Avoiding Unnecessary Antibiotic Exposure in Premature Infants: Understanding When (Not) to Start and When to Stop

Early antibiotic treatment is common among premature infants given their significant risk of early-onset sepsis (EOS) and its related consequences, but prolonged antibiotic exposure in the absence of proven infection is also associated with significant morbidity and mortality.¹ Flannery et al² present the results of a retrospective analysis of early antibiotic prescription practices in very low-birth-weight (VLBW) and extremely low-birth-weight (ELBW) infants from 297 hospitals across the United States. Three major themes arise from their results: (1) the majority of VLBW (78.6%) and ELBW (87.0%) infants are exposed to antibiotics in the first few days of life, (2) a significant proportion of VLBW (26.5%) and ELBW (37.8%) infants receive more than 5 days of early antibiotic therapy, and (3) there is marked center-to-center variability in the proportion of infants prescribed prolonged treatment regimens (ranging from 0%-80.4% of VLBW infants and 0%-92.0% of ELBW infants). While it is worth noting that prolonged antibiotic treatment in VLBW infants decreased over the 7-year study period and that the proportion of ELBW infants exposed to early antibiotics is less than previously observed,¹ the major findings underscore the need for intensified antimicrobial stewardship efforts in the neonatal intensive care unit (NICU), where antibiotic use remains high in a population at substantial risk of consequences from potentially avoidable exposures.

An ideal approach to early empirical antibiotic therapy would be one that accurately identifies and treats those at high risk while sparing those at low risk. Data from this investigation and others suggest that we are struggling to make this distinction. Flannery et al² observed that approximately 1 in 4 VLBW infants and 1 in 3 ELBW infants in their study cohort were exposed to prolonged early antibiotic therapy. Although the rationale behind each treatment course was unknown, the proportion of premature infants treated greatly exceeds published estimates of culture-proven EOS, which occurs in approximately 1 in 90 VLBW infants and 1 in 40 ELBW infants.^3,4 A recent retrospective cohort study assessed early antibiotic treatment in VLBW infants who were deemed to be at low risk for EOS (ie, born via cesarean delivery for maternal indications and without documentation of preterm labor or prolonged rupture of amniotic membranes).⁵ While 22% of the VLBW cohort met low-risk criteria, 99.8% had a blood culture drawn and 85.3% were treated with antibiotics despite only 1 (questionable) case of culture-positive EOS.⁵ The SCOUT study revealed that more than 93% of all antibiotic use in a single level III NICU occurred in infants without culture-proven infection, necrotizing enterocolitis, cellulitis, or congenital infection.⁶ Investigators from the Neonatal Research Network examined a retrospective cohort of infants born at 22 to 28 weeks’ gestation who were considered to be at low risk for EOS.⁷ Criteria included delivery based solely on maternal conditions (eg, preeclampsia) or chronic fetal conditions (eg, growth restriction). Infants failing to meet these criteria were included for comparison. In the initial cohort, 5759 infants (37.3%) were deemed to be at low risk, but 33.2% of the low-risk cohort received 5 or more consecutive days of antibiotic treatment in the absence of culture-proven EOS, necrotizing enterocolitis, or spontaneous intestinal perforation. For each documented case of EOS, 66 infants at low risk had prolonged antibiotic exposure, with the ratio of prolonged treatment to proven EOS cases 3 times that in the comparison group.⁷

These data illustrate some of the barriers faced by NICU medical professionals when attempting to assess risk for EOS among premature infants. In times of uncertainty, many look to national organizations and expert opinion for guidance. The current Committee on Fetus and Newborn management guidelines for EOS recommend empirical treatment with broad-spectrum antibiotics for all critically ill infants, irrespective of gestational age and risk factors.⁸ The authors acknowledge the difficulties in distinguishing clinical signs of EOS from those related to other noninfectious causes. This ambiguity lends itself to overtreatment and the variability in prescription practices observed in this and other investigations.⁹ It therefore seems apparent that a modified approach to risk assessment and early antibiotic prescription in premature infants is needed, but how best to proceed remains a challenge.

Recent antimicrobial stewardship efforts conducted in late-preterm and term infants may provide some guidance. The creation and implementation of a multivariable risk estimate and clinical assessment tool for EOS has been shown to significantly reduce early antibiotic exposure compared with the use of national guidelines.¹⁰ A similar assessment tool could assist in accurately identifying and potentially reducing antibiotic prescription among the one-quarter to one-third of VLBW infants estimated to be at low risk for EOS.^5,7 It is also imperative that NICU medical professionals understand and seriously consider the risks of unnecessary antibiotic exposure when deciding to initiate and continue antibiotic treatment in premature infants. Perhaps a fair comparison can be made to lessons learned from the use of supplemental oxygen. There was a time when early and prolonged oxygen therapy was administered to nearly all premature infants, irrespective of individual risk assessment for respiratory morbidity, until data revealed a causal relationship between overexposure and serious harm. Once this knowledge was gained, practice was changed to limit treatment to only those infants who required it. Flannery et al² have demonstrated that most VLBW and ELBW infants continue to be exposed to early antibiotic treatment despite data suggesting that unnecessary antibiotic exposure can increase the risk of serious morbidity and death. It is therefore the responsibility of individuals who prescribe antibiotics to premature infants to ensure, to the best of their ability, that treatment is only administered to those who need it.

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2018 Bizzarro MJ. JAMA Network Open.

Corresponding Author: Matthew J. Bizzarro, MD, Yale University School of Medicine, Division of Perinatal Medicine, Department of Pediatrics, PO Box 208064, New Haven, CT 06520-8064 (matthew.bizzarro@yale.edu).

Conflict of Interest Disclosures: None reported.