parameters to be estimated from the incomplete data typical of ward-based epidemiological studies. The approach is used to estimate parameters using S. aureus and VRE transmission data. With the latter data, transmission rates were related to patient antibiotic use. All antibiotic combinations considered were associated with increased acquisition rates, the effect being strongest for cephalosporins.