Fig. 3: Changes in AMRFinderPlus defined resistome following antimicrobial exposure.

A Prevalence of AMRFinderPlus-defined resistance gene subclass with exact binomial 95% confidence intervals. B Selected AMRFinderPlus-defined gene subclass prevalence stratified by visit and arm with exact binomial 95% confidence intervals. Colours represent different study arms. Cephalosporin and quinolone resistance gene prevalence shows little relationship with visit/arm, but macrolide and aminoglycoside resistance subclasses have a higher prevalence at visit 1 in the hospitalised/antimicrobial exposed group, consistent with an association with antimicrobial exposure. C Parameter values from modelling antimicrobial resistance gene subclass presence as a function of antimicrobial exposure. Parameter values (and 95% CrI) can be interpreted as logged odds ratio. Parameter values with a clear association between exposure and outcome (defined at lower bound of 95% CrI > 0) are coloured red. D, E simulated AMR gene subclass prevalence in stool for selected subclasses, following a 10-day hospital admission and 7-day exposure to a given antimicrobial agent. Lines show median posterior prediction, shaded area 95% credible interval. Colours show different antimicrobial exposures. Ceftriaxone is associated with an increase in aminoglycoside and clindamycin-erythromycin-streptogramin B subclass genes not seen with the other agents. F Associations of exposures to presence of AMR gene subclass restricted to beta-lactamases with models fit separately to Bacteroides-associated genes (defined as cfiA, cblA, crxA, cepA or cfxA beta-lactamases) or all others (defined as all other genes), expressed as parameter values and 95% credible intervals. The number of participants included in the analysis is given in Supplementary Table.