Association between time-to-antibiotics and clinical outcomes in neutropenic fever after hematopoietic stem cell transplant

Abstract

BACKGROUND: Guidelines recommend empiric antibiotics within one hour for neutropenic fever, extrapolated primarily from sepsis literature. Whether modest variation in time-to-antibiotics (TTA) affects outcomes in antibiotic-naïve hematopoietic stem cell transplant (HSCT) recipients remains unclear. METHODS: We conducted a retrospective cohort study of neutropenic fever episodes occurring within 30 days of HSCT at a single academic center (2018-2022). We included antibiotic-naïve episodes (no intravenous antibiotics in the 24 h preceding fever) with timestamp-validated TTA. Primary outcome was clinical deterioration (new vasopressor requirement ≥ 6 h, mechanical ventilation, or 90-day mortality). Secondary analyses examined bacteremia and blood culture yield. RESULTS: Among 224 antibiotic-naïve episodes, 17 (7.6%) experienced clinical deterioration. Median TTA was 1.2 h (IQR 0.8-2.2), with 86 (38.4%) receiving antibiotics within one hour. In the primary categorical analysis, TTA was not associated with deterioration (≤ 1 h vs > 1 h, adjusted OR 1.18, 95% CI 0.37-3.78, P = .78); a per-hour continuous parameterization was similarly null (adjusted OR 1.03, 95% CI 0.85-1.24, P = .80). Recognized pathogen bacteremia occurred in 59 episodes (26.3%) and was not associated with 90-day mortality (adjusted OR 1.98, 95% CI 0.50-7.90, P = .33); a sensitivity analysis using any-positive cultures (69/224 [30.8%]) yielded similar findings (OR 1.15, 95% CI 0.31-4.28, P = .84). Of 11 deaths, 10 (91%) occurred without preceding acute deterioration. Pre-antibiotic and post-antibiotic cultures had similar recognized pathogen positivity (27.1% vs 23.5%; adjusted OR 1.03, 95% CI 0.28-3.85, P = .97). CONCLUSION: In antibiotic-naïve adult HSCT recipients with neutropenic fever, TTA was not associated with deterioration and bacteremia was not associated with mortality. Per-hour TTA effects were bounded below the minimum detectable effect size, challenging the extrapolation of the 1-h antibiotic rule from septic shock cohorts to this population. Multicenter validation is needed.