Background: In the absence of procalcitonin, the American Academy of Pediatrics' clinical practice guideline (CPG) for evaluating and managing febrile infants recommends using previously untested combinations of inflammatory marker thresholds. Thus, CPG performance in detecting invasive bacterial infections (IBIs; bacteremia, bacterial meningitis) is poorly understood.
Objective: To evaluate CPG performance without procalcitonin in detecting IBIs in well-appearing febrile infants 8 to 60 days old.
Methods: For this cross-sectional, single-site study, we manually abstracted data for febrile infants using electronic health records from 2011 to 2018. We used CPG inclusion/exclusion criteria to identify eligible infants and stratified IBI risk with CPG inflammatory marker thresholds for temperature, absolute neutrophil count, and C-reactive protein. Because the CPG permits a wide array of interpretations, we performed 3 sensitivity analyses, modifying age and inflammatory marker thresholds. For each approach, we calculated area-under-the-receiver operating characteristic curve, sensitivity, and specificity in detecting IBIs.
Results: For this study, 507 infants met the inclusion criteria. For the main analysis, we observed an area-under-the-receiver operating characteristic curve of 0.673 (95% confidence interval 0.652-0.694), sensitivity of 100% (66.4%-100%), and specificity of 34.5% (30.4%-38.9%). For the sensitivity analyses, sensitivities were all 100% and specificities ranged from 9% to 38%.
Conclusion: Findings suggest that the CPG is highly sensitive, minimizing missed IBIs, but specificity may be lower than previously reported. Future studies should prospectively investigate CPG performance in larger, multisite samples.