A population survey on behavioural insights on COVID-19 in Sri Lanka: A quantitative analysis on perceptions, behaviours, opinions, trust and affect to inform pandemic outbreak response


 Background The COVID-19 pandemic outbreak is placing insurmountable challenges and burden on health systems and authorities demanding responses with effective and appropriate policies and interventions. Wrong decision making, lack of foresightedness and preparation in the acute and transition phases can threaten the gains collectively achieved. Understanding the behaviours, perceptions and feelings of the population is imperative in management decisions. Not much evidence is available on such behavioural insights to inform the COVID-19 response in countries. This study is expected to contribute to the evidence-informed understanding on effective management of the outbreak.Methods This was a cross sectional observational study with entirely voluntary online participation. The study protocol and the questionnaire, originally prepared by a WHO collaboration, were adjusted to match the local setting and resource availability. The study explored 14 variables including perceptions, trust and behaviours in relation to COVID-19. Data was collected during the first wave of COVID-19 in early May 2020 and statistically analysed to see the associations and comparisons between chosen variables.Results N = 1001 with 65.63% from the age range 18–29 years and 65% from the non -health care worker category. Basic or postgraduate degree holders were 64.7%. Participants were from 23 out of the 25 districts in SL with 34.07% from high-risk districts. They were in general, knowledgeable (e.g., > 95% correctly identified common symptoms of COVID19) and majority (99.5%) reported conforming to the health recommendations. They trusted government health bodies including hospitals (90.8%) but majority (63.8%) had poor confidence in media in the ability to manage the outbreak. Stigma around COVID-19 was not a concern for the majority of 78.9%. Participants less than 30 years believed in conspiracies around COVID-19 more than older counterparts.Conclusion The perceptions, affect and behaviour of the study population gives insights on information dissemination, gaps in health message delivery, alternative approaches in responding and policy making helping to improve the pandemic response.

Our sample consisted of a majority who believed either they or their immediate environment are free from the infection of the COVID-19 virus (80.63% and 76.53% respectively) In perception of self-infected status the number of participants in categories, infected and con rmed + suspected to be infected but not con rmed, were negligible. (Fig. 2) C. Knowledge Out of the participants 83.3% perceived their overall knowledge on prevention as to be good and very good. (Fig. 3). The perception of overall knowledge on prevention was not signi cantly different according to age, district, education level or the occupation. More than 95% of participants had correctly identi ed individual symptoms, fever, cough, shortness of breath and sore throat as related to COVID-19 infection (Fig. 4). HCW had statistically signi cant higher knowledge about some of the symptoms not so commonly discussed (body aches, headache, fatigue, diarrhoea and inability to sense smell and taste), compared to non-HCW (Table 2). A majority as high as 92.90% correctly knew that there is no drug or vaccine available against the COVID-19 infection during the study period. That knowledge was not signi cantly associated with age, living district, education and occupation of the participants. Both 14 and 21 days were marked by our population as the incubation period (Fig. 5). The knowledge about incubation period was signi cantly higher in participants who were below 30 years compared to participants who were older than 30 years of age.

D. Probability and severity perceptions
Out of the participants 50.9% believed that they are likely to get infected with COVID-19( Fig. 6). Those above 30 years of age and HCW had perceived their probability of getting infected with COVID-19 as signi cantly likely, compared to those who are below 30 and non HCW ( Table 3). The single highest category of participants belonged to the undecided category in relation to the severity perception. Yet the study sample swayed towards perceiving COVID-19 as severe (Fig. 7). There was a signi cant association between the perceived severity of the infection and presence of a chronic illness. (Table 4) Bold values = (p < 0.005), Unlikely = Extremely unlikely + Unlikely + Somewhat unlikely, Likely = Somewhat likely + Likely + Extremely likely. Bold values = (p < 0.005), Severe = Less severe + severe + very severe, Not severe = Very mild disease + Not severe + Not severe at all.

