Synthesising findings from Stage 1 to identify the most common reasons for vaccine hesitancy

The evidence emerging from the survey, rapid literature review and PPI findings were then triangulated through discussion between the two behavioural scientists (RJ, KA) contributing to this stage of the work. The aim of these discussions was to identify the most common COVID-19 vaccine concerns. This was based in part on the frequency with which concerns were identified in the survey, review, and PPI findings, ensuring that all three domains of the WHO 3C model were represented and that any unique perspectives raised by ethnic minority participants were also captured.

This led to the identification of nine core COVID-19 vaccine concerns. Concerns that were endorsed by fewer than 0.5% of the sample and did not align with concerns identified within the literature and PPI groups were not included within the intervention (i.e. vaccination is ‘inconvenient’; Table 1a.). In keeping with the most frequently cited concerns being related to ‘confidence’, 5/9 concerns related to ‘confidence’ (i.e. generalisability of evidence on vaccine safety and effectiveness to diverse populations; side-effects; rapid nature of vaccine development; clinical effectiveness and vaccine scepticism). Two out of nine concerns related to ‘complacency’ (i.e. low perceived risk of COVID-19 and belief in the ability to fight off the infection naturally). A further two concerns related to ‘convenience’ (i.e. perceived lack of knowledge about COVID-19 vaccine and altruistic beliefs regarding others having a greater need). A tenth concern was subsequently added when the UK government decided to alter the dosing schedule from 3/4 weeks to up to 12 weeks between the two doses recommended for the Astra Zeneca and Pfizer vaccines. In keeping with the WHO 3C model, this latter issue is also related to the issue of ‘confidence’. Each theme/concern was given equal weighting within the subsequent development process.

Synthesising the evidence-based views of independent experts.

Following the identification of 10 core vaccine concerns (Table 3 ) we sought to gather evidence-based responses to these concerns. This was achieved through semi-structured interviews with six academic and clinical experts from the fields of public health, general medicine, respiratory medicine and immunology with particular expertise in COVID-19 and/or COVID-19 vaccines. Each expert was presented with the list of 10 concerns and asked to provide an evidence-based response to each concern based on their knowledge of the scientific literature at that time. Interviews with experts were subjected to rapid thematic and content analysis after each interview, and interviews continued until saturation in responses was achieved (i.e. no new responses emerged).25

Expert responses to 10 most common reasons for vaccine hesitancy.

The expert responses demonstrated significant thematic overlap and consistency. Table 3 summarises the areas of evidence cited by experts in response to each concern.

Developing therapeutic dialogues to address common vaccine hesitancy concerns.

Our approach to developing the intervention was predicated on two main observations of the existing evidence. First that psychoeducation alone (i.e. provision of information gathered in Stage 2) is unlikely to be an effective way to address COVID-19 vaccine concerns. Second that a central pillar of our approach should be to acknowledge and engage with individuals' concerns in a supportive context. To achieve this, we sought to develop ‘therapeutic dialogues’ based on the communication principles of motivational interviewing (MI), including:

MI was considered an appropriate approach because individuals who are vaccine-hesitant are, by definition, not ready to, or ambivalent about, changing their cognitions and behaviour and MI is known to be effective in such contexts.27 ^, 28 Thus, for each of the most common vaccine concerns identified in Stage 1 we developed a therapeutic dialogue, which would both impart information relevant to the individual concern, but do so using the communication principles of MI with a view to facilitating cognitive and, in turn, behaviour change, i.e. reduce hesitancy and improve vaccine uptake. An online format was chosen to deliver the therapeutic dialogue to maximise audience reach and engagement, supported by substantial evidence based on the use of this modality to promote vaccine uptake.29 ^, 30

Development of the therapeutic dialogues occurred through several expert workshops with behavioural scientists with expertise in MI, therapeutic interventions, digital interventions, behaviour change and COVID-19. First, key themes identified in the expert interviews (Stage 2) were discussed and translated into conversational language. The investigators chose a conversational approach to align with the online delivery format and ensure inclusivity for all reading/English levels (see stage 4 below). Second, the dialogues were reviewed to identify points at which MI techniques could be integrated throughout. This process drew on contributors' experience in behaviour change research and adopted the approach proposed by Rollnick and colleagues.26 This included expressing empathy through the use of accepting and non-judgemental language. By developing discrepancy by simultaneously providing information related to the concern and presenting a rationale for vaccine uptake. The latter were derived from survey respondents willing to accept a COVID-19 vaccine (see Table 1b) and sought to develop a discrepancy between the individual's cause for concern and their wider personal values and goals. By embracing resistance by acknowledging that their concerns are shared by others and are legitimate and supporting self-efficacy by reinforcing the individual's personal agency in making their decision to accept a vaccine or not. See Table 4 for illustrative examples of how MI principles were embedded within the therapeutic dialogues.

Exemplars of how MI principles were included within the therapeutic dialogues.

Finally, we hosted a PPI workshop to discuss the resulting dialogues. Participants were members of the general public recruited through the University Hospital Southampton NHS Foundation Trust PPI team. The workshop was advertised as an opportunity to provide feedback about an online tool designed to answer the public's questions about the COVID-19 vaccines. Four individuals responded to the advertisement and attended the workshop. The group, while small, included two adults less than 30 years (two greater than 50 years); three women and one man and all reported interest in vaccine hesitancy and had some experiences of it among friends and family. All participants were paid for their time. The feedback obtained through this workshop fostered changes to their readability, along with an expansion of the information conveyed and greater consideration of specific groups within the population (i.e. those who have allergies or specific religious and cultural needs). No additional vaccine concerns were identified by the group.

The digital intervention.

The script from each of the 10 therapeutic dialogues provided the architecture for our digital, web-based vaccine hesitancy intervention. Given high rates of internet usage throughout the Uunited Kingdom (92% of adults)31 and other similar developed countries, it was felt that the use of a digital platform would maximise reach and accessibility. The research team worked with a digital development company to design and build a conversational interface through which individuals identify the issue that most closely underpins their reason for being hesitant (from the issues stated above, e.g. concerns about side effects). This identification triggers an MI driven therapeutic dialogue relevant to the selected concern, with opportunities for the individual to further explore the content as they progress through the dialogue, as well as to access responses to more than just their initial concern.

Once developed, the digital intervention was piloted with 18 members of the public (nine male/nine female) who had no previous experience with the dialogues. Participant feedback on the dialogue content, user interface, accessibility, and general presentation led to a final iteration of the intervention, which can be viewed here: www.covidvaxfacts.info. For illustrative screenshots, see Supplementary Figs. 1 and 2.