Antimicrobial selection for resistance in four major pathogens in the US Veterans Affairs Healthcare System, 2007-2021
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Systematic evidence on antimicrobial selection for antimicrobial resistance (AMR) is scarce. We estimated the effect of prescribing key antibiotic classes on AMR across U.S. Veterans Affairs Medical Centres (VAMC).
Methods
We analysed clinical isolates of Staphylococcus aureus, Escherichia coli, Klebsiella pneumoniae , and Pseudomonas aeruginosa from 138 VAMC from Feb 1, 2007 to Dec 31, 2021. Antimicrobial prescribing was measured as inpatient days of therapy per 1000 patient-days; multidrug resistance as number of resistant phenotypes per 1,000 admissions. Temporal trends were modelled using generalized estimating equations and average annual percentage changes (AAPC). Multilevel multinomial logistic regression related facility-level antibiotic prescribing (days of therapy per 100 patient-days in the last 14d) to the relative odds of resistant phenotypes.
Findings
Hospital-onset infection incidence declined for all pathogens, except third-generation cephalosporin (3GC)-resistant E coli . Antimicrobial prescribing remained stable or decreased, except 3GC prescribing, which increased from 2007 until 2019 (AAPC=2·4%, 95% CI 1·3%–3·5%, p-value<0.0001). Fluoroquinolone (FQL) use was associated with resistance across all pathogens. In S aureus , each day of FQL treatment was linked to a 4·6% (95CI: 1·5, 7·7, p-value=0.0127) increase in the relative odds of isolating FQL-resistant, macrolide-susceptible, methicillin-resistant S aureus . Anti-staphylococcal beta-lactams were not linked to MRSA. Each day of 3GC treatment increased the odds of isolating 3GC- and beta-lactam/beta-lactamase-resistant E coli by 5·2% (95%CI: 1·3%, 9·4%, p-value=0.0079) and K pneumoniae by 3·0% (95% CI: −0·1%-6·2%, p-value=0.0600). Each day of carbapenem treatment increased the odds of carbapenem-resistant, FQL- and BL/BLI-susceptible P aeruginosa by 15·7% (95%CI: 9·4%, 22·4%, p-value<0.0001).
Interpretation
Higher facility-level antimicrobial use increased the odds of corresponding resistant phenotypes, with important exceptions. FQLs selected for resistance across multiple pathogens. Increased 3GC prescribing likely offset reductions in FQLs and was associated with co-resistance in E coli . These findings underscore the need for comprehensive stewardship that coordinates strategies across antimicrobials.
