Time from symptom onset may influence C-reactive protein utility in the diagnosis of bacterial infections in the NICU

Abstract

BackgroundTaking into account the timing of C-reactive protein (CRP) testing may improve the performance of the test in diagnosing bacterial infections in the neonatal intensive care unit (NICU). We aimed to examine the yield of CRP, relative to time from symptoms onset.MethodsEnrolled were all NICU patients, for whom CRP was obtained as part of a sepsis workup. The time of symptoms onset and of blood draw was recorded. Patients were classified into bacterial and non-bacterial groups according to the National Healthcare Safety Network (NHSN) guidelines. The performance of CRP, CRP velocity, and CRP obtained before or after 6 hours from symptoms onset, was evaluated by receiver-operating characteristic (ROC) curve. Test characteristics were calculated using formulas based on Bayes’ theorem.ResultsOf 129 infants enrolled in the study, 21(16%) had a bacterial infection. A single CRP test and CRP velocity performed similarly in diagnosing bacterial infection, with area under ROC curve of 0.75 (95%CI: 0.61–0.89) and 0.77 (95% CI:0.66–0.88), respectively. The optimal cut-off value for a CRP test obtained <= 6 hours from symptoms onset was 1 mg/dL, whereas the optimal cut-off > 6 hours was 1.5 mg/dL. Using the optimal cut-off values increased the pre-test probability of 16%, to a post-test probability of 35–38%. For infants whose birth weight was < 1000 g, CRP performed poorly.ConclusionsThe optimal CRP cut-off used in the diagnosis of bacterial infections in NICU patients varies by the time from symptom onset. A “negative” CRP may support a clinical decision to stop empiric antimicrobial therapy, for infants whose blood cultures remain sterile.