Ethical considerations for clinical data sharing in craniofacial research: Recommendations from a multidisciplinary working group

Abstract

Objective:

To propose recommendations for ethical, participant-centered clinical data sharing in craniofacial research.

Design:

Series of deliberative multidisciplinary expert working group meetings to develop recommendations.

Setting:

Two one-hour virtual meetings; one all-day hybrid meeting.

Patients, Participants:

Working group (n=16) comprised of individuals with expertise in craniofacial research, bioethics, and patient/caregiver advocacy, as well as lived experience of craniofacial conditions.

Interventions:

The working group first reviewed prior empirical data about research participant attitudes about data sharing then drafted initial recommendations that were built on the data and the group’s collective expertise. Recommendations were iteratively refined until the group agreed upon their final presentation.

Main Outcome Measures:

Working group endorsement of recommendations.

Results:

The working group produced 16 recommendations that addressed considerations for primary and secondary researchers, data repositories, and the craniofacial research community across five domains. These domains address: (1) research team communication with participants, (2) data collection and protections, (3) data governance, (4) education for researchers, and (5) remaining research gaps. Recommendations highlight the importance of prioritizing the experiences of those with lived experience of craniofacial conditions in decision-making about data sharing, navigating varied perspectives on privacy protections for facial images, and striving to implement trustworthy data sharing and governance practices.

Conclusions:

This summary of recommendations offers guidance for the craniofacial research community to advance participant-centered clinical data sharing practices. Ethical data sharing that accounts for participants’ experiences and values has potential to advance scientific research and improve outcomes for individuals with craniofacial conditions, their families, and their communities.

Introduction

Sharing clinical research data has the potential to accelerate scientific research by making datasets accessible to future researchers to address a broad range of meaningful questions, reproduce prior findings, and enhance transparency of the research process.^1,2 Clinical research that aims to improve care for individuals with craniofacial conditions and their families may include a variety of data types and sources, including: health histories derived from electronic medical records, psychosocial well-being and treatment experience collected prospectively from patient/caregiver-reported survey or interview data, and medical images and/or facial photographs. Prior exploratory interviews with participants in clinical craniofacial research highlight the ethical complexities of sharing data in this context, where the potential for scientific advancement must be balanced with respect for participants’ preferences and preservation of participant privacy, particularly when sharing data that may include facial images.³

There is a growing recognition of the value of data sharing, reflected in recommendations from organizations such as UNESCO⁴ and data sharing policies from both commercial and non-commercial funders.⁵ For example, recent U.S. federal policy changes encourage increasingly broad sharing of clinical research data funded by the National Institutes of Health.⁶ However, detailed guidance for how to manage varied types of data—particularly sensitive data such as facial images that may raise unique privacy or other ethical considerations—are needed. Considerations for ethical clinical data sharing have been examined in clinical trials⁷ and qualitative research.⁸ More specific guidance has been developed for research that raises unique sets of ethical questions, such as research during the COVID-19 pandemic⁹ and research with individuals with dementia.^10,11 However, there are no existing recommendations that specifically address clinical data sharing in craniofacial research.

This paper aims to address this gap by proposing a set of recommendations to guide the ethical practice of clinical data sharing in craniofacial research. Prior qualitative interviews with individuals with craniofacial conditions and their caregivers³ were incorporated, alongside multidisciplinary expert working group members’ collective expertise in clinical research, bioethics, and craniofacial conditions, as well as lived experience. This paper describes the development process, summarizes the resulting recommendations, and discusses remaining questions and implications for practice.

Methods

Overview and objective

The goal of this work was to identify recommendations for best practices pertaining to ethical sharing of clinical data in craniofacial research through a series of deliberative working group meetings. Working group activities aimed to ensure robust engagement and partnership in developing these recommendations among representatives from the community of individuals with lived experience of craniofacial conditions as well as from multiple relevant research disciplines.¹²

Composition of working group

The research team, which included a craniofacial investigator and pediatrician, a bioethics investigator, and a craniofacial research manager, convened a multidisciplinary expert working group comprised of representatives of perspectives relevant to clinical data sharing in craniofacial research who were actively engaged in these topics. The core research team first identified categories of individuals who would potentially have an interest in data sharing policies, which included: individuals and parents/caregivers with lived experience of craniofacial conditions; advocacy organizations focused on craniofacial conditions; cleft and craniofacial researchers from different disciplines such as surgery, pediatrics, speech-language pathology, and psychology; and bioethicists with expertise in the broad ethics of data sharing. The core research team then invited individuals who could speak to the perspectives of one or more of these groups. Working group members were previously known to one or more individuals on the core research team, which included investigators with active research programs in craniofacial and bioethics research. Working group members (n=16, including the core research team) and their relevant areas of expertise are shown in Table 1.

Table 1.

