Podcast: Play in new window

Subscribe: RSS

Reference: Aronson PL, et al. Prediction Rule to Identify Febrile Infants 61–90 Days at Low Risk for Invasive Bacterial Infections. Pediatrics. September 2025

Date: January 6, 2026

Dr. Jillian Nickerson

Guest Skeptic: Dr. Jillian Nickerson is a pediatric emergency medicine attending at Children’s National Hospital and Assistant Professor of Pediatrics and Emergency Medicine at The George Washington University School of Medicine and Health Sciences in Washington, DC. Prior to completing her PEM fellowship, she completed an emergency medicine residency at Mount Sinai in New York. Now she is also the associate program director for the pediatric emergency medicine fellowship program at Children’s National Hospital.

Background:

Fever is a common complaint that we encounter in the emergency department. In general, we want to be careful in our counseling and our practice not to perpetuate many of the myths and misconceptions that contribute to fever phobia.

But there are certain populations where fever does get us a bit worried. When infants present with fever, we have to think about evaluating for other sources of infection such as bacteremia or meningitis, termed invasive bacterial infections (IBI). Fortunately, the prevalence of IBI tends to be low, but missing one could lead to significant morbidity or mortality. How do we determine whom to test and what tests to perform?

We’ve covered multiple clinical decision rules for risk-stratifying febrile infants before on the SGEM:

  • SGEM #171: Step-by-Step Approach to the Febrile Infant
  • SGEM#296: She’s Got the Fever but Does She Need an LP, Antibiotics or an Admission?
  • SGEM#341: Are the AAP Guidelines for the Evaluation and Management of the Well-Appearing Febrile Infant
  • SGEM#387: Lumbar Punctures in Febrile Infants with Positive Urinalysis
  • SGEM #474: Help! Which Clinical Decision Aid Should I Use to Risk Stratify Febrile Infants?

Some of these clinical decision rules like Step by Step can be applied to infants up to 90 days. Others like the 2021 American Academy of Pediatrics (AAP) clinical practice guideline and the Pediatric Emergency Care Applied Research Network (PECARN) clinical decision rule, only include infants up to 60 days.

Clinical Question: Is there an accurate prediction rule to identify well-appearing febrile infants 61–90 days old who are at low risk for invasive bacterial infection (IBI)?

Reference: Aronson PL, et al. Prediction Rule to Identify Febrile Infants 61–90 Days at Low Risk for Invasive Bacterial Infections. Pediatrics. September 2025

  • Population: Non-ill-appearing febrile infants 61–90 days who had evaluation with both urinalysis/urine dipstick and blood culture
    • Excluded: infants who were critically ill (ESI level 1, intubated, received vasoactive medication), death in the ED, prematurity ≤32 weeks, substantial pre-existing medical or surgical conditions, skin or soft tissue infections, home antibiotic use before ED visit
  • Intervention: Derivation of a clinical prediction rule using urinalysis, temperature, ANC, ± procalcitonin.
  • Comparison: none
  • Outcome:
    • Primary Outcome: Accuracy of the prediction rule to identify infants at low risk for IBI, defined as bacteremia or bacterial meningitis.
    • Secondary Outcomes: none
  • Trial: Retrospective cohort study

Dr. Nathan Kuppermann

Dr. Paul Aronson

Authors: Dr. Paul Aronson is a pediatric emergency medicine attending and Professor of Pediatrics and Emergency Medicine at Yale School of Medicine. He is the Deputy Director of the Pediatric Residency Program and leads the Research Track.

Dr. Nathan Kuppermann is executive vice president, chief academic officer of Children’s National Hospital and director of the Children’s National Research Institute. He also serves as chair of the Department of Pediatrics and associate dean of Pediatric Academic Affairs at the George Washington University School of Medicine and Health Sciences. Dr. Kuppermann is a pediatric emergency medicine physician, clinical epidemiologist and leader in emergency medical services for children.

Authors’ Conclusions: We derived two accurate clinical prediction rules to identify febrile infants 61–90 days at low risk for invasive bacterial infections when urine and blood testing are obtained. Prospective validation is needed.

Quality Checklist for Clinical Decision Rules:

  1. The study population included or focused on those in the ED. Yes
  2. Where was the study conducted (external validity). Conducted across 17 EDs in the PECARN Registry over 10 health systems (with many pediatric EDs).
  3. The patients were representative of those with the problem. Unsure.
  4. All important predictor variables and outcomes were explicitly specified. Yes
  5. This is a prospective, multicenter study including a broad spectrum of patients and clinicians (level II). No
  6. Clinicians interpret individual predictor variables and score the clinical decision rule reliably and accurately. Yes
  7. Is this an impact analysis of a previously validated CDR (level I study)? No
  8. For Level I studies, impact on clinician behavior and patient-centric outcomes is reported. N/A
  9. The follow-up was sufficiently long and complete. Yes
  10. The effect was large enough and precise enough to be clinically significant. Unsure.
  11. Funding of the Study: Eunice Kennedy Shriver National Institute of Child Health and Human Development. No financial conflicts of interest.
  12. Did the authors declare any conflicts of interest? The authors reported no conflicts of interest to disclose.

Results: They included 4,952 infants. The median age was 72 days, and 54% male. Median maximum qualifying temperature was 38.7°C. Urinalysis was positive in 18%, LP/CSF testing was performed in 10%, antibiotics were given in 26%, and 34% were hospitalized. Approximately 100 (2%) tested positive for IBI with 95 cases of bacteremia and 5 cases of bacterial meningitis. A little bit over half (57%) with bacteremia also had UTI.