E. Preparedness and perceived self-e cacy
Out of the participants 82.22% were having good con dence in the ability to protect self from COVID-19 ( Fig. 8). A high percentage of 99.5% claiming to have good compliance with the recommendations from authorities ( Fig. 9). There was no demonstrable statistically signi cant relationship between the compliance with recommendations and individual demographics. (Table 5) The majority (more than ≥ 90%) has followed the most recommended preventive measures indicated in bold text (Table 6) Six preventive measures commonly discussed were cross tabulated against the basic demographics. The use of herbal supplements was signi cantly higher in participants below the age of 30 and in non HCW. The disinfection of mobile phones was signi cantly higher in participants above 30 years of age, educated above Advanced Level and in HCW category (Table 7). We analysed the traditional and alternative preventive methods in relation to age and the living area. Drinking ginger tea was signi cantly higher in the less than 30 age group compared to the above 30-year group, and in the participants from rural areas compared to those in urban areas (Table 8). In our population 63.1% and 81.1% had good coping in not seeing family and friends respectively, during the pandemic (Fig. 10).  The recommendations in bold = the most frequently disseminated ones  Though the outbreak was worrying in nature to the majority (67.1%), many (61.2%), did not feel helpless either (Fig. 11). The feeling states were explored against the selected demographic variables for any associations and COVID-19 virus felt close to HCW more signi cantly than to the non HCW. None of the other associations were found to be statistically signi cant (Table 9). Health care workers were the best (76.1%) trusted source of information of our population with websites coming second (70 .7 %) ( Table 10). The associations between four information sources and the basic demographics were explored. Participants below the age 30 had a signi cantly higher trust towards private television stations than participants who were above 30 and we found no other signi cant associations (Table 11). More than 90% of the participants were keen to know speci c virus-related health concerns with only about a quarter of the population needing to know about grooming needs (Table 12). Majority (85.6%) of participants referred to the sources of information, highly frequently (Fig. 12).   The type of information more than ≥ 90% of the participants wanted is indicated in bold text.
H. Trust in the institutions /personals in their ability to manage COVID-19 Sri Lankan government sector health and other authorities have gained good trust of our study population. The religious places of worship have gained a lesser trust comparatively. A notable 63.8% had stated they had poor con dence in media (Table 13). Good trust = some con dence + high con dence + very high con dence, Poor trust = No con dence + very low con dence + l ow con dence The participants below the age of 30 were found to be placing more trust in certain institutions/ personnel (e.g., private hospitals, GMOA, SL Army) compared to the above 30 participants. While the trust in government hospitals was signi cantly higher among HCW compared to non HCW, the trust in private hospitals and ayurvedic medical council was signi cantly higher in non HCW than that of HCW. The participants who have got only school education have signi cantly more trust in mosques compared to those who have higher education levels (Table 14). Bold values = (p < 0.005), Poor trust = No con dence + low con dence + Very low con dence, Good trust = Some con dence + High con dence + Very high con dence.

I. Opinion on policies and measures
The high number of participants have agreed on the need for more screening tests (90.9%) and 3 weeks of quarantining (87.5%) and nearly 60 % of the participants did not believe the restrictions were greatly exaggerated. Majority (78.0%) were willing to take a vaccine against COVID-19 (Table 15). The below 30-year-old respondents have agreed with more coercive practices signi cantly than above 30 participants. The HCW and the participants with above school education have disagreed more signi cantly on allowing the government to restrict the access to the internet and social media (Table 16).  When it comes to opinions on conspiracies and secrecy 48.6 % was undecided on, whether the virus is a biological weapon by China (Table 17). The participants below 30 years of age believed on conspiracies to a signi cant level than the above 30-year counterparts (Table 18).  The majority (82.32%) agreed that the decisions taken by the Sri Lankan government to control COVID-19 were fair (Fig. 13). There was no signi cant association between opinion on fairness of the decisions and demographic variables (Table 19). Just above half of the population had a hard time making it through stressful events due to policies on COVID-19 (Fig. 14). The stressful experiences were not signi cantly associated with age, living district, education level or occupation (Table 20) and 56.4% of participants were ready to seek help for them (Fig. 15).  Our population was cautious and were supporting health recommendations even upon reopening (Table 21). Participants from high-risk districts were of the opinion that the restrictions should be upheld in urban but not in rural areas. Participants below 30 years as opposed to above 30 years people, signi cantly believed that the prevailed restrictions should not be lifted beyond May 2020, the country's reopening month of the rst wave. The non HCW agreed on obligatory mass testing to identify infected people in a statistically signi cant level than the HCW (Table 22).

J. Stigma
Majority (78.9%) of our study population did not consider contracting COVID-19 virus as stigmatizing (Fig. 16). There was no signi cant association between the stigma of COVID-19 and the selected variables (Table 23). Out of the total population, only 823 (82.22%) have indicated their basis for stigma (Fig. 17). Statistical signi cance = p < 0.005

K. Behaviours
The behaviours our population practiced most were the ones in relation to avoiding unnecessary contact with others and the majority (> 70%) have claimed that they were not interested or engaged in panic buying behaviours. The least common (3.69%) behaviour was "drinking alcohol more than usual" (Fig. 18).
The people below 30 years avoided family and friends and also blocked contact of the children with their friends to a signi cant level when compared with above 30 years participants (Table 24). Bold values = (p < 0.005), Already done or planned = I already did that + I plan to do that, Not done or not planned = I don't plan.

L. Worries and fears
The highest number of participants (85.1%) were worried about the wellbeing of their loved ones during the pandemic. Not being able to participate in social events was the least represented worry (41.8%) ( Table 25). Do not worry = Don't worry at all + Don't worry + don't worry much, Worry = Worry a lot + Worry + Somewhat worry Participants below 30 years were signi cantly worried about the government not disclosing some information to the public, the closure of universities and higher educational centres and not being able to pay their bills when compared to their older counterparts. The above 30 age group was signi cantly worried about limited freedom of movement and missing out on vacation opportunities. The worry about closure of universities and higher education centres was signi cantly higher in participants who lived in low-risk districts compared to high-risk districts. The non HCW had signi cantly worried about becoming unemployed and not being able to pay their bills than the HCW. The thought of having to defend a decision not to participate in a social event with family or friends has caused signi cant worry in participants who are less than 30 years, than in the above 30 participants (Table 26). Bold values = (p < 0.005), Do not worry = Don't worry at all + Don't worry + don't worry much, Worry = Worry a lot + Worry + Somewhat worry.

