Influence of Patient Characteristics on Antibiotic Use Rates among Preterm Infants

Background: The antibiotic use rate (AUR) has emerged as a potential metric for neonatal antibiotic use, but reported center-level AURs are limited by differences in case mix. The objective of this study was to identify patient characteristics associated with AUR among a large cohort of preterm infants. Methods: Retrospective observational study using the Optum Neonatal Database, including infants born from January 1, 2010 through November 30, 2016 with gestational age 23-34 weeks admitted to neonatal units across the United States. Exposures were patient-level characteristics including length of stay, gestational age, sex, race/ethnicity, bacterial sepsis, necrotizing enterocolitis, and survival status. The primary outcome was AUR, defined as days with ≥ 1 systemic antibiotic administered divided by length of stay. Descriptive statistics, univariable comparative analyses, and generalized linear models were utilized. Results: Of 17 910 eligible infants, 17 836 infants (99.6%) from 1090 centers were included. Median gestation was 32.9 (interquartile range [IQR], 30.3-34) weeks. Median length of stay was 25 (IQR, 15-46) days and varied by gestation. Overall median AUR was 0.13 (IQR, 0-0.26) and decreased over time. Gestational age, sex, and race/ethnicity were independently associated with AUR (P < .01). AUR and gestational age had an unexpected inverse parabolic relationship, which persisted when only surviving infants without bacterial sepsis or necrotizing enterocolitis were analyzed. Conclusions: Neonatal AURs are influenced by patient-level characteristics besides infection and survival status, including gestational age, sex, and race/ethnicity. Neonatal antibiotic use metrics that account for patient-level characteristics as well as morbidity case mix may allow for more accurate comparisons and better inform neonatal antibiotic stewardship efforts.