Model derived to rule out bacteria in feverish infants

A study by a nationwide network of pediatric emergency care specialists demonstrated that a protocol involving urinalysis, absolute neutrophil count, and serum procalcitonin levels helped identify febrile infants 60 days and younger who are at low risk for serious bacterial infections (SBIs).

The authors of the study, published today in JAMA Pediatrics, say the tool, if validated, could help reduce unnecessary antibiotic administration, lumbar punctures, and hospitalization in infants with fever.

97.7% sensitivity, 60.0% specificity

In the multicenter observational study, which was funded by the US Department of Health and Human Services, researchers from the Pediatric Emergency Care Applied Research Network (PECARN) enrolled 1,821 febrile infants from 26 emergency departments to derive a prediction rule to identify those at low risk for SBIs, which were defined as urinary tract infections (UTIs), bacteremia, and bacterial meningitis.

SBIs occur in 8% to 13% of febrile infants 60 days and younger, and while missed SBIs can lead to serious complications, fever is often the only sign of infection, and no single laboratory test can reliably identify them. As a result, nearly 500,000 febrile infants annually undergo spinal taps, are hospitalized, and receive broad-spectrum antibiotics while clinicians try to confirm the presence of bacterial infection.

The infants in the cohort were randomly divided into two groups, with the researchers deriving the prediction rule from 908 infants and validating it in 913. Overall, blood and urine culture results showed SBIs were present in 170 (9.3%) of the 1,821 infants, including 26 (1.4%) with bacteremia, 151 (8.3%) with UTIs, and 10 (0.5%) with bacterial meningitis. The researchers then evaluated several predictor variables, including patient demographics, fever height and duration, white blood cell count, absolute neutrophil count, urinalysis results, and serum procalcitonin levels.

Using binary recursive partitioning analysis, the researchers identified a prediction rule involving three variables: normal urinalysis, absolute neutrophil count of 4,090 per microliter or lower, and serum procalcitonin level of 1.71 nanograms per milliliter or lower.

In the validation group, the sensitivity for the prediction rule was 97.7% (95% CI, 91.3 to 99.6), specificity was 60.0% (95% CI, 56.6 to 63.3), negative predictive value was 99.6% (95% CI, 98.4 to 99.9), and negative likelihood ratio was 0.04 (95% CI, 0.01 to 0.15). Of the 170 infants identified with BSIs in both groups, 1 infant with bacteremia and 2 infants with UTIs were misclassified by the prediction rule.

May lead to fewer tests

"Our data contributes important information in the decades old debate about the necessity of lumbar punctures and hospitalizations for young babies with fevers," senior author Prashant Mahajan, MD, MPH, MBA, professor and vice-chair of emergency medicine at the University of Michigan Medical School and C.S. Mott Children's Hospital, said in a University of California-Davis press release.

"This study adds important information that we think will decrease the variability in current protocols and minimize unnecessary tests and hospital admissions, which can carry other risks for young patients."

Mahajan and his colleagues say further validation of the prediction rule is needed in cohorts with greater numbers of invasive infections before it can be implemented in clinical practice.

See also:

Feb 18 JAMA Pediatr study  

Feb 18 University of California-Davis Health press release

Newsletter Sign-up

Get CIDRAP news and other free newsletters.

Sign up now»


Unrestricted financial support provided by

Bentson Foundation 3MAccelerate DiagnosticsGilead 
Grant support for ASP provided by


  Become an underwriter»