Estimation of the Time-Varying Effective Reproductive Number of COVID-19 Based on Multivariate Time Series of Severe Health Outcomes

The time-varying effective reproduction number (Rt at time t) measures the transmissibility of SARS-CoV-2 and is conventionally based on daily case counts, which may suffer from time-varying ascertainment. We analyzed Rt estimates from case counts and severe COVID-19 (intensive care unit admissions, severe or critical cases, and mortality) across 2022 in Hong Kong's fifth and sixth waves of infection. Within the fifth wave, the severe disease-based Rt (3.5) was significantly higher than the case-based Rt (2.4) but not in the sixth wave. During periods with fluctuating underreporting, data based on severe diseases may provide more reliable Rt estimates. Variable case ascertainment may bias Rt estimates. We explored using more robust alternative data streams based on severe COVID-19 health outcomes to estimate Rt. We found that during periods of fluctuating case ascertainment, severe outcome time series could provide more reliable estimates.