Estimating the undetected burden of influenza hospitalizations in children.

During the 2004-2005 influenza season two independent influenza surveillance systems operated simultaneously in three United States counties. The New Vaccine Surveillance Network (NVSN) prospectively enrolled children hospitalized for respiratory symptoms/fever and tested them using culture and RT-PCR. The Emerging Infections Program (EIP) and a similar clinical-laboratory surveillance system identified hospitalized children who had positive influenza tests obtained as part of their usual medical care. Using data from these systems, we applied capture-recapture analyses to estimate the burden of influenza related-hospitalizations in children aged<5 years. During the 2004-2005 influenza season the influenza-related hospitalization rate estimated by capture-recapture analysis was 8.6/10,000 children aged<5 years. When compared to this estimate, the sensitivity of the prospective surveillance system was 69% and the sensitivity of the clinical-laboratory based system was 39%. In the face of limited resources and an increasing need for influenza surveillance, capture-recapture analysis provides better estimates than either system alone.