Variability in invasive bacterial infection proportions among febrile infants aged 8-90 days using administrative data

Abstract

ObjectiveDespite decades of research, knowledge is limited regarding sociodemographic risk factors (e.g., sex, rural/urban residence) for invasive bacterial infections (IBIs; bacteremia, bacterial meningitis) in young febrile infants and outcomes of current management strategies. Population-based administrative datasets can provide epidemiological insights not possible with clinical data but are limited because diagnosis codes alone may not accurately reflect culture-positive bacteremia or meningitis infections. Thus, using different IBI case definitions, we report IBI and missed IBI proportions in a population of febrile infants aged 8-90 days. MethodsFor this cross-sectional study, we used New York State’s all-payer database to identify healthy, full-term infants with fever aged 8-90 days evaluated in emergency departments from 2012-2023. We defined IBIs and missed IBIs using previously published diagnosis codes and then restricted original case definitions to inpatient encounters with variable lengths-of-stay. For each approach, we calculated total and age-stratified IBI and missed IBI proportions and used chi square statistics to compare proportions within and across age groups. ResultsOf 67,115 infants who met inclusion criteria (15,191 (23%) aged 8-28 days), total IBI and missed IBI proportions varied from 11.5-32.3/1,000 febrile infants and 4.2-8.0/100 IBIs, respectively. Although IBI proportions decreased significantly with advancing age, missed IBI proportions significantly increased. ConclusionsIBI and missed IBI proportions varied widely by case definition. Missed IBI proportions increased with advancing age in a step-wise fashion regardless of case definition. Validation studies are needed to compare IBI diagnosis codes with culture results to understand the accuracy of identifying IBIs with administrative data.