key: cord-0747128-9fyixsds authors: Lang, Raynell; Atabati, Omid; Oxoby, Robert J.; Mourali, Mehdi; Shaffer, Blake; Sheikh, Hasan; Fullerton, Madison M.; Tang, Theresa; Leigh, Jeanna Parsons; Manns, Braden J.; Marshall, Deborah A.; Ivers, Noah M.; Ratzan, Scott C.; Hu, Jia; Benham, Jamie L. title: Characterization of non-adopters of COVID-19 non-pharmaceutical interventions through a national cross-sectional survey to assess attitudes and behaviours date: 2021-11-05 journal: Sci Rep DOI: 10.1038/s41598-021-01279-2 sha: 69e415f9b51f62dfae344192b4a46bdf0e0f554c doc_id: 747128 cord_uid: 9fyixsds Adoption of non-pharmaceutical interventions (NPIs) remains critical to curtail the spread of COVID-19. Using self-reported adherence to NPIs in Canada, assessed through a national cross-sectional survey of 4498 respondents, we aimed to identify and characterize non-adopters of NPIs, evaluating their attitudes and behaviours to understand barriers and facilitators of adoption. A cluster analysis was used to group adopters separately from non-adopters of NPIs. Associations with sociodemographic factors, attitudes towards COVID-19 and the public health response were assessed using logistic regression models comparing non-adopters to adopters. Of the 4498 respondents, 994 (22%) were clustered as non-adopters. Sociodemographic factors significantly associated with the non-adoption cluster were: (1) being male, (2) age 18–34 years, (3) Albertans, (4) lower education level and (5) higher conservative political leaning. Participants who expressed low concern for COVID-19 and distrust towards several institutions had greater odds of being non-adopters. This information characterizes individuals at greatest odds for non-adoption of NPIs to inform targeted marketing interventions. The online survey was distributed to 14,887 participants and 5893 respondents (40%) clicked on the survey link, with 1395 (24% of respondents) excluded due to not completing and submitting the survey. A response rate of 40% is typical for the Angus Reid Forum 31 as they distribute broadly to their panel to promote quick responses. The proportion of individuals who completed the survey were similar in age, sex, ethnicity, highest level of education, and province of residence to those who did not complete the survey. Overall, 4498 participants completed the survey (Table 1 ) and were included in the analysis. The majority of participants were female (2298, 51%), Caucasian (3866, 85%) and educated with either an undergraduate or graduate university degree (1314, 29%). There were 1998 (44%) respondents from Alberta. The age of participants ranged from 18 to 94 years old, with a mean of 47 years ( ± 16 years). For information on sociodemographic factors associated with each NPI please see Supplemental Table S1 . There were 721 (16%) participants who reported that over the last few weeks they maintained physical distancing sometimes/rarely or never, 631 (14%) who masked sometimes/rarely or never, 986 (22%) who avoided crowded spaces sometimes/rarely or never and 646 (14%) who stayed home while sick, sometimes/rarely or never (Fig. 1 ). Adopter cluster vs non-adopter cluster sociodemographic factors. Using a cluster analysis, two distinct non-overlapping clusters were identified: 3504 (78%) were clustered as adopters and 994 (22%) were non-adopters (Supplemental Table S2 ). Adoption of NPIs differed by age, sex, province of residence, race/ethnicity, annual income, education level, and political leaning (Table 1) . A multivariable model, adjusted for all sociodemographic variables identified that annual household income and ethnicity was not associated with differences between adoption clusters and therefore not used in the subsequent adjusted analyses. Compared to those aged 18-34 years, participants age 35-54 years (aOR 0.64, 95% CI 0.53-0.79) and ≥ 55 years (aOR 0.25, 95% CI 0.20-0.31) and those of female sex (aOR 0.55, 95% CI 0.47-0.65) had lower odds of being in the nonadopter cluster. Albertans had greater odds of being non-adopters compared with all other provinces; however, when adjusted, this association was no longer significant for Quebec participants. Education was negatively associated with non-adoption, with respondents having a university degree (aOR 0.49, 95% CI 0.37-0.63) having lower odds of being non-adopters than respondents who graduated high school or less. People of Middle Eastern/Central Asian/South Asian (OR 0.39, 95% CI 0.18-0.86) and Chinese/Filipino/Other Asian (OR 0.52, 95% CI 0.30-0.88) ethnicity had lower odds of non-adoption compared with Caucasians; however, this association attenuated following adjustment for other demographics. People reporting slightly liberal (aOR 0.42, 95% CI 0.28-0.62), liberal (OR 0.42, 95% CI 0.30-0.58) or very liberal (aOR 0.23, 95% CI 0.15-0.35) political leaning had lower odds of non-adoption compared to people reporting moderate/middle of the road political leaning. Conversely, compared with moderate/middle of the road, people reporting slightly conservative (aOR 2.02, 95% CI 1. 