The Disputed Champion: Ampicillin and Gentamicin for Febrile Young Infants

The distribution of pathogens and approach to empirical antibiotics in febrile infants aged <90 days has been debated for decades. There is increasing disagreement as to whether the combination of those timeless workhorses ampicillin and gentamicin (A&G) remains appropriate empirical therapy given the changing epidemiology of bacteremia and meningitis in young infants.^1–8 In the article by Feldman et al, researchers used the Pediatric Health Information System to describe regional differences in both pathogens and empirical antibiotic use in previously healthy infants <90 days old who were seen in 1 of 8 US pediatric children’s hospital emergency departments and had a positive urine, blood, or cerebral spinal fluid (CSF) culture.⁹ Not surprisingly, urinary tract isolates (urinary tract infection [UTI] alone or with associated bacteremia) made up 87% of infections. There was no regional difference in UTI pathogens. Blood and CSF pathogens differed by hospital, but susceptibilities remained constant and Escherichia coli and group B Streptococcus were the most common organisms. Third-generation cephalosporins (3GCs) were used for empirical therapy in the majority of infants, either alone (43%) or with ampicillin (39%). The combination of A&G was used in only 11% of infants.

We struggle to understand why providers are not using A&G empirically in febrile infants <90 days of age. This choice flies in the face of data. In previous studies, researchers have consistently demonstrated equal, or in some cases superior, efficacy of A&G compared with 3GCs.^3,10 In this study by Feldman et al, for example, A&G would have been adequate in 95% of cases compared with only 90% for 3GC monotherapy.⁹ In addition to superior efficacy, A&G offers other advantages over 3GCs (Table 1), including the fact that it drives less bacterial resistance.^11,12 Still, many institutions continue to rely on empirical cephalosporin-based therapy. Aronson et al found low rates of empirical A&G use in febrile infants, with 11% overall (22% ≤28 days old, 3% 29–56 days old, and <1% 57–89 days old).¹³ In the current study by Feldman et al, even centers in which group B Streptococcus was the predominant cause of bacteremia had low utilization of A&G.⁹ We ask, why?

Is it fear of gentamicin toxicity that has limited its routine use? Gentamicin nephrotoxicity appears to be related to the duration of use, the total dose administered,¹⁴ the age of the patient,¹⁵ and factors associated with impaired renal function.¹⁶ With once-daily dosing, the transient high peak level does not cause toxicity.¹⁷ Therefore, the risk of 1 or 2 doses of once-daily gentamicin is highly unlikely to cause nephrotoxicity. Similarly, in numerous studies of infants exposed to gentamicin in the neonatal period, researchers have demonstrated no increased risk for ototoxicity compared with unexposed infants, but the perception of gentamicin as ototoxic has continued. As with nephrotoxicity, once-daily dosing has decreased the risk for gentamicin ototoxicity. Short-course gentamicin exposure did not increase rates of sensory neural hearing loss in a premature infant population.¹⁸ High peak levels are not ototoxic, but rather it is the total dose, concomitant use of other ototoxic medications (eg, loop diuretics and platinum-based chemotherapy agents), and genetic propensity.¹⁹ Therefore, we suggest caution in using gentamicin for those rare infants on other ototoxic medications or with a family history of hearing loss. Finally, the only infants who are exposed to prolonged gentamicin therapy are those with sepsis because of ampicillin-resistant, Gram-negative rods, which makes up a tiny fraction of the infants started on empirical therapy.

Are providers concerned about missing meningitis? To be fair, the increasing proportions of ampicillin-resistant, Gram-negative rods and the poor penetration of aminoglycosides into CSF make A&G suboptimal for suspected meningitis. Therefore, we fully endorse the use of 3GCs for suspected meningitis with 1 major caveat: infants with suspected meningitis must undergo lumbar puncture for CSF analysis and culture. Otherwise, empirical cephalosporins have a tendency to become 3 weeks of “definitive” treatment of suspected meningitis that is never proven. For young infants, lumbar puncture and appropriate diagnostic testing is a critical aspect of antimicrobial stewardship.

If providers are not convinced by the evidence supporting A&G and instead insist on 3GCs, we echo the conclusions by Feldman et al in not recommending the addition of ampicillin.⁹ Combining ampicillin (another β-lactam agent) with a third-generation cephalosporin adds to neurotoxicity and the possibility of antagonistic effects in certain situations.²⁰ The only additional pathogens covered by the addition of ampicillin are Listeria monocytogenes and Enterococcus spp. For a decade, researchers have reported on the rarity of L monocytogenes,^1–3,5–7 but the concern for this pathogen has persisted. In a 2016 meta-analysis, Leazer et al²⁰ reported on the rarity of L monocytogenes and Enterococcus spp. infections. In several large regional studies,^1,3,5–7 a national study,² the above meta-analysis,²¹ and the current article by Feldman et al,⁹ there were no cases of late-onset Listeria.

Optimization of antibiotic use in young infants must address not only overuse but also underuse of antibiotics. In the current study by Feldman et al, many febrile infants (hospital range, 20%–50%) with proven infection did not receive any empirical antibiotics at the time of presentation.⁹ This rate is higher than the 10.9% overall and 27.1% in those 57 to 89 days old who were not empirically treated in another national study.¹² While we await the American Academy of Pediatrics’ guidelines on the “Approach to the Febrile Infant,” we are reminded of risk stratification. For decades, providers were hesitant to withhold antibiotics in young febrile infants. The concept of stratifying selected infants into low-risk categories to safely withhold empirical antibiotic therapy emerged ∼25 years ago with the Philadelphia and Rochester criteria.^22,23 Risk stratification has evolved to include newborns²⁴ but remains elusive for infants 3 to 28 days old. Although we fully support assessing febrile infants and placing them in high- or low-risk strata, a meaningful percentage in this study who had serious bacterial infections were correctly stratified into a low-risk category and had antibiotics withheld, given that the negative predictive value of risk stratification exceeds 99%.^22,23,25

In conclusion, we prefer the use of empirical A&G in febrile infants because of its proven efficacy and narrower spectrum, with use of 3GCs reserved for suspected or proven meningitis. If providers insist on using a 3GC, there is no evidence to support the routine addition of ampicillin in infants 7 to 90 days old. Finally, we further support a standardized approach to empirical antibiotics. The similar distribution of infections (predominantly UTIs) and pathogens (predominantly group B Streptococcus along with E coli and other Gram-negative rods) across geographically diverse sites in this study and others, suggests that selection of empirical antibiotics in young infants with a suspected bacterial infection is a question in search of a national answer. We feel strongly that the answer should be A&G.

• Copyright © 2017 by the American Academy of Pediatrics