Communicate Hope to Motivate Action Against Highly Infectious SARS-Cov-2 Variants

Communicate Hope to Motivate Action Against Highly Infectious SARS-CoV-2 Variants

Michael Bang Petersen*a, Lasse Engbo Christiansenb, Alexander Bora, Marie Fly Lindholta, Frederik Jørgensena, Rebecca Adler-Nissenc, Andreas Roepstorffd, and Sune Lehmannb

aDepartment of Political Science, Aarhus University, Aarhus, Denmark bDTU Compute, Technical University of Denmark, Kgs Lyngby, Denmark cDepartment of Political Science, University of Copenhagen, Copenhagen, Denmark dInteracting Minds Centre, Aarhus University, Aarhus, Denmark

March 16, 2021

Abstract new variants, the B.1.1.7. variant, is now a global pres- ence and is expected to be the dominant strain in the The world is facing a race between controlling new and countries where it was first discovered within weeks [8]. more infectious variants of coronavirus and implement- These more infectious variants present a significant and ing vaccinations: How can health authorities and gov- new challenge. Because of higher infectiousness, new ernments most effectively communicate the need to en- variants strongly increase the risk of large outbreaks (wit- gage more strongly in protective behavior to avoid a col- nessed, e.g. in the United Kingdom and Ireland [7]), in- lapse of the healthcare system until vaccination programs tensifying the need for citizens to engage in protective are effective? In the first wave of the pandemic, citi- behavior [3], at a time where populations across the globe zens became engaged in ‘flattening the curve’ via pow- are strained with pandemic fatigue [4]. erful visualizations. Here, we use epidemiological modelling to develop a new visual communication aid, a This manuscript addresses the question of how health au- hopeful message of ‘buying time’, which reflects the pan- thorities and governments can most effectively commu- demic trade-offs currently facing governments, authori- nicate this new challenge to their citizens. ties and citizens. Using a population-based experiment In the spring of 2020, when the first wave of SARS-CoV-2 conducted in United States (N = 3, 022), we demon- spread across the globe, a figure illustrating the epidemi- strate that this hope-oriented visual communication aid, ological credo ‘flatten the curve’ became a global rallying depicting the competing effects on the epidemic curve of point. (1) more infectious variants and (2) vaccinations, moti- vates public action and communicates more effectively This illustration communicated the need for individual than fear-oriented visual communication, focusing exclu- and collective action within a conceptual framework of sively on the threat of the new variants. Finally, using epidemic modelling [1], in a way that resonated with a cross-national representative surveys from eight coun- large number of people. However, the ‘flatten the curve’ tries (N = 3, 995), we document the urgent need to mo- illustration, does not fit the current situation. Epidemi- tivate public action to halt the spread of the new, more ologically, the current goal is no longer to ‘flatten the infectious variants. These findings not only provide pub- curve’ in the sense of constantly keeping the number of lic health authorities globally with a validated blueprint hospitalized below the capacity of the health care system for health communication in a critical period of the pan- by moving infections from the present to the future. In- demic but also provide general insights into the impor- stead, there is a race between a variant-driven accelerated tance of hope as a health communication strategy. rate of infection versus vaccination-driven depletion of the pool of susceptibles. The goal of engaging in protec- After 10 months of the SARS-CoV-2 pandemic, the ar- tive behavior is, therefore, to decrease outbreak size (and rival of highly effective vaccines raised global expecta- associated hospitalizations and deaths) by buying time tions that a return to normal was within reach. As vac- until vaccines can take effect. Psychologically, research cines are rolled out, however, countries across the globe on health behavior prior to the pandemic has provided experienced the emergence of new, more infectious vari- increasing evidence that fear (e.g., fear of a collapse of ants of the coronavirus. The most wide-spread of these a hospital system) is not in itself enough for the genera- *To whom correspondence should be addressed. E-mail: tion of protective motivations [2]. Also, during the pan- [email protected]. demic, evidence has emerged that positive feelings such Author contributions: MBP, LEC, RAN, AR & SL conceptualized as empathy [5] and optimism [6] play a significant role, the study. LEC conducted epidemiological modelling. MBP, perhaps especially in the face of increasing fatigue. AB, MFL & FJ collected survey and experimental data. MBP, The challenges posed by more infectious variants call AB & MFL analyzed survey and experimental data. MBP, LEC, RAN, AR & SL wrote the paper. for intensified communication with the public, utilizing

