This is the submitted version of the article: Mogi, Ryohei; Spijker, Jeroen. «The influence of social and economic ties to the spread of COVID-19 in Europe». Journal of Population Research, 2021. DOI 10.1007/s12546-021-09257-1 This version is available at https://ddd.uab.cat/record/238307 under the terms of the license A peer-reviewed version of this preprint was recently published in the Journal of Population Research: Mogi, R., & Spijker, J. (2021). The influence of social and economic ties to the spread of COVID-19 in Europe. Journal of Population Research (online first). https://doi.org/10.1007/s12546-021-09257-1 For those citing our paper, please now use the above citation as reference. The preprint version can still be consulted for those who do not have access to the journal version. 1 The influence of social and economic ties to the spread of COVID-19 in Europe Ryohei Mogi*, Jeroen Spijker* Centre d’Estudis Demogràfics, Universitat Autònoma de Barcelona rmogi@ced.uab.es @rmogimogi @popageing *Equal contribution Background: By late January 2020, the 2019 novel coronavirus (SARS-CoV-2) had reached Europe and most European countries had registered cases by March 1. However, the spread of the virus has been uneven in both prevalence and speed of its propagation. Given the role of social factors in the spread of contagious diseases and the known demographic, economic and cultural differences in Europe, we analyse the contribution of such factors in the coronavirus disease COVID-19 caused by SARS-CoV-2. Methods: Diagnosed COVID-19 cases from the Center for Systems Science and Engineering at Johns Hopkins University and the European Social Survey for 20 European countries were used to estimate bivariate associations between cumulative reported case numbers ten-day intervals between March 1 and April 20, 2020 and ten social, demographic and economic variables. To avoid overfitting, we first reduce these variables to three factors (labelled “socially and economically vibrant”, “high-educated and not aged”, and “densely populated and traditional”) by means of factor analysis before conducting a multiple regression analysis. We also perform a sensitivity analysis using rates and new cases between two time periods. Findings: The socially and economically vibrant factor has a strong and positive association with COVID-19 throughout the study period. The association with the 2 densely populated and traditional factor was initially low but by mid-March similar and from April higher than the socially and economically vibrant factor. Conclusions: For future influenza-like pandemics where a vaccine is not yet available, early movement restrictions will be especially crucial to curb the spread of such diseases in economically, socially, and culturally vibrant and densely populated countries. Introduction As of late April 2020, there were more than 3 million confirmed cases worldwide of the coronavirus disease COVID-19 (1) that is caused by the 2019 novel coronavirus (SARS- CoV-2). It is thought that the virus was initially transmitted late 2019 to humans via live animals contaminated by bats that were sold at the Huanan seafood market in Wuhan, China, after which it began to be spread through person-to-person transmission. Person- to-person transmission occurs primarily through respiratory droplets produced when an infected person coughs or sneezes in the close presence of other people (2, 3). Although the epidemic initially only appeared to be affecting China and other countries quickly placed international travel restrictions to and from China, the first cases were being confirmed in Europe by early February and by late March about half of the world’s reported cases were in Europe (4). One well-established cause of global spread of previous influenza outbreaks is airplane travel (5) and without this mode of transportation the coronavirus would not have arrived in Europe so quickly. However, once someone brought the virus to Europe, human interaction and close contact was also required for the virus to spread quickly. Previous research on virus transmission, including on COVID-19, has shown that social contact is very important (6-10). At the same time, cultural differences in Europe are well established. In terms of social contact, in southern European countries intergenerational 3 contact is more frequent than in other parts of Europe, as social norms about providing support to family members and maintaining interpersonal familiar interactions are stronger than in the less family-oriented Western and Northern European countries (11- 12). Patterns in social mixing are therefore embedded in socioeconomic and cultural factors and are different across Europe, making this continent an excellent geographical area to the study of COVID-19 proliferation. This study aims therefore to ascertain whether countries where social interaction is considered to be more common have more confirmed cases of COVID-19. We examine this by analyzing statistical associations between different indicators of social ties and the reported number of confirmed cases of COVID-19 in 23 European countries between March 1 and April 20, 2020. Data and method Data on the number of confirmed cases of COVID-19 come from the Center for Systems Science and Engineering at Johns Hopkins University (1) (https://coronavirus.jhu.edu/map.html). Cross-sectional data for 18 countries from Round 9 of the European Social Survey (ESS), conducted in 2018, 4 from Round 8 (2016) and 1 from Round 7 (2014), was used to create variables to proximate different types of social ties, which we hypothesize to be positively associated with COVID-19. i) the average number of household members; ii) percentage living in a multi-generational household; iii) proportion of people who have a frequent social meetings with friends, relatives, or colleagues; iv) religious attendance. In addition, we also test the effect of the socioeconomic variables: v) tertiary education and vi) GDP per capita (current US$) as we assume that countries with higher proportions of tertiary educated or that are more economically developed are more likely to have people employed and pursue non- 4 employment activities (e.g. international business meetings; skiing) that require travelling, which increases person-to-person contact. Lastly, we test the effect of the following demographic variables: vii) the share of the population who live in (semi-) rural areas (expected to be negatively associated with COVID-19); viii) the share of the population that is 65 years or older and ix) population density and x) per capita number of beds in nursing and residential care facilities (expected to be positively associated with COVID- 19. More detail on the data sources (13-19) and the descriptive statistics of the covariates can be found in Table 1. We first analyze the bivariate associations between each variable and the ln of the cumulative number of confirmed COVID-10 cases. We do this for six different dates (March 1, 11, 21, 31, April 10 and 20) to check whether the type of predictor responsible for the diffusion changed over time (Table 2). March 1 was chosen as the first date as most of the 20 European countries had reported their first cases of COVID-19 (Table 1). April 20 is the last date analysed as by this time all studied countries had restrictions of movement and social interaction in place for over a month. This implies that country- differences in cumulative or new cases at later dates will be more and more influenced by the different