Predictive Model for Border Control During COVID-19: Experiences from Taiwan and New Zealand

Abstract

Background: Understanding the impact of imported coronavirus disease 2019 (COVID-19) cases on the subsequent cluster infections is important for establishing travel bubbles or travel corridors. In this study, we aimed to develop an epidemic surveillance model to detect cluster infections from imported cases.Methods: Data on imported and domestic COVID-19 cases from Taiwan and New Zealand between January and June 2020 were used. We applied Bayesian random-effect Poisson regression model, accounting for time lag of incubation period, to predicting the domestic cases following the imported cases. The upper limit of 95% credible intervals of estimated weekly-predicted domestic cases was treated as an alerting level to infer the odds of subsequent large-scale community-acquired outbreaks.Results: Empirical epidemic curves showed that as of 20th June, Taiwan and New Zealand had effectively contained COVID-19. An increase in one imported case increased the risk of domestic cases by 9·54% (95% CI: 6·44 to 12·59%) in Taiwan and 10·97% (95% CI: 10·30 to 11·67%) in New Zealand. The Taiwan epidemic curve revealed that imported cases did not lead to a large-scale community-acquired outbreak. In New Zealand, a community-acquired outbreak during 29th March-4th April could have been averted if control actions had been taken one-week earlier prior to the predicted cluster infection between 22nd and 28th March. Interpretation: An epidemic surveillance model was developed to timely predict domestic COVID-19 cases resulting from the transmission of imported cases. Such a quantitative surveillance model would be useful for monitoring, alerting and preventing large-scale community-acquired outbreaks during border reopening for selected countries with COVID-19 under control.Funding Statement: Ministry of Science and Technology, Ministry of Education (MOE)Declaration of Interests: All authors disclose no any potential conflicts (financial, professional, or personal) that are relevant to the manuscript and have nothing to disclose.Ethics Approval Statement: Publicly available case line list without any private and identifiable information does not require IRB approval.