M. Confusions and rumours
We considered 4 widely discussed areas of probable confusion in the time of the study and found the clarity in each area was more than the confusion (Fig. 19). There was no signi cant association between the confusions and basic demographics (Table 27). N. Probability and severity perception on cough cold and fever Vs probability and severity perception on COVID-19 The 57.34% of the study population thought it is likely for them to get infected with an episode of cough, cold and fever while 50.9 % who perceived it is likely to get infected with COVID-19. (Fig. 20) A fraction of 35.6% of the participants had perceived an episode of cough, cold and fever as severe whereas 40.3 % thought of COVID-19 as severe (Fig. 21).
The belief of the probability of getting infected with cough, cold and fever is statistically signi cant in HCW and those with chronic illnesses ( Table 28). The perceived severity of getting infected with the same is also signi cantly higher in participants with chronic illnesses than participants who are not (Table 29).
When compared with the probability and severity perceptions of cough cold and fever episode Vs the probability and severity perceptions of COVID-19 infection (Table 30) it was found our population considered COVID-19 to be more severe than common cold. However, they perceived that they are more susceptible to common cold compared to the COVID-19. Both of these differences are statistically signi cant.

Discussion
We aimed at a sample size of at least 1000 as it is a recommended number for surveys of large size populations to make the sample as representative as possible of the country's population [12]. The precision of estimates of surveys only increases very slightly beyond a sample size of 1000 and costs of inviting more than 1000 participants may exceed the statistical bene ts [13].
The study was conducted 8-10 weeks, into the pandemic during the rst wave of the COVID-19 in SL. By the time SL has been ranked 9th best country in the world for its successful immediate response on managing the virus [14]. With the medical and academic background of the authors it was expected more HCW, particularly the medical undergraduates and doctors to take the questionnaire and such a bias was one of the major concerns' authors had initially.
However, the sample had more non HCW with the non HCW: HCW ratio being 1.86:1. Colombo, Gampaha, Kurunegala and Kandy were represented more and it could be because they are highly populated districts [15] and also the research team has a higher contact base in them.
The majority of the participants perceived them (80.63%) and their immediate environment (76.53%) were free of the risk of the infection. That may be a re ection of the safer outbreak status of the country at the time or blissful unawareness or psychological denial in at least a few. It could also be that the study group possibly came from a higher socio-economic background with less congestion and more people compliant with preventive measures which lead to feel safe. However, it is noteworthy that 19.08% and 16.28% of participants were not sure whether they themselves were infected or anybody else in their immediate environment was infected, respectively. This response can be a result of the knowledge about the higher possibility of asymptomatic/less symptomatic cases and they probably lived with some uncertainty and related anxiety. Such anxiety can be alleviated if they are informed that kind-of disease patterns, which is the presentation of the majority are less harmful.
A percentage as high as 83.3% was having good con dence in their overall knowledge on prevention irrespective of their differences in socio-demographic characteristics. This positive response could be due to a certain degree of social desirability bias too. A high majority have shown a good knowledge about the common symptoms of the infection (Fig. 4) and this may re ect the success of the health messages disseminated in the country and or the knowledgeability of the study population; They could have educated themselves from online sources. It was expected to nd HCW and the more educated to have fared better on knowledge on preventive measures but there were no demographic boundaries. HCW, however, identi ed individual symptoms of COVID-19 which were given less publicity on media better than non-HCW probably because they have more exposure to actual patients as only designated health institutions were accepting COVID patients. They also possibly learnt on the job and / or by consulting scienti c literature on details of the COVID-19 as a necessity of the job. It was notable that a small percentage seem to believe there is a drug and /or a vaccine available against the COVID-19 infection. There were some media reports on the effectiveness of chloroquine against the virus and how some local as well as world leaders promoted the drug during the period of data collection. The hydroxychloroquine arm of the "Solidarity trial" was removed by health authorities since there was no signi cant difference of mortality and morbidity compared to the standard care on 17th of June [16]. The confusion must have sprung from such intermittent reports on various treatment trials.
There was some debate and unclarity regarding the exact incubation period with some research having shown it as 21 days. For example, the incubation period was found to be longer in older adults [17]. Although there were slight differences in evidence, the accepted incubation period by WHO was up to 14 days at the time of the study.
About half of our study population was thinking there is some self-susceptibility to the COVID-19 infection. It is understandable that the relatively more mature participants and HCW found the probability of them getting infected is signi cantly higher. There was no signi cant association between the probability of self susceptibility perception and the presence of a chronic illness. Probability of infection is found to have association with chronic kidney disease and Diabetes mellitus only [20] (Table 3).
There was an alarming media hype where Chinese patients seemingly dropped dead in the streets at the initial stages of the COVID-19. The media was blamed for fuelling panic and anxiety about the disease [21]. Yet it is interesting to see close to half of our population is inconclusive about the severity (Fig. 7). The fact that there is a signi cant association between the severity perception and having a chronic disease indicates, that particular health message was delivered to them quite successfully (Table 4) [22]. Despite substantial variability across cultures, individualistic worldviews, personal experience, prosocial values, and social ampli cation through friends and family are signi cant determinants of risk perception in a multi country study and risk perception correlated signi cantly with reported adoption of preventative health behaviours [23].
Our population, in general, was highly con dent in the ability to protect themselves and highly compliant with health recommendations. The recommendations were followed irrespective of the differences in their basic demographics. The scienti cally unproven methods such as sprinkling of turmeric water and use of "perumkayan" (Asafoetida) which is practiced by some in-Sri Lankan culture were not popular among the study group (Table 5). The younger (< 30 years) participants used more herbal supplements compared to the above 30 group as a preventive measure. It is possible their elders offered traditionally popular herbal supplements such as coriander to youngers during the lockdown periods which could have in uenced the response of younger participants or this may be a result of the developing nationalist ideas mainly among the youth, globally [24]. Non HCW also used herbal supplements signi cantly more than their HCW counterparts. The HCW participants are probably employed in the western medicine elds rather than the non-western health care elds such as Ayurveda. The participants who are younger, more educated and belonging to the HCW category disinfected the mobile phones more compared to older, less educated and non HCW respectively. Mobile phone use may be higher in the rst group with resultant higher risk perception of phones being contaminated and the behaviour of disinfecting them.
We inquired about coping with, not seeing family and friends. Sri Lanka being a collectivistic society, individual members often perceive themselves to be members of their ethnic, religious or linguistic group rather than individual and autonomous beings [25]. Claimed it can be done by understanding the need and respecting the recommendation to limit traveling. The authors have rst-hand experience where the HCW not visiting their elderly parents and children but living separately to protect them from the infection.