54-2.63) , conservative (aOR 4.12 95% CI 3. 26-5.21) or very conservative (aOR 7.00, 95% CI 5. 19-9.41 ) political views had greater odds of non-adoption (Table 1) . Table 1 . Sociodemographic factors associated with adoption clusters for COVID-19 non-pharmaceutical interventions. Odds ratios are the odds of being in the non-adopter cluster compared to odds of being in the adopter cluster. CI confidence interval, REF reference group, cOR crude odds ratio using logistic regression, aOR adjusted odds ratio, adjusted for sex, age, province of residence, annual household income, highest education, political leaning and ethnicity. + Prairie provinces included Saskatchewan and Manitoba; Atlantic provinces include Nova Scotia, New Brunswick, Prince Edward Island and Newfoundland and Labrador. 16 ) higher than respondents expressing concern about their friends or family becoming sick (Table 2 ). Participants who felt that they would have mild or no symptoms if they were to contract COVID-19 had increased odds (aOR 3.52, 95% CI 2.91-4.26) for non-adoption than people who felt that their symptoms would be manageable. Having known someone with COVID-19 or living with someone they considered high risk for severe outcome of COVID-19 were positively associated with adoption. Non-adopters were more likely to interact with more people outside their household, with 37% reporting having regularly interacted with more than 20 people over the past few weeks compared with 12% of adopters (Table 2) . Respondents who disagreed with the statement that "younger people are mostly to blame for the increase in cases" had greater odds of being non-adopters of NPIs (aOR 4.29, 95% CI 3.53-5.21); however, 36% of people who disagreed with this statement were of the age 18-34 years compared to 27% of people ≥ 55 years old. Therefore, the youngest age group were more likely to disagree that younger people were to blame for the increase in cases, yet had higher odds of being in the non-adopter cluster. Attitudes towards NPIs and reasons for non-adoption. Adoption of NPIs were positively correlated with adoption of contact tracing and exposure notification apps, with respondents who had downloaded an app having lower odds of being non-adopters (aOR 0.30, 95% CI 0.24-0.38) than those who had not downloaded an app ( Table 3 ). The odds of being in the non-adopter cluster was greater among people who disagreed with the statement that public health messaging had been clear and understandable (aOR 3.69, 95% CI 3.09-4.40) compared with those who agreed. There was also a positive association between non-adoption and disagreement that public health messaging had been consistent (aOR 3.69, 95% CI 3.06-4.45). The majority of respondents agreed that COVID-19 restrictions were harming the economy (3024, 77%), however people who disagreed with this comment had lower odds of non-adoption (aOR 0.43, 95% CI 0.32-0.59) compared with people who agreed. Individuals who responded that messaging or advertising made them less likely to physical distance (aOR 8.34, 95% CI 5.24-13.28), mask in indoor public spaces (aOR 9.49, 95% CI 6.44-13.99), avoid public spaces (aOR 4.34, 95% CI 2.96-6.36) and stay home when sick (OR 5.08, 95% CI 2.87-8.98) had significantly higher odds of non-adoption compared to people reporting no difference in their behaviors with messaging or advertising. Across all NPIs evaluated, people who reported not seeing any messaging or advertising had significantly higher odds of non-adoption (Table 3) . When evaluating messaging for physical distancing, 62 (1.4%) people reported they had not seen any messaging. While this is a low percentage, it represents a key target population that communication efforts had missed. Of these individuals, 58% were female, 48% from Alberta, 83% Caucasian race/ ethnicity and 84% were aged 18-55 years old. These demographic proportions were similar for people reporting not seeing messaging across each evaluated NPI. The greatest proportion of respondents in the non-adopter cluster reported that they did not believe that the recommendations work as the reason for having not followed the public health guidelines (333/994, 34%). Whereas the greatest proportion of adopters reported that they had intended to follow the guideline but simply forgot (78/3504, 2%) as the reason for having not followed public health guidelines (Fig. 2) . Other main reasons non-adopters reported not following NPIs included: they did not think the recommendations were important for their health (129/994, 13%) or the health of their friends and family (85/994, 9%), or