Evidence before this study

Following PRISMA guidelines, we searched MEDLINE, Embase, PsychInfo, and medRxiv from Jan 1, 2020, to Feb 24, 2021, using the search terms (COVID-19 OR Coronavirus OR SARS-CoV-2) AND (vaccine OR vaccination OR immunization) AND (hesitancy OR confidence OR uptake OR acceptance). We found 937 papers, of which four used a randomised controlled design to examine the effect of information provision on COVID-19 vaccine hesitancy. A US study done in August, 2020, with 1123 respondents found that providing information that a vaccine is safe and effective, compared with no information, increased vaccine acceptance. In a French study done in July, 2020, with a representative group of 1942 people, there was no overall effect on vaccine acceptance from the provision of information on the benefit of herd immunity. Within the group, however, those individuals who outright refused vaccination (28·8%) were not affected by information on the benefit of herd immunity, but for those who did not refuse vaccination outright (71·2%), information on the benefit of herd immunity increased acceptance slightly. In a German study with 1349 members of the general population done in November, 2020, vaccine acceptance was not influenced by information on the individual and prosocial benefits of herd immunity. In a study of 8000 people in the UK and the USA in September, 2020, provision of negative misinformation, compared with factual information, reduced acceptance.

Added value of this study

The overarching question addressed in our study was whether there is specific brief content about COVID-19 vaccination, above a statement of safety and effectiveness, that might reduce hesitancy or consolidate existing positive views in the general population. The study was done after the COVID-19 vaccination programme had started in the UK. It was a large study, testing multiple different conditions of interest, and included outcome tests, moderation, and mediation. The study shows that for individuals who are initially strongly hesitant (ie, will avoid being vaccinated for as long as possible or will never get vaccinated), the most effective message for reducing hesitancy highlighted the personal benefits of vaccination (eg, that vaccination can prevent serious illness or long-term COVID-19-related problems). Addressing personal benefit was better than addressing collective benefit. It is also helpful in this subgroup to directly address safety concerns about the speed of development of the vaccines. Mediation of effects by changes in underlying views about COVID-19 vaccines was not found. The various information conditions tested, including those that reduced hesitancy in the strongly hesitant, did not change hesitancy levels either in people who were already willing to be vaccinated or those who were doubtful (ie, those who would delay vaccination or who did not know whether they would agree to receive the vaccine). There might be differences in impact of messages by gender and ethnicity, with participants of Black ethnicity sometimes reacting oppositely (more negatively) to the information provision conditions in comparison with the statement of safety and efficacy.

Implications of all the available evidence

The overwhelming majority in the UK are willing to be vaccinated for COVID-19 and can appreciate the collective benefits. In the small proportion, approximately 10% at the time of the study, who are strongly COVID-19 vaccine hesitant, and less inclined to see the collective benefit of vaccination, it is likely to be better to highlight personal rather than collective benefit in information provision. This group also show greater concerns about the speed of development of the vaccines and are responsive to information designed to address these fears. The information emphasis that reduces hesitancy in this subgroup does not affect vaccination attitudes in the rest of the population. None of the messaging lowered hesitancy in those who were doubtful about vaccination; research is needed to identify messaging that is especially effective for this group. Research also needs to consider the moderation of messages by gender and ethnicity. This study provides clear text for information provision in UK vaccine programmes. It was not determined how the successful information worked or whether the study findings can be generalised to other countries.