The majority of the participants came up with understandable responses on exploring the effect of COVID-19 on the individuals' affect. However, it is interesting to see some, after all the wide spread knowledge to the contrary, believed the virus is spreading slowly. This shows the importance of reiterating the health messages in health education. The majority did not feel helpless and that may be due to the fact that SL has managed to contain the spread of the virus in the country successfully during the time of the study. One participant, a specialist colleague in public health, has criticized the response "close to me" in the questionnaire via personal communication (Fig. 11). The response came from the original mother study in Germany. We acknowledge that the use of the phrase in relation to a "deadly" virus is di cult to fathom in our culture. Sri Lankans use the phrase "close to me" to denote something dear to heart, not to an entity such as COVID-19 virus. Therefore, we wonder how the participants saw and processed the question. HCW category however, have a signi cant association with the feeling of the "virus being close to them", it could be they took the question in a more western context.
Only a 22.3% of the population had a good trust towards social media (e.g., Facebook, Twitter, YouTube, WhatsApp, Viber) despite the widespread use of them in the country with Facebook topping the list with 80.16% of Sri Lankans using the platform [26]. Maybe people use social media for pleasure and socio -political reasons and not as a trustworthy source of information. The newspapers and radio stations were not popular, either. This may be a re ection of an important shift in information seeking patterns in SL, with wider spread internet access, increasing electronic platforms disseminating information. The participants less than 30 years had a signi cantly higher trust in the private television channels compared to older group. It will be fascinating to explore in a qualitative approach whether they are ignorant, gullible or having some other driving force in order to have trusted private channels more compared to the older participants. It is common knowledge that the main private television channels are run by the owners from the two main political camps in the country and it is possible the trust on them was in uenced by the political views of the participants.
The information stated as most needed were those concerned with the disease itself; It's severity, treatment and prevention at country's level as well as personal level. This concern in our population may re ect how the Sri Lankan public was made to attune with the pandemic. It is a little surprising to see the participants did not have the need to know about the children's education as one of the highest needs. Education plays a major part in Sri Lankan culture and most of the Sri Lankan parents live to educate their children to their best. Relatively lower concern could be due to several reasons. The data collection was done fairly early into the outbreak and the worries probably have not started mounting and there were ongoing discussions to reopen the country too, which gave a sense of hope [27]. The teachers, in both the private and government sector, attempted to keep the education process continuing using their personal data and mobile apps in some settings and it must have helped the parents in our study not to worry too much. The population, nding the COVID-19 virus new and dangerous may have concerns over the safety of the children than their education. It is also quite noteworthy our population did not bother much about the personal grooming needs compared to some, for example some US citizens who rebelled against stay-at-home requests highlighting their grooming needs. Some communities apparently used demands on haircuts as a symbol of protest to win their "freedom". (" operation haircut" at Michigan, USA) [28].
Many Sri Lankans resorted to having haircuts at home by family members and considered it as the norm and the need of the era according to the expressions in social media and that compliance and understanding were represented by our study population too.
We did not explore the behaviour of checking individual data sources, but the overall checking. Majority looked for COVID-19 information more often than not.
In the face of stress and anxiety, individuals are often prone to use strategies that are designed to help but proven counterproductive. For example, checking social media frequently for COVID-19 related information can induce stress [29]. While COVID-19 fact checking is a necessity, ensuring there is some checking on checking and engaging in behaviours which help destress are important [30].
We modi ed the questionnaire to match the local COVID-19 response when assessing the trust in institutions/ personnel in their ability to manage the outbreak. Participants having more or less equal overall trust placed on HCW plus professional associations and Police plus military in the battle against COVID-19 in SL, could imply many aspects. Despite many criticisms of authorities for "militarizing "the COVID-19 operation, utilizing the armed forces to control spread by tracking contacts, building quarantine and other health care centres and running quarantine centres paid well to control the rst wave of the country. Health sector alone does not have the infrastructure, man power, skills or the means to manage the COVID-19 demands placed on the country. Some operations were technically supervised by content specialist consultants (e.g., epidemiologists) in armed forces. The carefully and strategically planned operations of rescuing the Sri Lankan students in Wuhan, China and Sri Lankan chef from a cruise ship are some extraordinary examples that gained the love as well as trust of the public in the authorities [31], [32].
The a uent Sri Lankan general public will rely on the private sector for ease, comfort and better care yet maybe it is worth exploring why the majority would not trust them in a condition where life and death infection happens.
It is noteworthy that most knew about the two professional colleges, the SL college of community physicians [33]and SL college of psychiatrists [34] ( Table 13) with only a small minority such as 12.8 % and 15.4% not knowing the two colleges respectively. We believe the active Facebook pages maintained by the both professional bodies and its members sharing college newsfeeds in their personal social media contributed to the raised awareness of the colleges [35]- [37]. Though the author characteristics may also have contributed to the nding, the knowledge of the two professional bodies among our participants who comprised a non HCW majority shows the potential of practical and convenient platforms such as Facebook for professional bodies to reach out to lay public in health education.
The majority of our population has trusted the newly elected President in his ability to manage the pandemic effectively [38]. Despite the accusations levelled at him by the local and international critiques about employing the military for the outbreak control, the rst wave of COVID-19 was successfully managed in the country. The government was also trusted by participants probably because of the trust they had in the president and at the time of the study the parliament was dissolved and the president was managing the situation with his trusted o cials, both civil and army. A notable percentage of our population have placed their trust on the ayurvedic council, though not to the extent they trusted the allopathic medical bodies. This insight is important and demands us to utilize the rich ayurvedic heritage and traditional ayurvedic practices in treatment following proper research either singly or in collaboration with western medicine.
SL has four main religions and Theravada Buddhism is the state religion with 70.2% of people being Buddhists. Many religious activities were continuing in mass gatherings and later regulations were brought in prohibiting them [39]. Our study population had comparatively less trust in religious institutes in the ability in managing the outbreak. Those who were educated only up to school level trusted the mosque in its ability to manage COVID-19 more than those who were educated above school level. Although there is a statistically signi cant association found between lower education levels and believing the mosque as a reliable source of information, strati ed analysis was impossible (since we did not collect the race and the religion) to control the confounding effects of the religion and the race.