the NPIs were too burdensome to follow (124/994, 12%). Fewer reported their behaviours were influenced by those around them not following recommendations (79/994, 8%) (Fig. 2) . www.nature.com/scientificreports/ Distrust across all institutions was associated with non-adoption of NPIs (Table 4 ). The most predictive trust factor was trust in government, with people who reported trusting government (1404, 31%) having lower odds (aOR 0.43, 95% CI 0.32-0.57) of being in the non-adopter cluster adjusted for sex, age, providence of residence, highest education and political leaning compared to people who reported neither trusting nor distrusting government. Whereas people who reported distrust in government (1685, 37%) had higher odds of being in the non-adopter cluster (aOR 3.71, 95% CI 3.05-4.52) compared to respondents who reported neither trusting nor distrusting government. People who had trust in healthcare (3373, 75%) also had lower odds of being in the non-adoption cluster (aOR 0.26, 95% CI 0.21-0.31) and those that distrusted healthcare had a higher odds of being in the non-adoption cluster (aOR 2.28, 95% CI 1.67-3.10) compared to people reporting neither trusting nor distrusting healthcare. Distrust in technology, financial and professional service institutions were also associated with reduced adoption of NPIs. Participants who expressed high trust in retail (1205, 27%) had greater odds of non-adoption for physical distancing and avoiding public spaces compared to people who reported neither trusting/nor distrusting retail (Table 4 ). Information and social media platform usage and trust. The most highly used communication channels and platforms for COVID-19 were public health websites (2897, 64%), health media briefings (2800, 62%), television/radio (2170, 48%) and physician/healthcare provider (1898, 42%). People who reported using Table 2 . Associations between attitudes and behaviours regarding COVID-19 and adoption clusters for COVID-19 NPIs. Bold signals statistically significant p-values (< 0.05). Odds ratios are the odds of being in the non-adopter cluster compared to odds of being in the adopter cluster. CI confidence interval, REF reference group, cOR crude odds ratio using logistic regression, aOR adjusted odds ratio, adjusted for sex, age, province of residence, highest education and political leaning. Odds of being a non-adopter compared to adopter www.nature.com/scientificreports/ or trusting Google or other internet search engines and information from friends and family had higher odds of non-adoption. Whereas those that reported using or trusting public health websites, health media briefings, television/radio or their physician for COVID-19 information were negatively associated with non-adoption (Fig. 3) . People who reported that they used none of these sources for COVID-19 information (206, 5%) had higher odds of non-adoption (OR 14.42, 95% CI 10.30-20.19) compared to adopters. Of these individuals who reported they used none of the listed sources for COVID-19 information, 188(91%) use some form of social media and 164 (80%) use Facebook with 85 (41%) reporting trust in Facebook for COVID-19 information. The most common social media platforms used by respondents were Facebook (3842, 85%), YouTube (2919, 65%), Instagram (2287, 51%) and Twitter (1433, 32%). There were 205 (5%) respondents who did not use any form of social media. Respondents who were more likely to use YouTube, Snapchat or a different social media platform than the ones listed had higher odds of non-adoption (Fig. 4) . www.nature.com/scientificreports/ Based on self-reported adoption of recommended NPIs intended on slowing transmission of COVID-19 including physical distancing, masking in public spaces, avoiding crowded public spaces and staying home when feeling sick, we segmented the population into adopters and non-adopters of NPIs through cluster analysis. Sociodemographic factors associated with non-adoption were: (1) being male, (2) age 18-34 years, (3) Albertans, (4) lower education level, and 5) a higher conservative political leaning. Participants who expressed low concern for COVID-19 had greater odds of being non-adopters. Non-adoption was associated with greater distrust among several institutions including technology, professional services, healthcare and government. Respondents who reported that public health messaging has been unclear and inconsistent and those where messaging has made them less likely to adopt NPIs had greater odds of non-adoption (Table 5) . Consistent with other research, males and young adults in our study reported lower adherence to public health recommendations 20, 21, 32 . In April 2020, a cross-sectional survey completed in Alberta and Ontario found that the highest non-adoption of NPIs occurred among males, age 16-29 years