Research team members: The work was led by JB, PAP, AC and RD as part of their contributions to the public engagement work of Health Data Research Network Canada (HDRN Canada).

Public/patient participants: A total of 20 Canadian participants (10 English speaking and 10 French speaking) took part in the process (Table 1 – inline). Previous studies have reported that health data sharing, linkage, use, and reuse is a complex topic that is not well understood by members of the general public [18, 34]. Therefore, recruitment focused exclusively on people who had prior experience as public or patient advisors in a health-related field and/or working with health data, e.g., in community organisations, hospitals, or health research institutes with the hope that such participants would be well-positioned to engage in dialogue with each other without extensive pre-reading or background preparation. Snowball recruitment was conducted beginning with emails sent to Canadian public or patient organisations and advisors by PAP, AC, RD. Participant selection focused on maximizing the diversity of participants across a range of factors including, but not limited to age, gender identity, highest education level, geography, ethnicity, and prior experience or advocacy positions. Participants were provided an honorarium of $100 Canadian, paid in two installments at the beginning and end of the process.

Public/patient peer-reviewers: Thirteen qualified applicants (eight English speaking, five French speaking) who were not selected to be participants were offered roles as peer reviewers for the report (Supplementary Material 1– Characteristics of peer reviewers). Peer reviewers were provided an honorarium of $25 Canadian, paid after they reviewed and provided comments on the draft report.

Participants were content creators and co-authors as opposed to being the objects of a study [37]. The process was facilitated by research team members employed by HDRN Canada (PAP, JB) and GRIIS (AC, RD). Participants’ perspectives were captured using a combination of written submissions and oral input during facilitated dialogue sessions. For written input (see Supplementary Material 2 – Step 1 document) participants were given flexibility regarding timelines and the format in which feedback was received. Participants also had the opportunity to change their written input at multiple points during the process.

Separate dialogues were held in English (morning) and French (afternoon) on the same dates. In the dialogues, a modified version of the “fist to five method” [38] was used instead of binary yes/no voting to determine if participants were close to agreement on a topic [39]. This entailed asking participants to indicate their “fingers” of support at the beginning and end of discussions about statements focused on various uses of health data. The “fingers” of support method created space for people in the minority to express their views by (i) providing a simple mechanism for people to signal disagreement or concern with majority views, and (ii) identifying people whom the facilitator should invite to contribute to speak to ensure that the full range of supportive and non-supportive views were expressed. The "fingers" of support displayed at the end of each discussion allowed the facilitator to determine whether there was a consensus regarding support or opposition to a use of health data (i.e., all participants displayed one or more fingers of support for a statement) or no consensus (i.e., any participant displayed a fist indicating opposition to the statement). The study followed inclusive design principles [16] and had the explicit aim to capture all participants’ views, including those held by a minority of participants.

Data were collected and analyzed in iterations according to the following process:

Additional details on the process can be found in Appendix B of the public report for the project [39].

Experienced patient and public advisors served as co-authors and peer reviewers for the report, Social Licence for uses of Health Data: A report on public perspectives, published on the HDRN Canada website [40]. Twelve of the experienced public/patient advisors (six participants and six peer reviewers who were part of the process) are also co-authors of this publication.

In brief, there was consensus support among participants for three uses of health data and consensus opposition to two uses of health data (see Text Box 1). Peer reviewers did not express disagreement with the consensus views of participants.

There was no consensus among participants about what constitutes an essential requirement for health data social licence. The original list of 40 potential requirements provided to participants in the first step of the process grew to include 85 requirements in seven categories by the end of the process (see Box 2 and Appendix C of the report [42]). Some requirements were supported by many people, others were seen as essential by just one or two participants. Out of the 85 requirements, 38 were perceived to be essential by just one participant.

Of the 20 participants, 15 people (four English speaking and seven French speaking participants, two English speaking and two French speaking peer reviewers) anonymously evaluated the process using the PPEET (Table 2 – inline, Supplementary Material 4 for peer reviewer results). Peer reviewers are not included in the main table because they had much less involvement throughout the process (about one to two hours of work total) compared to participants, who spent four hours in dialogue with each other, as well as time spent on independent reflection.

Overall, the participants and peer reviewers supported health data being used for public benefit, provided that their concerns were addressed. These findings are consistent with previous Canadian and international studies that have found conditional public support and social licence for uses of health data that produce public benefits [18–28].