Our population believes the government university and school system can manage the infection better than the private education institutes. This may be re ecting the trust in the free education providing institutions in the country as opposed to the private educational institutes. The government institutes are under strict regulations and rules and the authorities are responsible and accountable which may be giving the public some sense of security. The private institution probably has the reputation for aiming at more pro ts gaining than maintaining the quality. Particularly the medical education in SL lies almost solely in the purview of the government university system and the government sector doctors are trusted and respected in general, in any health emergency. Sri Lanka has proved it possessed considerable but unheralded expertise in managing deadly diseases, having eliminated malaria and polio, grappled successfully with AIDS, SARS, H1N1, Chikungunya and MERS and was substantially containing Dengue. In 2005, the health system avoided the muchanticipated epidemics following the Indian Ocean tsunami. The participants indicated their unwavering trust in the government healthcare system and its products.
Close to 2 /3 of people in the study have no con dence in the media in their ability to deal with the virus. This information presents an important take home message for the media authorities in the country. Our population has only 9.5 % who are school educated, all others having some post -AL education of varying degree. If the media authorities want to uplift the media conduct to a believable level by the educated public is a question, we researchers pose. We also believe they can contribute to increase the taste and expectations of the not so educated and / or questioning fractions in the community too. The media is a very important stakeholder in disasters to give correct health messages and information without making the public panic. Poor trust in mainstream media can in uence people to seek information from uncon rmed sources and then get misled. Lower trust in the media is alarming for both media authorities and health professionals involved in health promotion.
The younger participants have placed trust in some institutions and personnel more than their older counterparts. The professional colleges, medical associations and unarguably the most powerful trade union of the country, GMOA were trusted more signi cantly by them [40]. It is raising questions whether individuals are more cautious before trusting when they grow older. The difference in occupation health care Vs non health care also exposed some interesting ndings with HCW trusting the government allopathic health system while the Non HCW placed signi cant trust in private and ayurvedic hospitals. The HCW are probably from western medical disciplines and are biased towards the mainstream allopathic government medicine. Ayurveda may have much to offer in standalone or supplementary therapy yet the eld needs backing by quality research in order to be recognized by all fractions in a population.
We explored opinions on policies and measures related to covid-19 in SL. The whole country in general was compliant with whatever the health instructions disseminated by the authorities. The general belief of the majority was "One has to live rst, to safeguard human rights and respect traditions!". Whoever criticized and questioned the recommendations were scorned on social media and other media as such opposition was considered as an anti-government action, thus politicizing the issue. The concepts such as respecting autonomy and individual rights in relation to COVID-19 management were discussed among some fractions. We believe the highly positive and compliant approach of the study population on the policies and measures taken in the country re ects the country's majority's subservient opinion in general.
Sri Lanka ' s primary health care system is of high standards [41]. SL takes pride in prevention of communicable diseases by way of its very successful immunization program from birth [42]. The public is quite attuned to immunization programs run by the government health team which include the internationally renowned public health midwives who deliver the bulk of the service door to door [43]. Sri Lankans in general grow up listening to the importance of vaccination to prevent preventable infections and the anti-vaccination theme is almost not heard by an average Sri Lankan. Therefore, as opposed to some nations in the world SL is pro vaccination and the positive response of our study population is a re ection of that.
Avoiding certain people during COVID-19, based on their country of origin has many facets to it. SL has a reputation of being one of the best tourist friendly destinations in the world [44]. However, some fractions were panicking that the tourists would bring the virus to the country and closed the doors to foreigners [45]. The majority deciding it is appropriate to avoid foreigners probably thought prevention is better than cure. We believe that decision is based more on health risks rather than that of any discrimination. In self-preservation, people avoid the stimuli that cause danger and unpleasant sensation such as fear and pain. Thus, one naturally averts from dangerous situations and these responses are mostly instinctive with minimal deliberate control and overreacting is understandable [46].
We asked whether the government should restrict access to the internet and social media in order to stop misinformation and subsequent agitation of the public. It has been a practice to block Facebook for the same purpose following the easter attack in SL in April 2019 [47]. There is no proof that banning social media is a solution to dissemination of fake news. In fact, authorities can use social media to prevent adverse reactions as social media companies have information about user behaviour patterns that the governments do not have and the third-party researchers have additional information about how information moves across the ecosystem. Combating the spread of misinformation must be a joint effort between the social media companies, governments and individual users [48]. As a strategy, government institutions like health promotion bureau and department of government information entered to the world of social media and became popular among users with high level of post reach and engagement [49], [50] The participants below 30 years of age agreed signi cantly for coercive practices employed by the government compared to their older counterparts. The HCW and participants who had more than school education and not the younger participants, have opposed restricting access to internet and social media Thus the questions whether the younger fraction was ready to forgo individual rights /wishes and the HCW and more educated fraction relied on social media and internet more, in the face of the pandemic, arise. These are important insights for delivering health messages and policy making by the authorities in disasters like the COVID-19.
Some showed paranoia and indecisiveness on certain concerns emerged in the world about the pandemic. The majority did not believe China was deliberately placing the world in danger. China is a very close ally of SL to a point of concern in some fractions in SL. Participants below 30 years showed signi cant paranoia in relation to surveillance of citizens, hidden agendas and secret organizations behind government decisions. Why the below 30 group is more inclined to believe in conspiracies about secret activities and governments spying on them for their social media data are a question worth exploring. This can be due to a personality trait and we have a selection bias since the sample is less representative. A qualitative study would tell whether they are insecure, more knowledgeable or on a lighter note, watch too many American movies! There was no statistically signi cant association in relation to facing adversities against the age, living district, educational level and occupation; probably everybody had the same adverse experiences irrespective of their differences in demographics. It is noteworthy that the majority are willing to seek help in experiencing stress related to COVID-19 policies.
Non HCW were strongly of the opinion obligatory mass testing should be carried out upon re opening, compared to HCW. The different knowledge, attitudes and experience between two groups may be the reason for this difference in opinion. Either, even the non-health care fraction was in tune with scienti c preventive measures or the HCW perhaps knew more about PCR testing (cost, ability to generate false negative and positive results) and hence the neutral or low opinion of the effectiveness of mass testing.