old, Alberta residents, with low COVID-19 knowledge and low concern 20 . Despite extensive resources and effort put towards messaging campaigns, this non-adopter population appears to be quite similar in characteristics from this earlier survey in April 2020 20 to our current survey in November 2020. This demonstrates the challenges of changing individuals' attitudes and behaviours regarding NPIs. Political leaning was the strongest sociodemographic predictor of adoption. A recent survey found only very small differences between conservative and liberal supporters in Canada and Republican and Democrats in the US in behavioral responses to the pandemic; however, there were greater differences in confidence in governments and concern about COVID-19 33 . Using geotracking data in the US with 15 million smartphones per day, Gollwitzer et al. found that Republican-leaning counties exhibited lower physical distancing than Democraticleaning counties 34 . In an analysis of tweets, conservatives were more likely than liberals to believe and spread conspiracy theories and misinformation on the COVID-19 pandemic 35 . Outside of the setting of COVID-19, bipartisan support by government leaders has led to less partisan motivated reasoning. This strategy may be also be effective for combatting COVID-19 36, 37 . We found that NPI non-adopters were more likely to have little concern about themselves or their friends and family becoming ill from SARS CoV-2. Non-adopters were also less likely to live with someone high risk of the disease or know someone who had COVID-19. They were also much more likely to have multiple regular interactions with people outside their household. Prior studies have demonstrated that the strongest facilitators for adoption of physical distancing are wanting to protect oneself and others and feeling a responsibility to protect the community. Prior barriers identified included needing to help friends or family members with errands, feeling lonely, and not trusting the messages from the government about the pandemic 32 . Respondents who expressed distrust in government, healthcare, professional services and technology had lower odds of adoption of any of . Association between social media platforms used and trusted by non-adopters of COVID-19 NPIs compared to adopters. *Participants could pick more than one most trusted source from each list. Odds ratios are the odds of being in the non-adopter cluster compared to odds of being in the adopter cluster. Table 5 . Summary of key recommendations to improve NPI adherence. Targeted communication strategies Communication strategies targeted towards demographics most likely to be non-adopters of NPIs may be most effective, including males, those aged 18-34, individuals with lower education levels and those with a higher conservative political leaning Increasing individuals concern for COVID-19 infection Individuals who have little concern regarding COVID-19 infection and disease are less likely to adopt NPIs, therefore messaging that increases concern may be effective www.nature.com/scientificreports/ sources with evidence-based, up-to-date, valid information allow everyone to obtain, process, and understand this information in order to make appropriate health decisions. Lessons from previous infectious disease outbreaks and public health emergencies including HIV/AIDS, H1N1, SARS, and MERS highlight the importance of clarity in communication 41 . Trust in government has been correlated with willingness to adopt protective behaviours in the face of other health threats such as the 2009 H1N1 pandemic 42, 43 and the 2014-2016 West African Ebola epidemic 44 . A previous survey conducted in Canada at the end of April 2020 reported respondents from Ontario and Quebec as having the least amount of trust in the Canadian government 45 ; however, in our study we identified highest distrust among Albertans (48%) followed by Saskatchewan and Manitoba (40%). Other studies have also demonstrated that the higher the degree of trust in the health system, the greater the compliance with guidelines 30, 46, 47 . We identified a high degree of trust in healthcare among our respondents with 75% reporting they either trusted or completely trusted healthcare; however, there were greater odds of non-adoption in people expressing distrust in the healthcare system. Belief in the efficacy of NPIs has been found to be critical for compliance with NPI recommendations 48 . The greatest proportion of people in the non-adopter cluster reported that they did not believe that the recommendations work as the reason for having not followed the public health guidelines. People who felt that COVID-19 restrictions were harming the economy had lower odds of adoption of NPIs. Prior research has found that there is a positive association between risk