Delay in the Effect of Restricting Community Mobility on the Spread of COVID-19 in the United States

Abstract

Background: Ever since the introduction of COVID-19 into the US, governments from the local to the federal levels have implemented combinations of responses to COVID-19. It is unclear how changes in human mobility across the country in response to intervention policies have shaped the transmission dynamic of COVID-19.Methods: Coupling a Bayesian hierarchical spatiotemporal model with the Google mobility data, we aimed to quantify the impact of human movement on COVID-19 incidence rate at the county level while accounting for spatial autocorrelation. We used both single-lag and distributed lag models to assess how movement today can affect incidence rates in the following weeks.Findings: On average, an increase in travel from home to other places was associated with an increase in reported COVID-19 incidence rate about 5 to 7 weeks later. A 10% increase in workplace movement 7 weeks ago was associated with 5% [95% confidence interval (CI): 1-9%] increase in COVID-19 incidence rate, 8% [95% CI:4-12%] for retail and recreation stores, 5% [95% CI:1-9%] for transits, and 5% [95% CI:29%] for grocery and pharmacies in the single-lag model analysis. No significant impact on incidence was found for human movement in parks. A 10% increase in the total duration at residential places was associated with a 13% (95% CI:6-20%) reduction in incidence rate 7 weeks after.Interpretation: Human movement activities substantially affected COVID-19 incidence rates at local levels but with delays varying from state to state. Policymakers should anticipate such a delay when planning intervention strategies restricting human movement.Funding Information: YY was supported by grant R56 AI148284 from the US National Institutes of Health. PJ was supported by grant R01 HL150119 from the US National Institutes of Health.Declaration of Interests: The Authors declare that there is no conflict of interest.Ethics Approval Statement: The study only involves publicly accessible deidentified and aggregated data and is therefore exempt from IRB approval.