Show simple item record

dc.contributor.authorZawack, Kelson
dc.date.accessioned2017-07-07T12:48:41Z
dc.date.issued2017-05-30
dc.identifier.otherZawack_cornellgrad_0058F_10302
dc.identifier.otherhttp://dissertations.umi.com/cornellgrad:10302
dc.identifier.urihttp://hdl.handle.net/1813/51621
dc.description.abstractSurveillance is a key component of controlling antimicrobial resistance. In the United States this function is carried out by the National Antimicrobial Monitoring System. This effort combines the United States Department of Agriculture Food Safety Inspection Service collecting samples from slaughter facilities, the Food and Drug Administration collection samples from retail and the Center for Disease Control collecting samples from human medicine. In order to better understand single and multi-drug resistance as well as how their monitoring could be improved a comprehensive analysis of this surveillance data was undertaken. After an introductory chapter, the second chapter presents an analysis of single drug resistance measured both in terms of the amount of antibiotic required to prevent bacterial growth, minimum inhibitory concentration, and the proportion of isolates exceeding a given resistance cutoff. The effects of measuring resistance in these two different ways are compared along with the consequences of resistance cutoff choices. An analysis of variability is then carried out and used in an assessment of power and sample sizes. The examination of trends in single drug resistance reveals that these trends depend very much on the host, bacteria, and antimicrobial context regardless of whether one considers minimum inhibitory concentration or resistance proportion and that the resistance cutoff chosen has a dramatic impact on the nature of the trend observed. Measurements of single drug resistance are overdispersed which means large sample sizes are required to detect changes in resistance. The third chapter focuses on identifying multi-drug resistance associations by constructing contingency tables of resistance counts and modeling then with log-linear models. This approach uncovers associations that are in some cases so extreme they cannot be tested for using asymptotic or exact conditional methods and instead require a Bayesian approach. Interrogation into the nature of these interactions reveals a spectrum of interactions including a hierarchy among the -lactams. The fourth chapter explores the variability of interactions discovered in chapter three. As was the case with single drug resistance, multi-drug resistance also displays more variability than expected. This increased variability or overdispersion is likely due to unaccounted factors like antimicrobial use, husbandry practices and food handling hygiene procedures.
dc.language.isoen_US
dc.subjectComputational Biology
dc.subjectEpidemiology
dc.subjectStatistics
dc.subjectAntimicrobial Resistance
dc.titleA COMPREHENSIVE ANALYSIS OF THE UNITED STATES' NATIONAL ANTIMICROBIAL RESISTANCE MONITERING SYSTEM
dc.typedissertation or thesis
dc.description.embargo2019-06-08
thesis.degree.disciplineBiometry
thesis.degree.grantorCornell University
thesis.degree.levelDoctor of Philosophy
thesis.degree.namePHD of Biometry
dc.contributor.chairGrohn, Yrjo T
dc.contributor.chairBooth, James G
dc.contributor.committeeMemberWells, Martin T
dc.contributor.committeeMemberBien, Jacob
dc.identifier.doihttp://doi.org/10.7298/X42N50DQ


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record

Statistics