Study population and setting
Using an anonymized dataset of smartphone device location data (from Safe Graph, Inc.) from 208 USA counties covering 315 cities with a population over 100,000 people, the authors compared the degree of social distancing and COVID-19 case numbers before and after the protests between May 25 and June 20, 2020, and between counties that did and didn’t experience protests. Social distancing was measured in three ways: the proportion of the population that stayed at home all day; the median number of hours spent at home and the median proportion of time spent at home. Growth in COVID-19 cases were based on cumulative case growth at the county level using CDC data.
Summary of Main Findings
Mass protests were found to lead to an increase in stay-at-home behavior. Possible decline in social distancing among the protesting sub-population may have been offset by a direct increase in social distancing by the rest of the population, or even by the protesting population itself taking on mitigating measures in the aftermath (staying home after attending a protest). This increase in social distancing was statistically significant and observed 3 days post-protest, with effects increasing at a daily rate of 2% relative to the mean, and peaking one week after the protest onset. Social distancing effects are observed to plateau or decline after a week. Cities with persistent ongoing protests appeared to have greater increases in social distancing following the protests, compared to cities with shorter protests. The size of the protests may also lead to a larger effect in social distancing. Increases in social distancing associated with protests were seen regardless of whether the city had curfews imposed at a certain point. Protests did not appear to lead to any changes in COVID-19 case counts.
This is the first study to specifically investigate the 2020 Black Lives Matter protests and their effect on COVID-19 spread. Analyses considered both social distancing behavior and COVID-19 case counts, and used data from all large cities across the US. Supplemental analyses included additional counties to ensure rigor and capture any spillover effects of protests in an expanded geographical area.
As the authors acknowledge, data for the outcomes, social distancing and COVID-19 cases were measured at the county, not individual, level. Thus, conclusions are about the county as a whole rather than relating to individual behaviors and outcomes. In other words, certain subpopulations (e.g., those that attended protests) could have seen an increase in COVID-19 cases without this being seen at the overall county-level. Individuals attending protests may be younger and healthier, and thus more likely to be asymptomatic nor present themselves with severe enough symptoms to warrant testing. County-specific contexts pertaining to limited availability of public health resources may have reduced testing capacity and access, limiting case-detection. Authors used a media analysis of articles to assess for violence at protests but do not provide sufficient detail on how this was conducted. Moreover, avoidance of protests in anticipation of violence is likely influenced by other media not captured by a standard media analysis such as social media posts and recent historical protests.
Despite concerns about a rise in COVID-19 cases following the protests, this article provides some initial evidence that these concerns may be unfounded.
This review was posted on: 15 July 2020