The analysis showed that the co-design activities stimulated reflection and engagement with the ethical principles, with five key themes emergent: 1) clarity/unclearness, 2) operationalisation, 3) stakeholders, 4) impacts of not upholding principles, 5) interconnectedness. Concurrently, the co-design activities brought about reflection on how visuals could help and hinder the communication of ethical principles.
Reflection on the ethical principles
Reflection on the meaning of principles revealed first a need for clarity in the descriptions and terms used. In the show & tell exercise, one participant illustrated the challenge of grasping ethical concepts using the visual metaphor of a fish tank. She used the solid fish and corals to represent the clarity of data processing activities, whilst the water signified ethics. She also remarked that “you have to read the guidelines like five times to make sure you get all the points.” In the scenario-building activity, participants reiterated that some terms describing the principles were too vague and needed further clarification. For example, in relation to both Data Fairness and Accountability, participants asked, “what even is fair” and “fair to whom?” They also speculated whether the principle of Data Fairness had the same meaning as Fair Data (Findable Accessible Interoperable Reusable) (Reiser et al. 2018).
Participants continuously expressed a desire for more operationalisation of the ethical principles. One participant reflected, “for me, the gap is between the guidelines and the practical recommendation. What does it mean as a researcher? And how to implement that?” Similarly, another commented – “I did not really know what to look for, what to do with it (the framework)” and represented this using a figure scratching its head. The desire for operationalisation continued as a theme in the scenario building and discussions. For example, participants highlighted difficulties in implementing Privacy due to data types’ diverging requirements and privacy standards, alongside the lack of standardised terminology. Illustrating this was a research collaboration story where different terms complicated compliance, as the hospital used the term ‘medical data’ for which the institute used ‘health-related data.’ Despite such challenges, participants described Privacy as the most operationalisable principle as they often had to write about it in their ethical approval processes. This suggests the instrumental role of ethics committee reviews in sensitising researchers’ awareness and understanding of ethical principles.
The co-design activities resulted in the identification of the many stakeholders involved and impacted by data sharing within a health research context. Particularly in the scenario-building activity, participants identified health providers, research institutes, scientists, data managers, data harvesters (such as Google), private companies, hackers, journals, media, patients, and broader communities. The discussion moved from their different perspectives to their responsibilities, alongside the challenges of finding a balance between expectations. For example, in one scenario, a participant told of her first-hand experience juggling the diverging requirements for data processing and sharing of the research institute compared with the collaborating hospital. By sharing their personal experiences, participants started to connect ethical principles to real-world examples, allowing them to see their relation and relevance.
Impacts of not upholding principles
Participants also reflected on the consequences of disrespecting the ethical principles. Expected consequences ranged from damaging media attention, lawsuits and negative impacts on patients and the public, such as stigmatisation, decreasing trust or privacy due to insufficient data security and unauthorised data accessing. Harms to research were also mentioned, such as the increased need for time and resources, challenging collaborations, and ultimately unclear resolutions that impede research. One participant also noted that undermining Respect of Persons could lead to de-personalisation of participants and forgetting the wider context, as “when you work with numbers and databases you start seeing people only as organ donors.” In this sense, participants highlighted the significant impact of terminology describing stakeholder roles, specifically that of study participants. For example, a participant explained that calling participants ‘data subjects’ could reduce them to their functionality in a study, while calling them ‘patients’ would imply a health condition.
Finally, reflection on the ethical principles revealed their interconnectedness, with participants frequently discussing how the principles related to each other. One participant, for instance, considered Respect for Persons as the overarching principle, arguing it is the foundation for all other principles. Interconnectedness was also evinced through the scenario-building exercise, wherein groups used the same scenario to illustrate different principles. For example, they used the Covid-19 Vaccination Information leakage case in Switzerland (Fichter and Seemann 2021), to demonstrate the need for respecting all four principles.
Reflection on visualisation
The representation of the ethical principles also emerged as a key result of the co-design process. Even without instruction, participants approached making sense of the principles through imagery. “It draws a picture in my mind” remarked one participant. Data analysis further revealed that participants generally believed visuals could improve and motivate engagement with ethical frameworks. One participant commented, “the guideline was very informative… but very dry to read’ and signified this with an image of rice. Participants subsequently suggested visuals could promote engagement, understanding and memorisation by distilling the information, making abstract concepts more concrete, and providing potential structure to the document. Participants also reflected that visuals could do harm. For instance, visuals could convey the wrong message by oversimplifying the content, not telling the whole story, or causing suspicion when branding is inappropriate (for example, using a pharma industry logo). As well, they identified the challenge of visually representing diversity and ensuring inclusion. One participant reflected that she understood how difficult it is to visually include all stakeholders, saying “it’s hard to address all the aspects of diversity.” However, using a simplistic approach (for example, a non-gendered image) was also identified as exclusive, with one participant sharing that she did not relate to the icon of a male, able-bodied form often used to indicate a person.
