COVID-19 attack rate increases with city size

Andrew J Stier; Marc G Berman; Luis M. A. Bettencourt

is the (which was not peer-reviewed) The copyright holder for this preprint County level daily data from March 13-24 were aggregated to the city level (Metropolitan Statistical Areas, which are integrated labor markets) using delineation files from the US Office of Budget and Management (15). Counties with at least 1 reported case of COVID-19 are available in the data. We next excluded cities that had no COVID-19 cases on March 13. This excluded cities with low case counts, likely due to introductions from outside the MSA, for which accurate estimates of local case growth rate is unlikely. Results were similar for all contiguous subsets of the data of at least 7 days (see Supplmentary Figure 4a ). This left 163 cities for further analysis. We substracted deaths from cases in each city and found the slope of the ln(cases) ∼ ln(a) + r · t line for each resulting time-series of active COVID-19 cases. Finally we plotted the natural logarithm of r and the natural logarithm of city population from 2018 census estimates (16), and performed an ordinary least squares linear regression to determine the slope of the scaling line. Regression residuals were not related to city population (Supplementary Figure 2 ) and a q-q plot of the residuals indicated that they were well described by a normal distribution (Supplementary Figure 3) .