Estimating the rotavirus hospitalization disease burden and trends, using capture-recapture methods.

BACKGROUND: Rotavirus surveillance is needed to provide estimates of disease burden and to evaluate the effect of vaccination programs. Our objective was to use capture-recapture methods to estimate rotavirus hospitalization rates and to examine trends over time. METHODS: Children <3 years of age residing in Hamilton County, Ohio hospitalized with acute gastroenteritis, and laboratory-confirmed rotavirus between 1997 and 2008 were identified through 2 independent surveillance systems: an active system with prospective enrollment of children admitted with acute gastroenteritis and a passive system of children identified by rotavirus testing as part of their usual medical care. Capture-recapture methods compared cases from both systems to estimate the number of missed cases from either system. Using census data for Hamilton County, rates per 10,000 with 95% confidence intervals (CI) for rotavirus hospitalizations were estimated. RESULTS: Overall, 486 cases were identified using active surveillance and 244 using passive surveillance, with 127 cases captured by both. Using capture-recapture methods, the overall rate in children <3 years old was 26.9/10,000; CI: 24.1, 30.6. Rates varied by year: highest in 1998 (48.1/10,000; CI: 32.4, 92.2) and lowest in 2008 (3.2/10,000; CI: 2.1, 6.1) after rotavirus vaccine introduction. Among children <5 years old, rates were highest in <3-month-old children (51.8/10,000; CI: 39.4, 75.1) and lowest in older age groups: 24 to 35 months (20.5/10,000; CI: 14.7, 30.3) and 36 to 59 months (4.1/10,000; CI: 2.9, 7.2). Rates from capture-recapture methods and adjusted active system were comparable. CONCLUSIONS: Capture-recapture methods were a useful tool to estimate rotavirus disease burden and to monitor trends, especially in the era of rotavirus immunization.