Methods: We conducted a nested case-control study of ILI visits among children <18 years old in a large pediatric primary care network during 9 consecutive respiratory seasons (2005/06 – 2013/14). ILI visits were identified using ICD-9 codes validated for influenza syndromic surveillance. Clinical, provider and testing data were abstracted from the electronic health record. Multivariable logistic regression with generalized estimating equations to adjust for clustering within clinics measured patient and clinic factors associated with likelihood of influenza testing.
Results: Of 99,889 ILI visits, 1,029 (1.04%) had an associated influenza test and 1,170 (1.17%) had an associated antiviral prescription. In univariable analysis, a higher proportion of patients who were black (50.2% vs 19.1%) and received Medicaid (55.5% vs 22.4%) were tested for influenza. However, after adjusting for practice location, testing was significantly associated with history of any comorbidity (OR 1.15, 95% CI 1.26, 2.06), age <2 years (1.92, 95% CI 1.08, 1.13) and receipt of care in an urban practice (OR 9.7, 95% CI 2.60, 36.1). Results did not differ between annual and pandemic seasons.
Conclusion: In this cohort of children with outpatient ILI, we did not identify sociodemographic disparities in influenza testing. Instead, consistent with guidelines, the presence of risk factors for severe disease appears to have driven testing practices. Our results also suggest variability associated with clinic characteristics. Additional work is needed to ensure consistent diagnostic practices to optimize management of influenza especially for those who are most at risk for poor outcomes.
S. Coffin, None