severities and timing of these restrictions rather than due to differences in social and demographic factors. The supplementary material section provides robustness checks by repeating the analyses by using as the dependent variable the number of COVID-19 cases per 100.000 population and the ln of the change in cases between each period. Ordinary linear regression analysis was used to assess the unique association between confirmed cases of COVID-19 and the covariates. However, as covariate data could only be obtained for 23 countries, i.e. too few to test all covariates simultaneously without overfitting (21-22), we first opted to reduce the number of variables by 5 performing a factor analysis. This yielded three sociodemographic-like factors that explained 78% of the country-variation in the selected covariates. These have been labelled “socially and economically vibrant”, “high-educated and not aged”, and “densely populated and traditional” (see also Tables S1, S2, S3 and S4 and the explanation under Table S1 in the supplementary material). Correlation coefficients between COVID-19 and the obtained factors were subsequently calculated. As the rotated factors are orthogonal and thus not correlated, each covariate’s standardized coefficient is the same as the correlation coefficient in a multivariate regression analysis. However, multivariate regression analyses were still performed to obtain the adjusted R2 of the models as well as the unstandardized coefficients as they represent the slope of the factors (Table S5). Results Figure 1 and Table 2 present the association between the number of confirmed cases of COVID-19 and the different covariates for the six dates. The highest (and significant) associations are observed for the social meeting and population density variables. Particularly regarding the latter, this association gets stronger over time. As the relatively small sample of countries does not permit us to use all covariates in a multivariate analysis, we used the three factors that were extracted though factor analysis. Figure 2 and Table 3 present the association of each factor with the number of confirmed cases of COVID- 19 and are consistent with the results of the specific variables. The socially and economically vibrant factor has a strong and positive association throughout the study period and the highly-educated and not aged factor has a statistically insignificant negative and declining association with the number of confirmed cases of COVID-19. The association of densely populated and traditional factor was initially low but increased with time, becoming significant on March 11 and the most important factor by the end of 6 March. The three factors together explain 51% of the cross-country variation in the number of confirmed cases of COVID-19 on both March 31 and April 20, compared to just 21% on March 1 (Table 4). The slope of the socially & economically vibrant factor was greatest on March 11 and that of the densely populated and traditional factor on April 20. Our sensitivity analysis shows virtually identical statistical associations between the change in COVID-19 cases in a space of 10 days and the three sociodemographic factors (see Supplementary Table S5). The socially and economically vibrant factor is strongly significant across all time periods, except for the last period between April 10 and 21, while the densely populated and traditional factor was significant throughout the entire study period and became the most important explanatory factor at the end of March. The high-educated and not aged factor did not prove to be significant in any of the models. The proportion of the country differences in change in COVID-19 explained by the three factors was about the same during all but the last period (44-48%, dropping down to 38%). The factors explain much less of the country differences in the number of cases of COVID-19 per 100,000 population (see Supplementary table S6): the socially and economically vibrant factor from March 31 and the densely populated and traditional factor from April 10 but, again, becoming the most important explanatory factor. The explanatory power did increase steadily over time from 3% (March 1) to 40% (April 20). This is consistent with the fact that towards the end of April, the countries with high number of cases per 100,000 population included not only Italy, but also the densely populated Belgium and the Netherlands, while COVID-19 rates were (still) quite low in the sparsely populated Scandinavian and Baltic countries. 7 Discussion Confirmed cases of COVID-19 increased sharply across Europe during the course of March and April, 2020. Throughout most of the studied period, Italy was worst hit by the pandemic in absolute number of cases, but Spain surpassed Italy early April, while others also did so in terms of cases per 100,000 people (see Table 1). The question we posed is whether social, economic and demographic factors could explain the observed differences in Europe. Our results suggest that it is not so much how aged countries are but their (historical) level of economic development and (associated) social contact that determined the initial spread of the COVID-19 pandemic and continued to be important for its subsequent spread during late March and early April. However, population density and religious practice were particularly important demographic and cultural country- specific factors associated with frequency of human-to-human contact necessary for its subsequent spread in the most affected countries. Considering some specific examples, the Netherlands, Switzerland and Sweden all scored high on the socially and economically vibrant factor and saw their number of coronavirus infections quickly increase during March despite households being almost exclusively single-person or nuclear. An important component of this factor, however, is also the number of available beds in nursing and residential care facilities, in which all three countries score high. One recent study showed that nursery homes may be responsible for anything between 19% and 72% of COVID-19 deaths in, respectively, Australia and Canada, with the seven European countries included in the study observing proportions within this range; 23, see also 24). On the other hand, Italy having been the initial epicenter of the epidemic in Europe scored very low on “high-educated and not aged” factor, but high on “densely populated and traditional” factor, i.e. its high population density and traditional values 8 (approximated through the proportion attending church at least once a week) is likely to have contributed to their high levels. Interestingly, the relative position of other traditionally catholic countries in COVID-19, including Spain, Portugal and Belgium worsened markedly between March 21 and April 10 (Figure 2). Our sensitivity analysis shows almost identical results if change in COVID-19 cases are analyzed over 10-day periods, while the number of cases of COVID-19 per 100,000 population in the 23 European countries could only be weakly explained by these three factors from March 31 onwards. Again, on the last analysed date, 20 April, the population density factor proved to be the most discriminant one, although having a socially and economically vibrant population remained to contribute significantly to the explanation of the European country differences as the number of cases had increased markedly in the most socioeconomically developed European countries. Some limitations of our study should be