Diagnostic Performance of AAP Recommended Inflammatory Markers in Febrile Infants ≤60 Days

Introduction

Approximately 2% of febrile infants ≤60 days old have invasive bacterial infections (IBIs).¹ To risk stratify well-appearing febrile infants aged 22–60 days, the 2021 American Academy of Pediatrics (AAP) clinical practice guideline (CPG) recommends two inflammatory marker (IM) based strategies, based on availability of procalcitonin (PCT): PCT and absolute neutrophil count (ANC), or maximum temperature (Tmax), ANC, and c-reactive protein (CRP).² However, these combinations of IMs have not been directly compared in a multicenter study. We aimed to validate and compare the performance characteristics of the AAP CPG recommended risk stratification strategies for detecting IBI.

Methods

We conducted a secondary analysis of the Reducing Excessive Variability in Infant Sepsis Evaluation II (REVISE II), a quality improvement initiative that included infants 8–60 days with temperature ≥38°C who were evaluated at one of 106 sites from 11/1/2020–10/31/2022.³ Sites included freestanding (n=35) and non-freestanding children’s hospitals (n=37) and general hospitals (n=34). For inclusion in this analysis, infants required a documented temperature and performance of a blood culture.⁴ Exclusion criteria mirrored those in the CPG,^2,3 including ill-appearance which was based on emergency department physical examination documentation.⁵ Infants were classified as having IBI if an a priori defined pathogen was identified in blood culture (bacteremia) and/or in cerebrospinal fluid (CSF) culture (bacterial meningitis) with a treatment course for IBI. We classified indeterminate organisms by consensus. Additional details are described in the parent study.³

The data collection form listed normal and abnormal IM values per the AAP CPG; participating sites chose the appropriate option. We calculated the sensitivity, specificity, negative predictive value (NPV), and negative likelihood ratio (LR) for identification of IBI, stratified by AAP CPG age group, in two combinations: PCT+ANC and Tmax+ANC+CRP. As an exploratory analysis, we also assessed PCT+ANC+CRP. If all IMs in the combination were obtained, or at least one IM in the combination was abnormal (regardless of other IMs obtained), infants were included in that group. Infants could be included in more than one IM combination. Performance characteristics were calculated using Stata version 18.0 (StataCorp, Inc, College Station, TX). The study was deemed exempt by the AAP institutional review board.

Results

Of 13,262 infants who met eligibility criteria, 12,846 (96.9%) had blood cultures and were included, 292 (2.3%) of whom had IBIs. Characteristics of included infants are described in Supplemental Table 1. For identification of IBI in infants 22–60 days, both risk stratification strategies had high sensitivity (≥95%) and NPV (≥99.8%), and low negative LRs (≤0.12), though the sensitivity varied by age group (Table 1). Specificity was highest for PCT+ANC (63.2% in infants 22–60 days). Most false negatives were bacteremia (Supplemental Table 2). When limited to infants who had all IMs obtained in the combination, the specificity of both combinations was slightly higher (Supplemental Table 3). Addition of CRP to PCT+ANC reduced specificity (Supplemental Table 4). Two infants with meningitis aged 22–60 days would have been misclassified as low risk by either PCT+ANC (one infant) or Tmax+ANC+CRP (one infant); both infants were hospitalized (Table 2).

Table 1.

Performance characteristics of AAP CPG recommended risk stratification strategies

	Sensitivity, % (95% CI)	Specificity, % (95% CI)	Negative Predictive Value, % (95% CI)	Negative Likelihood Ratio (95% CI)
PCT+ANC1
All ages (N=10293; IBI=262)	95.8 (92.6–97.9)	58.7 (57.7–59.6)	99.8 (99.7–99.9)	0.07 (0.04–0.13)
8–21 days (N=2079; IBI=104)	97.1 (91.8–99.4)	40.1 (37.9–42.3)	99.6 (98.9–99.9)	0.07 (0.02–0.22)
22–28 days (N=1444; IBI=53)	100.0 (93.3–100.0)	62.3 (59.7–64.9)	100.0 (99.6–100.0)	0.00
29–60 days (N=6770; IBI=105)	92.4 (85.5–96.7)	63.4 (62.2–64.5)	99.8 (99.6–99.9)	0.12 (0.06–0.23)
22–60 days (N=8214; IBI=158)	94.9 (90.3–97.8)	63.2 (62.1–64.3)	99.8 (99.7–99.9)	0.08 (0.04–0.16)
Tmax+ANC+CRP2
All ages (N=11008; IBI=285)	96.8 (94.1–98.5)	35.3 (34.4–36.2)	99.8 (99.6–99.9)	0.09 (0.05–0.17)
8–21 days (N=2221; IBI=110)	99.1 (95.0–100.0)	26.5 (24.6–28.4)	99.8 (99.0–100.0)	0.03 (0.00–0.24)
22–28 days (N=1538; IBI=57)	94.7 (85.4–98.9)	41.2 (38.7–43.7)	99.5 (98.6–99.9)	0.13 (0.04–0.39)
29–60 days (N=7249; IBI=118)	95.8 (90.4–98.6)	36.7 (35.6–37.8)	99.8 (99.6–99.9)	0.12 (0.05–0.27)
22–60 days (N=8787; IBI=175)	95.4 (91.2–98.0)	37.5 (36.4–38.5)	99.8 (99.5–99.9)	0.12 (0.06–0.24)

