Preliminary evidence that higher temperatures are associated with lower incidence of COVID-19, for cases reported globally up to 29th February 2020

Melanie Bannister-Tyrrell; Anne Meyer; Celine Faverjon; Angus Cameron

All ADM1 associated with at least one confirmed case of COVID-19 in the source dataset were included in the analysis, excluding Hubei province in China where the outbreak emerged. Each case was classified as imported (using the proxy that travel history was reported in the line list data) or local (otherwise). We modelled the cumulative number of COVID-19 cases classified as local cases in each ADM1 during the observation period. The statistical model was based on the generalized linear regression framework [12] . We used a negative binomial distribution to account for the overdispersion of the number of local cases during the observation period. The primary exposure was the average temperature at the ADM1 centroid during the observation period. Temperature was included in the model both as a linear and a quadratic term to account for the non-linearity of the association observed during initial data exploration.