A Step-by-Step Approach to Infant Fever

William T. Basco, Jr., MD, MS


November 23, 2016

Approaches to Fever in Young Children

During the past two decades, several sets of objective criteria that included both clinical and laboratory evaluation have been validated to help clinicians in their approach to young febrile children. The goal of all of these evaluation approaches is to identify children at low risk who could subsequently avoid extensive evaluation for bacterial infection, including culture of cerebrospinal fluid, blood, and urine. Advances in conjugate vaccines, the availability of rapid viral testing, and even the advent of biomarkers such as procalcitonin (PCT) and C-reactive protein (CRP) have all been associated with a gradual decline in the number of children evaluated for invasive bacterial infection.

A Step-by-Step Strategy

A recent study[1] evaluated an approach developed by emergency physicians in Europe. The algorithm prioritized any ill-appearing patient as "high risk." Any patient aged ≤ 21 days was also by default flagged as "high risk." Then, in an approach that differed from previous decision structures, the algorithm looked at whether the patient had urinary leukocytosis, followed by PCT level ≥ 0.5 ng/mL, followed by whether they had a CRP level ≥20 µ/L or an absolute neutrophil count > 10,000/mm3. Any patient failing to meet these criteria was by default considered a "low-risk" patient and therefore could forgo extensive evaluation.

This study was an attempt to validate that approach applied prospectively, and 11 European emergency departments participated (most located in Spain). Patients <90 days of age were evaluated from 2012 to 2014. For all patients, the study centers obtained a urine dipstick, a urine culture, a peripheral white blood cell count, a blood culture, and CRP and PCT levels. Any other studies were at the discretion of the treating physicians. Similarly, the decision whether to treat with empirical antibiotics or to admit to the hospital was made at the local centers. The investigators attempted to follow up by telephone every patient who was not admitted to the hospital. Patients classified as having fever without a source were those with a temperature ≥ 38°C who had a normal exam, including no respiratory or gastrointestinal signs or symptoms. "Invasive bacterial infections" were defined by the isolation of either a blood or a spinal fluid bacterial pathogen. Urinary tract infections (UTIs) were considered noninvasive bacterial infections and were defined as any growth of ≥ 10,000 colony forming units/mL, along with leukocyturia.

Study Findings

The study enrolled 2185 patients, with a median age of 47 days (16.7% were ≤ 21 days of age). The median duration of fever was 5 hours. On initial examination, 87.7% were classified as "well appearing." Interventions included lumbar puncture (27.4%), empirical antibiotics (49%), and admission to the hospital (58.5%). Overall, 23.1% of the children had a bacterial infection, but only 3.9% had an invasive bacterial infection. The overwhelming majority of the noninvasive bacterial infections were UTIs (n = 409). There were five bacterial gastrointestinal infections and one case each of cellulitis, omphalitis, and myositis. The initial decision point criterion of "ill-appearing" was met by 269 patients. Among that group, 10.8% had invasive bacterial infections, and 14.5% had noninvasive bacterial infections.

Another 307 children aged ≤ 21 days were not initially ill appearing; of that group, 8.5% had invasive bacterial infections, and 17.9% had noninvasive bacterial infections. Results of other evaluations (leukocyturia, PCT, CRP, or absolute neutrophil count) left 999 patients (45.3% of the total) who still met "low-risk" criteria. Among that population, seven children (0.7%) had invasive bacterial infections, and 0.4% had noninvasive bacterial infections.

The step-by-step approach had a slightly higher sensitivity compared with other approaches (92% vs 81.6% for the Rochester criteria) and very good negative predictive value (99.3%). The researchers concluded that the step-by-step approach has a high sensitivity and appears to be more accurate than other systematized approaches to identifying infants at low risk for invasive bacterial infection.


In post hoc analysis, Gomez and colleagues[1] noted that a large majority of the children with invasive bacterial infections who were missed by the step-by-step approach had very short durations of fever, usually less than 2 hours. Although it was not formally part of the evaluation in this approach, they suggested that duration of fever be factored into the evaluation of febrile infants, mainly by considering a longer evaluation and observation time in the emergency department for children with short fever duration to make sure that their clinical appearance does not change before discharge.

In thinking about the past 20 years of evaluating febrile infants, many providers and centers have already evolved to this type of approach. Factors such as clinical appearance, vaccine history, and viral illness (either clinically apparent or as detected by rapid testing) have already made it into our pre-examination assessment of patient risk stratification. It's nice to see objective evidence that a systematized approach can validate the evolution of clinical practice.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: