Public perspectives on social distancing and other protective measures in Europe: a cross-sectional survey study during the COVID-19 pandemic

Objectives: The extent to which people implement government-issued protective measures is critical in preventing further spread of coronavirus disease 2019 (COVID-19) caused by coronavirus SARS-CoV-2. Our study aimed to evaluate the public belief in the effectiveness of protective measures, the reported implementation of these measures in daily life, and to identify communication channels used to acquire relevant information on COVID-19 in European countries. Design: A cross-sectional online survey available in multiple languages was disseminated on social media starting March 19th, 2020. After five days, we computed descriptive statistics for countries with more than 500 respondents. Each day, we compiled and categorized community containment measures enacted in each country by stringency (stage I-IV). Response collection continued for one week to explore possible dynamics as containment strategies intensified. Participants: In total, 9,796 adults responded, of whom 8,611 resided in the Netherlands (stage III), 604 in Germany (stage III), and 581 in Italy (stage IV). An additional 1,365 respondents completed the survey in the following week. Results: Participants indicated support for governmental measures related to avoiding social gatherings, selective closure of public places, and hand hygiene and respiratory measures (range for all measures: 95.0%-99.7%). Respondents from the Netherlands were less likely to consider a complete social lockdown effective (59.2%), compared to respondents in Germany (76.6%) or Italy (87.2%). Italian residents did not only apply enforced social distancing measures more frequently (range: 90.2%-99.3%, German and Dutch residents: 67.5%-97.0%), but also self-initiated hygienic and social distancing behaviors (range: 36.3%-96.6%, German and Dutch residents: 28.3%-95.7%). Respondents largely reported being sufficiently informed about the COVID-19 outbreak and about behaviors to avoid infection (range across countries: 90.2%-91.1%). Information channels most commonly reported included television (range: 53.0%-82.0%), newspapers (range: 31.0%-63.0%), official health websites (range: 39.0%-54.1%), and social media (range: 40.0%-55.8%). We observed no major changes in answers over time. Conclusions: In European countries, the degree of public belief in the effectiveness of protective measures was high and residents reported to be sufficiently informed by various communication channels. In March 2020, implementation of enacted and self-initiated measures differed between countries and were highest among Italian respondents, who were subjected to the most elaborate measures of social lockdown and greatest COVID-19 burden in Europe.