In particular, the requirement categories identified by participants (Box 2), were similar to key themes identified in the Aitken et al. systematic review of qualitative studies [24] that examined public attitudes towards sharing or linking health data. Furthermore, there were no themes identified in the systematic review that were contrary to the perspectives of participants in our process. In addition, our process also identified some requirements that were not reported, or emphasised as strongly, in previous studies, likely because of our inclusive design approach, which captured views that were held by a minority of participants as opposed to allowing them to be out-voted or hidden. For example, two of 20 participants thought it was essential that data collectors “mitigate the risk that patients (who are often in a vulnerable state) feel pressured to provide consent” for secondary uses of their data. Another example of a requirement expressed by a minority of participants was two participants’ view that it is essential that “Unless there is expressed consent for data sharing outside of Canada, health data are stored in Canadian jurisdictions and stay within Canadian law.” Other unique requirements highlighted by participants in our process included the role of accountable governance bodies in enforcing consequences for data theft and fraud, and the importance of access to personal health data in a useable and timely (or even real-time) manner.

The process of involving participants and peer reviewers as co-authors afforded them the opportunity to provide additional reflection on the process and findings presented in the non-peer reviewed report for the work [40]. For example, one co-author noted the importance of finding a balance between the sensitivity of health data that requires special data protection measures, and the effective use of health data to improve patient care and advancing research. Another co-author emphasised requirements related to equity and fairness, particularly around the importance of Indigenous data sovereignty and ensuring that people who contribute health data, especially those from marginalised and/or colonised populations, are not exploited. Another public/patient co-author noted the strong view among members of the public that they should have access to their health data and a say in determining how data are used with a focus on ensuring that individuals understand the potential risks inherent in sharing their data.

To our knowledge, the work presented here is unique among studies of public perceptions related to health data in its incorporation of inclusive design principles to capture views held by a minority of participants and points of disagreement [16]. As noted by Treviranus, the views of people in the minority are important because those people are often not well served by the status quo, and they may hold knowledge and innovative ideas that are essential to the design of processes, policies and products that work for the entire population, not just the majority. Therefore, while our sample size of 20 participants is small, we feel that publishing and highlighting views held by a minority of participants is an important step toward identifying more and different perspectives about health data social licence.

Our process also captured requirements that participants actively disagreed about. For instance, four of 20 participants believed it was essential that “Opt-in consent is obtained before health data are collected, used, shared, or re-used (i.e. people agree in writing to the use(s)/user(s) of their data)”. However, one participant disagreed because, from their perspective, “Too much protection, for example a regime that requires consent for each project (opt-in), can make research much more complicated if not impossible.”

Another point of disagreement related to patients’ access to their own data. Eight participants indicated that it is essential that “People (and caregivers that they authorise) can easily access health data that have been compiled about them in a usable format (e.g., one that can be read by a computer or electronic device) and in a timely manner so that they can make informed decisions about their health and health care.” However, one participant disagreed because, from their perspective “The security issue of individuals being able to access their health data is far too high and can jeopardise the privacy of individuals.”

There was no agreement among the 20 participants regarding (i) use of health data by companies and (ii) use of health data from or about systematically marginalised populations. Most participants supported both of those uses if specific conditions and safeguards are put in place; others were skeptical about whether safeguards and conditions could be operationalised and concerned that if harms were done by companies and/or to historically marginalised communities, they could not be undone.

Considered together, the outputs of the process suggest that there is no definable boundary when it comes to social licence for uses of health data because, even among the small number of participants in our process, there were divergent views regarding which users and users of health data should be allowed and prohibited. From this we infer that expansive plans to increase use, sharing and re-use of health data are likely to be opposed by some groups, no matter what requirements, conditions and safeguards are put in place.

On the other hand, the outputs of our process suggest that there may be agreement on a “centre” of social licence, which can be defined in terms of uses of health data that are likely to be supported by members of the public. We note that focusing initiatives, policies and processes related to health data on the three uses that participants of this process supported (Box 1) could yield many benefits including: better patient care, better health system planning, and better understanding of disease and wellness. We suggest that these and other benefits could be realised if there is a concerted and continuous effort to identify and act on increasing the health data uses and users that members of the public support.

