Background: The Febrile Infant Dilemma

Fever in a young infant is one of the most anxiety-provoking presentations in paediatric emergency medicine. Infants aged 60 days or younger with a rectal temperature of 38.0 degrees Celsius (100.4 degrees Fahrenheit) or higher occupy a unique position on the clinical risk spectrum. Their immature immune systems, incomplete transplacental antibody protection, and often subtle clinical signs make it difficult to distinguish benign viral illness from life-threatening bacterial infection on clinical grounds alone.

Serious bacterial infection (SBI), defined as bacteraemia, bacterial meningitis, urinary tract infection, bacterial enteritis, osteomyelitis, or septic arthritis confirmed by positive culture, occurs in approximately 7% to 12% of febrile infants in this age group across published series. Left untreated, bacteraemia can progress to septic shock, and bacterial meningitis carries a mortality rate of 5% to 10% with significant long-term neurodevelopmental morbidity. The clinical imperative, therefore, is to identify and treat SBI as early as possible.

However, the overwhelming majority of febrile infants do not have SBI. Most have self-limited viral infections that require only supportive care. The traditional approach of performing a full sepsis workup (blood culture, urinalysis, urine culture, lumbar puncture, and in some cases stool studies), admitting the infant to hospital, and administering empiric intravenous antibiotics pending 48-hour culture results is safe but carries substantial burdens: painful procedures, unnecessary antibiotic exposure, nosocomial infection risk, parental anxiety, healthcare costs, and disruption of early bonding and breastfeeding. The central clinical question is whether a reliable tool can identify the subset of febrile infants who are at sufficiently low risk for SBI that they can be safely managed as outpatients without empiric antibiotics.

Historical Context and the Origin of the Rochester Criteria

The Rochester Criteria emerged from work at Strong Memorial Hospital and the University of Rochester Medical Center in Rochester, New York, during the early 1980s. At that time, the standard of care for any febrile infant younger than 90 days was mandatory hospitalisation with empiric antibiotics. There was growing recognition among paediatric infectious disease specialists that this one-size-fits-all approach was overly aggressive for many well-appearing infants, but there was no validated tool to safely identify the low-risk group.

In 1985, Dagan, Powell, Hall, and Menegus published the original derivation study. Their prospective cohort enrolled 233 previously healthy febrile infants younger than 90 days who were admitted to Strong Memorial Hospital for evaluation of possible sepsis. Through systematic analysis of demographic, clinical, and laboratory variables, they identified a set of criteria that could distinguish infants at low versus higher risk for SBI. Of the 144 infants who met all low-risk criteria, only 1 (0.7%) was found to have SBI. In contrast, among the 89 infants with one or more high-risk features, 22 (24.7%) had SBI. This dramatic separation suggested that a clinical prediction rule could safely guide management decisions.

The derivation study, however, was a single-centre effort with a relatively small sample size, and the original age range extended to 90 days. Broader validation was needed before the criteria could be adopted into clinical practice.

Validation: The Febrile Infant Collaborative Study Group

In 1994, Jaskiewicz and colleagues published the landmark validation study through the Febrile Infant Collaborative Study Group, a multicentre collaboration that enrolled 1,005 febrile infants aged 60 days or younger from multiple emergency departments across the United States. This study narrowed the applicable age range to 60 days or younger (from the original 90 days), reflecting the recognition that infants beyond 60 days have a somewhat different risk profile and can often be assessed with other clinical tools.

The results were compelling. Of the 1,005 enrolled infants, 511 (50.8%) met all Rochester low-risk criteria. Among these 511 low-risk infants, only 5 (1.0%) were found to have SBI. The remaining 494 infants who failed to meet one or more criteria had an SBI rate of 12.3% (61 of 494). The negative predictive value (NPV), the probability that an infant classified as low risk truly does not have SBI, was 98.9% (95% CI 97.2% to 99.6%). The sensitivity, the ability of the criteria to correctly identify infants who do have SBI, was 92%. The specificity was 55%, and the positive predictive value was 12.3%.

