Assistant Professor of Pediatrics Children's Hospital of Philadelphia Philadelphia, United States
Background: Identification of meaningful metrics of neonatal intensive care unit (NICU) antibiotic use can allow hospitals and stakeholders to effectively audit antibiotic prescribing to improve the safety and quality of neonatal care. Currently, there are no uniformly accepted center-level metrics of NICU antibiotic use, and the relationships between proposed metrics are unclear.
Objectives: To calculate 3 center-level metrics of NICU antibiotic use and assess trends over time, center variation, and correlation between metrics.
Methods: This was a repeated cross-sectional study of NICUs in the Premier Database that continuously contributed annual data from 2017-2021 and had at least 20 annual NICU admissions. Median center-level metrics of antibiotic use (parenteral and oral) with IQR were calculated per 1,000 patient days (PD), overall and by year. Length of therapy (LOT) was defined as total days of antibiotics (yes/no for each day); days of therapy (DOT) was defined as aggregate antibiotic days, accounting for each individual antibiotic on a given day; and antibiotic spectrum index (ASI) was defined as the sum of assigned points for each antibiotic per day based on activity against clinically relevant pathogens (range 1 to 13), with higher ASI reflective of a broader spectrum agent (adapted from Gerber et al). The relative difference between 2017 and 2021 was calculated for each metric. Correlation between metrics was assessed using Spearman’s correlation coefficient.
Results: There were 272 NICUs and 438,156 infants included in the analysis. Median annual admissions per NICU was 217 (IQR 113, 401). Most NICUs were in urban settings (83%) and in non-teaching hospitals (62%). Half of the included infants were male and 0.9% died before discharge. Overall metric means (SD) per 1000 PD were: LOT 138.8 (73.2), DOT 233.7 (133.3), ASI 802.5 (320.9); and medians (IQR) per 1000 PD were: LOT 124.9 (90.5 - 166.9); DOT 208.6 (151.3 - 283.3), ASI 727.3 (502.8 - 962.7); all had a right-skewed distribution. All 3 metrics had a relative decrease from 2017 to 2021 (LOT by 16.8%; DOT by 18.9%; ASI by 18.9%) and each metric had an average annual relative decrease of ~5% during the study period. Metrics were all well correlated: LOT vs DOT (r=0.987; p< 0.001); LOT vs ASI (r=0.974; p< 0.001); DOT vs ASI (r=0.989; p< 0.001).
Conclusions: Antibiotic use in NICUs across the US from 2017-2021 decreased over time, as assessed by 3 center-level metrics (LOT, DOT, and ASI). The different metrics were all well correlated and variation in metric distribution across centers was identified. Further assessment of the relationship between NICU antibiotic use metrics and pertinent clinical outcomes is needed.