perception and economic threat perception, meaning that people that perceive a personal economic threat may as a result be less likely to adhere to guidelines 30, 49 . Individuals have access to a seemingly endless stream of information on COVID-19 through many different informational and social media platforms. One study evaluating the effect of information overload on the intention to self-isolate found that there was a negative impact of information overload on efficacy as information overload often does not allow for accurate understanding and therefore the uncertainty associated lowers efficacy 50 . Social media misinformation and lack of well-designed education programs without community engagement can impact on compliance and acceptability of NPIs 17 . Mobilizing an effective public health response during a pandemic requires clear communication and trust 48, 51 . Non-adopters were more likely to report that public health messaging has been unclear and inconsistent. As well, they were more likely to report that messaging has actually made them less likely to adopt NPIs. Our assessment of the trust people have in institutions (Table 4 ) and the "information sources" (from learned intermediaries (e.g. physicians, health professionals), news readers (celebrities and known people), social media (anonymous sources), and their government offers guidance for future communication and trust building approaches on COVID-19 NPIs and related efforts with vaccination 52 . The majority of respondents reported using public health websites, medical officer of health media briefings or TV to get their COVID-19 information; however, non-adopters were more likely to pick none of the provided options, meaning that there was possibly an information platform that we had missed listing or these individuals do not seek out COVID-19 information. Interestingly when respondents were asked if messaging or advertising had an impact on their likelihood to adopt NPIs, those who responded that they had not seen any messaging or advertising had much greater odds of non-adoption. Therefore, it is possible that there are a subgroup of people whom the public health messaging is not reaching, and these people are also more likely to be non-adopters. Future work is needed to identify effective methods and channels of messaging to this group of individuals as there is a shortage of qualitative research addressing this subject. Facebook, YouTube and Instagram were the most commonly used social media platforms. We found that individuals using YouTube and Snapchat for COVID-19 information had higher odds of non-adoption. Several prior studies have found a negative association between COVID-19 conspiracy beliefs, usage and trust in social media networks with adoption of NPIs 53,54 . One study evaluating conspiracy beliefs associated with COVID-19 on social media found that YouTube had the strongest association with conspiracy beliefs, followed by Facebook 53 . This suggests that targeting these social media platforms may be an effective option for distributing targeted messaging to promote uptake of NPIs for COVID-19 prevention. Overall, effective, credible, consistent and culturally informed health communication is vital in influencing positive health behaviours and building trust 41, 55, 56 , especially in terms of encouraging people to adhere to COVID-19 control measures and NPIs. There are several strengths and limitations to our work. This was a cross-sectional survey representing persons attitudes and behaviors at the time of this study, which will continue to change over time as the pandemic evolves. The survey recruited participants from an existing voluntary nationwide panel designed to be representative of the Canadian population; however, by using a panel there will be a component of selection bias as participants have volunteered to partake in research surveys and have access to electronic devices to do so. A non-response bias is possible as 1395 participants began but never fully completed the survey and were therefore excluded, however based on age, sex, ethnicity, education level and province of residence, these data were missing at random. Despite having a large sample size and being conducted in both English and French, generalizability may be limited. It is possible that some of the questions in the survey may have been interpreted differently by participants leading to variability in responses. It is also possible that individuals' motivations towards nonadoption of public health behaviours are complex and intertwined and not easily characterized or captured by survey tools. Future work including repeat surveys and qualitative studies to assess public attitudes and behaviors through this changing pandemic will be key in maintaining effective messaging promoting adoption of