mentioned. First, we did not consider country differences in (the timing of) government (and individual) responses to the COVID-19 pandemic. In particular, governments have differed in the timing of the implementation of measures such as public event cancellation, the closing of daycare centers, schools and universities, social distancing, or partial or total lockdowns (see 25 for details). This implies that the effect of social and demographic factors on COVID-19 cases may be confounded by these measures in those countries that were quickest at adopting them and had already past its peak of daily additional cases of COVID-19 (e.g. Italy). Given the estimated average latency period between the moment that someone becomes infected by the coronavirus and reported COVID-19, we think that only the last two data points may be affected by this. Other factors are also likely to be responsible for the spread of the coronavirus in Europe that we did not consider. We know that a French and Austrian ski resort were 9 responsible for initial infections in the UK and other Northern European countries (26- 27). However, the influence that winter holidays or international travel has on the country difference in COVID-19 during the studied period is likely to be captured by the tertiary education and GDP variables, two well-known proxies for average disposable income. Smoking is another variable associated with the proliferation of COVID-19. For instance, there is evidence that it is considered a social function (28), smokers are more likely to touch their face and mouth (29) and have chronic health conditions (29-30). However, we think that smoking is a more important factor to consider in individual or small-area studies and when studying COVID-19 mortality, as analysis showed smoking rates to be highest in Eastern European countries, which are least “socially and economically” vibrant and observed the lowest levels of COVID-19. Another issue of concern is country differences in the approach to testing for COVID-19. Some countries only test people admitted to hospitals or ramped up the testing program much later during the outbreak than other countries. This implies that particularly some of the earlier data points will be an underestimate of the real prevalence of COVID-19 (31-32). Finally, data from the ESS is not available for all European countries. This implies that different results may be obtained if data for more countries becomes available. Despite its availability, we did not use data from Round 9 (2018) as it contains fewer countries that the previous round. However, this is unlikely to have biased the results as cross-country demographic, social and cultural differences are unlikely to change in a few years. To conclude, our policy recommendation based on our results for similar future outbreaks of coronavirus-like epidemics is that very strict measures of social distancing would be required to be implemented quickly in countries with strong social ties in order 10 to minimize the proliferation of cases during the secondary transmission phase that occurs within households. These social ties are, however, not only related to population density, household characteristics and religious and cultural practices, but also to economic development. This is probably why it is particularly the north of Italy, Switzerland and some of the main urban centers of Spain (particularly Madrid) that have been hit the hardest by COVID-19. A recommendation for future research would therefore be to perform analysis at the sub-national level for the countries that have been studied here. 11 Table 1. Descriptive statistics of the number of reported confirmed COVID-19 cases and the covariates used in the analysis. Cumulative Confirmed Confirmed Date of cases of cases on cases on the first COVID-19 Country Area Code Population 31/03/ 20/04/ case on 100,000 100,000 confirmed 01/03 11/03 21/03 31/03 10/04 20/04 people people Austria West AUT 25/02 14 246 2814 10180 13555 14795 8858775 114.91 167.01 Belgium West BEL 04/02 2 314 2815 12775 26667 39983 11455519 111.52 349.03 Bulgaria East BGR 08/03 0 7 163 399 635 929 7000039 5.70 13.27 Czechia East CZE 01/03 3 91 995 3308 5732 6900 10649800 31.06 64.79 Denmark North DNK 27/02 4 444 1420 3039 6014 7711 5806081 52.34 132.81 Estonia East EST 27/02 1 16 306 745 1258 1535 1324820 56.23 115.86 Finland North FIN 29/01 6 59 523 1418 2769 3868 5517919 25.70 70.10 France West FRA 24/01 130 2293 14463 52827 91738 155393 67012883 78.83 231.89 Germany West DEU 27/01 130 1908 22213 71808 122171 147065 83019213 86.50 177.15 Hungary East HUN 04/03 0 13 103 492 1190 1984 9772756 5.03 20.30 Ireland North IRL 29/02 1 43 785 3235 8089 15652 4904240 65.96 319.15 Italy South ITA 31/01 1694 12462 53578 105792 147577 181228 60359546 175.27 300.25 Lithuania East LTU 28/02 1 3 83 537 999 1326 2794184 19.22 47.46 Netherlands West NLD 27/02 10 503 3640 12667 23249 33588 17282163 73.30 194.35 Norway North NOR 26/02 19 598 2118 4641 6314 7156 5328212 87.10 134.30 Poland East POL 04/03 0 31 536 2311 5955 9593 37972812 6.09 25.26 Portugal South PRT 02/03 0 59 1280 7443 15472 20863 10276617 72.43 203.01 Serbia East SRB 06/03 0 12 171 900 3105 6630 6963764 12.92 95.21 Slovenia East SVN 05/03 0 57 383 802 1160 1335 2080908 38.54 64.15 Spain South ESP 01/02 84 2277 25374 95923 158273 200210 46937060 204.37 426.55 Sweden North SWE 31/01 14 500 1763 4435 9685 14777 10230185 43.35 144.45 Switzerland West CHE 25/02 27 652 6575 16605 24551 27944 8544527 194.33 327.04 The UK North GBR 31/01 36 459 5067 25481 74605 125856 66647112 38.23 188.84 EU 23 total 24/01 2176 23047 147168 437763 750763 1026321 490739135 89.20 209.14 12 Table 1(Cont). Mean % living in % having %≥ Nursing/ Popula- % pop number of a multi- a frequent weekly % tertiary % people GDP per rest home Country tion per (semi) household generational social religious education aged 65+ capita beds / km2 rural members household meeting attendance 100,000 Austria 2.2 5 65.5 11.4 31.1 107.21 43.90 19.00 51462 862.4 Belgium 2.3 4.8 68.6 8.6 36.0 377.21 50.10 18.79 47519 1234.1 Bulgaria 2.4 24.7 51.7 9.3 24.7 64.70 32.20 21.02 9273 30.8 Czechia 2.4 7.6 50.8 5.8 21.6 137.60 29.30 19.42 23079 687.5 Denmark 2.0 1.2 71.2 4.3 33.1 138.07 26.00 19.81 61350 816.1 Estonia 2.2 8.9 43.5 3.7 36.5 30.39 28.00 19.63 23266 870.7 Finland 2.0 1.5 65.1 4.6 38.5 18.16 33.20 21.72 50152 1190.0 France 2.2 2.0 67.9 7.6 33.7 122.34 34.20 20.03 41464 981.5 Germany 2.0 5.3 59.7 7.1 25.9 237.37 32.80 21.46 47603 1152.2 Hungary 2.3 11.2 20.4 10.6 22.5 107.91 36.70 19.16 16162 853.3 Ireland 2.6 8.1 58.9 33.9 40.7 70.45 41.10 13.87 78806 639.3 Italy 2.3 14.2 60.1 27.4 17.4 205.45 46.80 22.75 34483 415.8 Lithuania 2.2 9.4 31.1 17.1 37.9 44.53 23.00 19.71 19153 726.3 Netherlands 2.2 2.7 74.5 9.8 34.8 511.46 45.40 19.20 53024 1379.6 Norway 2.0 2.3 78.5 3.9 37.7 14.55 35.30 17.05 81697 765.6 Poland 2.8 16.1 30.9 46.5 28.2 124.04 44.30 17.52 15421 195.3 Portugal 2.5 13.1 77.6 27.8 23.8 112.24 30.50 21.95 23408 555.7 Serbia 2.9 21.3 62.1 11.1 20.6 79.83 35.60 18.35 7247 291.5 Slovenia 2.5 16.1 53.1 12.6 29.3 102.64 53.00 19.61 26124 1012.4 Spain 2.5 17.9 70.2 15.2 35.1 93.53 47.30 19.38 30371 834.8 Sweden 2.0 1.2 74.3 5.7 37.8 25.00 28.00 20.10 54608 1388.0 Switzerland 2.2 2.3 71.6 8.3 38.6 215.52 55.00 18.62 82797 1174.2 The UK 2.3 3.9 62.4 11.7 40.6 274.83 25.20 18.40 42944 821.4 Source: Confirmed COVID-19 cases: the Center for Systems Science and Engineering at Johns Hopkins University: https://coronavirus.jhu.edu/map.html. Population on 1 January 2019 (13). Mean number of household members 2018 (Eurostat; 14), % living in a multi-generational household (people who live with i) parents, ii) parents and children, iii) parents and relatives, iv) parents, children, and relatives, v) children and relatives, or vi) relatives), % having a frequent social meeting (people who socially meet with their friends, relatives, or colleagues at least once a week), % ≥ weekly religious attendance, and % pop (semi) rural are created from European Social Survey (ESS) Rounds 7 (2014) (Denmark), 8 (2016) (Lithuania, Portugal, Spain and Sweden) and 9 (2018) (all other countries). % tertiary education of 25-64 year-olds in 2018; Eurostat (15), except BGR and SRB (ESS)). Population per km2 in 2018, % people aged 65+ in 2018 and GDP in 2018; World Bank (16-18). Available beds in nursing and residential care facilities per 100.000 inhabitants in 2017, except for Belgium (2012) and Denmark (2011) and Portugal (applied 2017 ratio of number long-term care recipients between Portugal and Spain to the beds available in Spain); Eurostat (19-20). 