In the 1960 film Spartacus, the Roman army succeeds in trapping the rebellious slave army led by the eponymous hero.1 The Romans know that Spartacus is a dangerous foe who must not escape; unfortunately, they have no idea what he looks like. To ensure that Spartacus is eliminated, the Romans elect to execute everyone. It is easy to see the parallels between Spartacus and our current approaches to sepsis in the newborn. Nursery clinicians are well aware that neonatal sepsis is dangerous and must not be missed, but the challenge of identifying the septic needle from the uninfected haystack of neonates is challenging. The most common approach is to adopt a “when in doubt, test and treat” tactic. As a result, clinicians (me included) are both inefficient and subjective in our diagnostic approach to sepsis. In this issue of Pediatrics, Schulman et al2 highlight this inefficiency in their study of antimicrobial use in >326 000 infants in 116 California NICUs. Clinicians administered antibiotics to an average of 95 infants (SD ±71) for each case of early-onset sepsis (EOS), and 8.4% of all newborns received ≥1 courses of antibiotics, >100-fold higher than the observed rate for EOS (0.075%).
There are 2 key drivers of this diagnostic inefficiency. First, identification of EOS is challenging. Newborns have a limited range of clinical responses to stress, so many conditions mimic sepsis. Noninfectious causes of respiratory distress, such as transient tachypnea of the newborn, are present in 1% of term newborns, which is 25 times higher than the incidence of EOS.3 Preterm infants without sepsis are still at risk for respiratory distress, hypotension, temperature instability, hyperglycemia or hypoglycemia, leukopenia or leukocytosis, and thrombocytopenia, all of which are drivers of empirical antibiotic therapy.4 Even vaccine administration is associated with increased risk for clinical signs of sepsis, with a corresponding increase in sepsis evaluations and antimicrobial therapy for 48 to 72 hours after immunization.5 Unsurprisingly, studies have consistently demonstrated poor specificity of sepsis evaluations in neonates.6,7
Second, nursery clinicians may not use objective risk factors or local epidemiology to guide decisions about sepsis evaluations. Schulman et al2 found no correlation between a hospital’s antibiotic use and incidence of sepsis or diagnostic efficiency for sepsis at that hospital. This is consistent with previous work that has revealed that the decision to begin antibiotic therapy or to extend therapy beyond 5 days is unrelated to maternal and infant risk factors for sepsis in both extremely preterm8 and healthy term infants.9 To quote Schulman et al,2 “the plausible correlations we explored provide no rational explanation for the observed variation in…antibiotic exposure.” Instead, nursery clinicians are more likely to use subjective or institutional factors rather than objective risk factors to determine need for treatment. Hopefully, a return to objectivity will be supported by the recent endorsement of multivariate risk assessment (ie, the neonatal EOS calculator) in the group B streptococcal and EOS guidelines from the American Academy of Pediatrics.10,11
We must do better, but how? The ideal situation (treating all infants with EOS while minimizing antibiotic exposure for those who are uninfected) requires a rapid test with near-perfect sensitivity and high specificity, which does not yet exist. In the interim, clinicians can limit antimicrobial exposure by observing (rather than empirically treating) well-appearing infants or low-risk infants with equivocal examination findings.12 When empirical antibiotic therapy cannot be avoided, limiting antibiotic duration to 24 hours should be considered because >98% (95% confidence interval, 95%–99%) of EOS cases are identified in the first 24 hours of blood culture incubation.13,14 Given a diagnostic accuracy of 1.1%,2 ∼4500 infants would require a second day of empirical therapy for EOS to capture the 1 infant with a positive result on the blood culture after 24 hours. Finally, development and validation of a multivariate risk assessment tool for the most preterm infants would be a marked leap forward in neonatal stewardship efforts. Preliminary work by Puopolo et al15 indicated that EOS risk stratification is possible, even among infants ≤28 weeks’ gestation.
More and more evidence is emerging regarding the short- and long-term complications of prolonged or unnecessary antibiotic use. Hopefully, the days of liberal antimicrobial therapy “just to be safe” are ending, and an era of more efficient, targeted approaches based on objective risk factors is beginning. Nursery clinicians can no longer indiscriminately put the microbiome of at-risk infants to the sword in search of Spartacus. Instead, research is needed to develop more precise diagnostic and therapeutic strategies for EOS and to reduce the unacceptably high number needed to treat of 95.2 Multivariate risk stratification, increased reliance on observation, and limiting the duration of empirical therapy are important steps forward in nursery stewardship. The work of Schulman et al2 is an excellent reminder of how far we have to go.
Opinions expressed in these commentaries are those of the author and not necessarily those of the American Academy of Pediatrics or its Committees.
FUNDING: No external funding.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2019-1105.
POTENTIAL CONFLICT OF INTEREST: The author has indicated he has no potential conflicts of interest to disclose.
FINANCIAL DISCLOSURE: The author has indicated he has no financial relationships relevant to this article to disclose.