Febrile infant: who has serious bacterial infection

A mother arrives at the emergency department with a five-week-old baby. He had a fever at home, measured with the family thermometer, and now he is breastfeeding and seems well. The temperature at the facility is normal. The resident looks at the child, who smiles, and thinks about sending him home. The preceptor asks to stop and estimate the risk before deciding.

This is one of the most classic scenarios in pediatrics. Most small febrile infants have a mild viral infection, but some have a urinary tract infection, bacteremia or meningitis, and the baby's appearance can be misleading. The goal of this text is to show how studies help estimate who has a serious bacterial infection, what tools exist and where they fail.

Key points

• Serious bacterial infection in the small febrile infant is rare, but the probability changes considerably with age.
• The first month of life concentrates the highest risk of bacteremia and meningitis.
• Urinary tract infection is the most common bacterial infection in this age group.
• Clinical rules combine age, urinalysis, temperature and blood markers.
• Current rules are very sensitive and not very specific, so they reduce few unnecessary tests.
• Fever reported by parents counts, even when the baby arrives without fever.

Why the small febrile infant is a separate problem

Babies in the first months of life respond to infection differently from older children. Signs are often subtle, fever may be the only finding, and the appearance may be good even with bacteremia. A recent clinical review summarizes this point by recalling that the evaluation of the febrile infant up to 90 days remains difficult and full of variation between services (Wilcox et al., 2024).

The terms used in studies help read the literature. Serious bacterial infection usually includes urinary tract infection, bacteremia and bacterial meningitis. Invasive bacterial infection includes only bacteremia and meningitis, which are the most dangerous forms. The review by Wilcox et al. describes invasive infection in 1 to 3 per 100 small febrile infants and urinary tract infection in 9 to 17 per 100 (Wilcox et al., 2024).

This difference between the two groups is important for reasoning. Urinary tract infection is much more common, and therefore urinalysis carries great weight in the evaluation. Bacteremia and meningitis are rare, but the harm of missing them is great, and that is why decision rules were designed to almost never miss these conditions.

The role of age

Age in days is the variable that most changes the baseline probability. A systematic review of 12 studies, with 15,713 cultures, showed that bacteremia occurred in 2.9 per 100 febrile newborns and meningitis in 1.2 per 100 (Biondi et al., 2019). In the second month of life, the values fell to 1.6 per 100 for bacteremia and 0.4 per 100 for meningitis (Biondi et al., 2019).

The authors propose that these numbers serve as pretest probability for clinical decisions (Biondi et al., 2019). It is the same reasoning discussed in the text on pretest probability: a test result only makes sense when you know where you start from. In the febrile infant, starting from age avoids both excessive testing in older infants and false security in newborns.

How much the fever reported by parents is worth

Many babies arrive at the facility already without fever. The family measured at home, gave a bath, and the baby is afebrile at triage. The review by Wilcox et al. recommends taking fever reported by parents seriously, treating the baby as febrile even when the temperature at the facility is normal (Wilcox et al., 2024).

This care relates to the baseline rate in diagnosis. Fever measured at home places the baby in the risk group, and normal temperature on arrival changes this probability little. Dismissing the family's report is a reasoning error that appears frequently in case discussion sessions.

Does high fever change the calculation?

A common question is whether a higher fever indicates bacterial infection. A study of 1057 febrile young infants found serious bacterial infection in 19.6 out of 100 (Tan and colleagues, 2022). In that study, a temperature of 39 °C or above had a sensitivity of 15.5 out of 100 and a specificity of 90.4 out of 100 for serious bacterial infection (Tan and colleagues, 2022).

In practical terms, a high fever raises suspicion, but a low fever hardly rules anything out. Most infants with serious infection had a temperature below that threshold. The text on signs, symptoms and history shows how a finding can carry different weight for confirming and for ruling out a hypothesis.

The clinical decision rules

To organize the assessment, research groups created rules that combine age, appearance, urinalysis and blood markers. A scoping review identified 32 clinical decision tools for the febrile young infant, and 57 out of 100 of them came from North America (Umana and colleagues, 2025).

The most validated tools were the PECARN rule, the American Academy of Pediatrics (AAP) guideline, the Step-by-Step approach and the Aronson score (Umana and colleagues, 2025). The most common variables were age, urinalysis, temperature, C-reactive protein and absolute neutrophil count (Umana and colleagues, 2025). The specificity of the PECARN rule ranged from 36 to 64 out of 100 in the validation studies (Umana and colleagues, 2025).

