Public risk perception and responsive behaviors in the United States, United Kingdom, and Brazil during the COVID-19 Pandemic: a cross-sectional study

Background: Understanding public perception and behaviors toward COVID-19 is valuable for mitigating the severe epidemic. Our study aimed to investigate differences of the public from the US, UK, and Brazil on the aspects of the living environment, behaviors, attitude and risk perception. Methods: Dataset were taken from the COVID-19 (Coronavirus) Survival Calculator project. 89554 respondents from the US (n=71812), UK (n=10392), and Brazil (n=7350) completed online questionnaire survey from April 28 to July 8. Frequencies and percentages were calculated to describe respondents’ responses. Chi-square tests with Holm-Bonferroni correction and binary logistic regression were used to determine the differences in environmental status, behaviors and worried degree of participants among the three countries, and further explore the sociodemographic characteristics related to protective measures. Results: Brazil participants had the highest risk perception to COVID-19, followed by the US, and the UK was the lowest (39.02±24.57 vs. 32.72±22.38 vs. 29.47+22.39, P<0.001). More Brazil participants expressed that they were worried about COVID-19. The proportion of respondents from the US and UK who reported they were not worried about COVID-19 was 2.00 times (95%CI: 1.80-2.23) and 3.24 times (95%CI: 2.88-3.65) that of Brazil. A higher percentage of Brazil respondents reported they were in close contact with more than ten people, using public transport and engaging in work. Regarding behaviors, social distancing (ranges:84.9%-86.8%) and washing hands (ranges:84.8%-90.3%) were the most frequently used protective measures, but respondents from the US (69.8%) and UK (15.8%) were relatively reluctant to wear masks. Moreover, it was found that participants from ethnic minorities were more likely to take protective measures. But males, respondents under 20 years, and respondents with poor economic conditions had various degrees of neglect to take steps. Meanwhile, healthcare workers also had a relatively stronger awareness of protection. Conclusion: There are This study was the rst attempt to use large-scale data to compare and analyze the performance and differences of ordinary people in the US, UK, and Brazil in terms of environmental status, responsive behaviors, attitudes, and risk perception towards COVID-19. We focused on the US, the UK, and Brazil, which were representative countries in the development of the COVID-19 epidemic in North America, Europe, and Latin America. The online survey began in the 7th week after the World Health Organization announced that the COVID-19 epidemic constituted a pandemic. Countries around the world have issued various policies and measures to respond to the epidemic, and the development trend of the COVID-19 epidemic has also been constantly changing. Respondents from the UK and the US were generally more optimistic, believing that their chance of contracting COVID-19 was relatively low, but Brazil respondents expressed more tension about the COVID-19 epidemic. Our nding observed that respondents living in the US and the UK did not pay much attention to masks use. Among sociodemographic subgroups analysis in the US, UK, and Brazil, it is also found that males, participants under 20 years, participants without private health insurance, and participants in a poor income are relatively less concerned about taking protective measures. At present, the global epidemic prevention and control has developed into a normalized and persistent battle. Governments and people of all countries should continue to strictly implement epidemic prevention and control measures. We believe that all countries should strengthen targeted publicity and assistance of public health intervention measures for different groups, and take effective steps to increase the protection awareness of ordinary people, thereby slowing down the process of the pandemic to a certain extent.


Statistical analysis
The statistical analysis software used in this study was IBM SPSS Statistics version 23.0. Frequencies and percentages were calculated to describe the sociodemographic characteristics, environmental status, and responsive behaviors of participants toward COVID-19 in the US, the UK, and Brazil. Mean ± standard deviation (Mean ± SD) and percentages of each interval were calculated to indicate the worried degree and risk perception of participants. Independent sample t test, one-way analysis of variance (ANOVA), and chi-square test were used to compare categorical variables between different groups, and using Holm Bonferroni correction to account for multiple comparisons. Odds ratios (ORs) and 95% con dence intervals (CIs), which obtained from binary logistic regression, were used to quantify the differences in environmental status and responsive behaviors of participants, and further explore the speci c performance of taking protective measures in groups with various sociodemographic characteristics. Multiple linear regression was utilized to explore the correlation between attitude, environmental status, responsive behaviors, and risk perception of participants towards COVID-19. Variance in ation factor was estimated to ensure the absence of multicollinearity. Results were considered statistically signi cant when P < 0.05.

