Effect of urine point-of-care Gram staining on broad-spectrum antibiotic use in febrile urinary tract infections

Febrile urinary tract infections (fUTIs) are commonly treated empirically with broad-spectrum antibiotics because of concerns regarding antimicrobial resistance. Although Gram staining is widely used in clinical laboratories, operational constraints often limit its use to regular working hours. Some facilities have implemented point-of-care Gram staining (PCGS) to enable timely results and treatment decisions, even outside regular hours. However, the effect of urine PCGS on reducing broad-spectrum antibiotic use remains unclear. This single-center observational study (May 2015–October 2024) included patients aged ≥ 15 years admitted with suspected fUTIs. Initial antibiotic therapy was chosen based on urine PCGS results (PCGS group) or empirically without Gram staining (Empiric group). The groups were compared regarding the proportion of Access antibiotics prescribed (WHO AWaRe classification), antibiotic spectrum, culture-based appropriateness, pharmaceutical costs of the first intravenous dose, and time to initial antibiotic decision. Among 251 patients (PCGS: 184; Empiric: 67), the PCGS group exhibited a higher use of Access antibiotics (15.2% vs. 4.5%, p = 0.022, risk difference: 0.11, 95% confidence interval [CI]: 0.035 − 0.18), narrow-spectrum agents (28.8% vs. 4.5%, p < 0.001, risk difference: 0.24, 95% CI: 0.16 − 0.32), and no difference in in vitro activity of the initial agent (94.5% vs. 89.5%, p = 0.16, risk difference: 0.050, 95% CI: -0.030 to 0.13); clinical effectiveness was not evaluated. In a multivariable analysis of Access antibiotic selection ( N = 251; 5 covariates), PCGS was associated with an adjusted odds ratio of 3.93 (95% CI: 1.12–13.7), though the wide CI indicated an imprecise result. The median antibiotic cost of the first intravenous dose was lower in the PCGS group (486 vs. 844 JPY, p < 0.001). Time from initial evaluation to the antibiotic decision did not differ between groups (median: 111; interquartile range [76–159] vs. 123 [92–162] min, p = 0.23). Using urine PCGS to estimate pathogens was associated with more narrow-spectrum prescribing and lower first-dose pharmaceutical costs, with no statistically significant difference detected in in vitro activity or time to antibiotic order.