Pre-intervention digital research experiences and their potential implications for digital health intervention participation: a qualitative observational study
This qualitative observational study of 50 adults with lower digital literacy reveals that pre-intervention digital research procedures often rely on relational support and accumulate interface friction that can cause distress and select for specific participants, suggesting that redesigning these processes to be more recoverable and transparent could improve inclusivity and representativeness in digital health interventions.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine a world where the first step to joining a medical study is not filling out a paper form at a clinic, but navigating a series of screens on a tablet or phone. This is the reality of modern digital health research, where technology acts as the gatekeeper, deciding who gets to participate and whose data ends up in the final results. For many, these digital tools promise to make research easier, removing the need to travel and allowing people to stay in their own homes. However, there is a hidden cost to this convenience. If the digital path is too confusing, too fast, or too unforgiving, it might silently filter out the very people the research needs to understand: those who are older, less familiar with technology, or living in difficult circumstances. The danger is that the studies we rely on to improve health might only reflect the experiences of the tech-savvy, leaving everyone else behind.
A team of researchers at the University of Southampton and the University of Aberdeen set out to uncover what actually happens when people with lower digital skills try to walk through this digital front door. They did not just ask people how they felt; they watched them. Between late 2024 and early 2026, the researchers invited fifty adults, mostly over the age of sixty, to sit down with a tablet and complete a series of tasks designed to mimic joining a real medical study. These tasks included watching an information video presented by a computer-generated avatar, signing a digital consent form, and logging into a separate system to report their health details. The participants came from diverse backgrounds, including rural and urban areas, and varied in their education and income, but they all shared a common trait: they found digital technology challenging or unfamiliar in their daily lives.
What the researchers saw was not a simple story of people failing to use technology. Instead, they found that participation was a deeply social act, often relying on the help of others. When a participant struggled, they did not always give up. Instead, a spouse, a child, or a friend would step in. Sometimes this help was gentle, offering reassurance or pointing out where to click. Other times, the helper took full control of the device, typing in answers or signing the form for the participant. In many cases, the person who actually completed the task was not the person the study was intended for. This "proxy" completion meant that a study might record a successful sign-up, while in reality, the participant had no idea what they were agreeing to, or had lost control over their own data. The researchers noted that this dynamic created a hidden burden; the person with the tablet might feel embarrassed or frustrated, while the helper took on the emotional and technical risk of making a mistake.
The journey through these digital tasks was often a series of small, accumulating frustrations. The researchers observed that the design of the screens often clashed with how these participants thought the world worked. For instance, when a keyboard appeared on the screen to type a password, it would cover the very question the person was trying to answer, forcing them to guess where to look. When asked to sign a document, some participants tried to type their name, while others tapped the screen with a finger, creating a small dot that looked nothing like a signature. These were not random errors; they were logical reactions to a system that assumed knowledge the participants did not have. As these small difficulties piled up, the participants began to feel a sense of panic or defeat. One person described the experience as "sweaty," while another said they just wanted to turn the device off and forget it. The researchers found that this emotional toll could happen even if the person eventually finished the task, potentially making them less likely to trust or use the medical intervention the study was testing.
The study also looked at how people reacted to a new trend in digital health: the use of artificial intelligence avatars to deliver information. The participants watched a video of a computer-generated character explaining the study. While some found the avatar friendly and clear, others were skeptical. A major factor in whether they trusted the information was knowing who was behind it. If they believed the avatar was speaking for a trusted hospital or a known doctor, they were more willing to listen. If the source was unclear, or if the avatar seemed to be a generic robot, they felt uneasy. Many participants expressed a strong desire to speak to a real human if they had a question that the computer could not answer. They were also sensitive to the appearance and accent of the avatar; while some welcomed the chance to see people who looked like them, others reacted negatively to diversity, viewing it as a sign of social change they did not understand. This suggested that simply putting a friendly face on a screen is not enough; the trust comes from the institution behind the technology and the assurance that a real person is available if things go wrong.
The most significant finding of this work is that the standard way of measuring success in digital research is flawed. Usually, researchers only count whether a task was completed or not. This study suggests that counting the "yes" answers hides the truth. A task completed with a helper, a task completed by someone else on behalf of the participant, and a task completed independently are all very different experiences. Treating them as the same thing masks the fact that many people are being excluded or are participating under conditions that compromise their privacy and autonomy. The researchers argue that for digital health to be truly fair, the design of these tools must change. They need to be more forgiving, allowing people to go back, try again, and recover from mistakes without feeling like they have failed. They also need to acknowledge that many people will need help, and that help should be designed to support the participant's own choices rather than taking over.
Ultimately, this research reveals that the digital tools we use to start medical studies are not neutral gateways. They are active filters that shape who gets to be part of the solution. If we do not design these tools with the reality of human relationships and the limits of digital confidence in mind, we risk building a future where the benefits of digital health are only available to a small, privileged group. By watching how people actually struggle and succeed, and by understanding the emotional weight of these interactions, we can build systems that are not just efficient, but truly inclusive. The goal is to ensure that when a new health innovation is tested, the people in the study truly represent the people who will one day use it.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.