Exploring Bystander Selection for Antibiotic Resistance

Abstract

Antibiotic consumption exerts selection for antibiotic resistance. Due to the wide spectrum of activity of many antibiotics, a single antibiotic use may impose selective pressures on a range of bacterial species, including organisms residing in the normal flora that the drug was not intended to treat. The extent of off-target selective pressures, also known as bystander selection, and its implications for interventions against antibiotic resistance have not been well-characterized. In the first two chapters, I define a metric to quantify bystander selection in the outpatient setting and extend this metric to estimate the potential impact of four hypothetical antibiotic stewardship programs. In the third chapter, I describe microbiology patterns in an inpatient setting and estimate the proportion of antibiotic use directed towards treatment of specific pathogens, the complement of which could be considered an upper bound on bystander selection. For most bacterial pathogens of interest, I estimate that the vast majority of outpatient antibiotic exposures (>80%) occur when the organism is asymptomatically colonizing the host, not causing disease. Lower levels of bystander selection tend to occur among organisms that are carried infrequently or are treated with a specific regimen when causing infection. Among inpatients, approximately one-third of antibiotic use could be linked to isolation of any bacterial organism, and only a small fraction could be presumptively attributed to any particular species or genus. Frequency of bacterial isolation was highly dependent on syndrome. Due to bystander selection, interventions that reduce antibiotic use may broadly impact a wide range of organisms. I estimate that if all unnecessary outpatient use was eliminated, up to half of antibiotic exposures experienced by some species could be averted. This finding also underscores the need for concurrent interventions to prevent infections, such as vaccination. Vaccines could similarly have far-reaching effects by decreasing the need for treatment. For example, pneumococcal conjugate vaccination programs may result in nearly the same proportional reduction in total antibiotic exposures of S. pneumoniae, S. aureus, and E. coli, despite the latter two organisms not being targeted by the vaccine.