Estimating the number of SARS-CoV-2 infections in the United States

Dayton G Thorpe; Kelsey Lyberger

We retrieved dates for policy interventions as follows. We took lockdown dates from the New York Times rounded to the nearest day (lockdowns going into effect past noon are recorded as the next day) [6]. School closure dates are from [7]. Following [3] , if a school closure began on a Monday, we set the date of the school closure to the previous Saturday. We set social distancing and the ban of public events to Mar 16, 2020 for all states, when federal guidelines were put into place [8] . See the dates we assign for each region in each state in the Appendix (Table A1 ). Age distributions for each state come from the US Census Bureau [9] . The time series of reported fatalities and cases in the United States was crowdsourced from Wikipedia [1]. The time series of reported fatalities and cases in the 11 European countries comes from the European Centre of Disease Control (ECDC) [10] . We change the name of the country "Georgia" to "Georgia(country)" in the ECDC data set to avoid confusion with the state of Georgia. In the ECDC data we also delete the entries for Puerto Rico, for which we get duplicate data as a US territory. For the 11 European countries, IFR comes from [3] . All other IFR's are age-adjusted from [11] .