key: cord-0739952-ocyy9ow3 authors: Mansilla Domínguez, José Miguel; Font Jiménez, Isabel; Belzunegui Eraso, Angel; Peña Otero, David; Díaz Pérez, David; Recio Vivas, Ana María title: Risk Perception of COVID−19 Community Transmission among the Spanish Population date: 2020-12-02 journal: Int J Environ Res Public Health DOI: 10.3390/ijerph17238967 sha: 354a728bb6e479c55a339a6b4fa286e1713921e1 doc_id: 739952 cord_uid: ocyy9ow3 On 11 March 2020 the SARS-CoV−2 virus was officially declared a pandemic and measures were set up in various countries to avoid its spread among the population. This paper aims to analyse the perception of risk of COVID−19 infection in the Spanish population. A cross-sectional, descriptive observational study was conducted with a total of 16,372 Spanish participants. An online survey was used to gather data for 5 consecutive days over the compulsory lockdown period which was established after the state of emergency was declared. There is an association between socio-demographic variables and risk perception, and a very strong relationship between this perception and contact and direct experience with the virus in a family, social or professional setting. We also found that compared to working from home, working outside the home increased the perception of risk of infection and the perception of worsening health. Understanding the public perception of the risk of COVID−19 infection is fundamental for establishing effective prevention measures. According to the timeline established by the World Health Organisation (WHO), on 5 January the first report was published of an outbreak of pneumonia in China caused by a new virus. On 13 January, cases began to appear in areas outside China. On 30 January the new virus, dubbed SARS-CoV−2, was declared a public health emergency affecting 18 countries, and on 11 March it was officially declared a pandemic. This novel coronavirus causes an illness called COVID−19, characterised as leading to serious pneumonia, although it is asymptomatic in 80% of cases [1] . Other typical symptoms include fever, dry cough, fatigue and aches [2, 3] , as well as gastrointestinal symptoms in 40% of cases, and other neurological and dermatological symptoms [4] . The patients considered to be the most vulnerable to A cross-sectional, descriptive observational study was conducted with the participation of a total of 16,372 residents of Spain, 171 of whom were excluded for not meeting the criteria for inclusion (either they did not accept the informed consent statement, or they were under 18). In the end, the sample was 16,201 people (98.9% of the participants): 51.5% women and 48.5% men after weighting. In order to get a large number of responses, we used non-probability consecutive sampling (snowball sampling). The initial bias of the sample was corrected by post-stratification, adjusting gross estimates according to gender and age, as proposed by Wang, Rothschild, Goel and Gelman [30] . Thus, in an appropriate statistical adjustment, non-random samples can generate accurate results, and refers to the importance of this type of survey for studying public opinion in different areas [30] . The adjustment method for the sample weighting was the iterative procedure with qualitative auxiliary information [31] . This procedure is used when there are qualitative auxiliary variables (gender and age groups) and maximum information is available on the univariate distributions of these auxiliary variables in the population and in the sample. Stratification is carried out by crossing these auxiliary variables, specifically from a double-entry contingency table. In the first iteration, the elevators corresponding to the gender variable are obtained and in the second iteration those corresponding to the age groups. Each iteration consists of two steps (as there are only two adjustment variables): (1) the distribution is adjusted with respect to the marginal population distribution of the first variable and (2) with this new weighted sampling distribution, the distribution is adjusted to the second characteristic. Finally, the assignment of elevators (Table 1) has enabled corrections to be introduced in other variables considered independent, such as educational level and marital status. The method also partially compensates for non-coverage [31] . An online questionnaire was used to gather data for 5 consecutive days over the compulsory lockdown period which was established after the state of emergency was declared in Spain. The questionnaire was distributed via social networks, scientific associations, and various healthcare-related institutions. A specific questionnaire was prepared for this project using the tool "Google Forms". The questionnaire comprised 59 items divided into 4 content areas (socio-demographic aspects, characteristics of the experience, health aspects, and risk perception) and included the project objectives, the requirements for participation and instructions for filling it in. The dependent variable: to be studied was the perception of the risk of infection, a nominal variable with three categories (low, medium and high). The baseline variables which were considered as possible