Data Ethics Club: Trust is a verb: how to reimagine confidence in health systems#

Article Summary#

Health systems are designed to deliver care, monitor risks, and offer guidance that improves population wellbeing. Today, however, the relationship that people have with their health is changing. The information environments we move through are shifting faster than ever before and offer many more places to seek advice or affirmation. People learn from one another online, form shared expectations, and build norms that feel responsive in ways which top-down communication from institutions often fails to achieve.

The paper highlights three shifts influencing the expectations of people seeking healthcare:

  • Shift 1: professional expertise must meet lived experience as a partner. People rarely arrive in health settings without a story. In an effort to make sense of their experience, they often have an interpretation of their symptoms gained from conversations with social peers and information offered by online influencers. Heavy workloads of clinicians and short appointments reduce the opportunities to grow trust with patients, so people look elsewhere for the time and attention. Distrust forms from the feeling that the authority someone has over their own experience has been dismissed.

  • Shift 2: conversation about health moves at a speed that strains understanding. People are now learning about health in places with constant flows of new information, where the pace of the information stream shapes what feels trustworthy. Influencers present confident views in ways that come across as personal and emotionally coherent, contrasting slower and more formal updates from public health organisations.

  • Shift 3: personalisation reshapes what feels credible. Almost everything today adjusts to our preferences. When looking for health information, the questions people ask shifts from “is this true?” to “is this for me?”. Evidence expressed in averages feels impersonal, probabilities sound dismissive, and uncertainty comes across as a failure of care. Personalisation feels and is sold as empowering in an unpredictable world. Yet, whilst greater access to health information and promotion of self-care by health systems perhaps should lead to empowerment, it also places the burden of wellness on individuals.

To adjust to the cultural shifts that are influencing heath and health care systems, the paper suggests three pivots:

  • Pivot 1: practicing trustworthiness. Look closely at how questions and disagreements are worked through, who has voice at different stages, and how responsibility is shared. Trust in institutions is strengthened when the institutions create the conditions that allow health workers to listen, explain, and respond with integrity, and when participatory practices are adequately resourced.

  • Pivot 2: renewing the social contract across generations. Reflect how trust differs across generations because each group carries its own history of how they have experienced health and authority. Generational differences need to be recognised, listened to, and intergenerational dialogue encouraged.

  • Pivot 3: rebuilding the structures that connect science and society. Make visible the process that translates evidence into guidance. To grow trust, people need to be able to clearly see the pathways that show how evidence moves from discovery to decision to public dialogue, and where uncertainty, judgment, and values enter along the way.

Discussion Summary#

What did you think of the 3 shifts the paper outlined? Are they a complete list?#

Shift 1: professional expertise must meet lived experience as a partner

We could see how mixing professional expertise with lived experience can help rebuild a system of trust, not just in healthcare, but also in other domains we work in such as children’s social care reforms.

However, whilst we agreed that incorporating lived experience is important, the reality of joining lived experience with medical advice isn’t always as straightforward as it sounds. Lived experience is grounded in history, looking backwards, instead of looking forwards to prognosis and treatment pathways. We’ve seen examples of these tensions arise in Patient and Public Involvement and Engagement (PPIE). PPIE is where members of the public actively participate in or conduct research themselves, providing a channel through which lived experience can be incorporated. Difficulties can arise because sometimes clinical judgement will conflict with outputs from PPIE.

Effectively combining the lived experiences of people and outputs of PPIE with clinical judgement requires engagement from both sides. Compromises will be required if PPIE and clinicians disagree. For example, if the outputs of PPIE reject solutions that need Bluetooth, but researchers are only able to use Bluetooth, researchers may need to work with PPIE to understand their concerns and improve gaps in education.

Alongside differences between public opinion and clinical judgement, there will also be variation within public opinion and people can have wildly different lived experience. It’s important to make sure that the loudest voices aren’t the only ones being heard. Those who are participating the most may not be representative of everybody and people most well set up to do co-production work aren’t necessarily the ones who need it most.

