Prior Antibiotic Exposure Linked to Same-Class Resistance Across Three US Health Systems

Researchers analyzed 1,165,151 culture episodes from Mass General Brigham, Stanford Health Care, and Beth Israel Deaconess Medical Center, using MIMIC-IV for the latter. The observational study evaluated seven antibiotic classes, estimating the effect of exposure within the preceding 90 days on subsequent same-class nonsusceptibility. Double machine learning adjusted for more than 35 confounders per site, including prior resistant isolates, organism group, and calendar period.

Prior fluoroquinolone exposure was associated with increased nonsusceptibility of 11.9 percentage points at Mass General Brigham, 12.6 at Stanford, and 8.6 at Beth Israel Deaconess (all P < .001). Associations were positive across classes and sites, except for an indeterminate sulfonamide result at Stanford. Associations were strongest at 30 days and attenuated through 365 days. Among bloodstream infection episodes treated empirically with fluoroquinolones, therapy was microbiologically inactive in 50%–71% with prior fluoroquinolone exposure versus 27%–37% without it.

Across classes, 3,309 inactive empiric prescriptions were associated with prior same-class exposure. Robustness assessments included alternative weighting, negative controls, and exposure-window analyses; the median E-value was 2.5. Although the causal-inference framework addresses measured confounding, observational findings do not establish causation or exclude residual confounding. The results support prospective evaluation of antibiotic-history-informed empiric prescribing. This report is based on the abstract; full text was unavailable.