Estimating the impact of physical distancing measures in containing COVID-19: an empirical analysis

BACKGROUND: Epidemic modelling studies predict that physical distancing is critical in containing COVID-19. However, few empirical studies have validated this finding. Our study evaluates the effectiveness of different physical distancing measures in controlling viral transmission. METHODS: We identified three distinct physical distancing measures with varying intensity and implemented at different times—international travel controls, restrictions on mass gatherings, and lockdown-type measures—based on the Oxford COVID-19 Government Response Tracker. We also estimated the time-varying reproduction number (R(t)) for 142 countries and tracked R(t) temporally for two weeks following the 100th reported case in each country. We regressed R(t) on the physical distancing measures and other control variables (income, population density, age structure, and temperature) and performed several robustness checks to validate our findings. FINDINGS: Complete travel bans and all forms of lockdown-type measures have been effective in reducing average R(t) over the 14 days following the 100th case. Recommended stay-at-home advisories and partial lockdowns are as effective as complete lockdowns in outbreak control. However, these measures have to be implemented early to be effective. Based on the observed median timing across countries worldwide, lockdown-type measures are considered early if they were instituted about two weeks before the 100th case and travel bans a week before detection of the first case. INTERPRETATION: A combination of physical distancing measures, if implemented early, can be effective in containing COVID-19—tight border controls to limit importation of cases, encouraging physical distancing, moderately stringent measures such as working from home, and a full lockdown in the case of a probable uncontrolled outbreak.