1 Figure 1: The effectiveness of public health communication with (Panel A) and without (Panel B) the hope associated with vaccines. Panels C-D report unstandardized coefficients of marginal effects from OLS regressions that compare responses in the Hope Condition (Panel A) and Threat Condition (Panel B) to a control condition, respec- tively. Hinges denote 95% confidence intervals and whiskers denote 99% confidence intervals. a visual communications aid that clearly expresses the tions about infection and vaccination rates and hospital present rationale for engaging in protective behavior. To capacity (see discussion below). sustain both public trust and public motivation, this vi- To assess the communicative efficacy of this visual aid sual communication aid should be anchored in available and associated text, we conducted an online population- epidemiological knowledge and sensitive to the full spec- based experiment with a sample reflective of the US pop- trum of psychological motivations, including the human ulation in the beginning of February 2021 (N = 3022) capacity for hope. (for details on all data and models, see code repository1). Testing the Effects of Hope and ‘Buying Time’ Participants were randomly assigned to one of three conditions: (1) A control condition, which simply informed To create a visual aid for a hope-oriented communication the participants that more infectious variants of the coro- strategy, we built realistic epidemiological models reflect- navirus were spreading in the US and that authorities ing possible trajectories of the epidemic in one of the first may urge citizens to more strongly adhere to the health countries where B.1.1.7. was identified (Denmark). We guidelines. (2) A hope condition that displayed the graph modelled three scenarios (Fig 1A): (1) A scenario without from Fig 1A to the participants and described the dis- B.1.1.7. and without the implementation of stronger epi- played epidemic scenarios, emphasizing how the ‘goal is demic control measures. In this scenario the epidemic is to buy time until vaccines against COVID-19 take effect’. effectively controlled until vaccines take full effect (blue (3) A threat condition that displayed and described only dashed curve). (2) A scenario with B.1.1.7. but without the initial half of the graph, focusing exclusively on the the implementation of stronger epidemic control mea- threat from the infectious variants without mentioning sures (the red curve). Here hospitalizations exceed capac- vaccines and associated hopes (see Fig 1B). After expo- ity as B.1.1.7. becomes dominant until the vaccines take sure, respondents were asked to rate whether they found full effect. (3) A scenario with B.1.1.7. but where stronger that the information conveyed ‘a fearful message’ and ‘a epidemic measures are implemented (the green curve). hopeful message’. Subsequently, respondents completed This reduces the infection rate and manages to keep hos- four 3-item batteries that measured whether the informa- pitalizations below capacity until the vaccines take full effect. The models were generated from realistic assump- 1https://github.com/SocialComplexityLab/pandemic\_hope