predictors of Perception of the risk of infection are as follows: The Affect index is an ordinal variable (ranging in value from 0 = No affect to 1 = Maximum affect) constructed by standardising a set of diverse situations: a family member testing positive; a friend testing positive; somebody at work testing positive; death of a family member due to Covid −19) . Income was originally measured with an ordinal variable with 11 response categories, with extremes <€12,000 and >€300,000 euros. These categories were grouped into income quartiles: Q1: ≤€12,000; Q2: €12,001 to €22,500; Q3: €22,501 to €40,000; Q4: €40,001 to >€300,000. The level of studies was classified into three levels: Primary studies (finished and unfinished); secondary studies (compulsory and post-compulsory) and higher studies (university). The dichotomous variable work inside/outside the home was considered a possible predictor of risk perception since, remember, the survey was carried out in a period of confinement. The coexistence variables (Who they live with) were also formulated dichotomously with response possibilities: Living alone and Living with others; if you live with a family member whose occupation is in healthcare (A family member works in healthcare, with Yes/No answers); Protective measures used (Protective measures, with Inadequate/Adequate categories); if Any family member has been affected (A member of the household is infected, with Yes/No answers). Interviewees' perception about whether their health had improved or worsened during confinement (Health status during confinement) was also considered a predictor variable, initially measured with a Lickert scale and later grouped into three categories (Good/Very good; Average; Poor/Very poor). Next, we conducted a preliminary analysis of the joint distribution of Perception of the risk of infection and the selection of potentially predictive variables ( Table 2 ). The frequencies were calculated for each bivariate relationship and the corresponding relative frequencies were expressed as percentages. The hypothesis of independence between the variables was resolved with a Pearson chi-square test and its significance for a two-tailed distribution scenario. We also calculated the size of the effect of each relationship with Cohen's d, as proposed by Rosenthal and DiMatteo [32] , Lenhard and Lenhard [33] and Ferguson [34] . This analysis was used to create the final selection of variables which became part of the multinomial logistic regression analysis. All analyses were carried out with a confidence level of 95% and, therefore, a margin of error p = 0.05. The multinomial logistic regression analysis was proposed in order to find predictors of the categories of Perception of the risk of infection, taking as the reference category Risk perception: Low. This type of analysis is especially suitable for modelling categorical and polytomous dependent variables, and is a multivariate extension of the binary logistic regression, admitting both continuous variables (covariates) and categorical variables (factors) as predictive variables [35] . The variables used in the model were gender, age, Affect index, place of work (working from home/outside the home), income (quartiles), health during lockdown (good/average/poor), protection measures (adequate/inadequate), whether a family member works in healthcare (yes/no), whether a member of the household is infected (yes/no) and where they get information on the disease (press/radio/TV; social networks/WhatsApp; official media/scientific documents). We decided the model would not include some of the variables studied in previous analyses, such as education level, marital status, working/non-working population, and who they live with (living alone/with others); despite presenting statistical significance in the chi-square test, the coefficients of the size of the effect measured with Cohen's d showed very small or negligible effects. The parameters corresponding to the predictive variables were tested against the Wald test and its p-values, and the odds ratio (OR) of the coefficients and their confidence intervals were calculated. We also measured the goodness of the overall fit of the model with the chi-square test and the likelihood ratio test, and the percentage of correct classifications for each category. The goodness of fit was measured using the Nagelkerke pseudo R-squared parameter. All the analyses were performed with the statistics program IBM SPSS v.21 (New York, NY, USA). The study was carried out in accordance with the principles of the Declaration of Helsinki and the laws and regulations in force in Europe and Spain and has been approved by the Research Ethics Committee University Europea (CIPI/20/138) and Research Ethics Committee with Medical Products of Cantabria (code: 2020.159). Given the exceptional circumstances of the pandemic, and in line with the indications of the European Medicines Agency and the Ministry of Health, Consumer Affairs and Social Welfare, written informed consent was requested at the start of the online survey. It was imperative that the user accepted it in order to continue with the survey. First, the sample distribution