13 Table 2. Correlation between the natural log COVID-19 at six different time periods and the covariates. Variable March 1 March 11 March 21 March 31 April 10 April 20 1. Mean number of household members -0.40 -0.38 -0.28 -0.20 -0.13 -0.08 2. % living in a multi-generational household (ln) -0.31 -0.42* -0.29 -0.24 -0.23 -0.21 3. % having a frequent social meeting (squared) 0.41 0.61** 0.56** 0.53** 0.52* 0.50* 4. % ≥ weekly religious attendance (ln) -0.07 -0.10 0.03 0.12 0.17 0.20 5. % tertiary education -0.01 0.06 0.05 0.05 0.05 0.05 6. Population density (people per km2) (ln) 0.25 0.37 0.44* 0.50* 0.53** 0.54** 7. % population in (semi-)rural areas 0.18 0.32 0.33 0.30 0.27 0.25 8. % people aged 65+ (squared) 0.35 0.22 0.23 0.18 0.14 0.10 9. GDP per capita 0.35 0.51* 0.44* 0.38 0.35 0.33 10. Nursing and rest home beds per capita 0.25 0.39 0.31 0.29 0.27 0.24 **p<0.01, *p<0.05 Table 3. Correlation between the natural log of COVID-19 at six different time periods and the extracted factors. Date of cumulative cases of COVID-19 March 1 March 11 March 21 March 31 April 10 April 20 Factor 1: Socially and economically vibrant 0.44* 0.61** 0.52** 0.47** 0.44* 0.41* Factor 2: High-educated and not aged -0.26 -0.16 -0.14 -0.11 -0.07 -0.04 Factor 3: Densely populated and traditional 0.25 0.38* 0.48** 0.55** 0.59** 0.60** **p<0.01, *p<0.05 Table 4. Multivariate regression analysis of social and demographic factors on cumulative cases of ln of COVID-19 on March 1, 11, 21 and 31, and April 10 and 20, 2020, for 23 European countries. Unstandardized coefficients. Unstandardized coefficients (p value) March 1 March 11 March 21 March 31 April 10 April 20 Factor 1: Socially & economically vibrant 0.91 (0.033) 1.30 (0.001) 0.93 (0.004) 0.81 (0.007) 0.75 (0.010) 0.70 (0.017) Factor 2: High educated & not aged -0.53 (0.194) -0.35 (0.304) -0.26 (0.377) -0.18 (0.506) -0.12 (0.649) -0.07 (0.795) Factor 3: Densely populated & traditional 0.53 (0.197) 0.82 (0.022) 0.86 (0.007) 0.96 (0.002) 0.99 (0.001) 1.01 (0.001) R2 adjusted 0.21 0.48 0.44 0.46 0.47 0.46 14 Mar 1: % Multi−generational household Mar 11: % Multi−generational household Mar 21: % Multi−generational household Mar 31: % Multi−generational household 15 (r = −0.17) (r = −0.12) (r = −0.08) (r = −0.09) ITA ESP ITA FRA 10 ITA FRA ESP GBR AUTNLD CHE BEL PRT GBR CHE SWE IRL NOR ITA FRA ESP AUTNLD NOR SWE BEL DNK CZE POL DNK CZEPRT FIN SVN SWENLD NOR GBR CHE DNK BEL FIN IRL EST POL SVN EST LTUHUN AUT 5 FRA ESP FIN IRL CZE PRT SVN LTUHUN GBR CHE POL AUTNLD NOR SWE EST HUN FIN DNK CZE BEL LTU 0 IRL EST LTUHUN PRT SVN POL 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 Mar 1: % Social meeting (squared) Mar 11: % Social meeting (squared) Mar 21: % Social meeting (squared) Mar 31: % Social meeting (squared) 15 (r = 0.44) (r = 0.66) (r = 0.65) (r = 0.63) ITA ESP ITA FRA 10 ITA ESP FRA GBR AUT BEL CHE NLD GBR CHE NORPRT SWE Log of reported confirmed cases of COVID−19 ESP FRA AUT BEL NLD NOR POL CZEDNK IRL ITA DNK CZEIRL SWE PRT FIN GBR DNKAUT CHESWE NORNLD POL FIN HUN LTU EST SVN BEL EST SVN 5 FRA ESP CZE HUN LTU GBR CHE NOR POL SVN IRL FIN PRT AUT SWE NLD HUN EST FIN CZE DNK BEL LTU 0 HUN LTU POL EST SVN IRL PRT 2000 4000 6000 2000 4000 6000 2000 4000 6000 2000 4000 6000 Mar 1: % Religious (ln) Mar 11: % Religious (ln) Mar 21: % Religious (ln) Mar 31: % Religious (ln) 15 European region (r = −0.04) (r = −0.06) (r = 0.08) (r = 0.17) ESP ITA 10 FRA ESP ITA FRA CHE GBR ● North ITA BEL NLD AUT PRT CHE GBR NOR NLD AUT BEL CZESWE DNK IRL POL ITA FRA ESP NOR CZESWE DNK PRT FIN ● West NORSWE DNKCHE NLD GBR BEL EST FIN SVN IRL POL EST HUN SVN LTU AUT 5 FRA ESP CZE FIN SVN PRT HUN LTU ● South CHE GBR IRL POL NORSWE NLD AUT EST HUN FIN CZE DNK BEL LTU ● East 0 EST HUN SVN LTU PRT IRL POL 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 Mar 1: % Tertiary education Mar 11: % Tertiary education Mar 21: % Tertiary education Mar 31: % Tertiary education 15 (r = −0.15) (r = −0.12) (r = −0.07) (r = −0.03) ITA ESP FRA ITA 10 ITA ESP FRA PRT AUT BEL NLD GBR CHE CHE GBR NOR AUT NLD BEL CZE DNK SWE IRL ITA ESP FRA CZEPRT NOR DNK SWE IRL FIN POL SVN EST NLD DNK SWE CHE NOR GBR FIN POL SVN EST HUN LTU AUT BEL 5 ESP FRA CZEPRT FIN SVN HUN LTU IRL AUT NOR GBR SWE CHE HUN POL EST FIN NLD CZE DNK LTU BEL 0 HUN PRT SVN POL EST LTU IRL 30 40 50 30 40 50 30 40 50 30 40 50 Mar 1: Pop density per km2 (ln) Mar 11: Pop density per km2 (ln) Mar 21: Pop density per km2 (ln) Mar 31: Pop density per km2 (ln) 15 (r = 0.18) (r = 0.33) (r = 0.4) (r = 0.47) ESP FRA ITA ITA 10 ITA ESP FRA AUT PRT CHEGBR BEL NLD CHEGBR NOR SWE ITA ESP FRA NOR SWE AUT BEL NLD IRL CZE DNK POL IRL PRTDNK CZE FIN EST SVN NOR SWE CHEGBR FIN SVNPOL LTU AUTDNK BEL NLD EST HUN 5 ESP FRA FIN CZE PRT LTU HUN CHEGBR IRL SVNPOL NOR SWE AUT NLD EST HUN FIN DNK CZE BEL LTU 0 EST LTU IRL SVN HUN PRT POL 3 4 5 6 3 4 5 6 3 4 5 6 3 4 5 6 Fig. 1. Association between the natural log of cumulative cases of COVID-19 on March 1, 11, 21 and 31 2020 and different covariates 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, European Social Survey Wave 8, World Bank Database, OECD, and WHO Database. 15 Mar 1: % Aged 65+ (squared) Mar 11: % Aged 65+ (squared) Mar 21: % Aged 65+ (squared) Mar 31: % Aged 65+ (squared) 15 (r = 0.34) (r = 0.23) (r = 0.2) (r = 0.17) ESP FRA ITA ITA 10 ITA ESP FRA GBR CHE BEL NLD AUT PRT CHENLD GBR NOR SWE ESP FRA NOR BEL AUT SWE IRL POL CZE DNK ITA DNK CZE PRT FIN NOR CHENLD DNK GBR SWE IRL POL FIN HUNSVN EST LTU BEL AUT SVN EST 5 ESP FRA CZE SVN FIN PRT HUNLTU GBR CHE IRL POL NOR AUT EST NLD SWE FIN HUN DNK BEL CZE LTU 0 IRL POL HUNSVN EST LTU PRT 200 300 400 500 200 300 400 500 200 300 400 500 200 300 400 500 Mar 1: GDP per capita Mar 11: GDP per capita Mar 21: GDP per capita Mar 31: GDP per capita 15 (r = 0.31) (r = 0.45) (r = 0.37) (r = 0.3) ITA ESP ITA FRA 10 ITA ESP FRA PRT GBR BELNLD AUT CHE GBR CHE SWE NOR Log of reported confirmed cases of COVID−19 ESP FRA BELNLD AUT NOR POL CZE DNK IRL ITA PRT CZE SWE DNK FIN GBR NLD SWE DNK NOR CHE POL EST FIN IRL EST LTU HUN SVN BEL AUT SVN 5 ESP FRA CZE PRT SVN FIN HUN LTU GBR CHE POL IRL AUT NLD SWE NOR HUN EST FIN DNK CZE BEL LTU 0 POL HUN LTU EST PRT SVN IRL 20000 40000 60000 80000 20000 40000 60000 80000 20000 40000 60000 80000 20000 40000 60000 80000 Mar 1: Mean household size Mar 11: Mean household size Mar 21: Mean household size Mar 31: Mean household size 15 European region (r = −0.07) (r = 0.12) (r = 0.12) (r = 0.07) ITA ESP 10 FRA