Results
Environmental status Table 2 shows the environmental status of respondents in the three countries, and the logistic regression results suggest several important differences, as shown in Meanwhile, most respondents living in these three countries also did not use public transport in the last week (ranges: 94.4%-97.9%), but the use of public transport for the US respondents and the UK respondents respectively decreased by 64% (OR = 0. 36   Note: *P < 0.05, **P < 0.01, ***P < 0.001

Responsive behaviors
The responsive behaviors of participants from the US, the UK, and Brazil are shown in Table 4,and there are some differences among the three countries, as shown in   Note: Boldface indicates statistical signi cance (*P < 0.05, **P < 0.01, ***P < 0.001)

Attitudes
The average response degree among participants in three countries on the view that they were worried about COVID-19 is shown in Fig. 1 behaviors, and risk perception is shown in Table 6. Gender, age, race, private health insurance status, income status, underlying medical conditions, and healthcare-related occupation information were used as the control variables. The overall performance of the three countries was that more close contacts, cohabitations, use of public transports, and the situation of going to school/work outside are positively correlated with risk perception. There was a positive correlation between participants' attitude and risk perception. Participants from the US and Brazil expressed that their cohabitants' emphasis on protective measures was negatively correlated with risk perception. In addition, the impact of personal protective measures on risk perception of UK participants was not signi cant, while the importance of washing hand and wearing masks of US participants was positively correlated with risk perception, the importance of social distancing was negatively correlated with risk perception, and the importance of using alcohol hand sanitizer was positively correlated with perceived risk. Note: Boldface indicates statistical signi cance (*P < 0.05, **P < 0.01, ***P < 0.001)

Responsive behaviors analyses under different sociodemographic factors
The binary logistic regression analysis results revealed the responsive behaviors towards COVID-19 under different sociodemographic factors (Table 7), and some differences of protective behaviors were also observed among subgroups in the three countries. Males were more likely to report disagreement with taking measures(social distancing, washing hands, using alcohol hand sanitizer, or wearing masks outside)to reduce their personal risk than females in all three countries. Due to the particularity of their jobs, healthcare workers from the US and the UK reported a signi cant reduction of disagreeing with using alcohol hand sanitizer and masks. However, the odds of healthcare Note: Boldface indicates statistical signi cance (*P < 0.05, **P < 0.01, ***P < 0.001)

Discussion
Main Findings To summarize, this study comprehensively analyzed the differences in environmental status, responsive behaviors, degree of worry about COVID-19, and risk perception of the public from the US, the UK, and Brazil during the COVID-19 pandemic. Compared to the US and UK, more Brazil respondents reported that they were in close contact with more than ten people, still using public transport, and engaging in work. Social distancing and washing hand were the most frequently used protective measures in the three countries, but respondents from the US and UK were relatively reluctant to using masks to reduce risk. Brazil participants had the highest average worried degree and self-assessed infection risk of COVID-19, followed by the US, and the UK was the lowest. Meanwhile, it was found that people with a higher degree of worry about COVID-19, more close contacts and cohabitants, using public transport, and going out to work in daily life tend to show a higher risk perception. And the active use of protective measures by cohabitants would reduce the participants' risk perception. On this basis, we further analyzed the speci c performance of participants with different sociodemographic characteristics in taking protective measures, and it was found that there were signi cant differences among participants of different gender, age, race, health insurance status, income status, and occupations. We believe that these insights are conducive to provide reference information for public health work during the COVID-19 epidemic and for the future.