Exploration of stigma in relation to COVID-19 was one major motive of the study. Stigma is common in relation to race, religion, culture, gender and health. Before COVID-19, there were other illnesses (e.g., Tuberculosis, HIV/AIDS) that carried social stigma, discrimination, and exclusion. [51], [52]. Stigmatizing language (e.g., "tuberculosis suspect") that has been criticized by advocates in the past, has been used during the current pandemic too, unfortunately. (e.g., "COVID-19 suspect"). Such judgmental terms in uence attitudes and behaviours not only preventing patients from seeking treatment but also in uencing the way authorities approach the disease. Research in China shows that the psychological impact of fear of COVID-19 is more dangerous than the disease itself [53]. Fear accentuates social stigma and discrimination. The frontline health care workers worldwide have been discriminated because of fears of transmitting the virus. The breakdown of social support structures such as religious institutes and family and friends, anxiety caused by curfews, many uncertainties around COVID-19 and fear of being infected and also being forced to quarantine has given rise to huge issue of stigma in relation to COVID-19 SL too. The media reports and authority statements gave the impression they are holding the COVID-19 infected people responsible for contracting the virus and the outlook was that of total blame. The involvement of police and military in COVID-19 management in SL pluralized criminalizing the infected. In the early phase of COVID-19 in SL, Sri Lankan media showed an alarming level of breaching privacy of the patients and contacts. Later it was identi ed as an issue and recti ed [54], [55]. Racial and religious minorities were discriminated against in media reporting and there was a huge issue of not allowing the burial of dead bodies of COVID-19 patients in SL which led to deep worries in Muslim community. Another fraction that was discriminated was patients with addictions to substances, they were double stigmatized when they were infected with COVID-19. Media should involve in combating stigma by joining hands with health authorities and more so by way of their conduct. The majority of our study population though reported not believing COVID-19 as stigmatizing, have indicated the basis of stigma. It is possible they were referring to others but not to themselves in the process. People with greater personal resources (income, education, social support) and good mental health have been shown to be less worried and less likely to stigmatize [56]. It is known that education, clear and honest communication and the use of non-discriminatory language have the power to signi cantly improve the knowledge, attitudes, and behaviours related to COVID-19 and reduce stigma [57]. The majority in our population did not think contracting COVID-19 is stigmatizing, probably they are a comparatively fortunate fraction that shows above protective qualities. Another possible reason could be that they did not feel the full impact of the disease as the pandemic was reasonably well contained in SL during the rst wave [58].
Study explored the COVID-19 related behaviours. Epidemics, natural disasters and extraordinary events trigger panic buying in general and during COVID-19 too people resorted to buying excessive amounts of toilet paper, hand sanitizers, dry rations etc. Panic purchases during a public crisis negatively impact social stability, economic orders, company management, and consumer psychology [59]. The need and obligation to protect and care for the family in uences the consumer's considerations, family being a social factor that in uences consumer decisions related to panic buying behaviour. Anxiety around the uncertainty, personal factors like self-con dence and self-e cacy also contribute to panic behaviours [60], [61]. Consistency in government policies on events such as re opening, giving correct and transparent information, rules restricting buying in bulks and organizing a system to have essentials provided without a break would reduce panic buying. The older method of door to door selling of products like sh, vegetables, milk etc. were revitalized in the country and the social media groups were formed and supported each other in local communities. The authorities promoted the mobile vendors by issuing curfew passes and recommended strict health guidelines to prevent the spread of the virus. Our population's majority (75.4%) that claimed they did not panic buy goods, probably have managed due to some of the personal and societal strengths such as economic stability and supportive environment. Drinking alcohol was marked as a least engaged behaviour in our sample. According to a national survey only 2.4% of females consume alcohol in SL [62] and 57.04% of our sample was females. SL, in general, looks down upon drinking alcoholic beverages and social desirability bias must have come in the marking of the response.
Worries and fears in relation to the pandemic were explored to nd more than the personal worries our population was worried about the country's capacity to handle the possible case load. This shows the population we studied were either more farsighted and knowledgeable about the depth of the pandemic or had basic needs of their personal life such as food, lodging and medicine etc. ful lled during the pandemic or both. Maslow postulated one has to have the basic needs met in order to concentrate on higher order issues [63]. Participants below 30 years of age were signi cantly worried about the government not disclosing some information to the public. This nding again throws the queries whether the younger population in our study is more mistrusting and questioning and why. The above 30 age group's signi cant worry about limited freedom of movement and missing out on vacations is probably because they had more responsibilities of employment and providing for the family and they have got used to a life style of taking break from employments respectively.
The younger participants were understandably more signi cantly worried about the closure of universities and higher educational centres. The low-risk district dwelling participants are more concerned about higher education, maybe participants from high-risk districts had more pressing health and basic survival related concerns comparatively. Less than 30, non HCW categories had signi cant worries about paying their bills. HCW most probably had stable government jobs compared to non HCW. Worries about bill payments may re ect the nancial hardships of younger participants and threat to job security in nonhealthcare participants. Worries could lead to mental health challenges at the time and in many years to come [64]. Our population worries least about maintaining social life in vivo. Yet the younger participants showed some signi cant worry about defending a decision not to take part in a social event. The younger the age more the need for social conformity [65]. It is important to consider these behavioural patterns in formulating regulations and punishing rule breakers in the pandemic response in the country.
Majority (above 61%) of our population have no confusion in the tested areas in the study. Health messages in SL were seemingly better grasped by them. Local COVID-19 authorities made sure the masks were mandated when the infection started creating an impact in SL [66]. We could not identify any statistically signi cant association between various confusions and selected socio-demographic factors.
We compared probability and severity perceptions on cough, cold and fever (the culturally relatable term for seasonal u, used in the mother study) and probability and severity perceptions on COVID-19. Our population perceived COVID-19 infection to be more severe than cough, cold and fever infection. However, probability wise they perceived that they are more susceptible to common cold than to COVID-19. Both of these differences are statistically signi cant. To our population "catching common cold is easier than catching COVID-19!" This perception was probably because SL did not have to face the full impact of COVID-19 as it was circumscribed to the few clusters and well controlled during the rst wave. The risk perception differs between different places, individuals, cultures and subcultures and it is potentially a strong modi er of the epidemic evolution, since it can in uence the community spread [67].
The population linked having chronic illnesses to a higher susceptibility to and severity of cough cold and fever (Table 28, Table 29). However, they associated chronic illness to higher severity of COVID-19 but not to higher probability of getting infected with COVID-19 (Table 3, Table 4). It is interesting to nd our study population thinking chronic illnesses do not increase the probability of getting COVID-19 but they do increase the probability of getting cough cold and fever.
They found HCW to be more susceptible to both conditions than non HCW (Table 3, Table 28). It is an understandable response. The symptom characteristics of the in uenza /common cold and COVID-19 are compared in a few studies but there is no work comparing severity and probability perceptions on the two conditions [68].

