The evaluation and management of fever in young infants has been a topic of great debate and considerable investigation for decades. The risk of delay in the diagnosis of treatable infections can lead to poor outcomes that might have been mitigated by earlier identification and intervention. Accordingly, many groups of investigators have made diligent efforts to create useful clinical tools intended to assist clinicians in all practice settings as they strive to provide the best care for their patients. Among those best known for this purpose are the clinical tools created in Rochester,1 Boston,2 and Philadelphia.3 Each tool differs from the others in some manner. These tools were originally intended to safely and reliably identify infants who are febrile and at a low risk of having a treatable, potentially serious or invasive bacterial infection as the cause of their fever. The creators of these tools intentionally valued sensitivity over specificity, looking to minimize failure to detect the presence of treatable bacterial infections. Each tool performs well in that regard.
More recently, health care has been focused on the cost effectiveness of medical practice as a prime directive. When viewed from this perspective, some layers of these clinical tools can be peeled away, trading off some degree of the sensitivity of detection of serious bacterial infections. In their current study, Aronson et al4 propose to eliminate routine use of lumbar puncture in the management of fever in young infants who do not appear to be ill. The original Rochester tool accommodates this strategy. However, the original Philadelphia tool includes lumbar puncture for all. Indeed, in their original data sets, the Philadelphia investigators observed infants older than 1 month whose bacterial meningitis was initially identified solely by information contained in the spinal fluid analysis.3,5 By modifying the original Philadelphia tool to exclude routine lumbar puncture for infants who are febrile but do not appear to be ill, its sensitivity to detect invasive bacterial infections in the current study sample was decreased to 91.9%, bringing it closer to the sensitivity of the Rochester tool (81.5%).4
Although the modified Philadelphia tool did not misidentify any infant older than 28 days with bacterial meningitis as low risk, study design limitations indicated by the authors need be recognized and weighed before adopting a change of clinical practice. Among those are nonuniform clinical documentation of appearance, frequent use of automated white blood cell differential counts, and exclusion of some eligible infants at some study sites. Not surprisingly, the modified Philadelphia criteria performed less well in Aronson et al’s4 study sample than did the original tool in the original prospective-design study, misidentifying 11 of 135 (8.1%) 1- to 2-month-old infants who were febrile with bacterial infections as low risk versus 1 of 65 (1.5%) in the original study. All 11 misidentified subjects in the current study had positive blood culture results, and 7 of those were positive for group B Streptococcus, a highly invasive organism.4
The authors’ conclusions are sound. The modified Philadelphia criteria, which does not include routine cerebrospinal fluid testing, identifies most infants who are febrile with invasive bacterial infections. More importantly, these modified criteria, within the framework of the current study design, inaccurately identify 8% of 1- to 2-month-old infants who are febrile with potentially serious bacterial infections as being low risk for such. In the current study, all of those misclassified low-risk infants had present bacteremia with potentially invasive organisms. The original Philadelphia criteria were intended to safely identify infants who were at a low enough risk of having concurrent bacterial infections to safely manage their febrile illnesses at home without the use of antibiotics. Those criteria performed well (approaching 100% sensitivity) when applied to different study populations.1,5,6 The modification of the original Philadelphia tool substantially lowers its sensitivity and jeopardizes safe use for its original purpose.
Whenever clinicians are formulating management plans to address their patients’ medical needs, it is incumbent on them to include the patients or their responsible guardians in the thought and decision-making process. That includes clearly stating the risk of including or excluding components of evaluation and management so that all principal stakeholders can make informed responsible decisions. In the case of the evaluation and management of fever in young infants, thoughtful omission of lumbar puncture in the initial assessment plans requires accurate disclosure of the likelihood and risk of delay in the diagnosis of bacterial meningitis, which is treatable and can have associated potentially lifelong morbidity. As is the case for any consequential health care decision, all stakeholders need to understand the data at hand and accept responsibility for the outcomes of their decisions.
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.2018-1879.
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.