These test characteristics established the Rochester Criteria as the first widely adopted clinical prediction rule for febrile infants. The high NPV of 98.9% provided clinicians with a quantitative basis for safely discharging low-risk infants to home with close outpatient follow-up within 24 hours, without empiric antibiotics.

The Criteria in Detail

The Rochester Criteria use a binary (all-or-none) approach: every criterion must be satisfied for the infant to be classified as low risk. If any single criterion is not met, the infant falls into the higher-risk group. This conservative design maximises sensitivity at the expense of specificity, which is the appropriate trade-off when the consequence of missing a serious infection is catastrophic.

Inclusion Criteria

Age 60 days or younger: The criteria are validated only for infants within the first 60 days of life. This age group has the highest baseline risk for SBI among febrile infants and is the population in which the risk-benefit calculus of full workup versus observation is most consequential. Infants beyond 60 days are generally managed using other clinical protocols and risk stratification tools.

Rectal temperature 38.0 degrees Celsius or higher: The definition of fever in young infants requires a rectal temperature measurement, which is the gold standard for core body temperature assessment in this age group. Axillary, tympanic, and temporal artery measurements are insufficiently reliable in neonates. A documented rectal temperature of 38.0 degrees Celsius (100.4 degrees Fahrenheit) or higher in the emergency department, or a credible parental report of an equivalent rectal temperature at home, qualifies the infant for evaluation.

Clinical and History Criteria

Infant appears generally well: This is the most subjective criterion in the Rochester framework. The clinician must assess overall appearance, noting alertness, responsiveness, skin colour, perfusion, hydration status, tone, and consolability. An ill-appearing or toxic-appearing infant (lethargy, mottled skin, poor feeding, prolonged capillary refill, weak cry, irritability that is not consolable) is automatically classified as higher risk regardless of all other criteria. While the Yale Observation Scales and other structured assessment tools exist, the original Rochester derivation relied on clinical gestalt rather than a formalised scoring system. This introduces interrater variability, particularly among less experienced providers, but reflects the reality that clinical appearance remains one of the most powerful predictors of serious illness in young infants.

Born at term (37 or more weeks of gestation): Preterm infants have immature immune function, reduced transplacental immunoglobulin G transfer (which is predominantly a third-trimester process), and a higher incidence of comorbidities that may predispose to infection. The Rochester Criteria exclude preterm infants from the low-risk category, recognising that the risk profile of a 34-week former preterm infant presenting at 45 days of life is fundamentally different from that of a healthy term newborn of the same chronological age.

No perinatal antibiotics: If the infant received antibiotics during the perinatal period (intrapartum antibiotic prophylaxis for group B streptococcus, or postnatal antibiotics for suspected early-onset sepsis), this suggests either a recognised maternal risk factor for neonatal infection or a prior episode of suspected or confirmed infection. Either scenario raises the baseline risk for SBI and excludes the infant from the low-risk group. It is worth noting that intrapartum GBS prophylaxis has become nearly universal since 2002 guidelines were implemented, which has complicated the application of this criterion in contemporary practice.

No antibiotics at any time in life, including at current presentation: An infant who has ever received antibiotics may have had a prior infection that was partially treated, potentially masking ongoing bacteraemia or seeding a secondary focus. Similarly, antibiotics administered at the current visit before cultures are obtained can sterilise blood cultures and prevent reliable microbiological diagnosis. This criterion ensures that the infant being evaluated has a clean microbiological baseline.

Not treated for unexplained hyperbilirubinaemia: Neonatal jaundice requiring treatment (phototherapy or exchange transfusion) that cannot be attributed to a benign aetiology (physiological jaundice, breast milk jaundice, ABO incompatibility) raises concern for an underlying infectious or metabolic process. Urinary tract infection, in particular, is a well-recognised cause of conjugated hyperbilirubinaemia in neonates. The presence of unexplained hyperbilirubinaemia therefore removes the infant from the low-risk category.

