(See the article by Bizzarro et al. on pages 1137–1143.)

The concern that antibiotic overuse contributes to the development of resistant pathogens is not new; Sir Ian Fleming famously warned of the dangers of antibiotic-resistant organisms in 1945, less than 20 years after he discovered penicillin. ¹ However, only in the last 20 years has the concept of antimicrobial stewardship gained traction as a critical tool in our armamentarium against the emergence of drug-resistant pathogens and as an important metric for patient safety and quality. Indeed, the necessity for rapid implementation of antimicrobial stewardship nationally and globally has emerged as an urgent priority.

Early efforts at developing antimicrobial stewardship programs (ASPs)—and thus much of the data available on the feasibility and outcomes of such programs—were concentrated in acute-care settings and established the utility of the core antimicrobial stewardship strategies of formulary restriction and prospective audit and feedback.Reference Barlam, Cosgrove and Abbo ² In recent years, as stewardship has expanded to ambulatory care and nursing homes, it has become clear that a “one size fits all” blueprint for antimicrobial stewardship cannot accommodate the needs of all healthcare settings. A prominent example of the necessity of modifying antimicrobial stewardship for special settings and populations is the Centers for Disease Control and Prevention’s campaign for ASP implementation in nursing homes. ³ Prioritizing antimicrobial stewardship initiatives for vulnerable populations who are at high risk of infections and who have high antibiotic utilization, such as infants in the neonatal intensive care unit (NICU), is crucial. Infants in the NICU have an increased risk of infections due to immature immune systems and the need for invasive medical devices. Infections are difficult to diagnose because of non-specific symptoms and laboratory findings.Reference Kaufman and Fairchild ⁴ Also, NICUs often have a unique ‘ecology’ including multibed rooms and frequent overcrowding, as well as infants in frequent physical contact with physicians, nurses, and other support staff. These factors contribute to the risk of infection transmission and, by extension, antibiotic use.

Concerns about the consequences of inappropriately treating or not treating an infection generally outweigh concerns about the consequences of antibiotic overuse. However, antimicrobial overuse in the NICU population can have grave unintended consequences for individual patients beyond the emergence of drug-resistant pathogens. These additional unintended consequences of prolonged, and potentially inappropriate, antibiotic use include an increased risk of invasive candidiasis or necrotizing enterocolitis (NEC)Reference Esaiassen, Fjalstad, Juvet, van den Anker and Klingenberg ⁵ as well as disruptions to the infant gut microbiome.Reference Fouhy, Guinane and Hussey ⁶ While the implications of antimicrobial-induced disruptions to the microbiome have yet to be fully elucidated, data suggest that the consequences may be long lasting and include asthma, obesity, and other chronic conditions.Reference Fouhy, Guinane and Hussey ⁶

Recognizing the need for measures to avert the consequences of inappropriate antibiotic use in the NICU population, Bizzarro et alReference Bizzarro, Nzegwu and Rychalsky ⁷ conducted a quasiexperimental study describing the implementation of an ASP in their 54-bed, level IV NICU. They formed a multidisciplinary team that included physicians, a nurse, a pharmacist, and an infection control practitioner and utilized a modified template from the Get Smart for Healthcare campaign ^{3 ,} Reference Patel and Saiman ⁸ to design a NICU-specific ASP. Their program consisted of 3 primary components: (1) design and implementation of clinical guidelines for common neonatal infections including clinical sepsis and post-operative prophylaxis, (2) educational sessions, and (3) prospective audit and feedback. Using an interrupted time series analysis, they evaluated antimicrobial utilization in the 18 months before (ie, the “preintervention” period) and the 48 months after (ie, the “stewardship” period) the implementation of their ASP.

Although they did observe a decrease in antibiotic utilization of 14.7 days of therapy per 1,000 patient days (DOT per 1,000 PD) during the stewardship period, this decrease was not statistically significant and not likely to be clinically significant (P=.669). While a cursory glance at their primary outcome might tempt the dismissal of the study, this result alone does not tell the whole story.

