By: Nancy Rixe, M.D.

Background: Febrile infants are one of the scariest populations that enter the emergency department for several reasons. First, they remind us all how old we really are and how much we have aged since starting medical training. Second, due to their cute and cuddly nature, it is often difficult to discriminate between those infants (especially ❤ months old) with a potentially devastating serious bacterial infection (SBI) versus a not-so-scary viral infection. The scariest part of all? Individual studies on febrile infants have shown the prevalence of SBI to be as high as 6-10%, making them a high risk population.

Given the risk of missing a potentially devastating SBI, why don’t we just tap all of the munchkins and give them antibiotics? For the same reason I don’t give every child with a runny nose a Z-pak: the antibiotic stewardship police have me on the grid and I don’t want to lose my e-prescribing privileges. But more importantly, we want to be judicious about the tests that we do and the medications that we prescribe, because nothing in medicine is without risk.

While institutions in Boston, Philadelphia, and Rochester have put forth algorithms to identify infants who are low risk, there is currently no universally accepted protocol for identification of SBI in infants. Regardless of the clinical criteria used, adherence to these algorithms is low. In a study examining case-based compliance with established criteria, PEM physicians reported using the Philadelphia, Rochester and Boston criteria 20%, 15% and 13% of the time, respectively (Meehan, Fleegler and Bachur 2010).

For this reason, Gomez et al. decided to create a new algorithm for identification of low risk febrile infants called the Step-by-Step Approach.


  • Multicenter prospective study conducted in Europe
  • 11 Pediatric Emergency Departments from 2012 – 2014
  • A urine dipstick, urine culture (catheterized specimen or suprapubic aspiration), WBC count, CRP, procalcitonin (PCT) and blood culture were obtained for each patient
  • Patients were classified as low risk by the Step-by-Step approach, the Rochester Criteria and the Lab Score

Low Risk Criteria:
1. Normal Pediatric Assessment Triangle (Appearance, Work of Breathing, Circulation to Skin)
2. Age >21 days old

  1. No leukocyturia
    4. Procalcitonin <0.5ng/mL
    5. CRP <20mg/L or ANC <10,000/mm3

Primary Outcome: prevalence of IBI (invasive bacterial infection as defined by positive blood or CSF culture) identified by the Step-by-Step approach vs. Rochester criteria vs. Lab Score.

Inclusion: 2185 febrile infants


  • Patients with a clear source of fever
  • Patients without fever on arrival and only tactile temperature in the history
  • Absence of one or more ancillary tests in the algorithm
  • Parent refusal


    • Performance of the Step-by-Step Approach:
      • Identified 991 patients as Low Risk, 7 (0.7%) of whom had an SBI (CI 0.2 – 1.2)
      • Sensitivity: 92%
      • Specificity: 46.9%
      • PPV: 6.7
  • NPV: 99.3
  • Performance of the Rochester Criteria:
    • Identified 949 patients as Low Risk, 16 (1.6%) of whom had an SBI (CI 0.9 – 2.5)
    • Sensitivity: 81.6%
    • Specificity: 44.5%
    • PPV: 5.7
    • NPV: 98.3
  • Performance of the Lab Score:
    • Identified 1798 patients as Low Risk, 35 (1.9%) of whom had an SBI (CI 1.3 – 2.6)
    • Sensitivity: 59.8%
    • Specificity: 84%
    • PPV: 13.4
    • NPV: 98.1


  • The results are not entirely generalizable given that there is a higher prevalence of IBI in Europe than in the United States (secondary to higher rates of UTI)
  • This provides no comparison to the Boston or Philadelphia criteria (both of which incorporate lumbar puncture into the algorithm)
  • The performance of the Rochester criteria was limited by the fact that the absolute band count was not available in all participating centers


  • The Step by Step was the most accurate of the 3 approaches for ruling out an IBI as evidenced by the best sensitivity, NPV and negative LR.
  • However, as expected due to the relatively low prevalence of IBI (4.0%), the specificity, PPV, and positive LR were poor for all the 3 approaches when considering all of the risk criteria of each protocol combined.
  • It is important to realize that 6 of the 7 infants with IBI that were misidentified as low risk by the Step-by-Step had a fever duration of < 2 hours, which is too short a duration for even the PCT to rise
    • This very short fever duration makes the evaluation of these patients even more challenging and highlights the important role of a short-term PED observation in the management of these patients.
  • Also note that 4 of those 7 infants were between 21 – 28 days old, a population in whom most practitioners would still obtain an LP based on the Philadelphia and Boston Criteria

Author Conclusion: The Step-by-Step Approach is a valid tool for the management of infants with fever without a source in the ED and is superior in identifying low risk patients as compared to the Lab Score and the Rochester Criteria

Take-home Points: Febrile infants are scary, but using the Step-by-Step approach can help to accurately identify low risk infants who may not need invasive diagnostic testing and treatment for fever without a source. However, remember to interpret these results with caution especially in patients between 21 – 28 days old and in those who have had short fever duration. Consider a period observation in those infants in whom there is still clinical concern.


  1. “Validation of the “Step-by-Step” Approach in the Management of Young Febrile Infants.” Gomez et al. Pediatrics, 2016. (108).
  2. “Predictive Model for Serious Bacterial Infections Among Infants Younger than 3 Months of Age.” Bachur & Harper. Pediatrics, 2001. (108): 311 – 316.
  3. “Adherence to guidelines for managing the well-appearing febrile infant: assessment using a case-based, interactive survey.” Meehan et al. Pediatr Emerg Care, 2010. (12): 875-80.

Faculty Review: Alex Sheng, MD