Exploring the Effect of Video Interventions on Intentions for Continued COVID-19 Vaccination

The COVID-19 vaccine has significantly reduced mortality from the COVID-19 disease, and relieved the pressure on society, the economy, and the healthcare system caused by the pandemic. However, waning immunity and the emergence of new, vaccine-evasive variants like Omicron have necessitated additional vaccine doses, or "booster shots". Similar to trends for COVID-19 vaccination and other routine vaccinations, young adults (aged 18-39) have demonstrated lower uptake of booster vaccination, and evidence suggests that public health recommendations are no longer translating to vaccine uptake. This study will examine the efficacy of altruistic and individualistic messages over informational messages in increasing vaccine intentions in this age group.

This study has four primary objectives:

1. Estimate pre- to post-intervention changes in COVID-19 vaccine intentions of a altruistic video intervention.
2. Estimate pre- to post-intervention changes in COVID-19 vaccine intentions of a combined altruistic + individualistic video
3. Estimate between-group (altruistic vs. vaccine-informational) differences in COVID-19 vaccine intentions post-intervention
4. Estimate the between-group (altruistic + individualism vs. vaccine-informational) differences in COVID-19 vaccine intentions post-intervention

In addition, there is an exploratory, multivariable objective:

1. To estimate the multivariable associations of vaccine intentions in each intervention group.

A survey will be administered online to a sample of Canadian residents, aged 18-39, in both English and French. Participants will be drawn from a database panel provided by Dynata, a multinational first-party data and insight platform. Dynata's use of Canadian census data from Statistics Canada and quota-based sampling will ensure that the sample will closely match the Canadian population in terms of age (18-29, 30-39), sex, first language English/French, population density (i.e., urban vs. rural), and household income. Dynata will send invitations in various forms (e.g., e-mail invitations, text messages and in-app alerts) to recruit eligible participants from their existing panel (individuals who have signed with Dynata and have expressed willingness to receive invitations to various appropriate surveys) until recruitment goals and quotas are met.

To calculate the required sample size for the within-participant change in vaccine hesitancy (i.e., pre-post intervention) hesitancy towards continuing to receive booster vaccination was estimated at about 70% in the sample population. This estimate of the hesitancy rate at time of writing was based on seasonal influenza vaccination uptake in the target age range (between 20-25%) and uptake of COVID-19 booster vaccination in the past 6 months (9.2 - 14.5%) accounting for a gradual increase in uptake observed over time. Informed by results of a previous two-arm RCT conducted by this research team (NCT04960228) that included a video-based intervention and an active control group, it is estimated that the intentions in the control group will increase from 30% to 33% and in the intervention groups from 30% to 39%. To detect a 6% difference in intentions between arms, at a 95% confidence level (for a two-sided significance level α=0.025) and a desired statistical power of 80%, it was calculated that the minimum required number of participants per arm would be N=1005. 1:1:1 randomization will be used the three between groups. Thus, one-third of the participants will be allocated to each intervention group. Including 95 additional participants per arm to account for possible inattentive responders, the total sample required for this study is 3300 participants i.e., 1100 in each of the two experimental groups and 1100 in the active comparator group.

Stratified randomization will be used for first language (English/French), sex (male/female), population density (Urban/rural) and household income (<$75,000 / >$75,000) to allocate participants to either intervention arm. Within each stratum, a random concept picker approach will be used to ensure a 1:1:1 allocation using groups of three participants. Correspondingly, the first participant of a group will be randomly allocated to one of the three study arms, the second randomly assigned to one of the two remaining arms, and the third to the remaining arm. If a participant does not finish the survey (incomplete data), that place in the group will be allocated to the next participant. Thus, the quota in each stratum will be filled in groups of three and ensure a balanced group allocation throughout data collection.

The initial online survey will be approximately 12 minutes in length and will be available in both French and English. After reading the introductory statement of the study, participants will be asked for their consent to take part in the survey. Those who consent to participate will be invited to go to the next page and start completing the survey. Participants will first be asked to answer several socio-demographic questions and will be asked to choose their current Precaution Adoption Process Model stage for continued COVID-19 vaccination. Participants will then watch the assigned intervention. After viewing this content, to ensure that participants paid attention, participants will answer questions about the interventions to check their attention. Finally, participants will: 1) once again be asked to choose their vaccine decision-making stage; 2) be asked questions regarding previous vaccination history (e.g., seasonal influenza, COVID-19), lifestyle factors, self-perceived health status, personal history of SARS-CoV-2 infection, and preferred health-information channels; and 3) complete validated scales regarding, empathy, distress, intolerance of uncertainty, individualism and collectivism, COVID-19 pandemic fatigue, intellectual humility, and social desirability.

To test within-participant outcomes (Objectives 1 and 2), McNemar's test will be used to compare paired (pre and post-intervention) proportions. Additionally, exact tests of symmetry comprising of multiple pairwise McNemar tests to estimate pre-post changes in vaccine intentions. Adjusted p values for multiple comparisons, odds ratios (OR) and Cohen's g effect sizes will be reported. For each study group, the significant transitions between vaccine intention stage pairs will be used for calculating the total number of participants that changed towards increased vaccination intentions (e.g., from undecided to decided to). To test between-group outcomes (Objectives 3 and 4), the Chi-square test will be used to examine differences in PAPM stage proportions between intervention groups.

For the exploratory objective, Structural Equation Modelling will be used to identify multivariable associations of vaccine intentions (e.g., collectivism, empathy, intolerance of uncertainty) and to control for the potential effect of other factors (e.g., sociodemographics, psychosocial factors, previous influenza vaccination, lifestyle factors).