Our take —
Adherence to social distancing recommendations in the United States, as measured by reductions in daily travel, was associated with socioeconomic status, race/ethnicity, and voting patterns. It is not clear how representative the data are, and unmeasured confounding may be present.
Study design
Ecological
Study population and setting
Cell phone records from 15-17 million users/day were used to estimate changes (from pre-epidemic to the period of March 19, 2020 to March 28, 2020) in county-level averages of daily distances traveled per person. Authors identified county-level, Census-derived socioeconomic and political characteristics associated with adherence to social distancing recommendations, defined by daily distance traveled.
Summary of Main Findings
Reductions in population-aggregated daily average travel were significantly higher in counties with higher per capita income and a larger proportion of college degrees. By comparison, counties with a higher share of registrants voting for President Trump in the 2016 election; a higher proportion of Black and Hispanic residents; and higher rates of employment in retail, transportation, and health/education/social services were significantly less likely to show declines in daily travel.
Study Strengths
This study integrates data from a number of sources (Unacast, American Community Survey, MIT Election Lab) in order to: 1) quantify potential engagement with social distancing recommendations at county level ;and 2) identify socio-economic and political characteristics rendering specific communities more vulnerable to COVID-19 transmission.
Limitations
There are no data presented on the representativeness of the mobility data; selection bias is possible. The social distancing measure is based on anonymized cell phone records and is highly sensitive to potential misclassification. Average daily travel estimates are derived from one time period, and do not account for geographic heterogeneity in the timing of control measures and social distancing messages. The ecological analysis had a limited ability to appropriately assess confounding.
Value added
The study identifies socio-economic factors (e.g., racial and education composition, labor and employment characteristics) and political characteristics (e.g., party affiliation) that may render counties more vulnerable to subsequent COVID-19 transmission based on population adherence to social distancing recommendations.
This review was posted on: 4 May 2020