ITA ESP FRA GBR CHE ● North ITA AUT NLD PRTNOR BEL GBR NLD CHE SWE IRL AUT BEL CZE DNK ITA FRA ESP DNKNOR CZE FIN SWE IRL PRT FIN POL SVN ● West GBR NLDSWE DNKNOR CHE BEL EST POL SVN HUN EST LTU AUT 5 FRA ESP CZE FIN PRT HUN LTU ● South GBR NOR CHE IRL POL SVN AUT NLDSWE HUN EST FIN CZE DNK BEL LTU ● East 0 LTU HUN EST IRL PRT POL SVN 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 Mar 1: % Living in (semi−) rural Mar 11: % Living in (semi−) rural Mar 21: % Living in (semi−) rural Mar 31: % Living in (semi−) rural 15 (r = 0.22) (r = 0.4) (r = 0.42) (r = 0.36) ITA ESP ITA FRA 10 ITA FRA ESP GBR NLDAUT BEL CHE GBR NLDAUT BEL CHE SWEPRTCZE DNK NOR IRL POL ITA FRA ESP SWEPRTCZE DNK NOR FIN SWE GBR DNK NOR NLD CHE FIN IRL POL LTU EST HUN SVN AUT BEL EST SVN 5 FRA ESP PRTCZE FIN SVN LTU HUN GBR CHE IRL POL SWE NOR AUT EST HUN FIN NLD DNK CZE LTU BEL 0 LTU EST HUN PRT IRL POL SVN 30 40 50 30 40 50 30 40 50 30 40 50 Mar 1: % Smokers Mar 11: % Smokers Mar 21: % Smokers Mar 31: % Smokers 15 (r = −0.05) (r = −0.24) (r = −0.09) (r = −0.03) ITA ESP ITA FRA 10 ITA ESP FRA GBR CHE BEL NLD AUT GBR CHENLD BEL AUT DNK NOR IRL PRT SWE CZE ITA FRA ESP DNK NOR SWE PRT FIN POL NOR CHENLD FIN IRL POL CZE SVN LTU EST DNK SWE GBR BEL SVN EST HUN AUT 5 FRA ESP FIN IRL PRT SVN CZE HUN LTU GBR CHE POL NOR SWE NLD AUT HUN EST DNK FIN BEL CZE LTU 0 SVNIRL PRT POL HUN LTU EST 15 20 25 30 15 20 25 30 15 20 25 30 15 20 25 30 Fig. 1. (Cont.) 16 Mar 1: Socially and economically vibrant Mar 11: Socially and economically vibrant Mar 21: Socially and economically vibrant Mar 31: Socially and economically vibrant 12 ITA ESP (r = −0.02) (r = 0.04) (r = 0.08) ITA (r = 0.13) FRA ESP GBR FRA CHE ITA BEL AUT NLD 9 CHE PRT GBR NOR SWE NLD CZE IRL DNK ESP FRA AUT BEL POL ITA NOR SWE DNK PRT FIN CZE IRL SVN NOR CHE EST GBRSWE DNK NLD POL FIN LTU HUN 6 BEL SVN EST AUT FRA HUN ESP CZE LTU SVN FIN PRT GBR IRL CHE POL 3 NOR AUT SWE EST HUN NLD FIN DNK CZE LTU BEL Log of reported confirmed cases of COVID−19 0 POL HUN LTU SVN EST IRL PRT Mar 1: High−educated and not aged Mar 11: High−educated and not aged Mar 21: High−educated and not aged Mar 31: High−educated and not aged 12 ITA ESP (r = −0.46) (r = −0.46) (rITA= −0.43) (r = −0.35) FRA ESP GBR FRA CHE ITA BEL NLD AUT 9 CHE PRT NLD AUT BEL GBR CZE SWE DNK NOR IRL European region FRA ESP NOR POL ITA SWE DNK FIN PRT CZE IRL SVN EST ● North CHENOR FIN POL LTU NLD SWE DNK GBR HUN 6 BEL SVN EST ● West AUT FRA ESP CZE HUN LTU ● South PRT FIN SVN GBR IRL CHE POL ● East 3 NOR SWE AUT HUN EST NLD FIN DNK CZE LTU BEL 0 PRT HUN SVN EST POL LTU IRL Mar 1: Densely populated and traditional Mar 11: Densely populated and traditional Mar 21: Densely populated and traditional Mar 31: Densely populated and traditional 12 ESP ITA (r = 0.29) (r = 0.3) (r = 0.34) ITA (r = 0.36) FRA ESP GBR FRA CHE ITA AUT BEL NLD 9 CHE PRT GBR NOR SWE NLD CZEDNK IRL FRA ESP AUT BEL POL ITA NOR SWE DNK PRT FIN CZE IRL SVN NOR CHE EST SWE DNK GBR NLD FIN POL LTU HUN 6 BEL EST SVN AUT FRA HUN ESP CZE LTU FIN SVN PRT GBR IRL CHE POL 3 NOR SWE AUT EST HUN NLD FIN DNK CZE LTU BEL 0 EST LTU HUN SVN IRL PRT POL −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 Fig. 2. Factor scores vs. natural log of COVID-19. March 1 and 31, 2020, for 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, European Social Survey Wave 8, World Bank Database, OECD, and WHO Database. Note: * p<0.05; ** p<0.01. 17 References 1. Johns Hopkins University Center for Systems Science and Engineering (2020). Novel Coronavirus (COVID-19) Cases. https://github.com/CSSEGISandData/COVID-19. Accessed: April 30, 2020. 2. Huang, C., Wang, Y., Li, W., & et al. (2020) Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet, 395, 497-506. 3. Peeri, N.C., Shestha, N., Rahman, M.S., & et al. (2020) The SARS, MERS and novel coronavirus (COVID-19) epidemics, the newest and biggest global health threats: what lessons have we learned? International Journal of Epidemiology. In press. 4. WHO (2020) Coronavirus disease (COVID-19) situation reports. www.who.int/emergencies/diseases/novel-coronavirus-2019/situation-reports. 5. Grais, R.F., Ellis, J.H., & Glass, G.E. (2003). Assessing the impact of airline travel on the geographic spread of pandemic influenza. European journal of epidemiology, 18, 1065-1072. 6. Bayer, C. & Kuhn, M. (2020). Intergenerational ties and case fatality rates: A cross-country analysis. Available from: https://voxeu.org/article/intergenerational-ties-and-case-fatality- rates. 7. Mossong J., Hens N., Jit M., Beutels P., Auranen K., Mikolajczyk R., . . . Wallinga J. (2008). Social contacts and mixing patterns relevant to the spread of infectious diseases. PLoS medicine, 5, e74. 8. Liu, Y., Eggo, R.M. & Kucharski, A.J. (2020). Secondary attack rate and superspreading events for SARS-CoV-2. Lancet, 395, e47. 9. Bi, Q., Wu, Y., Mei, S., & et al. (2020), Epidemiology and transmission of COVID-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts. medRxiv. 10. Wallinga, J., Teunis, P., & Kretzschmar. M. (2006). Using data on social contacts to estimate age-specific transmission parameters for respiratory-spread infectious agents. American Journal of Epidemiology, 164, 936-944. Doi:10.1093/aje/kwj317. 11. Reher, D.S. (1998). Family ties in Western Europe: Persistent contrasts. Population and Development Review, 24, 203-234. 12. Sánchez Rodríguez, M.M., Gierveld, J.D.J., & Buz, J. (2014). Loneliness and the exchange of social support among older adults in Spain and the Netherlands. Ageing & Society, 34, 330-354. 13. Eurostat (2020). Population on 1 January by age and sex. https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=demo_pjan&lang=en. Accessed: April 30, 2020. 18 14. Eurostat (2020). Average household size - EU-SILC survey. https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=ilc_lvph01&lang=en. Accessed: April 30, 2020. 15. Eurostat (2020). Population by educational attainment level, sex and age (%). http://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=edat_lfse_03&lang=eng. Accessed: April 30, 2020. 16. World Bank (2020). GDP per capita (current US$) https://data.worldbank.org/indicator/NY.GDP.PCAP.CD? Accessed April 30, 2020. 17. World Bank (2020). Population ages 65 and above (% of total population), 2019 https://data.worldbank.org/indicator/SP.POP.65UP.TO.ZS? Accessed April 30, 2020. 18. World Bank (2020). Population density (people per sq. km of land area), 2019 https://data.worldbank.org/indicator/EN.POP.DNST Accessed April 30, 2020. 19. Eurostat (2019). Available beds in nursing and residential care facilities. https://appsso.eurostat.ec.europa.eu/nui/show.do?dataset=hlth_rs_bdsns&lang=en. Accessed April 30, 2020. 20. Eurostat (2019). Long-Term Care Resources and Utilisation: Long-term care recipients. https://stats.oecd.org/Index.aspx?ThemeTreeId=9. Accessed April 30, 2020. 21. Harrell Jr, F.E., Lee, K.L., Califf, R.M., Pryor, D.B., & Rosati, R.A. (1984). Regression modelling strategies for improved prognostic prediction. Statistics in Medicine, 3, 143-152. 22. Peduzzi, P., Concato, J., Kemper, E., Holford, T.R., & Feinstein, A.R. (1996). A simulation study of the number of events per variable in logistic regression analysis. Journal of Clinical Epidemiology, 49, 1373-1379. 23. Comas-Herrera A, Zalakain J, Litwin C, Hsu AT, Lane N, Fernandez-Plotka J-L (2020) Mortality associated with COVID-19 outbreaks in care homes: early international evidence. International Long-Term Care Policy Network, CPEC-LSE. https://LTCcovid.org 24. Orange, R. (2020, Sunday 19). Anger in Sweden as elderly pay price for coronavirus strategy. The Guardian. 