Nationwide Survey in Greece about Knowledge, Risk Perceptions, and Preventive Behaviors for COVID-19 during the General Lockdown in April 2020

Background: The aim of this study was to investigate the knowledge, attitudes, and practices of the Greek general population toward coronavirus disease 2019 (COVID-19) during the lockdown period in April 2020, to examine factors associated with misperceptions and to determine behavioral patterns that may require interventions. Methods: A cross-sectional study of the general Greek population (N = 1858) was conducted. A geographically stratified cluster sampling was implemented. A questionnaire was composed consisting of 35 questions. Data collection took place from 15 April to 2 May 2020. A random-digit dialing survey was conducted by 29 interviewers. Results: The majority of respondents (62.7%) answered ≥12/17 questions correctly. Participants aged 18–44 years, male gender, specific occupations (freelancer, unemployed, housewife, retiree) and those who sought information about COVID-19 from less than two sources received lower aggregated scores on knowledge questions. Regarding attitudes toward future vaccination, 18.9% declared that were against it, while 81.1% that they may consider or will be vaccinated. About 40% were not using a face mask and only 42% washed their hands appropriately. Conclusion: Adjusting information campaigns targeting especially people below 45 years of age can help to sensitize them and realise their role to control the spread. Further targeted surveys are needed to adjust/design prevention campaigns.


Introduction
In the current phase of the coronavirus disease 2019 (COVID-19) pandemic where a vaccine is not available, the success of non-pharmaceutical prevention strategies depends to a great extent on people's behavior and their adherence to health advice. The European Centre for Disease Prevention and Control's (ECDC) rapid risk assessment (24 September 2020) highlights that the observed increase in transmission levels in the European Union/European Economic Area countries and the United Kingdom indicate that "the non-pharmaceutical interventions in place have not achieved the intended effect either because adherence to the measures is not optimal or because the measures are not sufficient to reduce or control exposure" [1]. Population surveys can provide insights into people's perception of risk, their practices, views on restrictions, misperceptions, information needs, and can further support the implementation of evidence-informed policies [2].
As of 15 November 2020, 220 countries areas or territories have reported to the World Health Organization (WHO) 53,507,282 confirmed cases and 1,305,164 confirmed deaths, while a total of 10,560,273 cases and 265,184 deaths have been reported in the European Union and European Economic Area and the United Kingdom [3,4]. On 23 March 2020 when a nationwide restriction of citizens' movements was enforced, Greece had reported a total of 695 confirmed COVID-19 cases and 17 deaths [5]. The general lockdown was gradually lifted starting on 4 May, when a total of 2632 confirmed cases and 146 deaths had been reported. Before and during this lockdown period, prevention and control measures for COVID-19 were supported by health communication strategies through media campaigns, emergency alerts, live broadcasted daily briefings, a COVID-19 call line, and other methods [6]. Between May and July 2020, the number of new cases reported daily remained low (less than 20), while in August, when Greece resumed its tourism activities, the daily number rose to between 100 and 300 cases [7]. In September 2020, in addition to the information campaigns for the general public, prevention strategies targeted people returning from holidays, as well as educational institution activities. The ECDC's rapid risk assessment report (24 September 2020) categorized Greece among the countries with stable trends; however, due to the strong increasing trend in intensive care unit admissions, the country may have the potential for a large resurgence [1]. The aim of this study was to investigate the Knowledge, Attitudes, and Practices (KAP) of the Greek population toward COVID-19 during the lockdown period in April 2020 to examine factors associated with misperceptions and to determine behavioral patterns that may require interventions.

