This study investigates how the design of weblinks in digital vaccination invitation emails influences recipient trust and their willingness to book an appointment. In this study, investigators compare three different link formats: a control third-party link previously used by the NHS, and two experimental weblinks: an improved version of the link, and a text embedded hyperlink.

The study tests primarily whether the two experimental weblinks will be perceived as more trustworthy and increase booking intention than the control weblink. Furthermore, the study examined whether the experimental weblinks are perceived to be more fluent (easier to read) and improve participants' ability to correctly identify the organisation (e.g., the NHS or a US pharmacy) that sent the hypothetical email.

To test these effects, investigators planned to gather data from 2,000 participants from the United Kingdom and 2,000 from the United States. They will be assigned to view one of the three hypothetical email versions. UK participants will see emails that appear to be from the NHS, while US participants see emails that appear to be from a local fictitious pharmacy. Due to US emails appearing to come from a fictitious pharmacy, investigators also expected that the benefit of correctly identifying the host organisation would be more pronounced in the United Kingdom than in the United States.

This research aims to provide evidence on how to design more effective and trustworthy digital health communications.

This randomised experimental study investigates how weblink design influenced the perceived trustworthiness of digital health communications. As digital invitations (e.g., emails and SMS) have become standard for public health initiatives such as vaccination programmes, the use of third-party or disfluent (hard-to-read) web links may inadvertently undermine trust.

Study Protocol and Intervention

Participants will be randomly allocated to one of three experimental conditions. In all conditions, participants will view a hypothetical email invitation for a COVID-19 booster vaccine. The content of the emails remains identical except for the format of the embedded booking link:

Control Condition: Features the "accurx" web link previously used by the National Health Service (NHS) in the UK.

Experimental Condition 1: Features an improved, transparent - "clear" - weblink that was designed to be easier to read and clearly identifies the host organisation (e.g., https://vaccine-booking.nhs.uk).

Experimental Condition 2: Features a text-embedded hyperlink.

To ensure relevance to the target populations, the stimuli are tailored to participants' countries of residence: UK participants see invitations appearing to come from the NHS, while US participants see slightly different versions of the email appearing to be sent by their local fictitious pharmacy. Unlike the UK, COVID-19 vaccination in the US was predominantly organised by local pharmacies.

Primary Hypotheses

The study is designed to test five primary hypotheses:

1. The email containing the two experimental weblinks will be rated as significantly more trustworthy than the control "accurx" weblink.
2. Participants will report being more likely to use the link to book a vaccine appointment based on the experimental weblinks than the control "accurx" web link.
3. Participants will find the two experimental weblinks more fluent (easier to read) than the "accurx" control link.
4. Participants will be more likely to correctly identify the organisation that hosts the link based on the two experimental weblinks than based on the "accurx" control web link.
5. Identifying the organisation hosting the link will have a more beneficial effect on perception in the UK than in the US.

Measurement

Following the presentation of the stimulus, participants will rate the email on three 5-point Likert scales: trustworthiness (1: Very suspicious to 5: Very trustworthy), fluency (1: Very difficult to 5: Very easy), and booking intention (1: Very unlikely to 5: Very likely). Host identification is assessed via a multiple-choice question with four options: "the NHS", "your local pharmacy", "accurx", or "Not sure".

Investigators will gather data from 2,000 participants from the UK and 2,000 from the US. They will be invited to complete the study via an online platform. Participants who complete the study too fast or fail the attention check will be excluded.

This study will be conducted as part of a larger research project exploring the broader relationship between the content of health text messages and trust perception.

The study hypotheses, design, measures and analyses were preregistered before the start of data collection on a different platform (AsPredicted). This time stamped preregistration protocol is linked as a document in this registration.

Inclusion Criteria:

• Residency: Must be a resident of either the United Kingdom or the United States

Exclusion Criteria:

• Attention Failure: Participants who failed the designated attention check included within the survey will be excluded from the analysis
• Speeding: Participants who completed the study "too fast" - indicating a lack of meaningful engagement with the experimental stimuli - were filtered out

Anonymised study data is shared on the Open Science Framework

The data and the study protocol were shared while the main scientific output was under review for publication. After acceptance for publication, all content will be made publicly available.

The data and study protocol will be freely accessible to the public after the main scientific output is accepted for publication.

This is a double-blind study in which participants were unaware of the specific experimental condition to which they had been assigned. Participants were randomly allocated to view only one of three possible versions of a hypothetical vaccination invitation email - featuring either a control link, a text-embedded hyperlink, or a transparent link - and were not informed of the existence or nature of the alternative stimuli. This masking ensures that their evaluations of trustworthiness, fluency, and booking intentions were not influenced by a direct comparison between the different link designs. As the study was conducted as an online survey via a fully automated platform, there was no direct interaction between the investigators and the participants during the intervention.