Quantifying the spatial spillover effects of non-pharmaceutical interventions on pandemic risk

Abstract

Abstract Background Non-pharmaceutical interventions (NPIs) implemented in one place can affect neighboring regions by influencing people’s behavior. However, existing epidemic models for NPIs evaluation rarely consider such spatial spillover effects, which may lead to a biased assessment of policy effects. Methods Using the US state-level mobility and policy data from January 6 to August 2, 2020, we develop a quantitative framework that includes both a panel spatial econometric model and an S-SEIR (Spillover-Susceptible-Exposed-Infected-Recovered) model to quantify the spatial spillover effects of NPIs on human mobility and COVID-19 transmission. Results The spatial spillover effects of NPIs explain $$61.2\%$$ 61.2 % [ $$95\%$$ 95 % credible interval: 52.8- $$84.4\%$$ 84.4 % ] of national cumulative confirmed cases, suggesting that the presence of the spillover effect significantly enhances the NPI influence. Simulations based on the S-SEIR model further show that increasing interventions in only a few states with larger intrastate human mobility intensity significantly reduce the cases nationwide. These region-based interventions also can carry over to interstate lockdowns. Conclusions Our study provides a framework for evaluating and comparing the effectiveness of different intervention strategies conditional on NPI spillovers, and calls for collaboration from different regions.

Keywords

• COVID-19
• Non-pharmaceutical interventions
• Spatial spillover effects
• SEIR
• Human mobility