2 sion the other outcome measures on observed feelings of hope and fear (controlling for the experimental treat- ments), demonstrate that feelings of hope (bs = [.49; .58]) predict these outcome measures substantatively and significantly stronger than feelings of fear (bs = [.13; .20]). These findings demonstrate that in the current phase of the pandemic, including the hope promised by the ad- vent of vaccines into the communication of future scenarios is effective in motivating stronger adherence to health guidelines and may facilitate a better public understand- ing of the pandemic situation that an exclusive focus on the threat from new, more infectious variants. Assessing the Need for Public Communication To assess the need for intensified health communication about the new variants, we ran online surveys in eight Western democracies that have already identified Figure 2: Knowledge, fear and behavior change be- these variants within their borders: The US, UK, Sweden, cause of new, more infectious variants. Proportions with France, Italy, Hungary, Germany, France and Denmark. associated 95% confidence intervals within each coun- In each country, we surveyed about 500 respondents dur- try (Denmark, Sweden, United Kingdom, United States, ing January 2021. Respondents were sampled to match Italy, France, Germany & Hungary), that agree ‘com- the population margins on age, gender, and geographic pletely’ or ‘somewhat’ with the statements: (1) ‘I have location for each country and imbalances were addressed heard about new variants of the corona-virus that are via post-stratification. Respondents were asked whether more infectious’, (2) ‘I am worried about the new, more they had heard about the new more infectious variants, infectious variants of the corona-virus’, (3) ‘I follow the whether they were worried about them and whether they health authorities’ advice to a greater extent because of as a result adhered more to the authorities’ advice (Fig 2). new, more infectious variants of the corona-virus’. Total Across all countries, 83% report that they have heard N = 3, 995. about new variants. A significantly smaller group (65%, t(3994) = 23.07, p < .001), are worried about them and an even smaller fraction (58%, t(3994) = 9.63, p < .001) tion ‘clearly explained’, helped ‘others understand’ and report to adhere more strongly to health advice as a con- helped ‘me understand’: (1) ‘why new coronavirus vari- sequence. These results show that knowledge about the ants are a public health threat’, (2) ‘why [people/they/I] new variants does not in itself fuel fear, which again does need to adhere more strongly to the health authorities’ not in itself motivate behavior change and, hence, high- guidelines because of new coronavirus variants’, (3) ‘how light the need for rethinking public communication in the we as a society can get safely through this pandemic’ and current phase of the pandemic. (4) ‘why strong measures are required until vaccines take full effect and drive down infections’. We averaged an- Assessing Required Levels of Public Action swers within each battery and z-scored the scales using While 58% across countries report changed behavior, this the control group’s mean and standard deviation. For the may not be enough for epidemic control, as the needed analyses, we report unstandardized OLS regression coef- level of behavior change is highly sensitive to the values ficients (bs) and associated p-values. of the specific parameters that underlie the three scenar- Fig 1C-H display the marginal effects of the Hope and ios displayed in the Hope Condition. The graph was pro- Threat Conditions relative to the Control Condition. The duced with an effective reproduction number of Rt = 1.0 Threat Condition (b = .094, p = .05) but not the Hope for the existing variants, a relative Rt of the new variants Condition (b = −.01, p = .78) is perceived as marginally of 1.6, the assumption that stronger measures would re- more fearful, whereas the Hope Condition is perceived as duce Rt to 0.9, and a daily vaccine rate of 500 per mil- substantially more hopeful (b = .39, p < .001). Further- lion population for the first 60 days and then increased more, the Hope Condition significantly increases the per- to 5 000. The models also assumed that 90% of the pop- ception of infectious variants as a health threat (b = .11, ulation wants to be vaccinated and that vaccines are 95% p < .05), the motivation to adhere to the guidelines of effective, which may be a valid assumption even in the the health authorities (b = .11., p < .05), the understand- face of the new mutations wu2021mrna. As a baseline, ing of how to get safely through the pandemic (b = .29, this graph is reproduced in Fig 3A. p < .001) and why stronger measures are needed (b = Fig 3B, 3C and 3D display the trajectory of the epidemic .23, p < .001). The effects of the Threat Condition are if the current measures are 10% less effective (Fig 3B), if for the most part insignificant (beyond effects in Fig 1D the relative Rt for the new variant is 0.1 higher (Fig 3C) and 1G) and for Fig 1D, 1G and 1H the marginal effects or if the vaccinations are rolled out more quickly in the are significantly stronger for the Hope than the Threat first 60 days (Fig 3D). The figures are based on a popula- Condition (all ps < .05). Furthermore, analyses regres-

3 C: 10 % pts higher transmission D: Additional vaccination A: Baseline B: 10% less effective restrictions for B.1.1.7 first 60 days 1200 1200 1200 1200 800 800 800 800 Hospitalized Hospitalized Hospitalized Hospitalized 400 400 400 400 0 0 0 0