NPIs. Throughout the vaccination roll-out, adherence to and adoption of NPIs to reduce the spread of COVID-19 will remain critical. NPIs are most successful when there is a greater uptake from the public. This work provides a significant contribution to the COVID-19 literature through characterization of individuals more likely to be non-adopters of NPIs. An in-depth review of these individuals' sociodemographic factors, behaviours and attitudes towards COVID-19 and the barriers for NPI adoption is presented. We deliver a unique perspective though a Canadian national survey at a critical time during the pandemic, just at the initiation of the second wave with rising COVID-19 cases occurring across the country. This information will be useful for developing effective www.nature.com/scientificreports/ communication strategies including; knowledge translation tools, marketing programs and community engagement, targeted toward non-adopters of NPIs, in both message content and effective platforms for dissemination. Fig. S1 ) was used to assess respondents' selfreported adherence to NPIs including physical distancing, masking, avoiding public spaces and staying home when sick as methods to reduce spread of COVID-19. This survey was informed by an online survey and focus groups conducted in Alberta, Canada by the research team 57, 58 . Information on sociodemographic factors, attitudes towards COVID-19 and NPIs as well as risk and trust measures were collected. Participants were also asked about their usage and trust in information and social media platforms for COVID-19 information. This study was approved by the University of Calgary Conjoint Health Research Ethics Board (REB20-1228). Informed consent was obtained, and participation was voluntary. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist was used to report our findings 59 . Participants and setting. The Angus Reid Forum was used for selection of participants 31 . Eligibility was defined as: (1) aged 18 years or older, (2) live in a Canadian province, (3) speak either English or French, and (4) have access to the internet. The survey was administered between October 27 and November 2, 2020, in both French and English by the Angus Reid Institute, a national, not-for-profit, research foundation. The survey was programed using Askia 60 and launched using Platform One 61 . The survey was distributed to 14,887 potential participants who were randomly selected from the Angus Reid Forum of 70,000 individuals who are representative of the Canadian population 31 , in order to obtain a sample size of 4500. Sample size calculated for a 95% confidence interval with a margin of error of 3% for the adult population of Canada was 1068. However, in order to be able to stratify for equal representation of Alberta residents and residents of the other Canadian provinces combined we used a sample size of 4500 respondents. This sampling strategy was used to allow for comparison of two Canadian applications used to facilitate contact tracing, ABTraceTogether (a contact tracing application which is only available in Alberta) and COVID Alert (an exposure notification application available in eight provinces and the Northwest Territories). Variables and measurement. The main outcome measure was adoption of NPIs assessed by respondents answering the question; over the last few weeks, how often have you been performing each of the behaviors: (1) social/physical distancing-keeping at least 2 m from other people who are not in your social bubble, (2) wearing a mask in public indoor spaces when physical distancing is not possible, (3) avoiding places & activities where you would interact with a large number of people outside your household, and (4) staying home if you were sick with any symptoms, even mild ones. Adoption of these NPIs was assessed on a Likert scale of; all the time, most of the time, sometimes, rarely and never. Sociodemographic factors including sex, age, province of residence, household income, highest level of education, ethnicity, and political leaning were collected and categorized. Questions identifying attitudes towards COVID-19, public heath recommendations and reasons for non-adoption of recommendations were asked. Likert scales were used to assess agreement with statements on public health messaging and restrictions, how effective people believe public health recommendations are at reducing spread of COVID-19, effectiveness of public messaging thus far, and trust in specific institutions. Respondents' usage and trust of information platforms and social media platforms were collected. Statistical analysis. Descriptive statistics (percentage frequencies) were calculated for all sociodemographic characteristics, attitudes toward COVID-19 and towards NPIs and adoption of NPIs. Categorical variables were compared