Written by Clay Smith
Infants ≤60 days are low risk for serious bacterial infection (SBI) if they have a negative urinalysis, ANC ≤4090/µL, and procalcitonin ≤1.71 ng/mL. NPV was 99.6%; negative likelihood ratio (NLR) 0.04. Keep in mind, this hasn’t yet been externally validated.
Why does this matter?
The big issue in febrile infants is LP and hospitalization. Do all these children need a LP? Do they all need antibiotics and hospitalization? Could we define a low risk cohort that could be screened in the ED and discharged with close follow up? More institutions are developing protocols to risk stratify infants with fever, especially those 29-60 days, as we will see tomorrow.
New landmark in risk-stratifying infant fever
The PECARN group derived and internally validated a new infant fever prediction rule in 26 EDs around the US. They included infants ≤60 days who underwent workup for SBI. Overall prevalence of SBI was 9.3% (170/1821); UTI was most common (8.3%), then bacteremia (1.4%), followed by meningitis (0.5%). There were 908 in the derivation set, and it was internally validated in 913. They found that a negative urinalysis, ANC ≤4090/µL, and procalcitonin ≤1.71 ng/mL identified infants as low risk for SBI. Diagnostic accuracy in the validation group was: sensitivity 97.7%, specificity 60%, NPV 99.6%, and NLR 0.04. Clinician pretest suspicion and Yale Observation Score weren’t predictive. The rule missed one patient with bacteremia, two with UTI, and none with meningitis. Only one patient with UTI also had concomitant meningitis.
This study is exciting, but I have some reservations. First, many centers don’t have a rapid procalcitonin assay. Also, keep in mind, sometimes with external validation, the accuracy may change. Finally, we need to be very cautious using risk stratification ≤28 days. For now, I think these children still need a full workup, including LP. My view in infants 29-60 days is evolving, and I now think checking blood, urine, +/- inflammatory markers, +/- multiplex PCR, and not performing LP on every infant is a reasonable approach in well appearing, term infants.
A Clinical Prediction Rule to Identify Febrile Infants 60 Days and Younger at Low Risk for Serious Bacterial Infections. JAMA Pediatr. 2019 Feb 18. doi: 10.1001/jamapediatrics.2018.5501. [Epub ahead of print]
Open in Read by QxMD
Reviewed by Thomas Davis