What the rules have in common

Almost all the rules follow the same logic. First, they separate the infant who looks ill, who needs full investigation and antibiotics. Then they look at age, because the newborn receives a more complete assessment. Finally, they use urinalysis and inflammatory markers to separate the low-risk infants.

The review by Wilcox and colleagues highlights that urinalysis is a good screening test for urinary infection in this age range (Wilcox and colleagues, 2024). The same review recommends lumbar puncture in infants younger than 28 days (Wilcox and colleagues, 2024). Between 29 and 60 days, the decision about the puncture depends on the markers and the child's appearance (Wilcox and colleagues, 2024).

Biomarkers in a trial

A randomized clinical trial tested the Lab-score, which combines procalcitonin, C-reactive protein and urinalysis, in 271 febrile children (Lacroix and colleagues, 2014). Antibiotic prescription was 41.2 out of 100 in the Lab-score group and 42.1 out of 100 in the usual care group (Lacroix and colleagues, 2014). A Lab-score of 3 or above had a sensitivity of 85.1 out of 100 and a specificity of 87.3 out of 100 (Lacroix and colleagues, 2014).

The result shows a common gap between the accuracy of a test and its effect on management. The score discriminated well, but its day-to-day use changed prescription little. This pattern appears in many studies of decision tools and deserves attention from anyone considering adopting a new rule.

What the FIDO cohort showed

The FIDO study, conducted in the United Kingdom and Ireland, was designed to test the rules in a population outside North America. The protocol recalls that British guidance was conservative and that serious bacterial infection appeared in about 10 to 20 out of 100 febrile young infants (Umana and colleagues, 2023).

The cohort included 1821 infants at 35 services, with a median age of 46 days (Umana and colleagues, 2024). Invasive bacterial infection appeared in 67 infants, or 3.7 out of 100 (Umana and colleagues, 2024). The most common germs were Escherichia coli and Streptococcus agalactiae (Umana and colleagues, 2024).

High sensitivity, low specificity

The AAP guideline and the British Society for Antimicrobial Chemotherapy (BSAC) rule had a sensitivity of 0.99 for invasive infection (Umana and colleagues, 2024). Specificity was 0.23 for the AAP rule and 0.20 for the BSAC rule (Umana and colleagues, 2024).

Reading these numbers is straightforward. The rules almost never let a baby with bacteremia or meningitis pass, but they classify the vast majority of babies who do not have invasive infection as high risk. In practice, many babies still receive a lumbar puncture, hospitalization and antibiotics without needing them.

How to read sensitivity and specificity in this age range

Sensitivity and specificity describe the test, but the decision also depends on how common the disease is in the group evaluated. When invasive infection is rare, even a very sensitive rule produces many false positives, and most babies classified as high risk end up without infection. This effect is predictable and explains much of the excess of procedures described in the cohorts.

The reverse reasoning also holds. A negative result on a very sensitive rule has high value for ruling out invasive infection, as long as the baby belongs to the population in which the rule was tested. That is why correct selection of the baby, with age, appearance and absence of prior disease, is part of the test, and applying the rule outside this group reduces confidence in the negative result.

Study	Population	Main finding	Source
Review of 12 studies	15,713 cultures	Bacteremia in 2.9 per 100 newborns and 1.6 per 100 in the second month	(Biondi and colleagues, 2019)
FIDO cohort	1821 babies in 35 services	AAP and BSAC rules with sensitivity of 0.99 and specificity of 0.20 to 0.23	(Umana and colleagues, 2024)
Scoping review	32 decision tools	PECARN specificity of 36 to 64 per 100	(Umana and colleagues, 2025)
Lab-score trial	271 children	Antibiotic prescription of 41.2 versus 42.1 per 100	(Lacroix and colleagues, 2014)
Boston 2.0	616 babies	Lumbar puncture from 66.2 to 31.9 per 100	(Dorney and colleagues, 2022)

Rules adapted to other places

The rules were created in specific contexts, and performance changes when the population changes. In Singapore, the FIRST score was derived in 1002 febrile young babies, with serious bacterial infection in 22.4 per 100 (Chong and colleagues, 2023). The area under the curve rose from 0.71 to 0.87 when the authors added laboratory tests to the clinical model (Chong and colleagues, 2023).

This result shows that history and physical examination alone discriminate little in this age range. The gain came from blood and urine tests. At the same time, the frequency of serious infection in this cohort was higher than in North American studies, which changes the predictive value of any rule.

