Guideline-Based Risk Stratification for Febrile Young Infants Without Procalcitonin Measurement

Modern risk-stratification strategies have been derived to identify infants at low risk of serious bacterial infections (SBIs; urinary tract infections, bacteremia, and bacterial meningitis) and invasive bacterial infections (IBIs; bacteremia and bacterial meningitis).1–3 These incorporate different diagnostic tests and low-risk cut-points. Procalcitonin testing offers the best discrimination of low-risk infants4; however, this test is not universally available.5,6 Recent guidelines from the American Academy of Pediatrics (AAP) recommend that if procalcitonin is unavailable, C-reactive protein (CRP), absolute neutrophil count (ANC), and temperature should be used5; however, these inflammatory markers have never been evaluated together in combination. This study sought to estimate the diagnostic accuracy of CRP, ANC, and temperature at AAP-recommended cut-points, prioritizing not missing IBIs.This was a secondary analysis of quality improvement data collected prospectively to standardize diagnostic testing for febrile infants ≤60 days of age at an urban tertiary pediatric emergency department between January 2018 and October 2021. All infants had blood and urine cultures, whereas cerebrospinal fluid testing was at the discretion of the treating physician. SBIs and IBIs were classified according to current definitions.5 Final SBI/IBI status was ascertained for all infants by chart review and by telephone follow-up 14 to 28 days postdischarge for those without cerebrospinal fluid testing. This was a complete-case analysis of all previously healthy, term-born infants, 8 to 60 days old meeting AAP inclusion/exclusion criteria,5 and for whom CRP, ANC, and temperature data were available. The primary outcome was accuracy for detecting IBIs (sensitivity and negative predictive value [NPV]). Infants with CRP ≤20.0 mg/L, ANC ≤5200/mm3, and a maximum home or emergency department rectal temperature ≤38.5°C were classified as low-risk.5 AAP guidelines define abnormal ANC as “>4000, >5200/mm3”; ≤5200/mm3 was selected for the primary analysis as this threshold was derived without procalcitonin,2 and sensitivity analyses were conducted by using a cut-point of ≤4000/mm3. This study received institutional research ethics board approval and followed the 2015 Standards for Reporting of Diagnostic Accuracy guidelines.There were 957 of 1041 infants (91.9%) meeting inclusion with complete data (Table 1); 779 of 957 were >21 days old (81.4%), 27 of 957 had IBIs (2.8%), and 164 of 957 had SBIs (17.1%). Among infants with IBIs, 22 had bacteremia and 5 had bacteremia with meningitis. Test characteristics using AAP criteria are summarized in Table 2. AAP-recommended cut-points of ANC/CRP/temperature classified 431 infants (45.0%) as low risk. In the full cohort, there were no cases of IBI misclassified as low risk (0 of 27); sensitivity for identifying IBI was 100.0% (95% confidence interval [CI] 87.2–100.0) and NPV was 100.0% (95% CI 99.2–100.0). A threshold of ANC ≤4000/mm3 (Supplemental Table 3) classified 373 infants (39.0%) as low risk, maintaining high sensitivity (100.0%; 95% CI 87.2–100.0) and NPV (100.0%; 95% CI 99.0–100.0) for IBI. Among all infants, sensitivity and NPV for SBI were 83.5% and 93.7%; 27 infants with positive urine culture results were classified as low risk, among whom 18 of 27 (66.7%) had a positive urinalysis result.This is the first assessment of recent AAP guidelines for risk-stratifying febrile young infants when procalcitonin is unavailable. The AAP-recommended combination of CRP, ANC, and maximal temperature resulted in 0 misclassified IBIs in a cohort of nearly 1000 well-appearing febrile infants 8 to 60 days old, yielding sensitivity and NPV of 100%. Overall, there were 526 of 957 infants who did not meet ANC/CRP/temperature low-risk criteria. However, applying AAP guidelines could reduce lumbar punctures, hospitalizations, and antibiotic exposure for nearly half (364 of 779) of all infants as young as 22 days. Findings suggest that clinicians without access to procalcitonin testing may safely follow AAP guidelines to assess the risk of IBI using these readily available, inexpensive inflammatory markers.Although accuracy was lower for identifying SBIs, recent efforts have prioritized identifying infants at risk for IBI. Most SBIs among febrile infants are urinary tract infections,7 which are less dangerous and identified with high sensitivity/specificity by urinalysis results.8,9 AAP guidelines consider inflammatory markers independently from urinalysis results.5 Infants with isolated urinary tract infections and low-risk inflammatory markers would have been identified by urinalysis in 18 of 27 cases. Nine were included as misclassified SBIs because their urine culture grew >50 000 colony-forming units of a uropathogen and they were managed clinically as UTIs9; however, even these more likely represented asymptomatic bacteriuria or contamination.5This study has limitations. This was a single-center study and the results may not be generalizable to other settings. Systematic CRP testing began in October 2019; consequently, complete data were not available for all eligible infants. Infants may have had higher unrecorded temperatures; however, this would lower specificity without affecting sensitivity. Only 27 infants had IBIs; however, the study cohort was similarly large compared with recent large prospective cohorts used to derive clinical prediction rules.1,3 AAP-recommended thresholds of CRP and ANC were used, although different cut-points may improve performance.10 Also, the analysis was limited to infants meeting all low-risk criteria and did not assess test characteristics when inflammatory markers were discordant.The authors thank Alexandra Yannopoulos and Emma Bainbridge for technical assistance, and Xun Zhang for statistical support.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.