Editor’s Note: Dr. Julie Evans (she/her) is a resident physician in pediatrics at the University of Virginia. She is interested in general pediatrics and global health. -Rachel Y. Moon, MD, Associate Editor, Digital Media, Pediatrics
Typically, fever in infants is a sign of an underlying infection, whether that be of the blood (bacteremia), spinal fluid (meningitis), or urine (urinary tract infection). Collectively, we call this an invasive bacterial infection (IBI). We don’t want to miss an IBI, as this is very serious in infants, but fever does not necessarily mean that there is an IBI. Pediatricians balance the harm of testing with that of missing an IBI.
Procalcitonin is an inflammatory marker that, when elevated, has high predictive value for IBIs. However, some institutions (including my institution) do not have access to procalcitonin.
It is important for providers who cannot look at procalcitonin results to have other options to help them assess the risk for IBI. Dr. Brett Burstein and colleagues from McGill University, University of British Columbia, University of Utah, and George Washington University looked into the optimal combination of C-reactive protein (CRP), absolute neutrophil count (ANC), and temperature to predict IBI in their paper, entitled “Optimizing Management of Febrile Young Infants Without Serum Procalcitonin,” which is being early released, with an accompanying video abstract, in Pediatrics this month (10.1542/peds.2024-068200).
The American Academy of Pediatrics (AAP) has clinical practice guidelines with recommendations of how CRP, ANC, and temperature can be used to determine risk of IBI with great sensitivity (ruling out IBIs), but with low specificity (ruling in an IBI). The AAP thresholds are:
- CRP less than or equal to 20 mg/L
- ANC less than or equal to 5200/mm3
- Temperature less than or equal to 38.5C
The authors conducted a secondary analysis of quality improvement data collected prospectively from 1987 infants under 60 days old at an urban tertiary pediatric emergency department to try to optimize diagnostic accuracy of the AAP recommendations for not missing IBIs and maximizing specificity.
By using a classification and regression tree analysis, a final decision rule was generated that looked at interactions between predictors of IBIs and identified optimal thresholds of CRP, ANC, and temperature for determining an infant at low risk for IBI. Based on this analysis, CRP, ANC, and temperature were all found to be predictors of IBI, but age and urinalysis result were not.
The decision rule determined optimal thresholds of:
- CRP less than or equal to 22.2mg/L,
- Temperature less than or equal to 39C, OR
- ANC less than or equal to 4500/mm3
With these new thresholds, more than 80% of infants in the study were categorized as low risk, with no missed IBIs. In addition, sensitivity and negative predictive value were upheld, and specificity improved from 50.7% to 83.8%.
Should this result become externally validated, this new decision rule could offer an alternative for pediatricians who do not have access to procalcitonin.