using chi-squared tests. Respondents who had not completed all survey responses were excluded, therefore there were no missing data. Cluster analysis was used as a data-driven method to identify the most important and meaningful patterns in the survey. This analysis highlighted how different or similar individuals are in their attitudes toward NPIs such that empirical patterns can be summarized in an insightful and concise manner. The goal of cluster analysis is to estimate a limited number of clusters with the most similarity within clusters but most dissimilarity between clusters. Clustering was based on individuals' attitudes and behaviors toward NPIs (limiting social gatherings, physical distancing, using face masks, avoiding public spaces and quarantining when sick), this includes reported actions and opinions on effectiveness of NPIs and on clarity of public messaging about NPIs. A series of statistical tests were conducted to determine the clustering method and the optimal number of clusters to explain the empirical variations in the data. The Kmeans algorithm was used for cluster analysis to partition the dataset into two distinct non-overlapping clusters labeled as adopter and non-adopter clusters. Kmeans is an iterative algorithm that assigns observations or data points to a cluster with the objective to minimize the sum of squared distance between the data points and the cluster's arithmetic mean of all the data points that belong to that cluster. The output of analysis in this part is a data-driven and detailed way to allocate each observation to their appropriate cluster and generate a clustering indicator variable (adopter and non-adopter cluster) to be used for further investigation. Logistic regression was used to calculate the odds ratio (OR) with 95% confidence interval (CI) for the nonadoption cluster compared to the adoption cluster as a reference. Logistic regression was used to identify risk factors for non-adoption of NPIs by sociodemographic factors, attitudes towards COVID-19, attitudes towards NPIs and towards public health messaging, and trust in specific institutions. Logistic regression was also used to identify communication channels and social media platforms at higher odds of being used and trusted by non-adopters. Backwards stepwise regression identified sociodemographic factors significantly associated with www.nature.com/scientificreports/ adoption clusters. These sociodemographic factors were included in multivariable regression models estimating the adjusted odds ratio (aOR) comparing adoption clusters with (1) attitudes towards COVID-19, (2) attitudes towards NPIs and (3) attitudes toward public health messaging. All P-values were two-tailed tests, and the statistical significance level was set at P < 0.05. All statistical analyses were performed using STATA version 15.1 (College Station, TX). All experiments were performed in accordance with relevant guidelines and regulations. Ethics approval. The study was approved by the University of Calgary Conjoint Health Research Ethics Board (REB20-1228). Consent for publication. All authors give their consent to publication of this work. The summary dataset used and or analyzed during the current study are available from the corresponding author on a reasonable request. Anxiety, distress, and turnover intention of healthcare workers in Peru by their distance to the epicenter during the COVID-19 crisis Economic consequences of the COVID-19 outbreak: The need for epidemic preparedness. Front Public Health 8, 241 Covid-19-Implications for the health care system Impact of COVID-19 pandemic on mental health in the general population: A systematic review The Peru approach against the COVID-19 infodemic: Insights and strategies Public health interventions for COVID-19: Emerging evidence and implications for an evolving public health crisis Effect of non-pharmaceutical interventions to contain COVID-19 in China Public perceptions and attitudes toward COVID-19 nonpharmaceutical interventions across six countries: A topic modeling analysis of twitter data Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe The impact of long-term non-pharmaceutical interventions on COVID-19 epidemic dynamics and control Association of public health interventions with the epidemiology of the COVID-19 outbreak in Wuhan, China The mental health consequences of COVID-19 and physical distancing: The need for prevention and early intervention Prevention and control of non-communicable diseases in the COVID-19 response The psychological impact of quarantine and how to reduce it: rapid review of the evidence G20 GDP Growth-Quarterly National Accounts. Record Fall in G20 GDP in First Quarter of Improving the impact of non-pharmaceutical interventions during COVID-19: Examining the factors that influence engagement