No previous hospitalisations: A history of hospitalisation after the birth admission suggests a prior medical event (illness, surgical condition, failure to thrive, apnoea event) that may reflect an underlying vulnerability to infection or a chronic condition that increases the risk of SBI. This criterion ensures that the low-risk group includes only previously healthy infants with an uncomplicated postnatal course.

No chronic or underlying illness: Infants with known congenital anomalies (especially cardiac defects), genetic syndromes, immunodeficiency, metabolic disorders, or other chronic medical conditions have altered baseline risk profiles that the Rochester Criteria were not designed to address. These infants require individualised evaluation and management that goes beyond the scope of a population-level screening tool.

Not hospitalised longer than the mother after delivery: If the infant remained in hospital after the mother was discharged, this typically indicates a complication of the birth or neonatal period: prematurity-related issues, hypoglycaemia, respiratory distress, feeding difficulties, suspected sepsis, or jaundice requiring monitoring. Any of these suggests a more complex neonatal course and excludes the infant from the straightforward low-risk category.

No evidence of focal infection on physical examination: A thorough physical examination must exclude localised signs of infection: cellulitis, abscess, omphalitis (umbilical infection), paronychia, pustulosis, osteomyelitis (bony tenderness, swelling, pseudoparalysis), septic arthritis (joint swelling, limited range of motion), or acute otitis media (bulging, erythematous tympanic membrane). The presence of a focal bacterial infection source increases the probability of concurrent bacteraemia and mandates directed antimicrobial therapy rather than observation.

Laboratory Criteria

White blood cell count between 5,000 and 15,000 per cubic millimetre: Both leukocytosis (WBC above 15,000) and leukopenia (WBC below 5,000) are associated with increased risk of SBI. Leukocytosis reflects a robust inflammatory response to bacterial invasion, while leukopenia, particularly in a young infant, may indicate overwhelming sepsis with bone marrow suppression or peripheral consumption. The Rochester Criteria define a relatively narrow normal range that excludes infants at either extreme.

Absolute band neutrophil count 1,500 per cubic millimetre or fewer: Band forms (immature neutrophils) are released from the bone marrow in response to acute bacterial infection. An elevated absolute band count, sometimes expressed as a band-to-total-neutrophil ratio (immature-to-total ratio, or I:T ratio), is a marker of acute bacterial stress. The Rochester threshold of 1,500 bands per cubic millimetre is designed to identify infants with a significant left shift that suggests active bacterial infection. It should be noted that band counts are subject to substantial interobserver variability in manual differential interpretation, which is a recognised limitation of this criterion.

Urine white blood cells 10 or fewer per high-power field: Urinary tract infection (UTI) is the most common SBI in febrile infants, accounting for 5% to 9% of cases. Enhanced urinalysis with microscopic assessment of unspun or centrifuged urine for white blood cells is the rapid screening component of the Rochester laboratory criteria. More than 10 WBCs per high-power field is considered positive and moves the infant to the higher-risk group. It is important that the urine specimen be obtained by catheterisation or suprapubic aspiration in this age group, as bag specimens have unacceptably high contamination rates.

Fecal leukocytes 5 or fewer per high-power field (if diarrhoea is present): This criterion applies only when the febrile infant also has diarrhoea. Fecal leukocytes suggest bacterial enteritis (e.g., Salmonella, Shigella, Campylobacter), which qualifies as an SBI. If diarrhoea is not present, this component is not assessed and is considered met by default.

Understanding the Test Characteristics

The clinical utility of the Rochester Criteria is best understood through its four key test performance metrics.

Metric	Value	Clinical Meaning
Sensitivity	92%	Of all infants with SBI, 92% were correctly classified as higher risk (not low risk)
Specificity	55%	Of all infants without SBI, 55% were correctly classified as low risk
Positive Predictive Value	12.3%	Among infants classified as higher risk, 12.3% actually had SBI
Negative Predictive Value	98.9%	Among infants classified as low risk, 98.9% truly did not have SBI

The high NPV of 98.9% is the cornerstone of the Rochester Criteria's clinical value. It provides strong reassurance that an infant meeting all criteria has a very low probability (approximately 1%) of harbouring SBI. The 92% sensitivity means that 8% of infants with SBI were misclassified as low risk (5 out of 66 infants with SBI in the validation cohort). While this is a small fraction, it represents the fundamental trade-off of any screening tool: no clinical prediction rule achieves 100% sensitivity.