19 April. https://www.theguardian.com/world/2020/apr/19/anger- in-sweden-as-elderly-pay-price-for-coronavirus-strategy 25. Flaxman, S., Mishra, S., Gandy, A., & et al. (2020). Estimating the number of infections and the impact of non-pharmaceutical interventions on COVID-19 in 11 European countries. London: Imperial College London. 26. Hodcroft, E.B. (2020). Preliminary case report on the SARS-CoV-2 cluster in the UK, France, and Spain. Swiss Medical Weekly, 150. doi:10.4414/smw.2020.20212. 19 27. Hruby, D. (2020). How an Austrian ski resort helped coronavirus spread across Europe. https://edition.cnn.com/2020/03/24/europe/austria-ski-resort-ischgl-coronavirus- intl/index.html. 28. Paul, C.L., Ross, S., Bryant, J., Hill, W., Bonevski, B., & Keevy, N. (2010). The social context of smoking: A qualitative study comparing smokers of high versus low socioeconomic position. BMC public health, 10, 211. 29. Science Media Centre (2020). Expert reaction to questions about smoking and COVID-19. https://www.sciencemediacentre.org/expert-reaction-to-questions-about-smoking-and- covid-19/ 30. GBD 2015 Tobacco Collaborators (2017). Smoking prevalence and attributable disease burden in 195 countries and territories, 1990–2015: a systematic analysis from the Global Burden of Disease Study 2015. Lancet, 389, 1885-1906. Doi: 10.1016/S0140- 6736(17)30819-X. 31. Farge, E. & Revill, J. (2020). ‘Test, test, test’: WHO chief’s coronavirus message to world. Reuters. https://www.reuters.com/article/us-healthcare-coronavirus-who/test-test-test-who- chiefs-coronavirus-message-to-world-idUSKBN2132S4. 32. Wikipedia (2020). COVID-19 testing. https://en.wikipedia.org/wiki/COVID-19_testing. 20 Supplementary material Table S1. Correlation matrix between the covariates Variable 1 2 3 4 5 6 7 8 10 1. Mean number of household members 2. % living in a multi-generational household (ln) 0.77** 3. % having a frequent social meeting (squared) -0.32 -0.54** 4. % ≥ weekly religious attendance (ln) 0.69** 0.60** -0.27 5. % tertiary education -0.40 -0.59** 0.29 -0.26 6. Population density (pop per km2) (ln) 0.19 0.10 0.05 0.32 -0.25 7. % population in (semi-)rural areas 0.29 0.19 0.14 0.34 -0.05 0.40 8. % people aged 65+ (squared) -0.34 0.02 0.08 -0.14 -0.45* 0.05 -0.18 9. GDP per capita -0.53** -0.73** 0.62** -0.27 0.62** -0.02 0.22 -0.35 10. Beds in nursing/residential care homes per 100,000 -0.64** -0.70** 0.40 -0.49* 0.54** 0.11 0.14 0.04 0.54** **p<0.01, *p<0.05 The Shapiro-Wilk test for normality was used to check the cross-country distribution of the variables. As a result, variables 2, 3, 4, 6 and 8 were transformed prior to analysis. Two variables were not included in the final Factors Analysis: the mean number of household members because of its high correlation with multigenerational household but a slightly lower communality; and % population in (semi-) rural areas because the factor loadings were almost identical to the population density variable but its association with COVID-19 was the opposite than expected. The remaining variables were used to obtain the final set of factors. The KMO test score after excluding variables 1 and 7 equals 0.65. Table S2. Communalities, factors, eigenvalues and percentage of explained variance in unrotated solution % of Variable Communality Factor Eigenvalue Cum % variable 2. % living in a multi-generational household 0.858 1 3.59 44.82 44.82 3. % having a frequent social meeting (squared) 0.582 2 1.42 17.78 62.61 4. % ≥ weekly religious attendance (ln) 0.766 3 1.23 15.40 78.01 5. % tertiary education 0.794 4 0.72 9.03 87.03 6. Population density (pop per km2) (ln) 0.854 5 0.44 5.53 92.56 8. % people aged 65+ (squared) 0.860 6 0.27 3.39 95.95 9. GDP per capita 0.829 7 0.20 2.50 98.46 10. Beds in nursing/residential care homes per 100,000 0.697 8 0.12 1.54 100.00 21 Table S3. Rotated factor matrix Variable Factor 1 Factor 2 Factor 3 2. % living in a multigenerational household -0.888 -0.105 0.241 3. % having a frequent social meeting (squared) 0.749 -0.020 0.146 4. % ≥ weekly religious attendance (ln) -0.565 0.249 0.620 5. % tertiary education 0.540 0.652 -0.279 6. Population density (pop per km2) (ln) 0.120 -0.142 0.905 8. % people aged 65+ (squared) 0.072 -0.924 -0.031 9. GDP per capita 0.772 0.482 0.024 10. Beds in nursing/residential care homes per 0.833 0.005 -0.050 100,000 Note: The principal component matrix was rotated using the Varimax method. Given the factor loadings (highest values in bold), Factor 1 could be labelled as “socially and economically vibrant”, Factor 2 as “high-educated and not aged”, and Factor 3 as “densely populated and traditional”. Factor 1 is also typified by a high number of nursing care facilities per population. The countries that score highest/lowest in Factor 1 are the Netherlands, Sweden and Switzerland/Eastern European countries; Factor 2: Ireland/Italy, Germany and Portugal; Factor 3 Netherlands, Italy/Estonia, Nordic countries except Denmark. See also Table S4. Table S4. Factor scores for each country Country Factor 1 Factor 2 Factor 3 Austria 0.2338 0.2490 0.2696 Belgium 0.8971 0.1933 1.0795 Bulgaria -1.5110 -0.9075 -0.6210 Czechia -0.4680 -0.7845 -0.1445 Denmark 1.1588 -0.2939 -0.1596 Estonia -0.4807 -0.1151 -1.9281 Finland 0.9651 -0.6853 -1.9351 France 0.7554 -0.3312 -0.0198 Germany 0.6553 -1.2374 0.5822 Hungary -1.0652 -0.3820 -0.2253 Ireland -0.4701 3.1165 0.5764 Italy -0.6188 -1.6056 1.3142 Lithuania -1.0465 0.4921 -0.8800 Netherlands 1.4106 -0.0443 1.5242 Norway 0.7975 1.2269 -1.7080 Poland -1.9381 1.1215 0.6966 Portugal -0.2882 -1.1869 1.0072 Serbia -1.3220 -0.2815 -0.0398 Slovenia -0.4741 -0.1887 0.1143 Spain -0.2511 0.2381 0.3171 Sweden 1.3230 -0.1462 -1.3008 Switzerland 1.3879 0.7528 0.7532 The UK 0.3494 0.7999 0.7274 Note: See Table S3 22 Table S5a. Correlation between the ln of change in cases of COVID-19 between March 1-11, 11-21, 21-31, March 31-April 10, and April 10-20, 2020 and the extracted factors. March 1-11 March 11-21 March 21-31 Mar 31-April10 April 10-20 Factor 1: Socially & economically vibrant 0.61** 0.50** 0.44* 0.41* 0.32 Factor 2: High educated & not aged -0.16 -0.14 -0.08 -0.04 0.00 Factor 3: Densely populated & traditional 0.39* 0.50** 0.58** 0.60** 0.60** **p<0.01, *p<0.05 Table S5b. Multivariate regression analysis of social and demographic factors on the ln of change in cases of COVID-19 between March 1-11, 11-21, 21-31, March 31-April 10, and April 10-20, 2020, for 23 European countries. Unstandardized coefficients. Unstandardized coefficients (p value) March 1-11 March 11-21 March 21-31 Mar 31-April10 April 10-20 Factor 1: Socially & economically vibrant 1.31 (0.001) 0.88 (0.006) 0.77 (0.011) 0.69 (0.017) 0.57 (0.010) Factor 2: High educated & not aged -0.35 (0.194) -0.25 (0.392) -0.15 (0.599) -0.07 (0.799) 0.01 (0.983) Factor 3: Densely populated & traditional 0.84 (0.020) 0.88 (0.006) 1.01 (0.001) 1.02 (0.001) 1.07 (0.002) R2 adjusted 0.48 0.44 0.47 0.46 0.38 Table S6a. Correlation between the natural log of cumulative cases of COVID-19 per 100,000 people at six different time periods and the extracted factors. March 1 March 11 March 21 March 31 April 10 April 20 Factor 1: Socially & economically vibrant -0.03 0.29 0.35 0.40* 0.44* 0.42* Factor 2: High educated & not aged -0.32 -0.17 -0.08 0.01 0.10 0.22 Factor 3: Densely populated & traditional 0.24 0.14 0.28 0.39 0.46* 0.50** **p<0.01, *p<0.05 Table S6b. Multivariate regression analysis of social and demographic factors on cumulative cases of COVID-19 