Reviewed by Erica Stohs, MD, MPH, University of Nebraska
Retrospective analysis of 20 years’ worth of Veterans Health Administration (VHA) hospitals’ antibiograms for E. coli and Klebsiella sp. poorly predicted antimicrobial resistance for individual patients, highlighting limitations of antibiograms’ broad application when it comes to antibiotic selection.
Stewards frequently encourage clinicians to consult their local antibiogram when selecting empiric coverage for individuals and use them to keep tabs on local antimicrobial resistance development. Investigators aimed to assess the diagnostic accuracy of the standardized hospital antibiograms from VHA from the previous calendar year as a prediction tool for E. coli and Klebsiella sp. at the patient-level.
Using the VHA’s integrated electronic medical record infrastructure, they compiled data from over 16 million unique patients from over 250 million patient-days and 1.6 billion clinic visits from 127 hospitals from 2000-2019. Investigators generated receiver operating characteristics (ROC) curves for commonly used antibiotics and categorized AUCs into ranges of discriminative ability (poor to excellent). The predictive performances for commonly used antibiotics for E. coli and Klebsiella sp. (ceftriaxone, fluoroquinolones, and trimethoprim-sulfamethoxazole) were poor with the exception of a moderate performance of ceftriaxone and Klebsiella sp.
Generalizing results from a study population of largely adult males with uropathogens is challenging but emphasize that facility-level antibiograms should not replace clinical judgment and local epidemiology when selecting antibiotics for individual patients.
Reference:
Hasegawa S, et al. Diagnostic accuracy of hospital antibiograms in predicting the risk of antimicrobial resistance in Enterobacteriaceae isolates: A nationwide multicenter evaluation at the Veterans Health Administration. CID 2023; 77 (11):1492-500. https://doi.org/10.1093/cid/ciad467