2
The recent pandemic of COVID-19 (coronavirus disease 2019) caused by SARS-CoV-2 (Severe 8 3 Acute Respiratory Syndrome Coronavirus 2) has infected more than 1,000,000 people worldwide in only a 8 4 few months' time and caused more than 51,000 deaths as of April 2nd, 2020 [1]. This rapidly spreading 8 5 virus imposes a tremendous burden on national healthcare systems, as they lack sufficient material and 8 6 human resources to respond to the rapidly increasing number of patients requiring intensive care[1,2]. 8 7 Worldwide, public health organizations, as well as national and international government bodies, have 8 8 suggested systematic implementation of protective, public health measures in an effort to delay the spread 8 The full survey was initially piloted on a sample of 50 respondents. After minor modifications to 1 4 4 the structure and language, the survey was actively disseminated through (social) media channels, such as 1 4 5 WhatsApp, Telegram, Facebook, LinkedIn, Instagram, and Twitter, and in professional networks via 1 4 6 electronic mailing lists. The survey was further promoted via a number of local and national news 1 4 7 websites and radio stations. On the landing page, participants were briefed about the study and only those 1 4 8 providing informed consent for participation were guided to the actual 5-minute survey. On the final page, 1 4 sociodemographic characteristics using descriptive summary statistics for the countries having more than 1 7 2 500 responses during this primary data collection period (the Netherlands, Germany, and Italy). Nominal 1 7 3 variables were described and visualized using frequencies and percentages. We also reported frequencies 1 7 4 of missing responses. We present stratified results for the assessed sociodemographic variables only for 1 7 5 the Netherlands, as the number of responses was sufficient per individual subgroup. No formal statistical 1 7 6 comparisons were made between countries since the primary aim was descriptive and there were no a 1 7 7 priori testable hypotheses. 1 7 8 As a secondary analysis, we explored changes in responses for items about the beliefs in the 1 7 9 effectiveness of these measures and their implementation over time. As for items about implementation of 1 8 0 protective measures, we reported the proportion of positive answers ("Yes") out of all responses, 1 8 1 excluding responses indicating the question was not applicable to their situation. Similarly, for items about 1 8 2 the belief in effectiveness of these measures, we considered the proportion of positive answers ("Probably 1 8 3 true"). To easily visualize the change over time, we modeled the proportions for each item and for each 1 8 4 country separately, using generalized additive models with time as the independent variable, using a 1 8 5 shrinkage version of cubic splines with three knots. In addition, we computed and presented visualizations 1 8 6 of the differences in proportion between the responses recorded during the primary data collection period 1 8 7 and the weeklong extension only for the Netherlands. 1 8 8 Data management, analyses and visualizations were conducted using Stata 16.1 (StatCorp LP, 1 8 9 College Station, TX) software and R 3.6.3 / RStudio 1.2. 1 9 0 1 9 1 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101/2020 effective, compared with 76.6% of respondents from Germany and 87.2% from Italy ( Figure 2 and Table  2  2  5 2). 2 2 6 2 2 7 Individual implementation of protective measures 2 2 8 2 2 9 For all items, the percentages reported in the text and Table 3 excluded respondents to whom the 2 3 0 item did not apply, which was especially important in the interpretation of three items (keeping children at 2 3 1 home before any mandates were put in place, range: 41.0%-75.1%; reducing the use of public transport, 2 3 2 range: 1.9%-28.5%; and going to school/university/work, range: 2.4%-14.9%). With regard to personal 2 3 3 protective behaviors, a high number of respondents from the Netherlands and Germany reported to have 2 3 4 washed their hands with soap and water more often than usual (range: 95.0%-95.7%). In general, 2 3 5 respondents from Italy reported applying all proposed personal protective behaviors more often than those 2 3 6 from the Netherlands or Germany, except for following a healthy diet or using vitamin supplements 2 3 7 (36.3%, Netherlands 54.5%, Germany 54.4%) ( Figure 3 and Table 3). 2 3 8 Behavior related to limiting interactions with people was fairly similar between countries, 2 3 9 although respondents from Italy reported more frequently cancelling or postponing social events (98.8%, 2 4 0 compared with 94.8% in the Netherlands and 97.0% in Germany) and avoiding crowded places more 2 4 1 frequently (99.3%, compared with 92.4% in the Netherlands and 93.8% in Germany). Respondents living 2 4 2 in Germany reported avoiding people with cold or flu-like symptoms (81.1%) less frequently than 2 4 3 respondents living in Italy (90.2%) or in the Netherlands (89.0%). Regarding behaviors related to avoiding 2 4 4 travel, respondents from Italy more often reported to have reduced the amount they went to school or 2 4 5 work (94.4%, compared to 88.0% in the Netherlands and 84.9% in Germany, of public transport use 2 4 6 (98.6% compared to 89.6% in the Netherlands and 91.3% in Germany), and of going to shops (97.7%, 2 4 7 compared to 81.4% in the Netherlands and 67.5% in Germany) ( Figure 3 and Table 3). Responses to 2 4 8 questions concerning limiting interactions and avoiding traveling may reflect both imposed restrictions 2 4 9 and respondents' awareness and willingness to follow protective measures. Therefore, we additionally 2 5 0 asked respondents whether they kept children at home prior to formal mandates to assess the percentage of 2 5 1 respondents that applied measures on their own accord. Of those indicating the question was applicable to 2 5 2 their situation, in Italy, 60.6% kept their children at home before protective measures compared to 53.6% 2 5 3 in the Netherlands and 58.1% in Germany. 2 5 4 2 5 5 Subgroup analyses among respondents living in the Netherlands 2 5 6 2 5 7 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the .

