Duration of same-species antimicrobial non-susceptibility persistence in urinary gram-negative bacilli: a 12-year...

Prior urine-culture results can guide empiric antimicrobial therapy, but their value depends on whether the same organism recurs and the prior phenotype remains informative. We estimated the duration of same-species antimicrobial non-susceptibility persistence among urinary gram-negative bacilli in a tertiary-care cohort.

We conducted a retrospective longitudinal cohort study of urine cultures from patients aged ≥12 years at Aseer Central Hospital, Saudi Arabia, from 2013 to 2024. Consecutive within-patient, same-species isolate pairs were formed for gram-negative bacilli. Among pairs in which the prior isolate was non-susceptible and the next same-species isolate was class-evaluable, persistence was defined as non-susceptibility of the next isolate to the same drug class. Estimates were stratified by organism, drug class, and time since the prior isolate; uncertainty was addressed using patient-clustered bootstrap or Wilson intervals, and temporal decay used modified Poisson regression.

The source cohort included 14,587 eligible urinary gram-negative bacilli cultures from 8,045 patients. The trajectory cohort included 2,401 patients with ≥2 eligible cultures, yielding 4,332 consecutive same-species pairs from 1,745 patients; 3,380 pairs from 1,368 patients entered class-evaluable persistence analyses. Escherichia coli and Klebsiella pneumoniae met support thresholds for organism-specific inference. Fluoroquinolone non-susceptibility persistence declined from 96.8% within 30 days to 79.9% after 365 days in E. coli and from 93.2% to 77.1% in K. pneumoniae. Extended-spectrum cephalosporin non-susceptibility declined from 90.0% to 48.2% in E. coli and from 94.9% to 63.6% in K. pneumoniae. Carbapenem non-susceptibility persistence in K. pneumoniae declined from 86.8% to 41.9%. All supported organism-class decay models had risk ratios below 1.0 per doubling of time.

Prior same-species urinary gram-negative bacilli non-susceptibility carried substantial empiric memory, especially for fluoroquinolones, but persistence declined with time and differed by organism and drug class. Prior results should be interpreted as time-dependent, species- and class-specific evidence, not as an indefinite resistance flag.