Materials and Methods
A nationwide cross-sectional study was conducted. The representative target sample size that was needed in order to achieve the study objectives and sufficient statistical power was calculated with a sample size calculator RAOSOFT [8]. The sample size calculator arrived at 1537 participants, using a margin of error of ±2%, a confidence level of 95%, an 80% response rate, and 8,693,742 people (adult population of Greece). A geographically stratified sampling plan based on regional units, which are categorized as level 3 in accordance with the Nomenclature of Territorial Units for Statistics (NUTS) was applied to produce a representative sample. The sample size of each regional unit was calculated according to the population distribution of regional units in Greece. Moreover, the sample was also stratified based on gender and three proportionally equal to adult population age groups ("18-39" (34%), "40-59" (34%), "60+" (32%)). Data about population, age, and gender distributions were according to the 2011 census [9]. A questionnaire was composed considering the WHO Regional Office for Europe survey tool and guidance for rapid, simple, flexible behavioral insights on COVID-19 [2]. The questionnaire consisted of 35 questions about (1) demographic characteristics; (2) knowledge (COVID-19 transmission and symptoms); (3) perceptions (risks, effectiveness of preventive and control measures); (4) practices (prevention and control of infection); and (5) self-rating health and financial status. Most of the questions were closed, asking the respondent to evaluate by giving the answer in a quantitative value four or five-level item (Strongly Agree/Agree/Disagree/Strongly Disagree, Definitely Yes/Yes/Maybe/No/Definitely Not, Very Good/Good/Average/Bad/Very Bad, Much Better/Better/The Same-No Change/Worse/Much Worse). Pilot testing of the draft questionnaire was conducted by dialing 20 randomly selected telephone numbers from the national telephone directory, interviewing respondents and completing the questionnaire. Considering the pilot-testing results, the final version of the questionnaire was composed. The list of questions can be found in Table 1. Data were entered into the database that was developed using the lime survey software [10] of the university-secured server. Anonymity, privacy, and confidentiality were maintained during data collection, entry, analysis, and storage. The study was approved by the Steering Committee of the Postgraduate Program of Applied Public Health and Environmental Hygiene of the Medical Faculty, University of Thessaly (Assembly of April 2020; Project Identification Code 11/2019-2020). Data collection took place from 15 April to 2 May 2020.
Qualitative variables were presented as frequencies with percentages and 95% Confidence Intervals (CI), and quantitative variables were presented with mean and standard deviation. For a univariate analysis, the chi-square test was applied to associate demographic characteristics and other factors with KAP responses, as well as scores of responses to knowledge questions calculating the Relative Risks (RR), with corresponding 95% CI. Multivariable logistic regression models were used to identify independent risk factors for the KAP and scores of responses to knowledge questions to calculate the Odds Ratios (OR) and the corresponding 95% CI. Factors with a p-value less than 0.20 in univariate analysis were included in multivariable analysis. A result with a p-value < 0.05 was considered to be statistically significant. All statistical analyses were conducted taking into account the clusters of the study through the complex sample module of SPSS 19.0 (IBM SPSS Inc., Armonk, NY, USA).
Participants were asked to rate their own income as "Low/Intermediate/High" without considering a specific numerical threshold. KAP were compared among the population of the geographical region of Attica (the major urban area that represents about 35% of the total Greek population) and the other regions of Greece. Three age categories were used in the analysis: 18-44, 45-60, and 61-92. Each correct response to a KAP question was scored with one point. Correct responses in the four-level quantitative items options for answers provided were considered both values Agree/ Strongly agree or Disagree/Strongly disagree depending on the question. Three aggregated scores were calculated: (a) questions 1-12, 14-15 and 22-24 with a maximum score of "17", (b) questions 1-12 with a maximum score of "12" and (c) questions 14-15 and 22-24 with a maximum score of "5" (questions are listed in Table 1).
A random-digit dialling survey was conducted by 29 interviewers who were trained in communication and data collection methods. To validate the effectiveness of interviewers' training, results were analyzed per interviewer according to the questionnaire response rate and the missing values per question. No significant associations were found. All candidate respondents were informed regarding the study's research objectives, the absence of any commercial purposes of the survey, and how their privacy and the confidentiality of data would be ensured. After receiving respondents' verbal consent to participate in the survey, the interview began by posing the 35 questions. At the end of the survey, respondents were asked about any potential questions they might have in order for interviewers to provide appropriate answers or clarifications.

Results
Interviewers dialled 27,241 random digits and 12,396 of them corresponded to a telephone number. A total of 5852 did not answer the phone, and 774 were business phone numbers. A total of 1858 (32.2%) individuals responded to the telephone survey, while 3912 refused to participate (most of them due to lack of time and approximately 1.5% of them were COVID-19 deniers).
The demographic characteristics of participants are presented in Table 2. The mean age of participants was 49.2 years (standard deviation: 17.4, minimum: 18, maximum: 92). The majority of respondents (98.3%) held Greek nationality. Fifty-eight of the 1822 respondents (3.2%) had people in their immediate social environment who were infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), while 11 respondents (0.6%) were confirmed cases. Approximately 39.6% of the 1813 respondents declared that they were not using a face mask to protect themselves from SARS-CoV-2, while 31.9% of respondents sometimes wore a face mask and 28.5% always used a face mask when outside of their home and before entering indoor areas. Moreover, approximately 31% of the 1757 respondents believed that wearing a face mask at the supermarket was not protecting them against SARS-CoV-2.
Out of 1785 respondents, 1316 (73.7%) consider information provided about COVID-19 to be sufficient, while 404 out of the 1785 (22.6%) characterized the information they receive to be excessive, and 3.4% rated the information as insufficient.