per 100,000 people on March 1, 11, 21 and 31, 2020, and April 10 and 20, for 23 European countries. Unstandardized coefficients. Unstandardized coefficients (p value) March 1 March 11 March 21 March 31 April 10 April 20 Factor 1: Socially & economically vibrant -0.02 (0.886) 1.36 (0.197) 8.04 (0.103) 23.06 (0.050) 38.76 (0.021) 48.78 (0.020) Factor 2: High educated & not aged -0.18 (0.146) -0.81 (0.434) -1.82 (0.702) 0.57 (0.959) 8.7 (0.580) 25.47 (0.199) Factor 3: Densely populated & traditional 0.14 (0.261) 0.65 (0.528) 6.44 (0.186) 22.84 (0.052) 40.86 (0.016) 58.10 (0.007) R2 adjusted 0.03 -0.01 0.08 0.21 0.33 0.40 23 Mar 1: % Multi−generational household Mar 11: % Multi−generational household Mar 21: % Multi−generational household Mar 31: % Multi−generational household 200 (r = 0.21) ESP (r = 0.12) (r = 0.14) (r = 0.1) CHE ITA 150 AUT BEL 100 ITA CHE FRANLD NOR PRT IRL ESP EST 50 NOR SWE GBR DNK SVN AUT DNK EST BEL FIN CZE ITA FRANLDIRL SWE SVN LTU SWE FRA NOR CHE DNKLTUHUN ESP FIN GBR PRT CZE HUN POL 0 SWE FRA AUT NOR FINNLDIRL GBR CHE EST DNKLTUHUN CZE PRT BEL ITA ESP SVN POL FINNLDIRL AUT GBR EST BEL CZE PRT SVN POL LTUHUN POL 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 Mar 1: % Social meeting (squared) Mar 11: % Social meeting (squared) Mar 21: % Social meeting (squared) Mar 31: % Social meeting (squared) Reported confirmed cases of COVID−19 per 100,000 people 200 (r = 0.04) ESP (r = 0.29) (r = 0.37) (r = 0.5) CHE ITA 150 AUT BEL 100 ITA NOR CHE FRA NLD PRT IRL 50 ESP EST DNK NOR SVN GBR SWE ITA EST SVN DNKAUT BELFRA SWE NLD LTU CZE FIN ITAAUT SVN DNKAUT ESP FRA BEL NOR CHESWE NLD HUN LTU POL CZEIRL FIN GBR PRT HUN POL 0 HUN LTU POL EST CZEDNK SVN IRL FIN GBR CHESWE ESP FRA BEL NORNLD PRT HUN LTU POL EST CZEIRL FIN GBR PRT 2000 4000 6000 2000 4000 6000 2000 4000 6000 2000 4000 6000 Mar 1: % Religious (ln) Mar 11: % Religious (ln) Mar 21: % Religious (ln) Mar 31: % Religious (ln) 200 (r = 0.22) (r = 0.01) (r = 0.03) (r = 0.08) CHE ESP European region ITA 150 ● North BEL AUT 100 ITA ● West NOR FRA CHE NLD PRT 50 ESP EST DNK IRL ● South NOR SWE SVN GBR EST BEL DNK AUT CZE FIN FRA NORSWE DNK FRA CHE SVNESP NLD AUT ITA CZESWE FIN NLD SVN HUN GBR LTU PRT IRL HUN LTU POL ● East 0 EST NOR CZE FIN SWEFRA CHE DNKNLD AUT BEL HUN GBR SVNESP LTU ITA PRT IRL POL EST FIN CZE BEL HUN GBR LTU PRT IRL POL POL 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 Mar 1: % Tertiary education Mar 11: % Tertiary education Mar 21: % Tertiary education Mar 31: % Tertiary education 200 (r = −0.45) ESP (r = −0.26) (r = −0.14) (r ITA = −0.01) CHE 150 AUT BEL 100 ITA NOR CHE PRT FRA NLD IRL 50 ESP EST DNK SWE AUT EST NOR CZE SVN GBR ITA SVN BEL DNK FRA NLD SWE FIN LTU NOR CHE PRT FIN POL IRL 0 ITA HUN CZEPRT AUT FIN POL SVN EST ESP NOR GBR DNK FRA SWE BEL NLD CHE LTU IRL HUN CZEPRT AUT SVN DNK FRA ESP FIN POL EST SWE NLD BEL GBR IRL LTU HUN CZE GBR LTU HUN POL 30 40 50 30 40 50 30 40 50 30 40 50 Mar 1: Pop density per km2 (ln) Mar 11: Pop density per km2 (ln) Mar 21: Pop density per km2 (ln) Mar 31: Pop density per km2 (ln) 200 (r = 0.13) ESP (r = 0.04) (r = 0.18) (r = 0.28) CHE ITA 150 AUT BEL 100 ITA NOR CHE FRA PRT NLD IRL ESP EST 50 NOR SWE DNK SVN CZE GBR EST AUTDNK BEL NLD FIN ITA IRL SVNFRA LTU ITA NOR FIN SWE IRL ESP SVN DNK FRA AUT CHEGBR BEL NLD FIN SWE LTU PRTCZE GBR POL HUN 0 NORFIN SWE EST LTU IRL ESP AUT SVNFRA HUN PRT POL DNK CZE CHEGBR BEL NLD EST LTU PRT HUN CZE POL POL HUN 3 4 5 6 3 4 5 6 3 4 5 6 3 4 5 6 Fig. S1. Association between cumulative cases of COVID-19 per 100,000 population on March 1, 11, 21 and 31 2020 and different covariates 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, European Social Survey Wave 8, World Bank Database, OECD, and WHO Database. Note: Population data in 2019 is used. 24 Mar 1: % Aged 65+ (squared) Mar 11: % Aged 65+ (squared) Mar 21: % Aged 65+ (squared) Mar 31: % Aged 65+ (squared) 200 (r = 0.46) ESP (r = 0.31) (r = 0.22) (r = 0.12) CHE ITA 150 AUT BEL 100 ITA NOR CHE IRL NLD FRA PRT ESP EST 50 NOR GBR DNK SVNSWE AUT EST BEL DNK CZE FIN ITA IRL NLDSVNFRA SWE LTU NOR CHE DNK CZE PRT FIN 0 IRL NOR CHE POL GBRAUT BEL ESP HUN NLD CZE SVN ESTFRA SWE LTU DNK FIN PRT ITA IRL BEL POL GBRAUTESP NLD HUNSVN EST CZE SWE FRA LTU FIN PRT POL GBR HUN LTU POL HUN 200 300 400 500 200 300 400 500 200 300 400 500 200 300 400 500 Mar 1: GDP per capita Mar 11: GDP per capita Mar 21: GDP per capita Mar 31: GDP per capita Reported confirmed cases of COVID−19 per 100,000 people 200 (r = 0.03) ESP (r = 0.37) (r = 0.43) (r = 0.4) CHE ITA 150 AUT BEL 100 ITA NOR CHE PRT FRA NLD IRL 50 ESP EST DNK NOR SVN GBR SWE AUT CZE FIN ITA EST SVN FRA BELNLD SWE DNK IRL LTU NOR PRT 0 LTU POL HUN EST CZESVNESP ITA FRAGBR PRT FIN AUT BELNLD SWE DNK IRL NOR CHE POL HUN EST CZE LTU SVNESP PRT FRAGBR BEL AUT FIN DNK NLD SWE IRL CHE LTUCZE POL HUN GBRFIN POL HUN 20000 40000 60000 80000 20000 40000 60000 80000 20000 40000 60000 80000 20000 40000 60000 80000 Mar 1: Mean household size Mar 11: Mean household size Mar 21: Mean household size Mar 31: Mean household size 200 (r = 0.12) (r = 0.23) (r = 0.32) (r = 0.32) CHE ESP European region ITA 150 ● North AUT BEL 100 ITA ● West FRA NOR CHE NLD PRT 50 ESP EST IRL DNK ● South NOR GBR SWE SVN AUT DNK BEL CZE FIN FRA NLD EST PRT FRA NLDSWE DNKNOR ITA CHE ESP POL SVN FIN SWE IRL CZELTU GBR SVN LTU HUN POL ● East 0 FRA AUT GBR CZELTUNLD HUN FIN SWE DNKNOR ITA EST IRL PRT CHE BEL ESP POL SVN AUT FIN CZELTU GBR HUN EST IRL PRT BEL HUN POL 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 2.25 2.50 2.75 3.00 Mar 1: % Living in (semi−) rural Mar 11: % Living in (semi−) rural Mar 21: % Living in (semi−) rural Mar 31: % Living in (semi−) rural 200 (r = 0.22) ESP (r = 0.32) (r = 0.56) (r = 0.6) CHE ITA 150 AUT BEL 100 ITA NOR CHE PRT FRA NLD IRL 50 ESP EST DNK NOR SWE GBR SVN EST DNK FRA AUT BEL CZE FIN ITA SWEPRTCZE IRL NLD SVN LTU SWE NOR DNK FRA ESP CHE LTU GBR HUN FIN HUN POL 0 LTU SWE GBR EST HUN NOR DNK FRA PRTCZE FIN IRL POL ITA ESP NLDAUT BEL CHE SVN LTU GBR EST PRTCZE HUN FIN NLD IRL POL AUT BEL SVN POL 30 40 50 30 40 50 30 40 50 30 40 50 Mar 1: % Smokers Mar 11: % Smokers Mar 21: % Smokers Mar 31: % Smokers 200 (r = 0) ESP (r = −0.24) (r = −0.02) (r = 0.08) CHE ITA 150 BEL AUT 100 ITA NOR CHE NLD FRA IRL PRT ESP EST 50 NOR DNK SWE GBR SVN DNK BEL EST AUT FIN CZE ITA SVN SWE NLD FRA LTU ITA POL DNK NOR SWEFIN IRL CHE SVN BELNLD POL ESP LTU FRA AUT GBR FIN IRL PRT LTU CZE POL HUN 0 DNK GBRNOR SWEFIN IRL CHE SVN BEL PRT NLD HUNFRA ESP LTU AUT EST CZE GBR PRT HUN EST CZE POL HUN 15 20 25 30 15 20 25 30 15 20 25 30 15 20 25 30 Fig. S1. (Cont.) 25 Mar 1−11: % Multi−generational household Mar 11−21: % Multi−generational household Mar 21−31: % Multi−generational household 15 (r = 0.24) (r = 0.26) (r = 0.18) ITA FRA ESP ITA 10 ITA FRA ESP GBR AUT NLD CHE BEL GBR CHE PRT FRA ESP AUT NLD NOR BEL SWE NOR IRL DNK CZE POL SWE IRL DNK PRT CZE FIN SWE NLD NOR GBR CHE DNK FIN POL EST LTUHUN SVN AUT BEL EST SVN 5 FIN CZE PRT SVN LTUHUN IRL POL EST HUN LTU 0 Log of new reported confirmed cases of COVID−19 between two periods 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 Mar 1−11: % Social meeting (squared) Mar 11−21: % Social meeting (squared) Mar 21−31: % Social meeting (squared) 15 (r = 0.08) (r = 0.16) (r = 0.28) ITA ITA ESP FRA 10 ITA ESP FRA GBR AUT BEL CHE NLD CHE PRT ESP FRA AUTGBR BEL NLD NOR PRT POL CZE IRL DNK SWE NOR DNK CZE IRL SWE FIN DNK AUTGBR CHE SWE NORNLD POL FIN HUN LTU EST SVN BEL EST SVN 5 CZE SVN FIN PRT HUN LTU POL IRL HUN EST LTU 0 2000 4000 6000 2000 4000 6000 2000 4000 