Conclusion
The study presents a girth of information on the perceptions, behaviours and concerns in relation to the COVID-19 pandemic of a considerably large community sample spread across the country. The majority study population is highly educated and young (less than 40 years), probably had internet access enabling them to seek information from many sources both national and international, did not have to rely on local media for information, and had important and very relevant perceptions and behaviours in relation to the pandemic over all, such as the country's economy and overloading of the health resources etc. for example. The participants below 30 years of age agreed signi cantly for coercive practices and believed in conspiracy theories more than the older counterparts. The total population in general was compliant with health recommendations and the majority did not worry about autonomy rights and individual freedom. The knowledge about various aspects of the COVID-19 infection was good in the majority. According to our study the effective risk communication, clear/ transparent policy decisions and responsible/ non stigmatizing and believable media conduct are major areas warranting improvement in COVID-19 management in SL. Thus, the ndings offer important messages for the authorities to shape their response to future pandemics as well as the current pandemic and future waves of the same pandemic. Researchers hope to provide local data for international comparisons informing WHO policies and interventions for current and future reactions for the pandemic/s.

Limitations Of The Study
The study using online platform limited the participation of certain important population groups; the elderly, a recognized risk group for COVID-19 and disadvantaged population groups such as ethnic minorities, nancially disadvantaged with no internet facilities, individuals who are not used to electronic devices/ internet, young people below 18 years, and other vulnerable groups such as medically ill etc. Phone interviews either alone or as a supplement were suggested by the mother study to be used to mitigate this limitation. However, the lock down, travel restrictions, nancial restraints (the study was unfunded) did not allow phone or face to face data collection. Therefore, representativeness of our population of the general population is reduced which in turn affects the generalizability of the study ndings. The social bene t of the study may consequently be reduced. The ndings of the survey need to be interpreted in this context. Conducting, more tailored and targeted surveys with those speci c population groups will be supplementary to the study.
The people who do not have the habit of using the internet for seeking or verifying factual information resort to mainstream media, mainly the television news.
While they can have a commendable role to help combat the pandemic [69] the television channels may represent politically biased viewpoints and reporting may not be independent at times [70]. The perceptions, attitudes and behaviours of the viewers could be greatly in uenced by those of the TV channels [71].
On one hand our population may not represent the general population because of the speci cs of the demography such as 64.7 % having quali cations of a basic degree or above, yet on the other hand that very same speci cs offer a tremendous opportunity for the authorities to shape COVID-19 response based on study results as the majority of our participants does not seem to base their thinking and behaviours on what is shown on TV!
In addition, the complexity of the pandemic and crisis and the public response is considerable, and an online survey can only serve to monitor a few key issues -not explore them in-depth. Another limitation of the study is that, while validated for other scales and well-grounded in robust behavioural research, the items in the study tool have not been validated through a rigorous process for COVID-19 speci cally. This is due only to the fact that we have never experienced this virus before. Self-reported behaviours are known to differ from actual behaviour, due to the social desirability effect. Finally, as each country needs to adapt the questionnaire, not all data collected with this tool can be compared across countries for future evaluation purposes. The hope is that each country will collect and analyse at least several variables in common that may provide useful insights for cross-country comparisons. Potential risks identi ed included only the inconvenience of the time taken to respond to the survey. However, given the prevailing restrictions of locomotion people faced, many individuals had more available time for a task like this. The cross-sectional design will not allow the assessment of actual causal relations and will only be snapshots of a current state of the public perceptions and behaviours. To explore the causal relations and also to see the pattern changes in behaviours and perceptions the survey is best conducted repeatedly. For such an intervention, nancial and manpower issues need to be resolved via national or international level collaboration.