Overall, the 11 anonymous responses to the PPEET were positive (see Table 2 for participants’ responses, Box 1 for example comments and Supplementary Material 4 for peer-reviewers’ responses). For instance, only one person who completed the evaluation selected the “strongly disagree” response for any of the 16 positively-worded statements of the PPEET questionnaire. The involvement of participants and peer reviewers as co-authors in this paper confirmed the generally positive assessment of the process. For example, four out of 11 participants indicated that they learned from the process, and five participants appreciated the inclusive design and variety of perspectives that were heard, including voices in the minority. Box 3 includes some example of anonymous comments captured in the PPEET selected by the public/patient co-authors of this paper.

Participants and peer reviewers were most likely to select “Disagree” or “Neither Agree nor Disagree” to PPEET statements related to the impacts and influence of the engagement. For example, three out of 11 participants who completed the evaluation selected “disagree” or “neither agree or disagree” for the statement “I am confident the input provided through this initiative will be used by HDRN Canada, GRIIS and other organisations engaging with the public about health data”. This more negative to neutral impression of the impact of the engagement was also reflected in respondent comments. Three of 11 participants wanted to know more about the report’s intended audience and use(s) and were hopeful that the findings would contribute to future work in the area of health data social licence, whereas two participants were not confident that the report would be shared widely or used at all.

Most of participants’ suggestions for process improvement focused on how the dialogues were conducted. Two participants suggested conducting mixed dialogue sessions with English and French participants together (as opposed to separate English and French dialogue sessions). Four participants suggested the addition of more meetings and/or face-to-face meetings and wanted the facilitators to intervene more so that no individuals dominated the dialogue, and everyone had a chance to speak. For example, one participant noted “I often felt like there wasn’t enough time, sometimes because one person spoke up repeatedly or took a lot of time to talk, and not always in line with the topic…”. A co-author involved in the writing of this paper elaborated on this point, noting that, while the process seemed very democratic, some of the “bolder participants got more airtime.”

In terms of other areas for improvement, some participants commented that there should have been more time for discussion, and one person associated the overall shortage of time with the fact that significant time was allocated to the discussion of views held by a minority of participants.

The research team had hypothesised that peer reviewers might have a less positive perception of the process because (i) they were less involved in the work than participants and (ii) the final report did not reflect their views to the same extent as it presented participants’ views. However, peer reviewers’ assessment of the process was similar to that of participants. For example, across the four peer reviewer respondents to the PPEET, only one “strongly disagree” response was received for the 16 positively-worded PPEET statements, and that response was for the PPEET statement “The individuals participating in the project represented a broad range of perspectives on the topic” (potentially related to the fact that peer reviewers had no visibility on the characteristics of other reviewers or participants). Overall, the peer reviewer PPEET responses suggest that members of the public can have positive views of processes even when they have small roles if there is clarity regarding their contributions and compensation from the outset.

This work has various limitations. Foremost, the participants and peer reviewers were experienced Canadian patient and public advisors who are more likely to have knowledge of the benefits, risks, and opportunities associated with data use, sharing, and re-use than other members of the public. Further, participants were generally highly educated (the majority with at least post-secondary education) and older (all but one of the participants was over the age of 35). For these reasons, their perspectives may be different from people with no prior exposure to the topic of health data and from people who are younger, have less formal education, or are members of different ethnocultural groups. It is also possible that the examples of health data uses that participants perceived to be within social licence would be perceived as being outside of social licence by other people, and that there are other uses of health data that are within social licence that were not identified by the participants of our process. It is possible that repeating our process with a different group of people and/or with additional time for dialogue would result in different (and possibly contrary) results. In addition, while participants in our process were not asked to read the systematic review by Aitken et al. (2016) [24], the original list of 7 requirement categories supplied to participants in Step 1 of the process was informed by the Aitken review, which could account for some of the consistency between our findings and the Aitken review.

Though participants discussed the needs and interests of historically colonised and/or marginalised populations, additional work would need to be undertaken to understand health data social licence from the perspective of people and groups that are different from mainstream society because of their ethnicity, abilities, language, gender, and other characteristics. Further, while the participant group included one person who is First Nations, and other people who were not Indigenous that emphasised the importance of Indigenous rights, separate Indigenous-led work would be required to identify the perspectives and concerns of First Nations, Inuit, Métis Peoples and other Indigenous peoples.

The facilitated process and dialogue between participants generated many unanswered questions that would require future work and definition. For example, questions were raised about the practical meaning of certain words or phrases related to health data social licence including what constitutes, “legitimate” uses of health data, or data accessed in a “timely manner”.

Given the small number of people who completed the PPEET, it is possible that the results do not present a complete picture of participants and peer reviewers’ views.