Validation of A Prediction Rule For Serious Bacterial Infections (SBIs) In Febrile Infants < 60 Days In A Multicenter Network

Background: We previously derived and validated a prediction rule for SBIs (urinary tract infections [UTI], bacteremia, or bacterial meningitis [BM]) in 1821 febrile infants < 60 days in the Pediatric Emergency Care Applied Research Network (PECARN). The rule includes 3 variables (using modified cutoffs): A normal urinalysis (UA), defined by absence of nitrites, leukocyte esterase and < 5 WBCs/hpf, absolute neutrophil count (ANC) < 4000 cells/mm3 (empirically derived at 4090) and procalcitonin (PCT) < 0.5 ng/ml (empirically derived at 1.71). With these modified, easier-to-apply cutoffs, the test accuracy was similar to the empirically derived rule: sensitivity 97.7%, specificity 56.3%, NPV 99.6%, PPV 19.5%. No infants with BM were misclassified. Objective: We sought to validate this modified PECARN SBI rule in a new cohort of febrile infants < 60 days. Design/Methods: We enrolled a convenience cohort of febrile (> 38.00 C) infants ≤ 60 days evaluated for SBIs in PECARN from 6/2016 to 4/2019. We collected the same variables as in the initial study and applied the UA and modified ANC and PCT cutoffs to determine the test characteristics for SBI. We defined bacteremia and BM by growth of a known pathogen, and UTI using standard definitions. We excluded infants missing any of the 3 variables or with unknown outcomes. We report point estimates and 95% binomial CIs. Results: 1363 eligible infants were analyzed. Of these, 127 (9.3%) had SBIs including 113 (8.3%) with UTIs, 24 (1.8%) with bacteremia, and 6 (0.4%) with BM. 16 infants had multiple sites of infection. Infants with SBIs were more likely to have positive UAs and higher levels of blood inflammatory markers (Table 1). The rule test characteristics for SBIs (Table 2) were similar to those of the original study. The sensitivity for bacteremia or BM was 92% in both the 1st and 2nd month of life (Table 2). The rule misclassified 3 infants with SBIs: 1 with UTI and 2 with bacteremia; all 3 had good outcomes. No patient with BM was misclassified. The original empirically derived rule performed similarly, misclassifying one additional infant with bacteremia (E.coli), who was 49 days and had a PCT of 0.7 ng/mL. Conclusion: The PECARN SBI rule for febrile infants < 60 days, consisting of the UA, and modified, easier-to-apply cutoffs for ANC and PCT, maintained its substantial test accuracy in a new cohort. Because the rule was not derived to identify herpes infections, the rule should only be implemented in the second month of life.