Fewer punctures without missing cases

In Boston, one service implemented a revised local guideline, called Boston 2.0, and compared results before and after in 616 babies (Dorney and colleagues, 2022). Lumbar puncture fell from 66.2 to 31.9 per 100 babies, and antibiotic use fell from 62.4 to 36.2 per 100 (Dorney and colleagues, 2022). The authors did not identify invasive infection left untreated after the change (Dorney and colleagues, 2022).

The study is from a single center and with a before and after design, so it needs confirmation. Even so, it shows that it is possible to reduce procedures without losing safety when the rule is well implemented and monitored. The text on clinical decision making discusses how to weigh the harm of a test against the harm of an undiagnosed disease.

Does a positive viral test rule out bacterial infection?

During the pandemic, many febrile babies tested positive for SARS-CoV-2. One study compared babies with and without COVID-19 and found serious bacterial infection in 24 of 134 babies without COVID-19 and in none of the 30 babies with COVID-19 (Guernsey and colleagues, 2022).

The number of babies with COVID-19 was small, so the result suggests lower risk, but it does not serve to skip the evaluation. The review by Wilcox and colleagues discusses the topic and recalls that bacterial coinfection can occur, especially urinary infection (Wilcox and colleagues, 2024). A positive viral test reduces the probability, but the urine test remains indicated.

In practice, the viral test is one more piece of information that adjusts the probability, alongside age, appearance and tests. The most common mistake is to treat it as a final point, ending the investigation as soon as a diagnosis that explains the fever appears. This shortcut has a name in the clinical reasoning literature, premature closure, and it is especially risky in an age range where a urinary tract infection can coexist with a respiratory virus. Keeping urine in the routine and reviewing the baby when the course departs from what is expected are simple ways to avoid the shortcut, and recording the reason for each decision makes later review by another colleague on the team easier.

Care for preterm infants

The rules were derived in term infants. The review by Wilcox and colleagues notes that preterm infants were excluded from the main studies (Wilcox and colleagues, 2024). For these babies, corrected age and neonatal history enter the calculation, and the decision is usually more conservative.

The same applies to babies with chronic disease, recent antibiotic use or prolonged hospitalization in the neonatal period. These groups were left out of the derivation cohorts, and applying the rules to them goes beyond what was studied. Differential diagnosis with clues helps remind you that context changes the weight of each finding.

What changes for the clinician

The studies point to a reasoning path that can be trained. The sequence below summarizes the most consistent points from the reviewed literature.

• Start with age in days, which defines the baseline probability of bacteremia and meningitis (Biondi and colleagues, 2019).
• Consider febrile the baby with fever measured at home, even if afebrile on arrival (Wilcox and colleagues, 2024).
• Order a urine test, because urinary tract infection is the most common bacterial infection (Wilcox and colleagues, 2024).
• Use blood markers to separate low-risk babies between 29 and 60 days (Umana and colleagues, 2025).
• Remember that current rules err little on the low side and a lot on the high side (Umana and colleagues, 2024).

This list does not replace bedside assessment. The baby's appearance, the family's ability to return to the service and the ease of reassessment also weigh in. The text on diagnostic uncertainty shows how to plan reassessment when the decision is to discharge.

How to talk with the family

The decision about the young febrile baby almost always involves the family. Explaining why age weighs so much, why a urine test is ordered even without urinary symptoms and why a lumbar puncture may be indicated helps parents take part in the choice. When the option is to observe at home, the guidance must be clear about the signs that call for immediate return, such as refusal to feed, unusual drowsiness, grunting or worsening skin color.

This conversation also protects against another common bias, anchoring on the first impression. A reassessment plan agreed with the family, with a set time to review the baby and the culture result, makes discharge a stage of reasoning, and the initial decision becomes a hypothesis to confirm. The text on affective bias shows how emotion also weighs in at that moment.

Where reasoning usually fails

Three errors appear often. The first is trusting the good appearance of a newborn. The second is ignoring fever reported by the parents. The third is failing to order a urine test because the child has an obvious viral respiratory picture.

Discussing these errors in morbidity and mortality sessions helps the team recognize blind spots. Recording age in days, the source of the fever and the urine result in the chart, as in the text on organizing reasoning in the chart, also makes case review easier. For those who want to practice cases in this age range, the text with clinical reasoning exercises for pediatrics brings other scenarios.

Overdiagnosis and excess procedures

The low specificity of the rules has a consequence. Many infants without invasive infection receive a lumbar puncture, hospitalization and intravenous antibiotics, with discomfort for the child and worry for the family. The text on overdiagnosis discusses this balance between safety and excess.