The relatively modest specificity of 55% means that nearly half of infants without SBI are classified as higher risk. This leads to a substantial number of unnecessary workups and hospitalisations among infants who ultimately have viral infections. The positive predictive value of 12.3% confirms this: among all infants flagged as higher risk, only about 1 in 8 actually has SBI. This overtriage is a deliberate feature of the rule's conservative design, reflecting the medical principle that the cost of missing a serious bacterial infection in a neonate far outweighs the cost of over-investigating a well infant.

Clinical Application: The Management Pathway

Low-Risk Infants (All Criteria Met)

When all 12 Rochester criteria are satisfied, the infant can be considered for outpatient management without empiric antibiotics. This management strategy was supported by the validation study and by the subsequent work of Baker, Bell, and Avner, who demonstrated in a 1993 randomised trial that outpatient observation without antibiotics was safe for low-risk febrile infants. The key elements of outpatient management include:

• Blood and urine cultures should still be obtained before discharge so that any positive result can be acted upon promptly.
• Reliable caregiver follow-up within 24 hours must be ensured. The caregiver must be able to return to the emergency department immediately if the infant's condition changes.
• Clear return precautions should be provided: worsening fever, poor feeding, decreased activity, increased irritability, rash, or any other signs of clinical deterioration.
• A follow-up telephone call or in-person visit at 24 hours allows reassessment and review of pending culture results.

It is critical to emphasise that "low risk" does not mean "no risk." The 1% SBI rate in the low-risk group means that approximately 1 in 100 low-risk infants will have a serious bacterial infection. Clinicians must exercise judgement and consider factors beyond the Rochester Criteria (parental reliability, geographic access to care, institutional protocols, and their own clinical instinct) when deciding on disposition.

Higher-Risk Infants (One or More Criteria Not Met)

Infants who fail to meet one or more Rochester criteria are classified as higher risk, with an SBI rate of approximately 12.3%. Standard management for these infants typically includes:

• Full sepsis workup: complete blood count with differential, blood culture, urinalysis and urine culture (by catheterisation or suprapubic aspiration), and consideration of lumbar puncture for cerebrospinal fluid analysis and culture.
• Empiric intravenous antibiotics, typically ampicillin plus gentamicin (or ampicillin plus cefotaxime/ceftriaxone depending on institutional preference and infant age) to cover group B Streptococcus, Escherichia coli, Listeria monocytogenes, and other common neonatal pathogens.
• Hospital admission for clinical observation and monitoring pending 24- to 48-hour culture results.
• Consideration of acyclovir if there are any risk factors or clinical features suggestive of herpes simplex virus infection.

Age-Based Risk Stratification: A Critical Nuance

One of the most important limitations of the Rochester Criteria is that they do not differentiate between neonates (0 to 28 days) and older young infants (29 to 60 days). This distinction matters enormously. Neonates in the first 28 days of life have the highest baseline risk for SBI, the highest prevalence of bacteraemia and meningitis, the greatest vulnerability to herpes simplex virus, and the least reliable clinical appearance as a predictor of illness severity. Many neonates with bacterial meningitis appear well at the time of initial evaluation.

For this reason, many experts and institutional protocols recommend a full sepsis workup with empiric antibiotics and hospital admission for all febrile neonates aged 28 days or younger, regardless of whether they meet Rochester criteria. The Rochester Criteria are most commonly applied in practice to the 29- to 60-day age group, where the risk-benefit calculus of outpatient management is more favourable.

The 2021 American Academy of Pediatrics (AAP) Clinical Practice Guideline for the Evaluation and Management of Well-Appearing Febrile Infants 8 to 60 Days Old formalised this age-based stratification, dividing the population into three subgroups (8 to 21 days, 22 to 28 days, and 29 to 60 days) with distinct evaluation and management algorithms that incorporate inflammatory biomarkers. This represents a significant evolution beyond the Rochester Criteria's single-threshold approach.

