Monitoring the spread of antibiotic resistance in wastewater
Lepper, Hannah Catherine
BACKGROUND: Antibiotic resistant bacterial infections are causing a growing amount of morbidity and mortality. Effective control and prevention relies on good data on the current burden of antibiotic resistance (ABR). Traditional ABR surveillance from phenotypic, passive, hospital-based testing may not adequately represent the resistome of the general population. Wastewater metagenomics has been proposed as a new type of surveillance to overcome this limitation. It generates rich, quantitative information on the bacterial species and resistance genes of a whole community. Large wastewater metagenomic datasets are now available to monitor and explore drivers of ABR in the community. However, questions remain about how to collect, analyse, and interpret these novel datasets. In this thesis, I aimed to 1) address key unknowns in wastewater data, including sources of resistance, environmental resistance dynamics, and what statistical models describe the distribution of the data well, and 2) investigate global and local patterns in wastewater resistance and identify potential community and hospital drivers. METHODS: I used a systematic review to find evidence in the literature for dissemination of ABR from hospitals to wastewater. I next developed a compartmental transmission model to investigate environmental resistance dynamics and its impact on human ABR levels. I implemented a multi-response statistical model to correlate hospital-based surveillance (EARS-Net) data with resistance gene abundance in sewage samples from around the world analysed with metagenomics by the Global Sewage Surveillance Project. Finally, I used a paired sampling design and multiple statistical methods to compare the resistome of sewage from hospitals, communities, and wastewater treatment plants (WWTPS) in Scotland. I also investigated the links between ABR in humans and antibiotic consumption in the modelling and data analysis chapters. RESULTS: I found increasing evidence in primary research that resistant bacteria and resistance genes can be disseminated from hospital patients to wastewater and into natural water sources. Modelling the dynamics of ABR in an environmental reservoir indicated that the environment can theoretically influence human ABR levels as much as or more than an animal reservoir, and mitigate intervention impacts. Combining EARS-Net and sewage metagenomic data indicated that some types of ABR are positively correlated in sewage and hospitals (such as aminoglycosides), but many are not (such as vancomycin and aminopenicillins). The paired sampling study demonstrated that hospital and community sewage resistomes are distinct, and WWTPs mostly reflect community sewage resistomes. I found mixed evidence for an impact of antimicrobial consumption on human ABR levels. Overall, the impact of antibiotic consumption at the population level appears to be small in these datasets. CONCLUSIONS: Wastewater metagenomics is a valuable way of monitoring ABR in the community. It can indicate the composition of the reservoir of ABR in the general population and what drives it. However, hospital rather than mixed municipal effluent may need to be collected to monitor clinical resistance patterns. To make the most of this new source of data more flexible modelling frameworks that account for wastewater metagenomics specific factors such as high dimensionality and overdispersion. Comparing resistance patterns in hospitals to community sewage implied that patients and/or the hospital environment may present unique and strong selection pressures for resistance. Finally, we also show that differential antibiotic consumption alone cannot explain the observed patterns in resistance abundance on the national or international level.