PECARN febrile infant rule

The diagnosis and risk stratification of febrile young infants continues to present a clinical challenge. Serious bacterial infection (SBI) rates in infants ≤60 days have continued to be reported between 8-13%. Despite several different classification rules and pathways, we continue to struggle to accurately delineate which infants have SBI and which do not. A paper titled “A Clinical Prediction Rule to Identify Febrile Infants 60 days and Younger at Low Risk for Serious Bacterial Infections” was published in JAMA Pediatrics in February of 2019.​1​ The authors sought to derive a new clinical prediction rule for infants with fever. The research was conducted as part of the Pediatric Emergency Care Applied Research Network (PECARN). We discussed this publication with lead author Dr. Nathan Kuppermann on a podcast and summarize our discussion below. 

Podcast

 

Clinical Objective

To empirically derive and validate a prediction rule to identify febrile infants ≤60 days at low risk for SBI.

Methods

Subjects

The data was collected from 26 EDs involved in the PECARN collaborative. The data was collected prospectively as part of a parent study evaluating the RNA microarray analysis for detecting of bacterial infections. Infants were eligible if they were ≤60 days of age, with fever, who had blood cultures collected. Fever was defined as rectal temperature of at least 38℃ in the ED, a prior health care setting, or at home, within the last 24 hours. Exclusion criteria were those who were critically ill in appearance, received antibiotics within the preceding 48 hours, history of prematurity, pre-existing medical conditions, indwelling devices, or known soft tissue infection. Notably infants were NOT excluded if they had otitis media.

Variables

The potential variables that were analyzed were clinician unstructured gestalt, Yale Observation Scale (YOS), and lab results including CBC  (including total WBC and ANC), urinalysis (UA), and procalcitonin. CRP was not included due to limited blood availability and a literature review showing that procalcitonin was a superior marker to CRP for evaluating febrile infants. Viral testing was not included as these were not typically available for decision making in the ED at the time of this study. CSF results were not included as predictors as part of the goal of these rules is to decrease the need for lumbar puncture. CSF culture was included to identify meningitis.

A urinary tract infection (UTI) was defined as ANY ONE of the following:

  1. At least 1,000 CFU/mL for cultures obtained by suprapubic aspiration
  2. At least 50,000 CFU/mL from catheterized specimens
  3. 10,000 to 50,000 CFU/mL from catheterized specimens in association with an abnormal UA

An abnormal UA was defined as ANY ONE of the following (LE and nitrite were considered positive if any amount including trace):

  1. Positive leukocyte esterase (LE)
  2. Positive nitrite
  3. >5 WBC/HPF

Data Analysis

The authors applied recursive partitioning analysis to both important variables and optimal thresholds for the variables. The strength of the recursive partitioning analysis is that it allows the data to show the important variable and cutoffs rather than the authors setting pre-determined lab cutoffs. It also develops a sequential rule where the most important variable is considered first, followed by the second most important variable, and so on. The authors of this paper describe this well in the methods and visually present this in their figures.

Results

A total of 1,896 febrile infants were included in the data set (908 derivation, 913 validation). The overall SBI rate was 9.3%. This was primarily composed of UTI (8.3%), with only 0.5% having meningitis and 1.4% with bacteremia.

The decision tree retained 3 variables at the end of recursive partitioning analysis as important for identification of the low risk cohort (in order):

  1. Negative UA
  2. ANC ≤ 4,090/mL
  3. Procalcitonin ≤ 1.71 ng/mL

With these variables, in the derivation set, the rule has a sensitivity of 98.8% and a specificity of 63.1%.

Interestingly the authors also considered altering the cutoffs of the ANC and procalcitonin to historically accepted cutoffs of ANC of 4,000 and procalcitonin of 0.5 (available in the electronic supplement to the article). Altering the ANC cutoff did not change the results. Lowering the procalcitonin level reduced the specificity to 53%.

Three febrile infants were misclassified by the rule and are discussed in detail in the article.

Authors Conclusions

“We derived and validated an accurate prediction rule to identify febrile infants 60 days and younger at low risk for SBIs using 3 easily obtainable, objective variables: the urinalysis, the ANC, and serum procalcitonin. Once further validated, implementation of the rule has the potential to substantially decrease the use of lumbar punctures, broad-spectrum antibiotics, and hospitalization for many febrile infants 60 days and younger.”​1​

Our Conclusion

This is an interesting and important contribution to the care of febrile infants ≤60 days old. We found it interesting that only 3 variables eventually were included in the rule and all are simple to obtain and interpret. The authors (in the article and in direct discussion with Dr. Kuppermann) do advise caution in using this rule in infants ≤28 days of age due to the risk of herpes encephalitis. We feel this can be implemented in infants age 29-60 days with confidence, and look forward to future studies that evaluate the safety of this rule in the youngest infants.

Listen to other PECARN podcasts hosted by ALiEM and read the AAP’s algorithm for the ED evaluation and management of UTI.

Reference

  1. 1.
    Kuppermann N, Dayan P, Levine D, et al. A Clinical Prediction Rule to Identify Febrile Infants 60 Days and Younger at Low Risk for Serious Bacterial Infections. JAMA Pediatr. February 2019. https://www.ncbi.nlm.nih.gov/pubmed/30776077.
Jason Woods, MD

Jason Woods, MD

ALiEM Podcast Editor for ACEP E-QUAL Series
Assistant Professor
Department of Pediatrics, Section of Emergency Medicine
University of Colorado, School of Medicine
Jason Woods, MD

@jwoodsmd

PEM physician. Creator of @littlebigmed podcast. Views are my own and do not represent medical advice. RT≠endorsement