Comparison with Other Febrile Infant Risk Stratification Tools

The Philadelphia Protocol

Developed around the same era as the Rochester Criteria, the Philadelphia Protocol applies to febrile infants aged 29 to 56 days. It uses similar clinical criteria (well-appearing, no focal infection) but defines different laboratory thresholds: WBC below 15,000, band-to-neutrophil ratio below 0.2, urinalysis with fewer than 10 WBC/hpf, negative Gram stain on urine, CSF with fewer than 8 WBC/mm3, negative Gram stain on CSF, and (if diarrhoea) no blood in stool and few or no fecal leukocytes. Notably, the Philadelphia Protocol requires lumbar puncture for all infants being evaluated, which is more invasive than the Rochester approach. Its NPV for SBI is reported at approximately 99.7%, slightly higher than Rochester's 98.9%, at the cost of the mandatory LP.

The Boston Criteria

The Boston Criteria, published by Baskin and colleagues in 1992, apply to well-appearing febrile infants aged 28 to 89 days. They require WBC below 20,000, CSF with fewer than 10 WBC/mm3, urinalysis with fewer than 10 WBC/hpf, and (if applicable) normal chest radiograph. Like Philadelphia, the Boston Criteria mandate lumbar puncture. Low-risk infants receive a single dose of intramuscular ceftriaxone and are discharged for re-evaluation at 24 hours. The Boston approach trades a broader age range and the safety net of a ceftriaxone dose against the burden of universal LP.

The Step-by-Step Approach

Developed by Mintegi and colleagues in Europe and published in 2014, the Step-by-Step approach introduced a sequential algorithm for febrile infants younger than 90 days. It begins with clinical appearance assessment, followed by age stratification (younger or older than 21 days), and then sequential evaluation of urinalysis, procalcitonin (greater than 0.5 ng/mL), C-reactive protein (greater than 20 mg/L), and absolute neutrophil count (greater than 10,000/mm3). The Step-by-Step approach was the first major febrile infant algorithm to formally incorporate procalcitonin, which has emerged as a superior biomarker for early bacterial infection compared to WBC or CRP alone. It achieved a sensitivity of 92% and an NPV of 99.3% in the derivation cohort.

The PECARN Febrile Infant Rule

The Pediatric Emergency Care Applied Research Network (PECARN) published its febrile infant prediction rule in 2019, derived from a large prospective multicentre cohort of over 1,800 febrile infants aged 60 days or younger. The PECARN rule uses urinalysis, absolute neutrophil count (ANC), and procalcitonin as sequential screening variables to classify infants as low risk for SBI (and specifically for invasive bacterial infection, i.e., bacteraemia or meningitis). The PECARN rule achieves a sensitivity of 97% and an NPV of 99.6% for invasive bacterial infection, representing a significant improvement over the Rochester Criteria's test characteristics. Notably, the PECARN rule does not require lumbar puncture for initial risk stratification.

The 2021 AAP Clinical Practice Guideline

The AAP guideline represents the most comprehensive and current evidence synthesis for the management of febrile infants 8 to 60 days old. It integrates the best available evidence, including the PECARN data, procalcitonin-based algorithms, and decades of accumulated clinical experience. The guideline stratifies infants into three age subgroups, each with specific evaluation and management recommendations. For infants 8 to 21 days old, full workup with hospitalisation and empiric antibiotics is recommended regardless of clinical appearance. For infants 22 to 28 days old, inflammatory markers (procalcitonin, CRP, ANC) guide the decision to hospitalise or discharge. For infants 29 to 60 days old, a well-appearing infant with normal inflammatory markers and negative urinalysis may be managed as an outpatient with close follow-up.

The Rochester Criteria in the Context of Contemporary Practice

Given the availability of newer, more refined risk stratification tools, one might question whether the Rochester Criteria remain relevant. They do, for several reasons.

