Retrospective validation of the Step-by-Step approach for febrile infants younger than 90 days in the emergency department

ABSTRACT

PURPOSE: To retrospectively validate the Step-by-Step approach, a sequential algorithm for prediction of serious bacterial infections (SBI) using the appearance, age, and inflammatory markers, in febrile infants younger than 90 days. METHODS: The presence of SBI was reviewed in febrile infants younger than 90 days undergoing blood and urine cultures (using perineal adhesive bags), assays for procalcitonin, C-reactive protein and absolute neutrophil count, and urinalysis at the emergency department from September 2015 through August 2017. The low-risk infants were classified according to the Step-by-Step approach. SBI was defined as urinary tract infection (UTI), bacteremia or meningitis. We measured the sensitivity and negative predictive value (NPV) of the approach in predicting SBI, and compared the values to those of the Rochester criteria and the Lab-score. RESULTS: Of 488 febrile infants (44.7% underwent lumbar puncture), 71 (14.5%) had SBI, including 67 UTI, 5 bacteremia, and 3 meningitis (mutually inclusive). Of 208 low-risk infants (42.6%), no SBI was found. The Step-by-Step approach showed a 100% sensitivity (95% confidence interval [CI] 94.9–100.0) and NPV (95% CI not applicable). The Rochester criteria showed a 98.6% sensitivity (95% CI 92.4–100.0) and 99.6% NPV (95% CI 97.1–99.9), and missed 1 meningitis. The Lab-score showed a 59.2% sensitivity (95% CI 46.8–70.7) and 93.2% NPV (95% CI 91.2–94.8), and missed 2 meningitis and 27 UTI. CONCLUSION: The Step-by-Step approach showed a 100% sensitivity and NPV in predicting SBI. This approach may help predict SBI without lumbar puncture in febrile infants younger than 90 days.