0
Although we conducted no formal comparisons between the sociodemographic subgroups of 2 5 8 participants, some patterns were evident (Supplemental Tables 2a-2r). In general, while there were no 2 5 9 meaningful differences in the degree of belief in the effectiveness of protective measures among gender 2 6 0 groups, females applied these measures most frequently. Among the different age groups, the belief in the 2 6 1 effectiveness of a complete social lockdown differed (e.g. ≤ 20 years: 47.5%, 21-40 years: 62.7%), as well 2 6 2 as among subgroups with different daily activities (e.g. retired: 55.4%, homemaker/unemployed: 64.7%) 2 6 3 and different education levels (primary/secondary: 54.6%, tertiary academic: 61.1%). Chronically ill 2 6 4 patients more frequently reported exhibiting protective measures than respondents without any chronic 2 6 5 diseases. Different sociodemographic subgroups used different sources of information to obtain 2 6 6 information related to the COVID-19 pandemic. With higher age, the percentage of respondents who 2 6 7 agreed they felt sufficiently informed was higher (for example: ≤ 20 years: 87.0%, versus > 60 years: 2 6 8 94.9%) 2 6 9 2 7 0 2 7 1 Change in responses over time 2 7 2 2 7 3 Immediately following the primary data collection period, we continued collecting data over the 2 7 4 next seven days (March 30th, 2020, 11.40 AM, UTC+0) and received responses from 1,588 additional 2 7 5 participants, of whom 1,365 reported living in the Netherlands (n=858), in Germany (n=413) and in Italy 2 7 6 (n=94). In general, we observed no substantial changes over time (Supplemental Figures 1a-1b), except 2 7 7 for a decrease in the belief of effectiveness of a complete social lockdown in Germany (Supplemental 2 7 8 Figure 1a). Furthermore, among respondents from the Netherlands, we observed a small increase in the 2 7 9 percentages of respondents indicating they believe in the effectiveness of preventive measures (range: 0%-2 8 0 5%) and those indicating they implemented these measures (range: 0%-10%) across both data collection 2 8 1 periods (Supplemental Figures 2a-2b). 2 8 2 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the .
Our findings indicate that in three European countries, the Netherlands, Germany, and Italy, the 2 8 4 public belief in the effectiveness and the actual implementation of certain protective measures during the 2 8 5 ongoing COVID-19 pandemic in March of 2020 was high. Furthermore, residents reported to be 2 8 6 sufficiently informed about the ongoing pandemic using various communication channels. 2 8 7 The public belief in the effectiveness of protective measures was highest among respondents 2 8 8 residing in Italy, which had the most extensive measures of social lockdown as well as the highest 2 8 9 numbers of COVID-19 cases and deaths in Europe in March 2020. Compared to the Netherlands and 2 9 0 Germany, respondents living in Italy most often reported not only exhibiting behaviors related to 2 9 1 government imposed restrictions but also voluntary hygienic and social measures. Although in general, 2 9 2 more than 90% of respondents indicated belief in the effectiveness of imposed measures of social 2 9 3 distancing, a complete social lockdown was deemed effective by only 59% of respondents residing in the 2 9 4 Netherlands (compared to 77% in Germany and 87% in Italy), where at the time of survey completion, 2 9 5 only lighter social distancing measures were enforced. The results of our study suggest that the level of 2 9 6 community containment measures implemented by national governments may be rapidly visible in the 2 9 7 public beliefs about protective measures, the extent to which people actually exhibit these relevant 2 9 8 behaviors, and reflect the severity of the outbreak situation in a given country.

9 9
To the best of our knowledge, to date, few data on this topic are available. Results from two 3 0 0 recently published survey studies conducted in the USA, the UK, and China primarily focus on the 3 0 1 respondents' knowledge about COVID-19 and assess understanding pertaining to the disease 3 0 2 course [13,14]. Furthermore, another study conducted between January 24th and February 13th, 2020 3 0 3 among 1715 Hong Kong residents showed that most respondents obtained information on the COVID-19 3 0 4 pandemic from social media and websites [15]. We found that traditional information sources (e.g. 3 0 5 television and news) were used most frequently among our respondents. Our study further corroborates 3 0 6 and adds to these first findings with similar results regarding beliefs in the effectiveness of hygienic and 3 0 7 social distancing measures and the extent to which these measures were exhibited in a European study 3 0 8 population.

