Quantifying the impacts of human mobility restriction on the spread of COVID-19: an empirical analysis from 344 cities of China

Abstract Objective Since the outbreak of novel coronavirus pneumonia (COVID-19), human mobility restriction measures have raised controversies, partly due to inconsistent findings. Empirical study is urgently needed to reliably assess the causal effects of mobility restriction. Methods Our study applied the difference-in-difference (DID) model to assess declines of population mobility at the city level, and used the log-log regression model to examine the effects of population mobility declines on the disease spread measured by cumulative or new cases of COVID-19 over time, after adjusting for confounders. Results The DID model showed that a continual expansion of the relative declines over time in 2020. After four weeks, population mobility declined by 54.81% (interquartile ranges, −65.50% to −43.56%). The accrued population mobility declines were associated with significant reduction of cumulative COVID-19 cases throughout six weeks (i.e., 1% decline of population mobility was associated with 0.72% (95%CI 0.50% to 0.93%) reduce of cumulative cases for one week, 1.42% two weeks, 1.69% three weeks, 1.72% four weeks,1.64% five weeks and 1.52% six weeks). The impact on weekly new cases seemed greater in the first four weeks, but faded thereafter. The effects on cumulative cases differed by cities of different population sizes, with greater effects seen in larger cities. Conclusion Persistent population mobility restrictions are well deserved. However, a change in the degree of mobility restriction may be warranted over time, particularly after several weeks of rigorous mobility restriction. Implementation of mobility restrictions in major cities with large population sizes may be even more important.