Abstract Rationale Because optimal antibiotic duration treatment for bacterial tracheostomy-associated infections (bTRAINs, e.g., pneumonia or tracheitis) in children is unknown, we evaluated the comparative effectiveness of short versus long antibiotic courses for bTRAIN on 30-day bTRAIN readmission. Methods We conducted a prospective cohort study in six freestanding children’s hospitals, including patients 21 years old with tracheostomy hospitalized and treated for bTRAIN between 2020-2024. A bTRAIN was defined as an attending physician diagnosis of bacterial pneumonia or tracheitis treated with complete antibiotic course. We excluded children pre-treated with antibiotics, had length of stay 45 days, or antibiotic duration 3 or 45 days. The primary predictor was total inpatient + outpatient antibiotic duration categorized as short (7 days) versus long (10 days), excluding 8 and 9 days. Our primary outcome was 30-day bTRAIN readmission. To adjust for confounding by indication, we applied stabilized inverse probability of treatment weighting (IPTW) to balance demographic and clinical characteristics between antibiotic duration groups. Propensity scores were derived from demographic and clinical severity variables associated with antibiotic duration (Table). To test for non-inferiority of short- versus long-course antibiotics on bTRAIN readmission, we used mixed effects logistic regression to estimate the IPTW-adjusted absolute risk difference between short course and long course antibiotics, setting an a priori non-inferiority margin of + 3.5%. Results Among 823 hospitalizations, the median age at admission was 5 years (IQR 2-12), 45% (n = 370) were female, 75% (n = 617) had public insurance, and 59.5% (n = 490) had home nursing. For comorbidities, 75% (n = 617) had neuromuscular disease and 66.3% (n = 546) were on home positive-pressure ventilation. 40.7% (n = 335) of encounters received short course antibiotics. Longer antibiotic courses were more common among children with public insurance (77.5% vs short: 71.3% p = 0.05), with neuromuscular disease (79.7% vs 68.1%, p = 0.0001), and presenting with hypoxemia (65% vs 53.4%, p = 0.0009). It was less common in children with respiratory distress (38.9% vs 53.1%; p 0.0001). Thirty-day bTRAIN readmission was 9.7% (n = 80), with 8.4% (28 of 335) in the short-course group versus 10.7% (52 of 488) in the long-course group. After applying IPTW, the 30-day bTRAIN readmission absolute risk difference for short versus long antibiotics was -2.1% (95% CI: -6.0, 1.9), meeting our a priori non-inferiority margin of + 3.5%. Conclusion While randomized trials are needed to confirm our findings and account for unmeasured confounding, this work provides the strongest evidence currently available to support shorter antibiotic courses in children hospitalized with bTRAIN. This abstract is funded by: Gerber Foundation, AHRQ (1R01HS027619)
Jin et al. (Fri,) studied this question.