is presented taking the variables considered in the analysis processes into account. Table 3 shows the variables which were considered to be possible predictors of Perception of the risk of infection. This first analysis could indicate an association between a higher perceived risk of infection and having a family member, friend or workmate test positive; working outside the home; perceiving one's own health as poor during lockdown; someone in the household becoming infected; using adequate protective measures; and getting most of one's information from official media and scientific documents. Next, a prediction analysis of the categories of the dependent variable was performed through a multinomial logistic regression. In the analysis, a series of variables with statistical significance in the bivariate tests were selected, taking them as independent variables, and risk perception as the dependent variable. The multinomial model shows different OR values for the predictor variables, which offers an indication of the importance of these variables in determining risk perception. The results of the model show an uneven predictive value: 82.2% for the category "Perception of the risk of infection: Low", 15% for the category "Perception of the risk of infection: Medium", and 49.0% for "Perception of the risk of infection: High". The total percentage of correct prediction was 52.4% (Table 4) , with a Kappa coefficient value of 0.223 (p < 0.001). The results show a significant model (−2log likelihood = 0.821; χ 2 = 2193.43; p < 0.001), with a moderate and acceptable goodness of fit Nagelkerke R 2 = 0.204. Tables 5 and 6 show the regression coefficients of the predictive variables for the Medium and High categories, respectively. The comparison of the OR values (Exp(B)) of the two tables shows a higher predictive value in the independent variables considered in a high perception of risk. In fact, it is 4.439 times more probable that perception of the risk of infection will be high among those whose health has been poor or very poor during lockdown, compared to those who experienced their health as good or very good. It is also 4.251 times more probable that people with family members working in healthcare have a higher perception of the risk of infection than those without. Other relevant OR values correspond to people who work outside the home, compared to those working from home (OR = 3.057), those with someone infected at home compared to those without (OR = 2.167), and to a lesser degree, being a woman rather than a man (OR = 1.531), and getting most of their information from official media or scientific documents, compared to the conventional media, press, radio and TV (OR = 1.258). It is also important to mention the affect index variable, which presents an OR = 2.770, and which reflects that higher values of affect coincide with a higher perception of risk of infection. However, as the scanty association between age and perception of the risk of infection shown in Table 2 seems to indicate, the model does not produce a significant regression coefficient for age. All of these tendencies can be seen in the predictions for the category Perception of the risk of infection: Medium, but with lower intensity in all the variables. To sum up, the multinomial model highlights better prediction in the category "Perception of the risk of infection: High", compared to the middle category, which is more poorly defined, probably due to the tendency for those surveyed to place some items in middling positions when they have to do with assessments and/or perceptions of situations experienced during the COVID−19 lockdown. This study presents the risk perception of the Spanish public during the lockdown period coinciding with the maximum peak of infection of the pandemic. It shows an association of risk perception with socio-demographic variables and level of affect. Affect is understood as a variable of proximity to the disease, through direct contact or in the workplace, household or social circle. A preliminary analysis shows that people who put inadequate prevention measures in place scored higher in low risk perception. Many studies refer to the important role of risk perception when establishing behaviour which will protect health in a pandemic [36] [37] [38] [39] Specifically, the review by Bish and Michie [40] shows the association between perceived personal and family susceptibility to the development of a disease and the presence of preventive behaviour (hand washing, household hygiene, wearing masks, etc.) and avoidance behaviour (avoiding going to public places, restaurants, shops, etc.). Some of these papers place particular emphasis on the need to present concrete interventions and directives to the general public, or focused on specific groups, to help give them an outlook which strikes a balance between the risk they face and the risk they perceive [37] . For this reason, knowing what factors are associated with a greater perception of risk is fundamental when taking effective measures in a pandemic. The data indicate that women perceive more risk, in line with the study by Dryhurst et al. [22] which concludes that men