The Boston experience suggests that revising local rules can reduce this excess (Dorney et al., 2022). Other cohorts, such as FIDO, show that a rule that keeps sensitivity and gains specificity is still missing (Umana et al., 2024). This is an active field of research.

How NEXO helps train this reasoning

In NEXO, the case of the day is fictional, with a real clinical presentation and up to six progressive clues, and you propose the diagnosis by searching for the ICD-10 code.

The limits of use of the game are described on the NEXO Economy page. If you research the teaching of clinical reasoning, the page for Researchers explains how to get in touch. Professionals who write or review cases find information on Curator and on Creators, and institutions can learn about the proposal on the partnerships page. Training with differential diagnosis and clues and dealing with diagnostic uncertainty complete this practice.

Frequently asked questions

What is the chance that a febrile young infant has bacteremia?

In a review of 12 studies, bacteremia appeared in 2.9 per 100 febrile newborns and in 1.6 per 100 febrile infants in the second month of life (Biondi et al., 2019).

Does a fever measured at home count as fever?

It does. The review by Wilcox et al. recommends taking fever reported by parents seriously, even when the infant arrives at the service without fever (Wilcox et al., 2024).

What is the most common bacterial infection at this age?

Urinary tract infection, described in 9 to 17 per 100 febrile young infants, versus 1 to 3 per 100 with invasive infection (Wilcox et al., 2024).

Are the decision rules safe?

In the FIDO cohort, the AAP and BSAC rules had a sensitivity of 0.99 for invasive infection, with a specificity of 0.20 to 0.23 (Umana et al., 2024).

Does high fever indicate bacterial infection?

In 1057 infants, a temperature equal to or above 39 °C had a specificity of 90.4 per 100 and a sensitivity of 15.5 per 100 for serious infection (Tan et al., 2022), so it helps to raise suspicion, but does little to rule it out.

Take-home messages

• Age in days is the data point that most changes the probability of invasive infection.
• The first month of life concentrates the highest risk of bacteremia and meningitis.
• Fever reported by parents should be taken seriously.
• The urine test is central, because urinary tract infection is the most common.
• Current rules are very sensitive and not very specific.
• Preterm infants and infants with chronic disease were left out of the main studies.

References

1. Biondi EA, Lee B, Ralston SL, et al. Prevalence of Bacteremia and Bacterial Meningitis in Febrile Neonates and Infants in the Second Month of Life: A Systematic Review and Meta-analysis. JAMA Netw Open. 2019. PMID 30901044.
2. Chong SL, Niu C, Ong GY, et al. Febrile infants risk score at triage (FIRST) for the early identification of serious bacterial infections. Sci Rep. 2023. PMID 37740004.
3. Dorney K, Neuman MI, Harper MB, et al. Boston Febrile Infant Algorithm 2.0: Improving Care of the Febrile Infant 1-2 Months of Age. Pediatr Qual Saf. 2022. PMID 36337736.
4. Guernsey D, Pfeffer M, Kimpo J, et al. COVID-19 and Serious Bacterial Infection in Febrile Infants Less Than 60 Days Old. West J Emerg Med. 2022. PMID 36205666.
5. Lacroix L, Manzano S, Vandertuin L, et al. Impact of the lab-score on antibiotic prescription rate in children with fever without source: a randomized controlled trial. PLoS One. 2014. PMID 25503770.
6. Tan VSR, Ong GY, Lee KP, et al. Pyrexia in a young infant - is height of fever associated with serious bacterial infection? BMC Pediatr. 2022. PMID 35395789.
7. Umana E, Mills C, Norman-Bruce H, et al. Applying clinical decision aids for the assessment and management of febrile infants presenting to emergency care in the UK and Ireland: Febrile Infant Diagnostic Assessment and Outcome (FIDO) Study protocol. BMJ Open. 2023. PMID 37730397.
8. Umana E, Mills C, Norman-Bruce H, et al. Performance of clinical decision aids (CDA) for the care of young febrile infants: a multicentre prospective cohort study conducted in the UK and Ireland. EClinicalMedicine. 2024. PMID 39677360.
9. Umana E, Norman-Bruce H, Waterfield T. Scoping review of clinical decision aids in the assessment and management of febrile infants under 90 days of age. BMC Pediatr. 2025. PMID 40181355.
10. Wilcox H, Umana E, Fauteux-Lamarre E, et al. Conundrums in the Management of Febrile Infants under Three Months of Age and Future Research. Antibiotics (Basel). 2024. PMID 38247647.