Possible Explanations
Participants' self-assessed infection risk can comprehensively re ect participants' perceptions and attitudes towards the COVID-19 epidemic. Among the three countries, participants from the UK and the US generally represented a more optimistic anti-epidemic performance, but Brazil participants' performance was relatively nervous. Our ndings also indicate that participants' risk perception in the three countries is affected by living environment, responsive behaviors and worried degree towards COVID-19 to a certain extent. The worried degree and risk perception of participants from Brazil were much higher than those from the US and the UK, which might be closely related to Brazil's long-standing lack and uneven distribution of medical resources, the rising number of con rmed cases and deaths, and the inactive national prevention and control policies [12,13]. At the same time, due to relatively complicated political, economic, and geographic factors, Brazil's quarantine measures have not been implemented thoroughly.
Only 40% of Brazils are in formal employment, and there were a large number of informal workers who were still working outside during the COVID-19 epidemic [14]. The survey results indicated that a higher proportion of respondents from Brazil was in close contact with too many people, still using public transport and engaging in work. The excessive number of close contacts and the crowded living environment will increase the chances to be exposed to the SARS-Cov-2 virus, and also limit the quarantine options of family members [15,16], and a higher use proportion of public transport can also easily lead to COVID-19 infection [17], thus causing greater worried and panic among the public. The Brazil Government needs to take more economic interventions to ensure that people comply with quarantine measures, while also pay attention to reducing the infection risk during public transport.
The less situation of working outside and lower risk perception of UK participants might be related to the national lockdown measures promulgated by the British government [18,19]. The time nodes of this survey (from April 28 to July 8) were within the period of the nationwide lockdown in the UK. These measures have reduced the spread of the SARS-Cov-2 virus in the community to a certain extent. On June 19, the British government announced that the threat level of the COVID-19 epidemic was lowered from level 4 to level 3 [20], indicating that the COVID-19 in the UK was in a state of widespread transmission, but the transmission rate is no longer high or rising. The average risk perception of US participants was slightly lower than that of Brazil, but the development of the COVID-19 epidemic in the US was also severe. Although the United States Federal Government has also announced social distancing recommendations, the actual and speci c decisions related to restricted behaviors were left to the state governments. Researches have shown that due to various economic and political factors, there was much difference in the time interval between governors' response to quarantine, which would cause delays in epidemic prevention and control [21]. It was not until April 7, 2020, that 42 states in the United States and Washington, DC issued "Home Orders" one after another. But by mid-May, although the incidence of COVID-19 in many states continued to increase, most states had begun to relax lockdown measures to support economic recovery [22], which might result in 18.2% of participants from the US still reporting a close contact with more than 10 people in the last week and 26.2% of participants still going to work or school.
Respondents from the US and UK were even less optimistic about the use of masks. Many studies have con rmed that wearing masks is important for mitigating pandemics caused by viruses, including SARS, H1N1, and SARS-CoV-2 [23,24], but the mask usage rate in the US and the UK has been low, and the UK was even worse. The "International COVID-19 tracker" jointly launched by Imperial College and YouGov [25] also showed that as of early June, only about 66% of Americans and 21% of British expressed they would wear masks in public places. In Italy, where the epidemic was equally severe, reached 82% in early April. The low usage rate of masks is related to the government's failure to implement the mandatory mask policy. With the COVID-19 epidemic becoming ever more severe, the attitudes of the US and UK governments towards masks use have gradually changed. As of August 1, 2020, 33 states in the US and Washington, DC have announced statewide mask orders [26]; The British government rst forced people to wear masks on public transportation on June 15 and issued a "National Mandatory Mask Order" on July 22. These policies may help the subsequent increase in the use of masks.
Our research also found that gender, age, race, income status, health insurance status, and occupation were all related to the enforcement degree of preventive behaviors. Men, participants under 20 years, participants without private health insurance, and participants in a poor income showed varying degrees of inattention to take protective measures in these three countries, with the performance of US participants was more obvious. These ndings are consistent with other research on COVID-19 between March and April this year. Males compared with females [27,28], young people compared with elderly people [16,18,19], and low-income groups compared with high-income groups [29,30] are less likely to adopt preventive behaviors. Some studies considered that COVID-19 has a certain gender orientation, and males are more susceptible than females [31]. Our nding results also support to a certain extent that males may increase the infection risk due to insu cient protective measures. While participants without private health insurance and with relatively poor income conditions would be unable to take effective protective measures due to the living environment and economic conditions. The incidence of COVID-19 is inversely proportional to the economic situation. Therefore, governments should also strengthen the publicity of protective measures and equipment for indigent areas and people.
In terms of race subgroups, participants with non-white backgrounds were more likely to report taking behaviors to reduce infection risk, such as the Hispanic, Asian, black, and mixed-race participants from the US, the Asian, black, and mixed-race participants from the UK, and Brazil Hispanic participants were more willing to use masks outside than white participants. Ethnic minorities in the US, UK, and Brazil are generally more susceptible to more serious impacts and are accompanied by a higher risk of death [32][33][34][35], compared with white people. These situations may affect minority groups to take protective measures relatively actively to a certain extent. Rubin et al. conducted a survey of the swine u outbreak in the UK showing that people with non-white backgrounds are more likely to take protective measures to reduce infection than white [36]. This nding was also consistent with the recent research on COVID-19 [28][29][30]. In addition, participants in healthcare work were also less likely to disagree with using alcohol hand sanitizer and masks. A survey of non-pharmaceutical interventions against the American public indicated that ordinary people not sure about the effectiveness of masks use in slowing down the disease spread, which is related to the spread of publicity and the level of information transmission by public health authorities [37]. Due to the particularity of their work, healthcare workers are bound to understand the protective effects of alcohol and masks more precisely than non-healthcare staff, and have a stronger awareness of using alcohol hand sanitizer and masks. Our ndings are conducive to the public health authorities to carry out more targeted publicity work of COVID-19 protection measures and improve the de ciencies.