perceive lower levels of risk than women. Women appear to be more susceptible to suffering from virus-borne diseases, and to present a greater incidence and severity of the disease [41] . Additionally, other studies conclude that more women are infected with SARS-CoV−2 via the main causes of infection: being a healthcare worker, visiting a medical centre, close contact with a person with an acute respiratory infection, direct contact with probable or confirmed cases of COVID−19 [42] , or being an informal caregiver [43] . All of this seems to justify women having a greater perception of risk. Bearing in mind that the elderly population are at greater risk of death [6, 44] , it is notable that this group, to a lesser degree, also has a high risk perception. These results are in line with the findings of Barber and Kim [25] , in which they explain how elder adults report less concern and present less anxiety over health than younger adults; and they contradict other studies which determine that the elderly are one of the groups with a higher perception of threats [36, 45] . A higher level of affect, understood as direct contact, correlates to an increased risk perception. Direct contact with SARS-CoV−2 relates to a greater perception of risk; more specifically, having a family member working in healthcare, or having lived with somebody infected. Other studies support the results obtained, concluding that people who have had a direct personal experience with the virus, and people who have received more information about it, present a higher risk perception [22] . Meanwhile, it can be observed that workers who work outside the home have a high-risk perception. This is attested by studies such as Ruiz-Frutos et al. [46] , which affirm that people working outside the home present higher levels of loss of sleep, unhappiness and depression. The authors associate this fact with the existence of a greater perception of risk of infection for the subjects or their families, compared to people working from home. Many studies discuss the fact that healthcare workers' heavy exposure to SARS-CoV−2 means they have an increased perception of the risk of infection [47] [48] [49] , with nurses as a group showing a higher perception than doctors [50] . We observe that people who perceive their own health as poor or very poor are the ones with the highest perception of risk [51, 52] . However, they also present fear of going to hospital, demonstrated in a decrease in hospital admissions for such significant pathologies as myocardial infarction [53] and stroke during the lockdown period [54] . Additionally, multiple studies show that the presence of comorbidities increases the severity and risk of death of COVID−19 [55] [56] [57] , a fact which may justify a sense of greater risk among the more vulnerable. The main limitation of this paper is that the sample was not randomised, limiting the general applicability of the results. Even so, the large size of the sample means we can regard it as fairly representative. The circumstances of the pandemic prevented the use of more accurate sampling methods, but nevertheless the results obtained are highly significant and could be the basis for future papers and work with randomised samples. The fact that the sample is not comparable has likely led to bias in the response, highlighting the fact that more women than men and more people with university studies responded to it. This has partially been corrected using post-stratification methods, although the response bias is not completely eliminated. There continues to be a slight over-representation of people with higher education. This fact must be taken into account when interpreting the results. This work has attempted to measure risk perception that each of the respondents considered they had and, therefore, the following question was formulated: What do you consider is your risk of infection? Where 1 is the minimum and 5 the maximum. It is a subjective measurement and, therefore, it is understood that it could vary depending on the knowledge and individual characteristics of each one of them. Therefore, even with no objective measurement of risk, this work focused on individuals' subjective perception in an extreme situation of confinement. This variable reflected in the questionnaire thus enables the necessary data to carry out this study on the subjectivity of the impact of the disease to be obtained. To carry out this study, our own survey was used, designed ad hoc and, therefore, with no specific sensitivity and specificity analysis. This survey does not contain any validated tool. Despite the design of the study does not allow to extrapolate the results to all the Spanish population, we consider that the population sample used makes that the data can be used as a guide to perform effective prevention strategies and developing new research projects. Finally, it should also be noted that despite the statistical significance shown by some crossovers of variables, effect sizes <0.30 indicate that caution must be exercised when determining clear associations, so this represents an added limitation to the time to build strong relationships. In our