BACKGROUND: National influenza surveillance in the outpatient setting consists of clinician-reported influenza-like illness (ILI) and influenza testing results from a variety of clinical laboratories; these reports are neither directly linked nor population-based. Further, influenza disease burden estimation requires a more sensitive case definition than ILI to avoid missing cases. We piloted an enhanced surveillance system designed to determine the incidence of medically-attended acute respiratory illness (ARI) and associated laboratory-confirmed influenza, while maintaining the ability to estimate ILI for national surveillance reporting.
METHODS: The Acute Respiratory Infection Epidemiology and Surveillance program (ARIES) was conducted in 21 clinics in 5 geographically diverse sites representing a population of 166,758 persons. Clinics reported weekly counts of all-cause, ARI visits (≥2 respiratory symptoms), and ILI visits (fever with cough or sore throat). From the first 10 ARI patients presenting each week, symptom information and an upper respiratory specimen was collected for influenza RT-PCR testing. To estimate the number of ILI visits, we used the reported symptoms from patients tested for influenza to determine the percent meeting ILI criteria, and then multiplied by the ARI visit count. Visits associated with influenza were extrapolated from the percent of test-positive patients. Incidence calculations used the clinics’ patient population size as the denominator.
RESULTS: From October through December 2016, 5.9% of 51,626 outpatient visits were for ARI and the 3-month cumulative incidence of ARI was 2490 per 100,000 population (95% confidence interval [CI] 2410-2564 per 100,000). Of 361 ARI patients (average 6.4 per week [range 1 – 28]), 63% met ILI criteria; influenza was detected in 3.0% of ARI and 3.5% of ILI patients. Counted and estimated ILI accounted for 1.1% and 2.9% of visits, respectively; weekly proportions were weakly correlated (Pearson 0.7, p <0.05). The estimated cumulative incidence of influenza-associated ARI visits was 56 per 100,000 population (95% CI 33-85 per 100,000). Point estimates for the influenza-associated ILI visit incidence differed between counted and estimated ILI (13 per 100,000 [95% CI 7-21 per 100,000] and 37 per 100,000 [95% CI 18-58 per 100,000], respectively), though confidence intervals overlapped slightly.
CONCLUSIONS: Enhanced surveillance established through ARIES enabled incidence estimation for influenza-associated clinical visits, with the ARI being more sensitive than ILI. Estimated weekly ILI percent correlated with the counted ILI percent, but influenza-associated ILI incidence differed during the initial weeks of surveillance when specimen volume and influenza detection was low.