The authors drilled down into the utilization of individual antibiotics and found that the use of gentamicin/tobramycin, vancomycin, and cefotaxime all decreased over the stewardship period, although none exhibited a statistically significant decline. Ampicillin utilization did, however, decrease significantly (−22.5 DOT per 1,000 PD; P=.037). Yet, no corresponding decrease in the incidence of early onset sepsis (EOS) or NEC was observed during the stewardship period; ampicillin was included in the empiric treatment regimen for these 2 indications at the study site. Although the data are not provided, the authors attribute the decreased use of ampicillin to the decreased duration of therapy for EOS and NEC as well as decreased use for urinary tract and postsurgical prophylaxis. These findings suggest that the treatment guidelines for discontinuing antibiotics were effective.

The findings of Bizzarro et alReference Bizzarro, Nzegwu and Rychalsky ⁷ evoke another question: Can ampicillin (and by extension, gentamicin, which is often the additional agent used for empiric treatment of EOS) utilization be decreased even further by refining the criteria that prompt initiation and continuation of antibiotics for EOS? Determining best practices to identify those infants at highest risk for EOS (ie, infants born in the setting of maternal chorioamnionitis) has been challenging given the absence of sensitive and specific biomarkers to predict infection for this population. Currently, there are guidelines for EOS evaluations and initiation of antibiotic therapy.Reference Polin ⁹ However, there is much concern in the neonatology community that strict adherence to these guidelines will result in otherwise well-appearing infants with sterile blood cultures but with other lab abnormalities (eg, elevated C-reactive protein or immature to total neutrophil ratio) being unnecessarily treated for ≥7 days for EOS.Reference Kiser, Nawab, Mckenna and Aghai ¹⁰ Given the growing evidence that the use of even ampicillin and gentamicin may adversely impact a neonate’s future health,Reference Fouhy, Guinane and Hussey ⁶ a more accurate way to identify infants truly at risk for EOS is urgently needed.

Other findings described by Bizzarro et alReference Bizzarro, Nzegwu and Rychalsky ⁷ include the reduction in late-onset sepsis (LOS) evaluations by an average of 2.65 evaluations per NICU attending physician per year as well as the decrease in the physician-to-physician variability in conducting LOS evaluations during the stewardship period. These declines likely represent the confluence of multiple factors beyond the 2 core AS strategies including the use of clinical guidelines, feedback to individual practitioners regarding guideline compliance, and an ongoing concerted educational effort. Granted, the decline in LOS evaluations did not impact overall antibiotic utilization, but the fact that LOS prescriptions only accounted for ~20% of antibiotic use at the study site explains, in part, the lack of an overall effect.

While these secondary outcomes are of value, the question remains: How is the study’s primary finding explained when implementation of an ASP would be expected to have resulted in significant declines in overall antibiotic utilization? We highlight some factors that may have been at play. First, the study NICU started off with a low rate of antibiotic utilization (270.4 DOT per 1, 000 PD) compared to peer NICUs (mean, 526.8 DOT per 1,000 PD), based on data from the Pediatric Health Information System. Prior to rolling out their ASP, the study site already utilized a restricted formulary. They had also successfully implemented an initiative to decrease central line-associated bloodstream infections.Reference Bizzarro, Shabanova and Baltimore ¹¹ These factors likely contributed to the low preintervention antibiotic utilization rate that left less room for improvement during the stewardship period. And, as noted previously, their stewardship intervention could not fully address one of the major drivers of antibiotic use: treatment of suspected EOS. Finally, the optimal outcome metric(s) for antimicrobial stewardship in the NICU is unknown. Although the authors used one of the most commonly employed metrics of DOT per 1,000 PD for their primary outcome measure, they recognized that the lack of an evidence-based metric that best correlates with the success or failure of an antimicrobial stewardship initiative could be a limitation.

In conclusion, optimizing antibiotic use in the NICU is challenging given the vulnerability of neonates to infection but necessary given the adverse effects of antibiotic misuse. Bizzarro et alReference Bizzarro, Nzegwu and Rychalsky ⁷ demonstrate that the implementation of a NICU-specific ASP is feasible, and their findings support the need for a multipronged approach to stewardship that includes the use of clinical guidelines, educational efforts, and audit and feedback of both adherence to guidelines and antibiotic use. Their work provides a blueprint that can guide future NICU-specific ASP efforts and can be modified to meet the needs of other NICUs. Still, ongoing work is needed as impactful stewardship interventions should not only target antibiotic prescribing, but also address the diagnostic uncertainties that drive antibiotic initiation and continuation; better tools to more accurately identify those neonates who truly need treatment for culture-negative EOS are particularly needed.