Background The reasons for the large differences between countries in the sizes of their SARS CoV2 epidemics is unknown. Individual level studies have found that the use of face masks was protective for the acquisition and transmission of a range of respiratory viruses including SARS CoV1. We hypothesized that population level usage of face masks may be negatively associated SARS CoV2 spread. Methods At a country level, linear regression was used to assess the association between COVID19 diagnoses per inhabitant and the national promotion of face masks in public (coded as a binary variable), controlling for the age of the COVID19 epidemic and testing intensity. Results Eight of the 49 countries with available data advocated wearing face masks in public: China, Czechia, Hong Kong, Japan, Singapore, South Korea, Thailand and Malaysia. In multivariate analysis face mask use was negatively associated with number of COVID19 cases/inhabitant (coef. -326, 95% CI -601- -51, P=0.021). Testing intensity was positively associated with COVID-19 cases (coef. 0.07, 95% CI 0.05-0.08, P<0.001). Conclusion Whilst these results are susceptible to residual confounding, they do provide ecological level support to the individual level studies that found face mask usage to reduce the transmission and acquisition of respiratory viral infections. 2 Abstract 32 33 34 Background 35 The reasons for the large differences between countries in the sizes of their SARS 36 CoV-2 epidemics is unknown. Individual level studies have found that the use of face 37 masks was protective for the acquisition and transmission of a range of respiratory 38 viruses including SARS CoV-1. We hypothesized that population level usage of face 39 masks may be negatively associated SARS CoV-2 spread. 40 41 Methods 42 At a country level, linear regression was used to assess the association between 43 COVID-19 diagnoses per inhabitant and the national promotion of face masks in 44 public (coded as a binary variable), controlling for the age of the COVID-19 epidemic 45 and testing intensity. 46 47 Results 48 Eight of the 49 countries with available data advocated wearing face masks in public 49 – China, Czechia, Hong Kong, Japan, Singapore, South Korea, Thailand and 50 Malaysia. In multivariate analysis face mask use was negatively associated with 51 number of COVID-19 cases/inhabitant (coef. -326, 95% CI -601- -51, P=0.021). 52 Testing intensity was positively associated with COVID-19 cases (coef. 0.07, 95% CI 53 0.05-0.08, P<0.001). 54 55 Conclusion 56 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 6, 2020. ; https://doi.org/10.1101/2020.03.31.20048652doi: medRxiv preprint 8 keeping with its population size and relative autonomy as regards public health 160 responses. Because Czechia was the only country to introduce universal face masks 161 late in the epidemic (18 March 2020), we repeated the analyses excluding Czechia 162 [19, 20]. 163 164 Results 165 Forty-nine countries were found with complete data, epidemics older than 7 March 166 2020 and more that 500 cases/million inhabitants (STable 1). Large variations were 167 evident in the number of COVID-19 cases per million inhabitants (median 158, 168 interquartile range [IQR] 20-486), testing per million (median 1723, IQR 307-4802) 169 and the date of the first case (median 24 February, IQR 28 January – 28 February; 170 STable 1). Only 8 of these countries advocated wearing face masks in public – 171 China, Czechia, Hong Kong, Japan, Singapore, South Korea, Thailand and 172 Malaysia. These countries tended to have older epidemics. Seven of the 8 were in 173 the group of countries with the 10 oldest epidemics (STable 1). 174 175 In multivariate analysis face mask use was negatively associated with number of 176 COVID-19 cases (coef. -326, 95% CI -601- -51, P=0.021; Table 1). Testing intensity 177 was positively associated with COVID-19 cases (coef. 0.07, 95% CI 0.05-0.08, 178 P<0.001). 179 180 Repeating the analyses excluding Czechia strengthened the association between 181 COVID-19 cases and face mask usage slightly (STable 2). 182 183 Discussion 184 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 6, 2020. ; https://doi.org/10.1101/2020.03.31.20048652doi: medRxiv preprint 9 In this ecological study we found that countries that promoted widespread face mask 185 usage had lower cumulative numbers of COVID-19 diagnosed after controlling for 186 testing intensity and age of the epidemic. It is important to note that this association 187 may be entirely explained by unmeasured confounders. For example, if countries 188 promoting universal face masking also conducted more effective contact tracing and 189 isolation than other countries and this was responsible for the slower spread, our 190 study design would have falsely attributed this effect to using face masks. We did not 191 have accurate data to control for these confounders. We did however control for 192 testing intensity which is an important potential confounder. We also controlled for 193 the age of the epidemic which is an obvious independent determinant of the size of 194 the epidemic. A further limitation of our study was that we were unable to quantitate 195 the intensity of face mask use per country. This resulted in a rather crude binary 196 classification of face mask usage per country. 197 198 At this albeit early phase of the pandemic it is important to explore why the virus has 199 spread more