Temporal Trends and Center Variation in Early Antibiotic Use Among Premature Infants.

Importance: Premature infants are frequently administered empirical antibiotic therapy at birth. Early and prolonged antibiotic exposures among infants without culture-confirmed infection have been associated with increased risk of adverse outcomes. Objective: To examine early antibiotic use among premature infants over time and across hospitals in the United States. Design, Setting, and Participants: This retrospective cohort study used a comprehensive administrative database of inpatient encounters from 297 academic and community hospitals across the United States to examine data concerning very low-birth-weight (VLBW) infants (<1500 g), including extremely low-birth-weight (ELBW) infants (<1000 g), who were admitted to the neonatal intensive care unit and survived for at least 1 day. Data collection took place in November 2015 and analysis took place from February 2016 to November 2016. Exposures: Antibiotic initiation within the first 3 days of age and subsequent antibiotic administration for more than 5 days. Main Outcomes and Measures: Temporal trends in early antibiotic initiation and duration from 2009 to 2015, and center variation in early antibiotic use from 2014 to 2015. Results: We identified 40 364 VLBW infants (20 447 female [50.7%]) who survived for at least 1 day, including 12 947 ELBW infants, from 297 centers. The majority of premature infants had early antibiotic initiation (31 715 VLBW infants [78.6%] and 11 264 ELBW infants [87.0%]), and no differences were observed over time in temporal trend analyses (P = .12 for VLBW and P = .52 for ELBW). The annual risk difference in the proportion of VLBW infants administered early antibiotic therapy ranged from -0.75% (95% CI, -1.61% to 0.11%) to -0.87% (95% CI, -2.04% to 0.30%); in ELBW infants the annual risk difference ranged from -0.34% (95% CI, -1.28% to 0.61%) to -0.38% (95% CI, -1.61% to 0.85%). There was a small but significant decrease over time in the rate of prolonged antibiotic duration for VLBW infants (P = .02), but not for ELBW infants (P = .22). The annual risk difference in the proportion of VLBW infants with prolonged antibiotic duration ranged from -0.94% (95% CI, -1.65% to -0.23%) to -1.08% (95% CI, -2.00% to -0.16%); in ELBW infants the annual risk difference ranged from -0.72% (95% CI, -1.83% to 0.39%) to -0.75% (95% CI, -1.96% to 0.46%). We also observed variation in early antibiotic exposures across centers. Sixty-nine of 113 centers (61.1%) started antibiotic therapy for more than 75% of VLBW infants, and 56 of 66 centers (84.8%) started antibiotic therapy for more than 75% of ELBW infants. The proportion of VLBW and ELBW infants administered prolonged antibiotics ranged from 0% to 80.4% and 0% to 92.0% across centers, respectively. Conclusions and Relevance: Most premature infants in this study received empirical early antibiotic therapy with little change over a recent 7-year period. The variability in exposure rates across centers, however, suggests that neonatal antimicrobial stewardship efforts are warranted to optimize antibiotic use for VLBW and ELBW infants.