Of those infants 1207 (24%) had procalcitonin and absolute neutrophil count (ANC) measured. That group had 27 with IBIs with 2 being bacterial meningitis.  Low risk predictors:

  • Procalcitonin <0.24 ng/mL
  • ANC < 10,710 cells/mm3

Key Results: This clinical prediction rule for risk-stratifying febrile infants 61-90 days based on urine and temperature >38.9°C had a sensitivity of 86%, specificity of 58.9%, NPV of 99.5%, and -LR of 0.24, but still needs external validation.

This decision rule missed 14 infants with IBIs (13 with bacteremia and one with Group B Strep meningitis).

There was a second decision rule that included procalcitonin ≤0.24 ng/mL and ANC  ≤10,710 cells/mm3. The derivation sensitivity was 100% but dropped to 85.2% on cross-validation. The specificity was around 65-68%. NPV ranged from 99.5-100%, Negative likelihood ratio was 0.22.

Tune in to the podcast to hear Drs. Aronson and Kuppermann answer our nerdy questions.

Selection Bias

This secondary analysis included only febrile infants aged 61-90 days who underwent both urine and blood testing. A total of 20,211 infants in that age range had fevers, but only 30% of them had urine and blood cultures obtained. It’s also mentioned that the included infants had higher maximum qualifying temperatures, more assigned ESI triage level 2, and received parenteral antibiotics or were hospitalized. It’s possible that these infants may have been deemed sicker than those who did not undergo testing. The study was unable to capture the clinical decision-making that determined which infants underwent testing and which did not.

How do you think this selection bias could impact your results?

Overfitting the Data

 The PCT and ANC rule showed perfect sensitivity in derivation but lower sensitivity on cross-validation (4 false negatives). This is a pattern that may represent model instability especially when dealing with uncommon outcomes. Increasing model complexity can improve apparent performance in the derivation set but worsen performance in validation because of overfitting.

What steps did you take to try to limit overfitting and what changes if any do you anticipate in making to this CDR for external validation?

The “Original” PECARN

 Although this new clinical decision rule has a high NPV, we must also recognize the limitation that the prevalence of IBI is low. As disease prevalence decreases, NPV increases. The study team did look at this with the “original” PECARN rule’s rounded cutoffs of procalcitonin ≤0.5 and ANC ≤4000 without urinalysis. The sensitivity was 100% (95% CI 87.2-100) and specificity was 49.7% (95% CI 46.8-52.6). This was in the supplemental section.

While we’re waiting for external validation of this new clinical decision rule, why not use the tried-and-true existing clinical decision rule? One less thing with new cutoffs for inflammatory markers to remember right?

90 Days and Beyond!

 The clinical decision rule in this study, if and when externally validated, would apply to infants up to 90 days. What about beyond that? There’s quite a bit of variation in practice when it comes to workup for infants 2-6 months presenting with fever to the emergency department.

How do you approach the workup of infants over 90 days? 

Prematurity

Many of the existing clinical decision rules exclude infants born prematurely. In reality, we also encounter these patients in the ED.

How do you approach the workup of a febrile premature infant? 

Bonus Question: Respiratory Virus Testing

You report that you were unable to assess the results of respiratory viral testing as a predictor because of missing data, but we do know that febrile infants with viral infections do seem to have lower prevalence of IBI compared to those without.

In your clinical practice, how do you manage infants with viral symptoms? Is there a role for obtaining a respiratory viral PCR test?

Comment on the Authors’ Conclusion Compared to the SGEM’s Conclusion: We agree with the authors’ conclusion.

SGEM Bottom Line: While this new clinical prediction rule for febrile infants 61-90 days has a high negative predictive value, it still needs to be externally validated.

Case Resolution: You acknowledge the parents’ concern and explain to them that their daughter’s age makes the approach to testing a little bit different compared to if she were still a really young infant. While there is still some risk of urinary tract infection, which is most common, and bacteremia, the risk of bacterial meningitis is lower.

You explain that you’d recommend at least checking the urine and engage in shared-decision-making about whether to pursue blood tests.

Clinical Application: Tune in to hear the response from Drs. Aronson and Kuppermann

Is this clinical prediction tool ready for us even without external validation?

What do you do if the infant does NOT meet low risk criteria based on this tool?

What Do I Tell the Patient? I’m sorry your daughter has a fever. Based on her age, we approach how we work-up a fever a little bit differently in comparison to if she were younger. We can check her urine for signs of a urinary tract infection. This is the most common bacterial infection in this age group. We could also obtain blood tests to determine whether bacteria are present in her blood. Her risk of infection, including bacterial meningitis, is lower given her age.

I would recommend at least checking her urine. Let’s talk more about doing the blood tests.

Remember to be skeptical of anything you learn, even if you heard it on the Skeptics’ Guide to Emergency Medicine.

References:

  1. Aronson PL, Mahajan P, Nielsen B, et al. Risk of bacterial infections in febrile infants 61 to 90 days old with respiratory viruses. Pediatrics. 2025;156(1):e2025070617. doi:10.1542/peds.2025-070617
  2. Mahajan P, Browne LR, Levine DA, et al. Risk of bacterial coinfections in febrile infants 60 days old and younger with documented viral infections. J Pediatr. 2018;203:86-91.e2. doi:10.1016/j.jpeds.2018.07.073
  3. Green RS, Sartori LF, Florin TA, et al. Predictors of invasive bacterial infection in febrile infants aged 2 to 6 months in the emergency department. J Pediatr. 2024;270:114017. doi:10.1016/j.jpeds.2024.114017
  4. Green RS, Sartori LF, Lee BE, et al. Prevalence and management of invasive bacterial infections in febrile infants ages 2 to 6 months. Ann Emerg Med. 2022;80(6):499-506. doi:10.1016/j.annemergmed.2022.06.014