Skip to main content

Effects of different types of written vaccination information on COVID-19 vaccine hesitancy in the UK (OCEANS-III): a single-blind, parallel-group, randomised controlled trial

Our take —

In this randomized controlled trial involving over 15,000 adults in the UK, participants were initially classified by general COVID-19 vaccine hesitancy (willing, doubtful, and strongly hesitant), and then were provided with a standard statement of COVID-19 vaccine safety and effectiveness. Participants then read a randomly assigned package of information related to COVID-19 vaccines, and responded to survey questions designed to measure their level of vaccine hesitancy. Among those who were strongly hesitant, information focused on personal benefits of COVID-19 vaccination was the most effective in lowering hesitancy relative to the control group (standard statement only). This finding may be useful in planning public health messaging to improve vaccine coverage, though the results may not be applicable in other contexts.

Study design

Randomized Controlled Trial

Study population and setting

This was a randomized controlled trial that tested whether the provision of specific types of information content related to COVID-19 vaccination, beyond a standard statement of safety and effectiveness, would increase the acceptance of vaccines. The study enrolled 15,014 adults in the UK from January 19 to February 5, 2021; participants were quota-sampled to be nationally representative of the UK with respect to gender, age, region, education, and ethnicity. An additional 3,841 vaccine-hesitant adults were recruited through February 19, 2021. Participants were recruited via a wide variety of online platforms (e.g., social networks, advertisements, and mobile games) and offline methods (television and radio ads, mail) through a market research firm. Participants were stratified into three levels of vaccine hesitancy (willing, doubtful, and strongly hesitant) by their response to a single initial question about vaccine intentions, and then were randomized within strata to one of ten “information conditions.” The control condition was a safety and effectiveness statement from the National Health Service; the others added more information to this statement addressing personal benefits, collective benefits, pandemic seriousness, and safety concerns in various combinations. Participants then read text corresponding to each information condition (without being aware that other conditions existed) and completed a mediation measure of COVID-19 complacency and confidence beliefs and an outcome measure of COVID-19 vaccine hesitancy (the 7-item Oxford Vaccine Hesitancy Scale, with scores ranging from 7 to 35). The authors performed linear regression on the vaccine hesitancy score to test for differences by information condition, with interaction terms for baseline vaccine hesitancy.

Summary of Main Findings

In the unstratified population, none of the information conditions were associated with differences in the vaccine hesitancy score compared to the control. However, within the strongly hesitant group, three conditions were associated with lower vaccine hesitancy: one addressing personal benefit (difference in vaccine hesitancy score: -1.49 (95% CI: -2.16 to -0.82)), one addressing safety concerns in relation to the speed of vaccine development (-0.91 (-1.58 to -0.23), and a full combination of all forms of information (-0.86 (-1.53 to -0.18)). In this group, provision of information related to personal benefits of vaccination was associated with lower vaccine hesitancy relative to either a) provision of information related to collective benefit, or b) a combination of the two. There were no statistically significant differences in vaccine hesitancy for any information condition relative to control in the willing or doubtful groups. No evidence was seen for mediation by COVID-19 vaccine views.

Study Strengths

This was a randomized controlled trial, and as such was less susceptible to confounding bias than other study designs. The sample size was large and participants were sampled to be representative of the UK population with respect to several demographic variables. The vaccine hesitancy score was developed in consultation with members of the general public, was validated against other scores, and had high internal consistency.


Self-reported intentions regarding COVID-19 vaccination may not correspond well with actual behavior. Even if the study findings are valid, it is not clear how the vaccine information designed to reduce hesitancy could be best delivered; participants received the information in the context of a study into which they had voluntarily enrolled, and this dynamic may not be replicable in the wider population. This was not a probability-based sample; those who participated in the study may have been systematically different from the general population in unpredictable ways. The representativeness of the study population was diminished by the inclusion of the second group of vaccine-hesitant individuals. Finally, because participants were classified at baseline by a single question (as opposed to the full outcome measure), no change in vaccine hesitancy scores could be calculated for participants.

Value added

This is the only randomized controlled trial to date of information-based interventions to decrease COVID-19 vaccine hesitancy.

This review was posted on: 28 May 2021