Participants then suggested various ways to address the potential shortcomings of visuals. Recommendations included using lay language, showing the perspective of both the researcher and study participant, and ensuring that audiences can easily interpret the visual. The voting exercise captured participants’ ideal design criteria (see supplementary file 1), and through group discussion agreement was reached as to what would be tested in sketching and prototyping. The following sketching and ideation process then resulted in an array of referents and ideas to communicate each principle. Respect of Persons was paired with images of hands to represent solidarity and human figures for human dignity. Privacy was represented with commonly used symbols, such as images of shields and locks to show protection, alongside a masked person to signify anonymity. Accountability was translated through law symbols, such as an official stamp to show quality, and a signature to indicate someone responsible. Data Fairness was conveyed using a scale for fairness, a server or paper records to show data, and the metaphor of a cake to convey sharing. Interestingly, participants found it challenging to visually represent data, as while some saw data in a computerised, digital form, others imagined data as stacks of paper with pie charts and analysis.
Feedback on the online workshop format
In complementation to reflections and discussion within the workshop, six out of the eight participants completed voluntary workshop evaluation questionnaires. One survey response was excluded from the analysis as it was incomplete. All completed responses (5/5) found the workshop met expectations, was interesting, enjoyable, useful, and stimulated reflection on both the ethical principles alongside the role of visual communication. In an open comment, one participant wrote, “I have them (ethics principles) more in mind for the day’s work. It was a refreshment or reminder of the pillars of our research.” Whilst another reported, “they feel less scary…”. Similarly, 5/5 reported having a good experience collaborating, and 4/5 felt the workshop succeeded in exploring how visuals could help or hinder communication. As to the activities themselves, the majority of responding participants enjoyed the sketching and voting activities most (4/5). The below anonymous comment demonstrates that although the online format was not ideal for all activities, namely sketching, the overall experience was still positive, providing both skills and learning.
“In-person meeting would have been better, especially for the sketching and for encouraging participation in the discussions. During in-person meetings, even if not everyone talks, it is possible to read people´s faces, but in Zoom with no camera (switched on) that gets lost. Saying that, you did a good job reminding participants to talk and giving positive reinforcement for people´s contributions. Overall, I'm really happy I participated, and I take home new skills to apply to my professional and personal life. I look forward to see the final design.” – Anonymously commented by one participant
Participation post-workshop took various forms. Following the INVVOLVE framework (Slattery et al. 2020), we gave co-design participants control and choice by asking them how they would like to be further involved in the project. One participant wanted to work as a designer, five participants chose the role of decision-maker, and four participants decided for no further involvement. The participant in the role of the designer, together with the principal researcher, drafted four visuals. Decision-makers then used five criteria identified from the workshops to evaluate the drafted designs. These criteria are outlined below. In an iterative process, visuals were created of the four principles Respect of Persons, Privacy, Accountability and Data Fairness. A knowledge visualisation expert also provided feedback based on these design criteria, with all resulting changes approved by the workshop participants.
Criteria guiding the decision makers’ and expert’s feedback.
The visual is easy to interpret
The visual conveys the right message
The visual makes the concept more concrete
The visual does not oversimplify the concept
The visual represents diversity and inclusion
The visual considers both the perspective of the researcher and researcher participant
As Fig. 2 shows, each resulting visual has its’ own unique shape aiming to aid visual memory encoding and retrieval. Perspective and shading draw the viewer’s attention to what’s inside the shapes, alluding to how the principles help frame issues. SPHN’s branding, along with the design criteria voted on during the workshops, determined the visual style, use of curved lines and the colour palette. Regarding the referents, participants decided on using metaphor alongside visually observable phenomena and icons. For Respect of Persons, heart-shaped hands embrace to visually signify an equal and respectful relationship between participants, researchers, and their institutes. Participant diversity gets conveyed by a crowd, and the need for researchers to respect participant’s privacy and confidentiality is shown by a lock. To convey the principle of Privacy, participants tossed up between imagery of locks and shields. In the end, they agreed on a vault to emphasise that privacy was linked to health data storage, access and sharing. For the principle of Accountability, participants agreed to graphically represent a person accountable. Icons surrounding this person then show the qualities of being fair, lawful, transparent, and are placed in cogwheels to indicate their relation to accountability mechanisms. Participants discussed at length how to visualise the principle of Data Fairness due to the difficulty of representing the broadness of data and data processing concepts. Finally, the group decided to communicate the principle through the metaphor of a data cake, an idea that emerged during the second workshop. In this way, the visual highlights how data fairness is primarily about sharing data. It is important to note that these visuals were not designed to function as stand-alone, rather they were designed to be positioned within the SPHN framework and read alongside the text to promote multi-modal communication.