First, the Rochester Criteria established the conceptual framework that underlies all subsequent febrile infant prediction rules: the principle that a well-appearing infant with a reassuring history and normal laboratory screening has a sufficiently low risk of SBI to permit outpatient management. Every subsequent tool, from Philadelphia to PECARN to the AAP guideline, builds on this foundational concept.

Second, not all clinical settings have access to procalcitonin. Many community emergency departments, urgent care centres, and resource-limited settings cannot obtain procalcitonin results in a clinically actionable timeframe. In these environments, the Rochester Criteria remain a practical and evidence-based tool for risk stratification using universally available clinical and laboratory data (history, physical examination, CBC with differential, urinalysis, and stool studies when applicable).

Third, the Rochester Criteria serve as an important educational benchmark in paediatric training. Understanding the development, validation, test characteristics, and limitations of the Rochester Criteria provides trainees with a framework for understanding evidence-based clinical prediction rules more broadly.

Fourth, in settings where institutional protocols have not yet been updated to incorporate the 2021 AAP guideline, the Rochester Criteria continue to serve as the basis for local clinical pathways. Many hospitals still use Rochester-based protocols, either alone or in combination with inflammatory biomarkers, as their standard approach to febrile infant management.

Special Considerations

Herpes Simplex Virus

The Rochester Criteria were designed to assess risk for serious bacterial infection and do not address neonatal herpes simplex virus (HSV) infection. Neonatal HSV is uncommon (estimated incidence 1 in 3,000 to 1 in 20,000 live births) but carries devastating consequences if untreated: mortality rates of 30% for disseminated disease and 4% for CNS disease, with significant long-term neurological sequelae among survivors. Risk factors for neonatal HSV include maternal primary genital herpes at the time of delivery, prolonged rupture of membranes, use of fetal scalp electrodes, and infant age younger than 21 days. Clinical features suggestive of HSV include vesicular skin lesions, seizures, hepatitis (elevated transaminases), coagulopathy, and CSF pleocytosis. Clinicians evaluating febrile infants must maintain a parallel clinical assessment for HSV and initiate acyclovir when the index of suspicion is elevated, independent of the Rochester Criteria result.

Urinary Tract Infection: The Most Common SBI

Urinary tract infection is by far the most common SBI in febrile young infants, accounting for roughly 60% to 80% of all SBI diagnoses in this age group. The Rochester Criteria screen for UTI through urine white blood cell count on enhanced urinalysis. However, it is important to recognise that a small proportion of culture-confirmed UTIs may have a negative screening urinalysis (pyuria is absent in approximately 10% to 20% of infant UTIs). This is one of the reasons why urine culture should be obtained on all febrile infants undergoing evaluation, even when the screening urinalysis is normal and the infant is classified as low risk. A positive urine culture result at 24 to 48 hours will trigger recall and treatment of the infant.

Male infants in the first few months of life, particularly those who are uncircumcised, have a substantially higher prevalence of UTI (up to 20% in some series for uncircumcised febrile males under 3 months). This baseline risk should be factored into the clinical assessment alongside the Rochester Criteria result.

The Evolving Role of Biomarkers

The most significant advancement in febrile infant risk stratification since the Rochester Criteria has been the incorporation of inflammatory biomarkers, particularly procalcitonin (PCT). Procalcitonin rises rapidly (within 2 to 4 hours) in response to bacterial infection and has demonstrated superior sensitivity and specificity for invasive bacterial infection compared to WBC, absolute neutrophil count, and CRP in multiple paediatric studies. A procalcitonin level below 0.5 ng/mL has a negative likelihood ratio of approximately 0.1 for invasive bacterial infection, making it one of the most powerful single biomarkers for ruling out serious illness in febrile infants.

C-reactive protein (CRP), while slower to rise than procalcitonin (peaking at 24 to 48 hours), provides complementary information. A CRP below 20 mg/L in combination with a normal procalcitonin and urinalysis offers a very high NPV for SBI. The Step-by-Step approach, the PECARN rule, and the 2021 AAP guideline all incorporate one or both of these biomarkers into their algorithms.