Shift 2: conversation about health moves at a speed that strains understanding

Information that is available for us to consume is rapidly growing, including our access to advice which looks convincingly expert. ChatGPT is being increasingly used as counsel for all domains in life, including healthcare. However, studies such as Bean et al. (2026) have found that whilst large language models (LLMs) can pass standardised tests of medical knowledge, the tools are no better than traditional methods like online searchers or a patient’s own judgement. In addition, Bean et al. found that participants often did not know what information to give LLMs so that outputs offer accurate advice and LLM responses frequently combined good and poor recommendations. LLMs are also inherently susceptible to hallucinations, where responses are completely made up.

Growth in propagation of misinformation and disinformation is not just down to LLMs, however. Also playing a role is the rising availability of the internet and people [active on social media](https://www.statista.com/statistics/454772/number-social-media-user-worldwide-region/. There is an incomprehensible amount of data on the internet and an increasing proportion of it is inaccurate.

The format of social media platforms also influences how we consume information. Short videos mean that information has to be presented convincingly and quickly, before the viewer moves on. However, the condensed timeframe of these videos means that caveats are often skimmed over or ignored, leading people to believe that they can find a quick solution and magic pill. Repeated exposure to short form content has downstream effects on our psychology, for example, TikTok has been linked to reduced attention spans in students. Younger generations are big adopters of TikTok so healthcare communicators need to take this into consideration.

Whilst the accessibility of information increases the risk of misinformation, it also means that people can connect with communities that they may previously been unable to access. In healthcare, this may lead to patients knowing more about their condition than their practitioner does. For example, a patient could have a rare disease a doctor has not encountered before, but the patient may have connected with others around the world who also have the condition and can share their knowledge and experiences.

Shift 3: personalisation reshapes what feels credible

Health systems are generally quite rigid, but we need to acknowledge that people are extremely diverse and there isn’t one size fits all. It’s important that practitioners are able to individualise their approach and figure out how to appropriately adjust themselves to who they are interacting with, which is something that we all do anyway whether or not we do it consciously. However, hyper-personalisation is impractical and potentially undesirable. Healthcare systems need to find a suitable balance of personalisation.

Gauging the balance is a tricky skill to develop. For example, medical training heavily emphasises asking before doing anything and making sure that everything is explained clearly, but sometimes older patients just want you to “stop talking and get on with it”. It’s important that practitioners are aware of the worst that can happen if you over or under explain.

What change would you like to see on the basis of this piece? Who has the power to make that change?#

We thought the paper was quite generalised and rested on some implicit assumptions, seemingly assuming that there is a consensus in public perception of healthcare. This could be improved by incorporating more perspectives, as the paper adopts a very global north and western perspective. Healthcare may be structured and valued very differently depending on where you are in the world. For example, where people go to seek healthcare advice in rural or nomadic populations might be very different to where people go in cities, such as by seeking the opinions of elders rather than specific healthcare practitioners. We can’t assume that everybody researches health online or has access to wearable devices. For populations who have been systematically mistreated by institutions, rebuilding trust may be completely off the table.

Attendees#

  • Huw Day, Data Scientist, University of Bristol: LinkedIn, BlueSky

  • Jessica Woodgate, PhD Student, University of Bristol

  • Amy Joint, Publisher, ISRCTN Clinical Study registry, LinkedIn

  • Paul Matthews, Lecturer in Data Science, UWE Bristol. LinkedIn

  • Natalia Kappos, Design and delivery lead, HMPPS, (https://uk.linkedin.com/in/nataliakappos)

  • Noshin Mohamed, Principal Social Worker and QA lead - Newham

  • Rachel Peck, haematology SpR, PhD, Bristol

  • Naomi Cornish, also a haematatology SPR and PhD student, Bristol

  • Rosie Jones McVey, anthropologist of ethics/minds/data/health based at University of Exeter

  • Kamilla Wells, Citizen Developer / AI Product Manager, Brisbane