Limitation
This research has several limitations. First, the COVID-19 (Coronavirus) Survival Calculator project is an online survey for the global public, not speci c to certain countries and regions, which results in different sample sizes in each country. Then, the survey recruited participants based on the willingness of individual users to visit the website, which may bias the results toward those who can use the internet.
Secondly, the study population cannot be representative of the individual countries, especially there are relatively small sample sizes for some race subgroups in the study. Limited by the setting of the questionnaire, the included race types contain white, Hispanic, Asian, black, mixed-race, and other race.
Participants living in the US and the UK are mostly white (87.3% vs. 92.6%), while in Brazil, except for 56.2% of white participants, 15.5% of mixed-race and 22.4% of other race are involved. Moreover, the Hispanic and black participants from the UK are relatively few, accounting for only 0.3% and 0.7% of the total. Finally, information related to COVID-19 is constantly evolving. There is a certain lag in the results of our study, and the online investigation is likewise ongoing, which is worthy of follow-up research.

Conclusions
This study was the rst attempt to use large-scale data to compare and analyze the performance and differences of ordinary people in the US, UK, and Brazil in terms of environmental status, responsive behaviors, attitudes, and risk perception towards COVID-19. We focused on the US, the UK, and Brazil, which were representative countries in the development of the COVID-19 epidemic in North America, Europe, and Latin America. The online survey began in the 7th week after the World Health Organization announced that the COVID-19 epidemic constituted a pandemic. Countries around the world have issued various policies and measures to respond to the epidemic, and the development trend of the COVID-19 epidemic has also been constantly changing. Respondents from the UK and the US were generally more optimistic, believing that their chance of contracting COVID-19 was relatively low, but Brazil respondents expressed more tension about the COVID-19 epidemic. Our nding observed that respondents living in the US and the UK did not pay much attention to masks use. Among sociodemographic subgroups analysis in the US, UK, and Brazil, it is also found that males, participants under 20 years, participants without private health insurance, and participants in a poor income are relatively less concerned about taking protective measures. At present, the global epidemic prevention and control has developed into a normalized and persistent battle. Governments and people of all countries should continue to strictly implement epidemic prevention and control measures. We believe that all countries should strengthen targeted publicity and assistance of public health intervention measures for different groups, and take effective steps to increase the protection awareness of ordinary people, thereby slowing down the process of the pandemic to a certain extent.