94 - Impact of viral infections on absolute neutrophil count in febrile infants

Poster Number: 94
Publication Number: 94.1458

Background: Febrile illnesses in infants are often caused by viral infections with low morbidity and mortality. However, it is important to identify infants at high risk for bacterial infection (BI) due to increased risks. Absolute neutrophil count (ANC) correlates to immune system activation in the setting of BI, but some viruses (i.e. influenza and human rhinovirus) can induce a large anti-inflammatory response, which may impact the accuracy of ANC in identifying infants at high risk for BI.

Objective: Assess the impact of viral infections on the accuracy of ANC to predict BI in febrile infants.

Design/Methods: This single center, cross-sectional study examined febrile infants 8-60 days presenting 1/1/2018-8/31/2022. Patients were identified via ICD-10 codes for fever or BI. Exclusion criteria: reported/recorded maximum temperature < 38°C, prematurity ( < 37 weeks), previous intensive care admission, prior antibiotic use, or complex medical conditions. Data collected: age, ANC, bacterial cultures, and viral polymerase chain reaction results. Confirmed viral infection (CVI) was defined as positive for one or more viruses. The Wilcoxon rank sum test compared ANC between infants with BI only (BI+/CVI-), both BI and CVI (BI+/CVI+), and neither CVI nor BI (BI-/CVI-). Logistic regression examined ANC (with cut-off of 4000) as a predictor of BI, separately for CVI- and CVI+; corresponding odds ratios compared using a Z-test.

Results: A total of 1220 patients met inclusion criteria. Median ANC was higher in BI+/CVI- compared to BI+/CVI+ subjects [6450 vs. 5400; p = 0.01] overall and when stratified by age (Table 1). Higher ANC was detected in BI+/CVI- vs. BI-/CVI- [6450 vs. 2300; p< 0.01] overall and in 8-29, but not 29-60 ages (Table 1). Comparing BI+/CVI+ vs BI-/CVI- groups detected higher ANC overall [5400 vs. 2300; p< 0.01] and when stratified by age (Table 1). At ANC cut-off of 4000, the presence of CVI reduced the odds ratio to detect BI presence [9.89 vs 4.72; p< 0.01], and corresponding sensitivity [77% vs. 60%; p=0.01], while specificity stayed consistent [75% vs. 76%; p=0.65].

Conclusion(s): ANC is a valuable tool used by clinicians in risk stratification of febrile infants. In the presence of a viral illness, the sensitivity of ANC is reduced while specificity remains unchanged. This supports the known idea that ANC cannot be used alone to rule out a BI in a febrile infant, and this holds even greater truth in the setting of a CVI due to the higher proportion of false negatives. However, febrile infants with an ANC > 4000 still warrant further clinical consideration for BI, regardless of viral status.