Factors Associated with Misperceptions
Multivariable logistic regression models were used to test for an association between the aggregate number of wrong answers (at least one incorrect answer versus no incorrect answers and three or more wrong answer versus less than three wrong answers) and the participants' characteristics. Moreover, multivariable logistic regression models were used to test for an association between individual question responses with participants' characteristics, including age ≥45, male gender, residence outside Attica, occupation in the public sector, level of education up to secondary school or Master's/PhD degree, middle or high income, and the use of ≥2 information sources. Table 3 presents results of the association of participants' characteristics with misperceptions. Table 4 presents the participants' characteristics that correlated with correct answers to the individual questions about knowledge, attitudes, and practices.

Discussion
Our study demonstrated that despite the fact that the majority of participants had a sound knowledge of COVID-19 transmission modes and prevention measures, good practices related to these topics were not reported by participants at the same level. In particular, about 96% of respondents acknowledged that SARS-CoV-2 is transmitted through respiratory droplets, but only 28.5% responded that they always used a face mask when visiting indoor spaces outside of their home (during the survey period, the use of a face mask was mandatory on mass transport and in taxis, medical facilities, supermarkets, and pharmacies) [6]. Moreover, 98.5% of participants recognized that hand washing could help prevent the transmission of COVID-19, but 58% reported washing their hands for less than 20 seconds and 35% did not carry an antiseptic with them when leaving the house. A great majority of respondents (96.8%) recognized that COVID-19 is transmitted when touching contaminated surfaces and then touching their eyes; however, 43% of respondents did not wash their hands before touching their eyes. Other studies in North America, China (>90%), and Taiwan showed that the majority of respondents were knowledgeable about COVID-19 [11][12][13]. Our findings demonstrate that additional surveys are needed to investigate the reasons why sound knowledge does not always translate into or ensure correct practice. Furthermore, the results of such surveys can be used to adjust the current or design new health information campaigns for COVID-19. Campaigns are important to be based on community engagement; it is important to establish multi-sectoral teams at central, peripheral, and local levels that are able to identify the needs of target groups and address any misinformation and disinformation timely [14]. Messages that include real stories in Greece and trusted messengers for each of the target audiences could also play an important role in the effort to change attitudes.
As shown in Table 3, participants aged 18-44 years, male gender, specific occupations (freelancer, unemployed, housewife, retiree), and those who sought information about COVID-19 from less than two sources received lower aggregated scores on knowledge questions. The ECDC's recent rapid risk assessment report (24 September 2020) highlighted that in several countries, the increasing reported number of COVID-19 cases correlates with high transmission among persons aged 15-49 years as well as with increased testing rates [1]. These epidemiological data and our study findings related to incorrect knowledge among persons 18-44 years of age demonstrate an urgent need for adjusting information campaigns targeting especially people below 45 years of age, in order to sensitize them to realize the role they can play in the spread of the epidemic and the importance of their contribution to control the spread [1]. Serial cross-sectional KAP studies are needed for the general population, as well as focused surveys for groups where minimal knowledge or incorrect practices have been identified [2]. Future studies could measure changes in the KAP of the population before and after governmental interventions.
Respondents in our survey rated their mental health as worse (41.6%) to a greater extent than they rated their physical health (12.6%), when comparing periods before and during the lockdown measures. Similar findings were also identified in other studies conducted in China and Spain [15,16]. Loneliness experienced throughout lockdown measures and anxiety about financial issues can affect mental health during the restrictive measures [1]. In our study, approximately 68% of respondents expected their financial status to worsen after the pandemic. Protection measures for mental health should also be part of the COVID-19 pandemic prevention strategies.
This was an observational study with voluntary participation in the general population with a relatively low response rate and relatively high mean for age of participants (about 49 years), and therefore, generalization of the results for the Greek population cannot be safely assumed. Selection bias and information bias might have occurred. Additional serial cross-sectional KAP studies with bigger sample are needed in order to be more representative for the Greek population. Moreover, it was not possible to test for ethnic or disadvantaged population groups [2].

Conclusions
In view of scenarios of sustained COVID-19 community transmission in several European countries in the coming months, understanding the perceptions of people and especially those below 45 years of age, their concerns and beliefs, as well as their knowledge and practices related to COVID-19 is essential to target communication strategies so they can be engaged and actively participate in the battle against the COVID-19 pandemic. Funding: The study was supported by the Laboratory of Hygiene and Epidemiology of the University of Thessaly.

Conflicts of Interest:
The authors declare no conflict of interest.