0 50 100 150 0 50 100 150 0 50 100 150 0 50 100 150 Days Days Days Days

Figure 3: Impact of restrictions, relative transmission and additional vaccination. Each display contains a baseline scenario with B.1.1.7 (red) and without B.1.1.7 (dashed blue) as well as a scenario with B.1.1.7 and 10% more effective restrictions (green). A) reference Rt = 1.0 and a vaccination rate of 500 persons per day for 60 days and then 5 000 per day. B) As A) but with reference Rt = 1.1. C) as A) but with relative Rt for B.1.1.7 equal to 1.7 instead of 1.6. D) as A) but with a vaccination rate of 2500 persons per day for the first 60 days. The hospital capacity is 200 beds related to covid-19 (dotted line). All displays are normalized to a population of 1 million. tion of 1 million and a hospital capacity of 200 COVID- theory. Journal of applied social psychology, 30(2):407– 19-related hospitalizations (the planned peak capacity in 429, 2000. December 2020 and coinciding with B.1.1.7 becoming a [3] N. D. Grubaugh, E. B. Hodcroft, J. R. Fauver, A. L. matter of public concern). Phelan, and M. Cevik. Public health actions to control new sars-cov-2 variants. Cell, 2021. Fig 3 demonstrates that if the current measures are less [4] W. H. Organization et al. Pandemic fatigue: reinvig- effective or the new variant is more infectious, the epi- orating the public to prevent covid-19: policy frame- demic will very quickly escalate out of control. If the vac- work for supporting pandemic prevention and man- cines can be rolled out more quickly, epidemic control is agement: revised version november 2020. Technical much easier to achieve. report, World Health Organization. Regional Office Conclusion for Europe, 2020. [5] S. Pfattheicher, L. Nockur, R. Böhm, C. Sassenrath, The current pandemic situation is a race between con- and M. B. Petersen. The emotional path to action: trolling emerging infectious variants and implementing Empathy promotes physical distancing and wearing vaccinations. In this situation, motivating the public to of face masks during the covid-19 pandemic. Psycho- engage in protective behavior to buy time is crucial. As logical Science, 31(11):1363–1373, 2020. we have demonstrated in this manuscript, a dual com- [6] A. Sheetal, Z. Feng, and K. Savani. Using machine munication strategy that emphasize both the threat from learning to generate novel hypotheses: Increasing the new variants and the hope induced by the vaccines is optimism about covid-19 makes people less willing an effective means to this end. The present findings pro- to justify unethical behaviors. Psychological Science, vide public health authorities globally with a validated 31(10):1222–1235, 2020. blueprint for health communication in a critical period [7] E. Volz, S. Mishra, M. Chand, J. C. Barrett, R. Johnson, of the pandemic and emphasize the general importance L. Geidelberg, W. R. Hinsley, D. J. Laydon, G. Dabr- of using hope in public health communication. As fu- era, Á. O’Toole, R. Amato, M. Ragonnet-Cronin, ture mutations may require individuals to be repeatedly I. Harrison, B. Jackson, C. V. Ariani, O. Boyd, N. J. vaccinated, this emphasis on using hope in public health Loman, J. T. McCrone, S. Gonçalves, D. Jorgensen, communication is not just of key importance now but R. Myers, V. Hill, D. K. Jackson, K. Gaythorpe, may be continuously important over the coming years. N. Groves, J. Sillitoe, D. P. Kwiatkowski, , S. Flaxman, Acknowledgements O. Ratmann, S. Bhatt, S. Hopkins, A. Gandy, A. Ram- Grant CF20-0044 from the Carlsberg Foundation funded baut, and N. M. Ferguson. Transmission of sars-cov-2 the study. Ene Gro Stubsgaard drew the graphs. lineage b.1.1.7 in england: Insights from linking epidemiological and genetic data. medRxiv, 2021. [8] N. L. Washington, K. Gangavarapu, M. Zeller, REFERENCES A. Bolze, E. T. Cirulli, K. M. Schiabor Barrett, B. B. Larsen, C. Anderson, S. White, T. Cassens, S. Jacobs, [1] T. R. Amidon, A. C. Nielsen, E. H. Pflugfelder, D. P. G. Levan, J. Nguyen, J. M. Ramirez, C. Rivera-Garcia, Richards, and S. H. Stephens. Visual risk literacy in E. Sandoval, X. Wang, D. Wong, E. Spencer, R. Robles- “flatten the curve” covid-19 visualizations. Journal of Sikisaka, E. Kurzban, L. D. Hughes, X. Deng, Business and Technical Communication, 35(1):101–109, C. Wang, V. Servellita, H. Valentine, P. De Hoff, 2021. P. Seaver, S. Sathe, K. Gietzen, B. Sickler, J. Antico, [2] D. L. Floyd, S. Prentice-Dunn, and R. W. Rogers. A K. Hoon, J. Liu, A. Harding, O. Bakhtar, T. Basler, meta-analysis of research on protection motivation

4 B. Austin, M. Isaksson, P. Febbo, D. Becker, M. Lau- rent, E. McDonald, G. W. Yeo, R. Knight, L. C. Lau- rent, E. de Feo, M. Worobey, C. Chiu, M. A. Suchard, J. T. Lu, W. Lee, and K. G. Andersen. Genomic epi- demiology identiﬁes emergence and rapid transmission of sars-cov-2 b.1.1.7 in the united states. medRxiv, 2021.