Using Capture-Recapture Methods to Estimate Local Influenza Hospitalization Incidence Rates

BACKGROUND: Accurate population estimates of disease incidence and burden are needed to set appropriate public health policy.

The capture-recapture (C-R) method combines data from multiple sources to better estimate prevalence than is possible using single sources.

This study used the C-R method to estimate influenza cases using research and administrative databases to calculate county-wide influenza hospitalization burden.

METHODS: Data were derived from a database of clinical virology test results and research data from an influenza vaccine effectiveness study from seasons 2015-2016 to 2018-2019.

Missed influenza cases were estimated using C-R method.

These estimates were used to calculate disease burden using the multiplier method to correct for underreporting due to curtailing data collection before the end of influenza circulation.

RESULTS: Over all seasons, 422 influenza cases were reported in the administrative database and 382 influenza cases in the research database.

Seventy-five cases (18%) reported in the administrative database were not captured in the research database, and 35 (9%) cases in the research database were not captured in the administrative database.

Completeness of the influenza hospitalization was estimated to be 76%.

Influenza hospitalizations were higher among unvaccinated (32%) than vaccinated (22%) in the current season and among unvaccinated (28%) than vaccinated (23%) in the previous year.

The incidence rates for influenza hospitalizations varied by age and season and averaged 421 cases/100,000 population annually.

CONCLUSION: The capture-recapture method offers a more accurate method for estimating influenza hospitalization than relying on a single data source.

Using the multiplier method with adjustments improves the detection of influenza disease burden through a matched database.

The incidence rates are consistent with national estimates.