Inflammatory Markers and the Youngest Febrile Infants

Throughout much of the 20th century, fever in neonates was considered a harbinger of sepsis, extended to include all infants up to 2 to 3 months of age. Complete “sepsis workups” were considered warranted, followed by hospitalization and empirical antimicrobial treatment. Then efforts began to identify those whose risk was sufficiently low to permit “safely doing less.”1 The American Academy of Pediatrics Clinical Practice Guideline (AAP Guideline) serves as a summary of such efforts as of 2021.2 When the AAP Guideline was being developed, data were emerging about a possible central role for inflammatory markers (IMs), but the availability of procalcitonin (PCT) was limited.3–5 Since then, PCT has been recognized as the best single IM, with even better performance when combined with other IMs, such as C-reactive protein (CRP) or absolute neutrophil count (ANC).6 Moreover, it is now more widely available. The provocative research brief by Itzkovitz et al in the current issue of Hospital Pediatrics prompts consideration of widening possibilities for IMs to aid clinicians faced with the daunting task of evaluating very young febrile infants.7 Accordingly, a status report is in order.In older infants and children, clinicians rely on physical examination to identify levels of illness. However, clinical decision-making on the basis of physical examination in very young infants, apart from those who are critically ill, is challenging due to their limited repertoire. The AAP Guideline was based on infants being considered “well appearing.” In 2 studies, clinicians were provided a third category for infants who could not easily be dichotomized as “well” or “ill.”8,9 In the office-based study with minimally/moderately/severely ill categories, clinical appearance was the best single predictor but only identified 59% of infants with bacteremia and 92% with meningitis when used alone.8 Notably, in the study that permitted a third designation of “can’t tell,” the rate of bacteremia in this group was the same as in the group considered “ill.”9 In the past decade, the inclusion of PCT and CRP in the development of decision rules has shown them to be more important than clinical appearance.3–6 A shift from clinical appearance to laboratory detection opens the door to reconsider how infants other than those obviously ill can be stratified and provides beneficial leeway when clinicians are uncertain about how “well appearing” an infant is.Clinicians also use age to stratify risk assessment. Traditionally, infants in the first month of life have been considered at “high risk,” in addition to being difficult to categorize by clinical appearance. Accordingly, the AAP Guideline excludes infants born prematurely and infants in the first week of life and recommends full sepsis “workups” for those aged 8 to 21 days. In their research brief, Itzkovitz et al challenge the exclusion of infants born prematurely.7 The presumption that such infants are at a particularly high risk of infection is not supported by the results of the current study. As only 5% to 10% of births are preterm, it is not surprising that the Itzkovitz study was hampered by a low number of subjects with only 3 invasive bacterial infections (IBIs) and 1 case of meningitis. In a large system, Kaiser Permanente of Northern California accumulated 39 659 preterm infants in a 12-year period yet only had 44 with IBI and 6 with meningitis.10 In a 6-year period, 51 children’s hospitals reported 28 cases of bacteremia and 1 case of meningitis in preterm infants.11 The results of IM accuracy in the Itkovitz study are comparable to numerous previous studies of preterm infants in neonatal intensive care units.12 Although the number of preterm infants and infections in the Itzkovitz study is too small to be conclusive, the question of whether IMs are accurate enough to make prematurity irrelevant in risk stratifying febrile infants is provocative.Studies attempting to identify low-risk infants initially used rates of SBI to assess strategies. SBI includes urinary tract infections (UTIs), bacteremia, and meningitis, and the rate of SBI is high enough to permit studies in single institutions. The downside, however, is that UTIs so outnumber bacteremia and meningitis as to obscure the rates of these infections. A presumptive link between UTIs, bacteremia, and meningitis was based on similar causative organisms, but data continue to demonstrate that infants with abnormal urinalysis but normal IMs have “just a UTI” and need not undergo lumbar puncture (LP).11,13–15 The AAP Guideline recommended retiring the term SBI, and subsequent authors support this recommendation.7,11,14,15Researchers such as Itzkovitz et al have turned their attention from SBI to IBI, consisting of bacteremia and meningitis without UTIs. Studies addressing IBI have generally included all febrile infants in the first month (or younger than 28 days), although the AAP Guideline identifies infants between 22 and 28 days of life to be at lower risk than those aged between 8 and 21 days. In the IBI studies, IMs have sensitivities of ∼95% and specificities using PCT and ANC of ∼60% and PCT combined with CRP of ∼80%.3–6 Comparable sensitivities are achieved combining CRP, ANC, and temperature but at considerably lower specificity.6 Infants with bacteremia will be “missed,” even with the best combination of IMs currently available, prompting consideration of ceftriaxone administration to all young febrile infants pending blood culture results. However, as bacteremia makes up the vast majority of IBI cases, it overshadows the information needed to accurately stratify the risk of meningitis. Therefore, the term IBI should be retired along with SBI and the focus put on bacteremia and meningitis separately.The question of greatest concern to clinicians at point of care is which infants should undergo LP. LPs in very young infants have high rates of failure and bacterial contamination, and parents have high rates of anxiety about the procedure. In addition, LPs are rarely performed in office settings anymore, requiring an emergency department visit if an LP is desired, with many febrile infants now bypassing office settings altogether.The overall rate of meningitis in young febrile infants is less than 1% for infants younger than 28 days and even lower for those aged 29 to 60 days. The vast majority (90%) of young infants with meningitis appear unwell or are abnormal by the Step-by Step criteria.2,7 As these infants are excluded from the AAP Guideline and most studies, the prevalence in infants considered well appearing appears to be approximately 0.1%, or a number needed to test of 1000. Sorting by clinical appearance may be challenging in very young infants, however, prompting the use of IMs as the foundation for risk stratification. The role of IMs is promising for reducing the number of LPs, especially because the recent combination of PCT and CRP promises higher specificities than documented in the AAP Guideline in 2021.6 In recent prospective studies of infants younger than 60 days with PCT or CRP, the sensiivity for bacterial meningitis in infants was 95.9% (94/98, CI = 89.3%, 98.7%).4,16,17 Two recent bacterial meningitis studies addressed infants younger than 28 days with sensitivity of 96.4% (54/56, CI = 87.9%, 99.0%).16,17 Even with the best combination of currently available IMs, there will still be some infants whose evolving meningitis is not detected on the initial visit, stressing the need for careful clinical follow-up.The ongoing accumulation of data reinforces the 2021 CPG recommendations, closes some gaps, and leads toward a potential set of recommendations that consider UTIs, bacteremia, and meningitis separately. It is tantalizing to say that IM combinations including PCT answer the most pressing questions about febrile infants. The initial step taken by Itzkovitz et al prompts consideration of whether IMs will permit extending the Guideline recommendations to infants presumed to be at too high a risk for inclusion. To determine optimal risk stratification will require large numbers of infants to be studied in ways that make the results able to be pooled; this is particularly true for meningitis because of the extremely low rate. Ongoing national and international collaboration will be required. SBI and IBI should not be used as outcome measures but results reported about the individual infections. And, as prompted by Itzkovitz et al, the data need to be available by age of the infants to determine whether exclusions are justified.