analyses, this fact has been taken into account and clearly specified when low effect sizes were obtained. Understanding the perception of the risk of COVID−19 infection is fundamental for establishing prevention strategies among the public. An association can be observed between socio-demographic variables and risk perception, and a very strong relationship between this perception and contact and direct experience with the virus in a family, social or professional setting. The logistic model also finds that working outside the home increases the perception of risk of infection, compared to people working from home. Additionally, the perception of worsening personal health is strongly associated with greater perception of the risk of infection, as is living with a family member who works in healthcare. After carrying out this work, it is considered appropriate to develop new research projects aimed at knowing in detail what the behavioural and emotional responses of the population are, as well as their evolution during an epidemic or pandemic period. Specifically, one aspect that can be addressed in depth in future work is the fear of being stigmatised or discriminated against for having a positive test for COVID−19. A multidisciplinary perspective combines biological, psychological, and sociological aspects. Therefore, a multidisciplinary approach which also includes the psychological aspect, should be considered by the worldwide health policy makers [58] . It is also important to note that the groups that present a greater risk and fear perception are those that need the most support [59] and, therefore, require more complex investigations that place the focus of their objectives on it. Dryhurst et al. identify the important role of risk perception in motivating health protection behaviours, especially during the COVID−19 pandemic [22] . They find that risk perception is positively and significantly correlated with an index of preventive health behaviours such as washing hands, wearing a mask and physical distancing and also point out that minimised as well as exaggerated risk perceptions can potentially undermine the adoption of behaviours to protect the [22] . Thus, higher risk perceptions lead to more protective behaviours, although taking effective measures can also reduce risk perception. Therefore, we highlight and suggest the importance of evaluating the accuracy of public risk perceptions and the correlation with taking positive protective behaviour measures. Among others, we should consider the vaccine acceptancy, being this one of the top 10 threats to global health in 2019 [60] . The providers of the different health systems in the different states must guarantee easy access to qualified information and support measures. Specifically, a transparent information policy that balances identifying risks, promoting appropriate behaviour and avoiding sensational reporting styles. Policy-makers often conceptualise risk as the probability of contracting a disease multiplied by the magnitude of the consequences. However, our findings, which present evidence of how people perceived the risk of COVID−19 infection in Spain in the first wave, clearly illustrate that risk perceptions are consistently correlated with a number of experiential and sociocultural factors. More specifically, it is shown that risk perception among the population is higher in those with direct personal experience of the virus. Therefore, health risk communication messages tend to be more effective when they include information on the effectiveness of measures aimed at protecting people from illness personally as well as socially. Cori et al. indicate that knowledge implies the growth of collective awareness, increased self-efficacy and empowerment to contribute to political decision-making [61] . Mutual trust and dependence on local communication networks among peers could exponentially increase the possibilities to apply flexible measures, introducing concepts that promote health and well-being linked to healing, collaboration and solidarity. Therefore, the implementation of community interventions is recommended mainly in population groups in which risk perception is poorer. Although this study is only observational and could be expanded with experimental studies, what does seem clear is that a better understanding of the knowledge that people have, as well as the social and cultural factors that accompany them, help to understand the risk perception of COVID−19 contagion in the world and its role in motivating preventive health behaviours. This information could help policymakers to design evidence-based risk communication strategies. Therefore, future research is advised to consider expanding this research. Funding: This research received no external funding. Unique epidemiological and clinical features of the emerging 2019 novel coronavirus pneumonia (COVID-19) implicate special control measures Clinical features of patients infected with 2019 novel coronavirus in Preventing 30-Day Hospital Readmissions COVID-19 pandemic-A focused review for clinicians Influenza-associated pneumonia as reference to assess seriousness of coronavirus disease (COVID-19) COVID-19-Toward a comprehensive understanding of the disease Coronavirus disease 2019: What we know? The transmission and diagnosis of 2019 novel coronavirus infection disease (COVID-19): A Chinese perspective Viable SARS-CoV-2 in the air of a hospital room with COVID-19 patients Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1 Chasing the ghost of infection past: Identifying thresholds of change during the COVID-19 infection in Spain Estimation of COVID-19 prevalence in Italy, Spain, and France Evaluation of the lockdowns for the SARS-CoV-2 epidemic in Italy and Spain after one month follow up The effect of large-scale anti-contagion policies on the COVID-19 pandemic COVID-19 in Italy: Impact of containment measures and prevalence estimates of infection in the general population Real Decreto 463/2020, de 14 de marzo, por el que se declara el estado de alarma para la gestión de la situación de crisis sanitaria ocasionada por el COVID-19. Diario La Ley 2020 Understanding coronavirus disease (COVID-19) risk perceptions among the public to enhance risk communication efforts: A practical approach for outbreaks The COVID-19 risk perception: A survey on socioeconomics and media attention Crisis Communication and Public Perception of COVID-19 Risk in the Era of Social Media Risk perceptions and their relation to risk behavior Risk perception of coronavirus disease 2019 (COVID-19) and its related factors among college students in China during quarantine Risk perceptions of COVID-19 around the world Effect of Income Level and Perception of Susceptibility and Severity of COVID-19 on Stay-at-Home Preventive Behavior in a Group of Older Adults in Mexico City Perceptions and behavioural responses of the general public during the COVID-19 pandemic: A cross-sectional survey of UK Adults COVID-19 Worries and Behavior Changes in Older and Younger Men and Women People at Risk of Influenza Pandemics: The Evolution of Perception and Behavior The Tripartite Model of Risk Perception (TRIRISK): Distinguishing Deliberative, Affective, and Experiential Components of Perceived Risk Developing a Broadly Applicable Measure of Risk Perception Changes in risk perception and self-reported protective behaviour during the first week of the COVID-19 pandemic in the United States Forecasting elections with non-representative polls Ajuste de Muestras con Información Auxiliar Meta-Analysis: Recent Developments in Quantitative Methods for Literature Reviews Computation of Effect Sizes An effect size primer: A guide for clinicians and researchers Applied Logistic Regression Pandemic influenza in Australia: Using telephone surveys to measure perceptions of threat and willingness to comply The coronavirus (COVID-19) fatality risk perception of US adult residents in March and Evaluation of Risk Perception of Covid-19 Disease: A Community-based Participatory Study Social Distancing and Stigma: Association Between Compliance with Behavioral Recommendations, Risk Perception, and Stigmatizing Attitudes During the COVID-19 Outbreak Demographic and attitudinal determinants of protective behaviours during a pandemic: A review Impact of sex and gender on COVID-19 outcomes in Europe COVID-19: The gendered impacts of the outbreak Cuidados y abordaje de la pandemia de COVID-19 con enfoque de género Variables asociadas con mortalidad en una población de pacientes mayores de 80 años y con algún grado de dependencia funcional, hospitalizados por COVID-19 en un Servicio de Geriatría Analysis of Risk Perceptions and Related Factors Concerning COVID-19 Epidemic in Chongqing Health-related factors of psychological distress during the COVID-19 pandemic among non-health workers in Spain Assessment of Knowledge, Attitudes, and Perception of Health Care Workers Regarding COVID-19, A Cross-Sectional Study from Egypt The psychological impact of COVID-19 and other viral epidemics on frontline healthcare workers and ways to address it: A rapid systematic review Risk perception of COVID-19 among Portuguese healthcare professionals and the general population Disease Perception and Coping with Emotional Distress During COVID-19 Pandemic: A Survey Among Medical Staff Attitudes toward COVID-19 and stress levels in Hungary: Effects of age, perceived health status, and gender Potential Indirect Effects of the COVID-19 Pandemic on Use of Emergency Departments for Acute Life-Threatening Conditions-United States Decrease in ST-segment elevation myocardial infarction admissions in Catalonia during the COVID-19 pandemic Has COVID-19 played an unexpected "stroke" on the chain of survival? Comorbidity and its impact on 1590 patients with COVID-19 in China: A nationwide analysis Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the Does comorbidity increase the risk of patients with COVID-19: Evidence from meta-analysis Stigma and Discrimination (SAD) at the Time of the SARS-CoV-2 Pandemic Emotional and Behavioral Consequences of the COVID-19 Pandemic: The Role of Health Anxiety, Intolerance of Uncertainty, and Distress (In)Tolerance com_content&view=article&id=14916:ten-threats-to-global-health-in-2019&Itemid=135&lang=es Risk Perception and COVID-19 The authors declare no conflict of interest.