List Of Abbreviations
WHO -World Health Organization Perception on probability of getting infected with COVID 19 Unlikely = Extremely unlikely + Unlikely + Somewhat unlikely, Likely = Somewhat likely + Likely + Extremely likely Figure 8 Con dence in the ability to protect self from COVID 19 Good con dence = I know + Very much so, Poor con dence =I know to some extent + I don't know at all+ I don't know much Compliance with recommendations from authorities Compliant = Follow to a some extent + I follow+ Very much so, Not compliant = Not at all + I do not follow much Figure 12 Frequency of seeking information High frequency = often + once a day + several times a day, Low frequency = sometimes + rarely +very rarely Figure 13 Opinion on overall COVID 19 decisions taken in Sri Lanka Disagree= Strongly disagree + Disagree + Somewhat disagree, Agree = Strongly agree + Agree + Somewhat agree.

Figure 14
Opinion on the di culty in managing stressful events due to COVID 19 policies Figure 15 Opinions on help seeking behaviours in relation to stressful experiences due to COVID 19 policies.

Figure 16
Opinion on contracting COVID 19 as stigmatizing  Confusions about the information on COVID 19 information Figure 20 Perception of the probability of contracting COVID 19 Vs. perception of the probability of contracting an episode of cough, cold and fever Unlikely = Extremely unlikely + Unlikely + Somewhat unlikely, Likely = Extremely likely + Likely + Somewhat likely.