Main content

Meeting the Challenge of Antibiotic Stewardship in Livestock: Unleashing the Power of Evidence Synthesis through Scoping Reviews, Machine Learning, and Meta-epidemiology

Show simple item record

dc.contributor.advisor Sargeant, Jan Wisener, Lee 2021-05-20T18:22:00Z 2021-05 2021-05-13 2019
dc.description.abstract This thesis is an investigation into the opportunities and challenges of evidence synthesis for advancing antibiotic stewardship in livestock. The thesis includes two scoping reviews (ScR) of non-antibiotic interventions, one on veal and beef production, and one on nursery pig production for disease control or prevention. The objectives were to describe the volume, breadth, and depth of the research; identify specific topics that may support systematic reviews (SR); and identify knowledge gaps. Multiple databases were searched, then screened for relevance by two reviewers followed by data charting. Both ScR revealed a large volume of research, and wide breadth in intervention and outcome types. The veal-beef ScR had 7 specific topic areas that may support SR, the nursery pig had 13. For both ScR there was a dearth of clinical trials evaluating management interventions. Many vaccine clinical trials failed to report clinically important outcomes. Evidence syntheses are labour intensive. Automation of the relevance screening step could reduce that burden. The objective for the third study was to evaluate the DistillerSR (EvidencePartners Ltd.) machine learning (ML) prioritization tool using the two ScR databases. Simulations of the ML tool were used to emulate its performance for our ScR at three levels of recall (sensitivity). The DistillerSR ML tool would have reduced the burden of screening by 70% for the veal-beef ScR and by 49% for the swine ScR at 95% recall. Bias from inadequate trial design and conduct may impact summary effect sizes (ES) and thus the quality of SR for non-antibiotic approaches. Trial characteristics such as random allocation and blinding have been shown to impact intervention ES in human and laboratory animal medical research. The objective for the fourth study was to use meta-epidemiological methods to assess the associations between inadequate random allocation or blinding on intervention ES using source data from six SR for livestock interventions for antibiotic stewardship. Inadequate random allocation or blinding has the potential to bias intervention ES in livestock intervention clinical trials in unpredictable ways. en_US
dc.description.sponsorship Ontario Ministry of Agriculture, Food, and Rural Affairs, Ontario Veterinary College en_US
dc.language.iso en en_US
dc.publisher University of Guelph en_US
dc.rights Attribution-NonCommercial-ShareAlike 4.0 International *
dc.rights.uri *
dc.subject antibiotic stewardship en_US
dc.subject livestock en_US
dc.subject machine learning en_US
dc.subject meta-epidemiology en_US
dc.subject scoping reviews en_US
dc.title Meeting the Challenge of Antibiotic Stewardship in Livestock: Unleashing the Power of Evidence Synthesis through Scoping Reviews, Machine Learning, and Meta-epidemiology en_US
dc.type Thesis en_US Population Medicine en_US Doctor of Philosophy en_US Department of Population Medicine en_US
dc.description.embargo 2021-05
dcterms.relation Wisener LV, Sargeant JM, O’Connor AM, O’Sullivan TL, McEwen SA, Nwosu A, Rossi TM (2019). Non-antibiotic approaches for disease prevention and control in beef and veal production: a scoping review. Animal Health Research Reviews 20, 128–142. en_US University of Guelph en_US

Files in this item

Files Size Format View Description
Wisener_Lee_202105_PhD.pdf 2.339Mb PDF View/Open PhD Thesis

This item appears in the following Collection(s)

Show simple item record

Attribution-NonCommercial-ShareAlike 4.0 International Except where otherwise noted, this item's license is described as Attribution-NonCommercial-ShareAlike 4.0 International