Climate affects global patterns of COVID-19 early outbreak dynamics

Gentile Francesco Ficetola; Diego Rubolini

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.23.20040501 doi: medRxiv preprint states or China provinces), whenever separate Covid-19 cases data for these regions are available. Initially, US data were reported by county but, as of March 9, they were reported at the state level. We therefore merged all US county data before March 9 to state level, and used statelevel time series for subsequent calculations. With the exception of US data, in all other cases we maintained the original country/region information adopted by the JHU-CSSE. The datafile 5 considered for the analyses was downloaded on March 19, 2020, and included confirmed Covid-19 cases until March 18, 2020. From this dataset, we selected data for all countries/regions for which local outbreaks were detected. We define a local outbreak event when at least 50 positive cases were detected in a given country/region, and calculated the growth rate of confirmed Covid-19 cases between day 1 and day 5, when day 1 was the day at which the 50 cases 10 threshold was reached. We calculate the daily growth rate r of confirmed Covid-19 cases for each country/region, assuming an exponential growth as: r = [ln(n cases day 5 ) -ln(n cases day 1 )] / 4. We checked the robustness of our estimates of growth rate by calculating daily growth rate after the first 25, 50 or 100 cases (r 25 , r 50 and r 100 , respectively). Growth rates estimated at different thresholds were strongly positively correlated (Pearson's correlation coefficients, r 25 vs. 15 r 50 : r = 0.74; r 50 vs. r 100 : r = 0.81).