Understanding and modelling reservoirs, vehicles and transmission of ESBL-producing Enterobacteriaceae in the community and long term care facilities

The continuing spread of extended-spectrum beta-lactamase-producing Enterobacteriaceae (ESBL-PE) is among the most important problems in antimicrobial resistance. It is also a good model to investigate the epidemiological complexity of resistance in Enterobacteriaceae. Available data on the transmission determinants of ESBL-PE in community settings are scarce, methodologically limited and mostly based on single centre studies. A comprehensive investigation using modern typing and modelling techniques is warranted to develop a sound quantitative understanding of the interactions involved. A consortium of investigators with diverse expertise from countries with high and low endemicity of ESBL-EP has been created. Transmission and persistence of ESBL-PE within households and long-term care facilities will be studied. Individual and group-level determinants for transmission and persistence will be quantified, together with other ecological variables including environmental, food and wastewater contamination. Advanced molecular typing techniques and state-of-the-art analytical methods will be used. Data generated in this project will directly inform a suite of mathematical models which, in addition to encapsulating current understanding of the processes, will be used to explore the potential effectiveness of different interventions to control ESBL-PE spread. The expected outputs are a comprehensive characterisation of ESBL-PE transmission considering bacterial clones and mobile genetic elements, as well as individual and ecologic-level factors in different settings, to inform public health authorities about interventions that should be prioritised to control transmission of these organisms.

This summary refers to methodology for WP5, Modelling transmission dynamics of ESBL-EP. It depends on other WPs as described in the full proposal.

We will initially adapt methodology previously developed to study the transmission of bacterial pathogens in hospital settings. Adapted methods will be used to analyse data from WP1-4. The transmission of ESBL-PE (whether at bacterial or gene levels) can be represented as a partially observed Markov process, where the state is defined by the number of hosts colonized at a given time point. Fitting transmission multistate Markov and hidden Markov models (HMMs) to such data will be carried out with standard software (e.g. the msm and JAGS packages within R for likelihood-based and Bayesian inference respectively) or using standard algorithms to fit HMMs. Previously described Bayesian data augmentation approaches offer more flexibility at the expense of greater computational cost; recent work has shown how these approaches can make use of WGS data without restrictive assumptions to reconstruct transmission trees. We will extend this approach to account for within-host evolution and to allow for incorporation of individual-level covariates from WP1 & 2. We will cross-validate inferences about transmission processes by using, in parallel, approaches to reconstruct transmission networks using WGS and exposure history data using a novel approach based on structured coalescent models. Such an approach employs an explicit evolutionary model, though requires more restrictive epidemiological assumptions. For objective 4, we will combine inferences from objectives 1 and 2 to inform both compartmental and individual based multi-setting models. Compartmental models will account for interactions in the community, nursing homes, LTCFs and hospitals. These models will be developed by extending existing individual-based model code. Compartmental models will be implemented in R, and individual-based models in C++ (with interface to R).

Control of ESBL-PE transmission remains a top public health priority. These bacteria are now frequent aetiological agents of complicated and invasive infections in the community and hospitals. Despite limitations in previous studies, the available evidence favours the concept that these infections are associated with increased mortality and therapeutic failure rates when compared with those caused by susceptible Enterobacteriaceae, probably in relation with delayed appropriate treatment. To overcome this problem, carbapenems (which are active against these bacteria) have been increasingly used over the last few years, which is probably fueling the spread of more resistant bacteria like the now emergent carbapenemase-producers. Therefore, control of transmission of ESBL-PE in the community and long-term care facilities (LTCFs) would have an enormous impact in reducing the number of infections caused by these bacteria and would also reduce the need for prescribing carbapenems. The primary impact of this research is expected to arise through identifying possible interventions that should be prioritised to control transmission of these organisms, and informing public health authorities of these results.

However, this is a complex task. To achieve it, a better understanding on the reservoirs, vehicles and transmission determinants in different epidemiologic settings is needed, allowing the developement of multi-setting transmission models which can be use to estimate the impact of potential interventions. Specifically, interruption of household transmission represents a hitherto largely neglected opportunity for interventions to reduce the community-based spread of antibiotic-resistant Enterobacteriaceae. Also, LTCFs may be playing a central role in the maintenance and amplification of the ESBL-PE pandemic, which needs to be understood in relation to other determinants in order to plan infection control measures. Finally, the characterisation of the epidemiological importance of food and wastewater in the epidemiology of ESBL-PE would lead to specific measures. Also, because the project includes a "One Health" approach by considering human samples, food products and wastewater, the information would inform policies in different areas for potential intervention.

The information obtained with ESBL-PE will serve as a model for the epidemiology of other plasmid-mediated beta-lactamases (and eventually other mechanisms of resistance) among Enterobacteriaceae, such as plasmid-mediated AmpC enzymes, but most importantly, the carbapenemases, which are yet not as frequent in the community but are a real, serious threat for the future.