Fast spread of COVID-19 in Europe and the US and its implications: even modest public health goals require comprehensive intervention

Ruian Ke; Steven Sanche; Ethan Romero-Severson; Nicholas Hengartner

Switzerland (SW), Netherlands (NT), United Kingdom (UK) and the US (US). 105 106 We included a subset of case confirmation and death count data for inference based on the two 107 following criteria. First, to minimize the impact of stochasticity and uncertainty in early data 108 collection, we used case confirmation incidence data starting from the date when the cumulative 109 number of cases was greater than 100, and used daily new death count data starting from the date 110 when the cumulative death count is greater than 20 in each country. Second, to estimate the early 111 outbreak growth in each country before control measures were implemented and at the same time 112 maximize the power of inference, we allowed a maximum of 15 days of data points for the two 113 types of data, leading to a maximum of 30 data points for each country. Note that the end date of 114 incidence data used for inference is at or close to the date when strong control measures were 115 implemented in each country (Table S1 ). We tested the sensitivity of model predictions when 116 only 10, 13 days of data points are included for inference. The results are robust to this variation 117 (Table S1 ), suggesting that the choice of 15 days is reasonable and that the data shows consistent 118 exponential growth during this period. 119 120