Where procalcitonin is available and results can be obtained within a clinically useful timeframe (typically 1 to 2 hours with modern point-of-care or laboratory immunoassays), its use substantially enhances risk stratification beyond what the Rochester Criteria alone can achieve. The combination of Rochester clinical criteria with procalcitonin has been evaluated in several studies and shown to improve both sensitivity and specificity compared to the Rochester Criteria alone.

Fever Without a Source Versus Focal Infection

The Rochester Criteria are designed for febrile infants without an identifiable source of infection on examination. When a clear focal source is identified (e.g., acute otitis media, cellulitis, omphalitis), the clinical approach shifts from screening for occult SBI to directed management of the identified infection, which may or may not require hospitalisation depending on the nature and severity of the source. It is important to recognise that the presence of a viral upper respiratory infection (URI) does not "explain away" the fever in a young infant. While URI symptoms reduce the pretest probability of SBI somewhat, they do not eliminate it. Published data suggest that febrile infants younger than 60 days with documented viral illness (e.g., positive respiratory viral panel) still have an SBI rate of approximately 2% to 4%, which is lower than the overall SBI rate but not negligible. Clinical protocols should not bypass evaluation for SBI solely on the basis of a positive viral test in this age group.

Practical Tips for Applying the Rochester Criteria

• Obtain a detailed birth and neonatal history. Many Rochester criteria relate to the perinatal and postnatal course (gestational age, perinatal antibiotics, prolonged nursery stay, prior hospitalisations, unexplained hyperbilirubinaemia). A thorough history from the caregiver is essential to assess each criterion accurately.
• Perform a complete physical examination. The examination must be systematic and unhurried, with particular attention to fontanelle tension, skin colour and perfusion, umbilicus, joints, and general responsiveness. Undress the infant fully and examine all skin surfaces for rashes, vesicles, or signs of focal infection.
• Use catheterised or suprapubic urine specimens. Bag urine specimens have unacceptably high contamination rates in infants and should not be used for either urinalysis or culture when evaluating for SBI.
• Assess band counts with caution. Manual band counts are subject to significant interobserver variability. Some laboratories no longer routinely report band counts. When band counts are available, they should be interpreted in the context of the overall clinical picture.
• Document the clinical rationale for disposition. Whether discharging a low-risk infant or admitting a higher-risk one, clear documentation of the clinical reasoning, the criteria assessed, and the plan for follow-up protects both the patient and the clinician.
• Ensure reliable follow-up. Outpatient management of a low-risk febrile infant is only as safe as the follow-up plan. If there is any doubt about the caregiver's ability to return promptly if the infant's condition changes, err on the side of admission.

Limitations in Contemporary Practice

While the Rochester Criteria established a foundational approach to febrile infant risk stratification, several limitations have become more apparent as evidence has accumulated over the past three decades.

The moderate specificity of 55% means that nearly half of well infants are classified as higher risk, leading to overtreatment. This is a particular concern as awareness of antimicrobial stewardship and the harms of unnecessary antibiotic exposure (including disruption of the developing neonatal microbiome) has grown.

The criteria do not incorporate inflammatory biomarkers (procalcitonin, CRP) that have been shown to substantially improve risk stratification. Reliance on WBC and band counts alone provides less precise discrimination than modern multimarker panels.

The binary (all-or-none) structure does not capture the gradient of risk across different clinical scenarios. An infant who fails a single criterion (e.g., mildly elevated band count with an otherwise entirely normal profile) is placed in the same higher-risk category as an infant who fails multiple criteria with clear clinical toxicity. More nuanced algorithms provide graded risk estimates.

The criteria were derived and validated in an era before widespread GBS prophylaxis, universal newborn screening, and the routine availability of rapid viral testing. Changes in the epidemiology of neonatal infection, the micro-organisms causing SBI, and the diagnostic tools available to clinicians all affect the performance and applicability of the criteria in current practice.

The lack of age subgroup stratification (particularly the failure to treat neonates younger than 28 days as a distinct higher-risk group) is a significant shortcoming that newer guidelines have addressed.