6000 Mar 1−11: % Religious (ln) Mar 11−21: % Religious (ln) Mar 21−31: % Religious (ln) 15 European region (r = 0.24) (r = 0.26) (r = 0.2) ESP 10 ESP ITA FRA GBR ITA ● North ITA FRA BEL CHE NLD AUT CHE GBR PRT NLD AUT NOR SWE FRA CHE ESP NOR CZE SWE BEL DNK PRT IRL CZE SWE NOR FIN DNK IRL POL ● West DNK NLD BEL GBR EST FIN SVN POL EST HUN SVN LTU 5 AUT CZEFIN SVN PRT IRL POL HUN LTU ● South EST HUN LTU ● East 0 1 2 3 4 1 2 3 4 1 2 3 4 Mar 1−11: % Tertiary education Mar 11−21: % Tertiary education Mar 21−31: % Tertiary education 15 (r = −0.44) (r = −0.36) (r = −0.16) ITA ITA ESP FRA 10 ITA ESP FRA AUT BEL NLD GBR CHE CHE GBR PRT ESP FRA AUT NLD BEL CZE POLDNK SWE NOR IRL CZE PRT NOR DNK SWE FIN NOR NLD DNK SWE CHE GBR FIN POL IRL SVN EST LTU AUT BEL SVN EST HUN 5 CZE PRT SVN FIN HUN LTU POL IRL HUN EST LTU 0 30 40 50 30 40 50 30 40 50 Mar 1−11: Pop density per km2 (ln) Mar 11−21: Pop density per km2 (ln) Mar 21−31: Pop density per km2 (ln) 15 (r = 0.21) (r = 0.25) (r = 0.27) ITA ESP FRA ITA 10 ITA ESP FRA AUT CHE GBR BEL NLD CHE GBR PRT ESP FRA AUT BEL NLD NOR SWE IRL CZE POL NOR SWE PRT DNK CZE FIN DNK NOR CHE GBR IRL SWE AUT DNK BEL NLD FIN EST SVNPOL EST LTU SVN HUN 5 FIN PRT SVN CZE LTU HUN IRL POL EST HUN LTU 0 3 4 5 6 3 4 5 6 3 4 5 6 Fig. S2. Association between the ln of new cases of COVID-19 during the periods March 1-11, 11-21 and 21-31 2020 and different covariates for 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, Eurostat, European Social Survey Wave 8, World Bank Database, OECD, and WHO Database. 26 Mar 1−11: % Multi−generational household Mar 11−21: % Multi−generational household Mar 21−31: % Multi−generational household 15 (r = 0.24) (r = 0.26) (r = 0.18) ITA FRA ESP ITA 10 ITA FRA ESP GBR AUT NLD CHE BEL GBR CHE PRT FRA ESP AUT NLD NOR BEL SWE NOR IRL DNK CZE POL SWE IRL DNK PRT CZE FIN SWE NLD NOR GBR CHE DNK FIN POL EST LTUHUN SVN AUT BEL EST SVN 5 FIN CZE PRT SVN LTUHUN IRL POL EST HUN LTU 0 Log of new reported confirmed cases of COVID−19 between two periods 10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 Mar 1−11: % Social meeting (squared) Mar 11−21: % Social meeting (squared) Mar 21−31: % Social meeting (squared) 15 (r = 0.08) (r = 0.16) (r = 0.28) ITA ITA ESP FRA 10 ITA ESP FRA GBR AUT BEL CHE NLD CHE PRT ESP FRA AUTGBR BEL NLD NOR PRT POL CZE IRL DNK SWE NOR DNK CZE IRL SWE FIN DNK AUTGBR CHE SWE NORNLD POL FIN HUN LTU EST SVN BEL EST SVN 5 CZE SVN FIN PRT HUN LTU POL IRL HUN EST LTU 0 2000 4000 6000 2000 4000 6000 2000 4000 6000 Mar 1−11: % Religious (ln) Mar 11−21: % Religious (ln) Mar 21−31: % Religious (ln) 15 European region (r = 0.24) (r = 0.26) (r = 0.2) ESP 10 ESP ITA FRA GBR ITA ● North ITA FRA BEL CHE NLD AUT CHE GBR PRT NLD AUT NOR SWE FRA CHE ESP NOR CZE SWE BEL DNK PRT IRL CZE SWE NOR FIN DNK IRL POL ● West DNK NLD BEL GBR EST FIN SVN POL EST HUN SVN LTU 5 AUT CZEFIN SVN PRT IRL POL HUN LTU ● South EST HUN LTU ● East 0 1 2 3 4 1 2 3 4 1 2 3 4 Mar 1−11: % Tertiary education Mar 11−21: % Tertiary education Mar 21−31: % Tertiary education 15 (r = −0.44) (r = −0.36) (r = −0.16) ITA ITA ESP FRA 10 ITA ESP FRA AUT BEL NLD GBR CHE CHE GBR PRT ESP FRA AUT NLD BEL CZE POLDNK SWE NOR IRL CZE PRT NOR DNK SWE FIN NOR NLD DNK SWE CHE GBR FIN POL IRL SVN EST LTU AUT BEL SVN EST HUN 5 CZE PRT SVN FIN HUN LTU POL IRL HUN EST LTU 0 30 40 50 30 40 50 30 40 50 Mar 1−11: Pop density per km2 (ln) Mar 11−21: Pop density per km2 (ln) Mar 21−31: Pop density per km2 (ln) 15 (r = 0.21) (r = 0.25) (r = 0.27) ITA ESP FRA ITA 10 ITA ESP FRA AUT CHE GBR BEL NLD CHE GBR PRT ESP FRA AUT BEL NLD NOR SWE IRL CZE POL NOR SWE PRT DNK CZE FIN DNK NOR CHE GBR IRL SWE AUT DNK BEL NLD FIN EST SVNPOL EST LTU SVN HUN 5 FIN PRT SVN CZE LTU HUN IRL POL EST HUN LTU 0 3 4 5 6 3 4 5 6 3 4 5 6 Fig. S2. (Cont.) 27 Mar 1: Socially and economically vibrant Mar 11: Socially and economically vibrant Mar 21: Socially and economically vibrant Mar 31: Socially and economically vibrant 200 ESP (r = 0.05) (r = 0.3) (r = 0.38) (r = 0.45) CHE ITA 150 AUT BEL 100 ITA NOR CHE FRA NLD PRT IRL Reported confirmed cases of COVID−19 per 100,000 people ESP EST DNK 50 SWE NOR SVN GBR AUT CZE EST BEL DNK FIN ITA NOR CHE SVN IRL PRT FRA SWE NLD LTU ESP DNK SWE NLD CZE FIN GBR POLHUN 0 POLHUN LTU SVN ITA ESP EST CZE IRL FIN AUT DNK PRT NOR SWE CHE FRA BEL GBR NLD LTU SVN POLHUN EST CZE AUT IRL FIN FRA BEL GBR PRT LTU POLHUN Mar 1: High−educated and not aged Mar 11: High−educated and not aged Mar 21: High−educated and not aged Mar 31: High−educated and not aged 200 ESP (r = −0.43) (r = −0.29) (r = −0.2) (r = −0.12) CHE ITA 150 European region ● North AUT BEL 100 ● West ITA NOR CHE FRA ● South PRT NLD IRL ● ESP EST East 50 DNK NOR SWE SVN GBR AUT CZE ITA EST FRABEL DNK NLD FIN SVN SWE IRL LTU DNK NOR CHE PRT CZE FIN GBR ITA PRT CZE FIN PRT CZE SWE ESP FRA NLD SVNAUT ESTBEL GBR IRL HUN LTU POL HUN POL 0 FIN HUNFRA ESP SVNAUT EST SWE NLD DNK NOR CHE BEL GBR POL LTU IRL HUN POL LTU Mar 1: Densely populated and traditional Mar 11: Densely populated and traditional Mar 21: Densely populated and traditional Mar 31: Densely populated and traditional 200 ESP (r = 0.23) (r = 0.12) (r = 0.24) (r = 0.34) CHE ITA 150 AUT BEL 100 ITA NOR CHE FRA NLD PRT IRL ESP EST DNK 50 SWE NOR SVN GBR AUT CZE EST FRADNK BEL FIN ITA SWE IRL NLD SVN LTU NOR DNK SVN CHE FIN CZE GBR PRT HUN POL 0 ESTNOR FIN SWE FRA LTU AUT CZE HUN CHE ESP DNK SVN GBR ITA NLD IRLBEL PRT POL FIN SWE EST FRA HUN LTU AUT CZE ESP GBR NLDPOL IRLBEL PRT LTU HUN POL −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 Fig. S3. Association between cumulative cases of COVID-19 per 100,000 population on March 1, 11, 21 and 31 2020 and factor scores for 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, Eurostat, European Social Survey Wave 8, World Bank Database, OECD, WHO Database, and UNDP. Note: Population data in 2019 is used. 28 Mar 1−11: Socially and economically vibrant Mar 11−21: Socially and economically vibrant Mar 21−31: Socially and economically vibrant 15 (r = 0.05) (r = 0.09) (r = 0.15) ESP ITA ITA FRA 10 ESP GBR ITA FRA BEL CHE NLD CHE AUT PRT GBR NLD ESP FRA AUT BEL CZE IRL DNK NORSWE POL Log of new reported confirmed cases of COVID−19 between two periods PRT NOR CZE IRL DNK SWE FIN DNK GBR SWE CHE NOR NLD POL SVN FIN LTU HUN EST SVN BEL AUT EST 5 CZE HUN LTU SVN PRT IRL FIN POL HUN EST LTU 0 Mar 1−11: High−educated and not aged Mar 11−21: High−educated and not aged Mar 21−31: High−educated and not aged 15 (r = −0.46) (r = −0.41) (r = −0.28) ITA ESP European region ITA FRA 10 ESP GBR ITA FRA BEL CHE CHE PRT NLD AUT ● North NLD GBR ESP FRA AUT BEL CZE SWE NOR IRL PRT SWE NOR DNK POL ● West CZE DNK IRL FIN NLD SWE CHENOR FIN POL LTU DNK BEL AUT GBR SVN EST EST HUNSVN ● South 5 PRT CZE FIN SVN HUN LTU ● East POL IRL HUN EST LTU 0 Mar 1−11: Densely populated and traditional Mar 11−21: Densely populated and traditional Mar 21−31: Densely populated and traditional 15 (r = 0.3) (r = 0.35) (r = 0.34) ESP ITA ITA FRA 10 ESP GBR ITA FRA CHEBEL NLD CHE AUT PRT GBR NLD FRA ESP AUT BEL SWE NOR CZEDNK IRL POL NOR SWE PRT CHE NLD CZEDNK IRL FIN NOR SWE DNK GBR FIN POL EST LTU HUN SVN AUT BEL EST SVN 5 CZE HUN LTU FIN IRL PRT POL SVN EST HUN LTU 0 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 Fig. S4. Association between the ln of new cases of COVID-19 during the periods March 1-11, 11-21 and 21-31 2020 and factor scores for 20 European countries. Sources: the Center for Systems Science and Engineering at Johns Hopkins University, European Social Survey Wave 8, World Bank Database, OECD, and WHO Database. 29