¹Infant classified as non-low-risk if PCT>0.5ng/dL and/or ANC>4000/mm³

²Infant classified as non-low-risk if Tmax≥38.6°C, ANC>4000/mm³, and/or CRP≥20mg/L (≥2 mg/dL);

Abbreviations: ANC, absolute neutrophil count; CRP, c-reactive protein; IBI, invasive bacterial infection; PCT, procalcitonin; Tmax, maximum temperature

Table 2:

Inflammatory markers and culture results of infants with bacterial meningitis who would have been misclassified as low risk

Age (days)	Inflammatory Markers1 (Tmax+ANC+CRP) and (ANC+PCT)				Urinalysis	Urine Culture	Blood Culture	CSF Culture	Disposition
	Tmax2	CRP	ANC	PCT
13	Normal	Normal	Normal	Not obtained	Normal	No Growth	GBS	GBS	Admitted
14	Abnormal	Not obtained	Normal	Normal	Normal	E. coli	No Growth	E. coli	Admitted
34	Normal	Abnormal	Normal	Normal	Abnormal	E. coli	No Growth	Haemophilus spp.	Admitted
38	Normal	Normal	Normal	Not obtained	Normal	Enterococcus spp.	E. coli	E. coli	Admitted

¹Inflammatory markers considered abnormal at the following values: Tmax ≥38.6°C; CRP ≥20 mg/L (≥2 mg/dL); ANC >4,000/mm³; and PCT >0.5 ng/dL

²Tmax defined as the highest temperature documented in the ED, at home, or in an outpatient clinic within 24 hours prior to presentation

Abbreviations: ANC, absolute neutrophil count; CRP, c-reactive protein; GBS, Group B streptococcus; PCT, procalcitonin; Tmax, maximum temperature

Discussion

This multicenter study validates the use of the AAP-recommended risk stratification strategies for well-appearing febrile infants 22–60 days and increases the generalizability of prior studies^4,6 by inclusion of infants evaluated at both children’s and general hospitals. Both AAP CPG recommended risk stratification strategies have high sensitivity for identifying febrile infants with IBIs. PCT+ANC had the highest specificity, with fewer infants classified as non-low-risk potentially resulting in fewer lumbar punctures and hospitalizations. Although two infants 22–60 days with meningitis would have been misclassified as low risk, both were hospitalized, which may indicate other factors raised the suspicion for meningitis (e.g., clinical appearance). Further investigation is needed to assess outcomes of the rare infants with IBIs misclassified as low risk and to further evaluate the use of these strategies in infants 8–21 days.

Limitations include those inherent with retrospective data collection, including potential infant misclassification as well-appearing. Laboratory results were collected in a binary manner of normal vs. abnormal, limiting additional analyses to assess different cutoff values. Additionally, we used ANC>4000/mm³ as our cutoff for the Tmax+ANC+CRP group. It is possible that sensitivity would be lower, and specificity higher, if 5200/mm³ was used, as recommended when PCT is unavailable.⁵

In conclusion, the AAP-recommended risk stratification strategies for febrile infants are excellent at ruling out IBI. Clinicians can have confidence in not obtaining lumbar punctures in low-risk infants. With higher specificity, use of PCT+ANC can potentially result in less resource utilization and harm.

Abbreviations:

AAP

American Academy of Pediatrics

ANC

Absolute neutrophil count

CPG

Clinical Practice Guideline

CRP

C-reactive protein

CSF

Cerebrospinal fluid

IBI

Invasive Bacterial Infection

LR

Likelihood Ratio

NPV

Negative Predictive Value

PCT

Procalcitonin

REVISE II

Reducing Excessive Variability in Infant Sepsis Evaluation II

Tmax

Maximum temperature