0 9
Social distancing and other behavioral measures 3 1 0 Individuals' adherence to country-specific mitigation measures has the potential to influence the 3 1 1 course of the COVID-19 pandemic. Social (physical) distancing has been proposed as one of the most 3 1 2 effective measures for mitigating pandemics caused by viruses, including 7,16]. Large-scale 3 1 3 simulation studies have found that closure of borders is only effective to prevent further spread of the 3 1 4 All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the Provision and acquirement of information during pandemics 3 2 7 Transparent, timely, and easy-to-understand information is essential to increase trust in national 3 2 8 governments during pandemics [23]. The increasing use of portable devices and social media is evident in 3 2 9 our findings, which indicate frequent use of social media to acquire pandemic-related information (range 3 3 0 across countries: 40.0%-55.8%). However, in recent epidemics and pandemics, a substantial amount of 3 3 1 online information, especially distributed via social media, was found to be incorrect and 3 3 2 misleading [24,25]. Environmental cues to follow behavioral recommendations, favorable attitudes 3 3 3 towards prevention measures, and knowledge about the virus were associated with exhibiting protective 3 3 4 behavior [5]. Therefore accurate information provision via social media channels is crucial, besides 3 3 5 information via traditional information sources. 3 3 6 Study strengths and limitations 3 3 7 Given the evolving pandemic situation, we felt it was important to develop, translate, and 3 3 8 disseminate our questionnaire rapidly to capture a snapshot of public perceptions and behaviors in 'real 3 3 9 time' as the COVID-19 crisis unfolded in Europe and as policy makers enacted formal containment 3 4 0 measures in several European countries. Many items in our survey were adapted from an existing 3 4 1 validated questionnaire created to assess perceptions and behaviors in response to influenza[10]. We 3 4 2 attempted to make our survey accessible to participants of diverse backgrounds by providing the survey in 3 4 3 eight languages. These translations could be readily adapted for use in future studies on other viral 3 4 4 pandemics.

4 5
All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101/2020 Readers should consider some important limitations when interpreting our findings. First, since 3 4 6 the survey was web-based and recruitment was largely through digital channels including social media, we 3 4 1 4 3 7 8 CONCLUSION 3 7 9 The extent to which individuals internalize and respond to (government-mandated) mitigation 3 8 0 measures and recommendations is critical to the control of the spread of the SARS-CoV-2 virus and to 3 8 1 optimize outcomes during the current COVID-19 pandemic. In our survey study of the general public 3 8 2 living in the Netherlands, Germany, and Italy, we found that approval and application of publicly enforced 3 8 3 and self-initiated protective measures were highest in Italy, the region with the most extensive measures of 3 8 4 social lockdown and highest burden (number of cases and deaths) in Europe, during the study time period 3 8 5 in mid-March, 2020. Media channels used to acquire information and the extent to which respondents felt 3 8 6 sufficiently informed about the COVID-19 pandemic differed per country and among sociodemographic 3 8 7 subgroups in the Netherlands. No substantial changes in the perceived effectiveness of behavioral 3 8 8 protective measures and the implementation of these measures in these countries were observed between 3 8 9 March 19th and March 30th, 2020, as the COVID-19 pandemic continued to evolve in Europe and formal 3 9 0 community isolation measures became stricter. We believe these insights are valuable to inform the 3 9 1 information dissemination and infection control strategies of governments and public health organizations 3 9 2 during the current crisis and also for future pandemics. 3 9 3 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.02.20049676 doi: medRxiv preprint 1 6 submitted work. TK reports outside of the submitted work to having contributed to an advisory board of 4 2 4 CoLucid and a research project funded by Amgen, for which the Charité -Universitätsmedizin Berlin 4 2 5 received an unrestricted compensation. He further reports having received honoraria from Lilly, 4 2 6 Newsenselab, and Total for providing methodological advice, from Novartis and from Daiichi Sankyo for 4 2 7 providing a lecture on neuroepidemiology and research methods. He is further a consulting clinical 4 2 8 epidemiology editor at The BMJ and has received compensation for editorial services. 4 2 9 and that any discrepancies from the study as planned have been explained. 4 3 7 Patient and Public Involvement: The target population, the general public, was actively involved in the 4 3 8 dissemination of this survey. No patients were involved in this study, as the target group was the general 4 3 9 public. 4 4 0 Dissemination declaration: Upon request, the data will be available for policy makers and government 4 4 1 bodies. 4 4 2 4 4 3 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  COVID-19 pandemic on March 23rd, 2020, by country. 5 1 8 Note: Response percentages are rounded and may not add up to 100%. Percentages below 5% omitted. 5 1 9 5 2 0 March 23rd, 2020, by country. 5 2 2 Note: Response percentages are rounded and may not add up to 100%. Percentages below 5% omitted. 5 2 3 5 2 4 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https: //doi.org/10.1101//doi.org/10. /2020  * Based on differences in the Italian education system, we considered primary, lower secondary school and upper secondary school as "Primary/Secondary" and university degrees as "Tertiary academic".

2 6
All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.04.02.20049676 doi: medRxiv preprint 1 author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10. 1101/2020