and the impact on individuals Non-compliance with COVID-19-related public health measures among young adults in Switzerland: Insights from a longitudinal cohort study Different correlates of COVID-19-related adherent and dysfunctional safety behavior. Front Public Health 8, 625664 What drives resistance to public health measures in Canada's COVID-19 pandemic? A rapid assessment of knowledge, attitudes, and practices Does high public trust amplify compliance with stringent COVID-19 Government Health Guidelines? A multi-country analysis using data from 102,627 individuals Institutional trust and misinformation in the response to the 2018-19 Ebola Conspiracy theories as barriers to controlling the spread of COVID-19 in the Conspiracy theories in the era of COVID-19: A tale of two pandemics Regulatory Affairs Professionals Society (RAPS) Social distancing remains key during vaccinations COVID-19 vaccines: The pandemic will not end overnight Pandemics: Avoiding the mistakes of 1918 Risk Communication recommendations and implementation during emerging infectious diseases: A case study of the 2009 H1N1 influenza pandemic Analysis of public perception of the Israeli Government's early emergency instructions regarding COVID-19: Online survey study Barriers and facilitators of adherence to social distancing recommendations during COVID-19 among a large international sample of adults Novel coronavirus, old partisanship: COVID-19 attitudes and behaviours in the United States and Canada. Can Partisan differences in physical distancing are linked to health outcomes during the COVID-19 pandemic Partisan public health: How does political ideology influence support for COVID-19 related misinformation? Using social and behavioural science to support COVID-19 pandemic response The influence of partisan motivated reasoning on public opinion COVID-19: effective policymaking depends on trust in experts, politicians, and the public. Policy Des Rhetoric By Aristotle: Book 1 (Written 350 B.C.E), the Classics Archive The influence of source credibility on communication Effectiveness* Exploring communication, trust in government, and vaccination intention later in the 2009 H1N1 pandemic: Results of a national survey Trust during the early stages of the 2009 H1N1 pandemic The role of public trust during pandemics: Implications for crisis communication Public health and public trust: Survey evidence from the ebola virus disease epidemic in Liberia A national cross-sectional survey of public perceptions of the COVID-19 pandemic: Self-reported beliefs, knowledge, and behaviors COVID-19 pandemic: A litmus test of trust in the health system The influences of patient's trust in medical service and attitude towards health policy on patient's overall satisfaction with medical service and sub satisfaction in China Predictors of COVID-19 voluntary compliance behaviors: An international investigation Citizens' adherence to COVID-19 mitigation recommendations by the government: A 3-country comparative evaluation using web-based cross-sectional survey data Impact of online information on self-isolation intention during the COVID-19 pandemic: Crosssectional study Monitoring the level of government trust, risk perception and intention of the general public to adopt protective measures during the influenza A (H1N1) pandemic in the Netherlands COVID-19: An urgent call for coordinated, trusted sources to tell everyone what they need to know and do Health-protective behaviour, social media usage and conspiracy belief during the COVID-19 public health emergency Association between public knowledge about COVID-19, trust in information sources, and adherence to social distancing: Cross-sectional survey Communicating in times of uncertainty: The need for trust Trust influences response to public health messages during a bioterrorist event Attitudes, current behaviours and barriers to public health measures that reduce COVID-19 transmission: A qualitative study to inform public health messaging Attitudes, behaviours and barriers to public health measures for COVID-19: A survey to inform public health messaging The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies Toward effective government communication strategies in the era of COVID-19 An analysis of government communication in the United States during the COVID-19 pandemic: Recommendations for effective government health risk communication This study was funded by a COVID-19 Rapid Response Funding Grant from Alberta Innovates (Grant #202100489). The authors declare no competing interests. Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598-021-01279-2.Correspondence and requests for materials should be addressed to R.L.Reprints and permissions information is available at www.nature.com/reprints.Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.