Working group members and expertise

Name	Relevant expertise
Kathleen Bogart, PhD	Research psychologist; lived experience
Canice Crerand, PhD	Craniofacial research advisory committee member; psychologist; researcher
Kelly Evans, MD	Craniofacial research study team member; craniofacial pediatrician and researcher
S. Malia Fullerton, DPhil	Bioethicist; researcher
Carrie Heike, MD, MS*	Craniofacial research co-principal investigator; craniofacial pediatrician and researcher
Alexis Johns, PhD	Craniofacial research advisory committee member; psychologist; researcher
Erica Klauber, MBA	Advocate; lived experience
Sara Kinter, PhD, MA, CCC-SLP	Craniofacial speech-language pathologist and researcher; epidemiologist; lived experience
Stephanie Kraft, JD*	Bioethicist; lawyer; researcher
Christy McKinney, PhD, MPH	Craniofacial research advisory committee member; epidemiologist
Amy Mendillo, MM, MPP	Parent advocate; health writer
Alexis Prepotente, BS	Craniofacial research study team member; lived experience
Amy Schefer, MEd	Craniofacial research advisory committee member; caregiver; parent advocate
Nicola Stock, DPhil	Craniofacial research co-principal investigator; research psychologist
Laura Stueckle, MPH*	Craniofacial research study team member; research manager
Melissa Tumblin, BA	Craniofacial research advisory committee member; caregiver; parent advocate

^*indicates working group member was part of the core research team

Process

Before the working group activities began, the core research team conducted a qualitative interview study to understand the views of individuals with craniofacial microsomia and their caregivers about clinical data sharing in craniofacial research. These interviews were conducted with individuals who had previously participated in research with the team. Five themes were identified, related to (1) hope to benefit the craniofacial community, (2) recognition of the importance of sharing images paired with discomfort for some, (3) general positive views on data sharing with concerns focused on harm to the craniofacial community, (4) the importance of trustworthy researchers and data protections, and (5) the complexity of pediatric decision-making. Results from these interviews are published elsewhere.³ Because the goal of this process was to develop recommendations that centered lived experience of craniofacial conditions, these themes served as the framework on which the group would build recommendations. Rather than building on a previously developed ethical framework, this inductive approach ensured participant perspectives were at the core of the process.

The core research team hosted a series of hour-long web-based meetings with the broader working group to: (1) introduce members of the group to each other and orient them to the project, including an overview of data sharing in clinical research and an ethical case for the centrality of participant views to evaluate risks and benefits, and (2) review the interview data. Finally, the team held an 8-hour hybrid virtual and in-person meeting to (3) develop recommendations through group deliberation. The latter meeting’s agenda and activities were informed in part by deliberative engagement methods, an approach intended to support incorporation of the group’s diverse areas of expertise into the synthesis of practice recommendations.¹³ Meeting activities were also designed to promote best practices for team science, including starting with an articulation and discussion of the project’s goals, incorporating team-building to promote a culture of trust, and providing transparency around authorship criteria,¹⁴ with input from an expert in team science methodology to optimize an agenda that would foster interdisciplinary collaboration. The agenda for the all-day meeting included: introductions; didactic presentations on craniofacial research, data collection, and key ethical issues arising in this context; discussion of the implications of each theme from the interview data; brainstorming about key topics for recommendations; and development of specific recommendations. The meeting used a combination of small breakout groups and large-group discussions to identify topics and develop recommendations. Small groups each discussed a key topic then reported out suggested recommendations to the full group, which provided feedback and additional suggestions. A facilitator from the core research team then represented the full list of suggested recommendations for continued group discussion. Group comments were incorporated into the list during discussion, and the day concluded with a draft list of recommendations with the group’s preliminary endorsement.

The core research team subsequently revised the list of recommendations, maintaining the key points from the group meeting and re-organizing to provide clarity and detail as needed. The revised draft list of recommendations, along with details about their interpretation and application, was then shared with the full working group for asynchronous review and feedback. Working group members provided written comments on the draft recommendations, and the lead author revised them to incorporate all group feedback. The revised list of recommendations was again sent to the group for additional feedback and was iteratively refined until all group members agreed upon the final version presented here. Figure 1 illustrates the recommendation development process.

Figure 1.

Steps in recommendation development

Results

Overview

The working group developed 16 recommendations across 5 broad categories: (1) research team communication with participants, (2) data collection and protections, (3) data governance, (4) education for researchers, and (5) remaining research gaps (see Table 2). Each recommendation statement reflects the full group’s endorsement and is followed by a summary of the key considerations that informed the recommendation. In some cases, considerations include suggestions for implementation that may only be applicable in certain circumstances.

Table 2.

Summary of recommendations and key responsibilities for clinical data sharing in craniofacial research

Domain/recommendation		Responsible party	When to implement
1. Research team communication with participants
1a	Communicate clearly about data sharing during the informed consent process	Primary researchers	Protocol development Informed consent
1b	Specifically address the possibility of sharing photographs and other sensitive data	Primary researchers	Protocol development Informed consent
1c	Discuss the possibility that participants’ views on data sharing may change	Primary researchers	Protocol development Informed consent Longitudinal engagement
1d	Share direct impacts and/or findings based on data contributions with participants	Primary researchers Data repositories	Protocol development Longitudinal engagement
2. Data collection and protections
2a	When available, use standardized measures to facilitate interoperability among datasets	Primary researchers Craniofacial research community	Protocol development Ongoing professional development
2b	De-identify data at the earliest reasonable opportunity	Primary researchers	Protocol development Data collection and preparation
2c	Consider heightened protections for sharing particularly sensitive data	Primary researchers	Protocol development Data collection and preparation
3. Data governance
3a	Implement clear standards for data interoperability	Data repositories Primary researchers	Data governance
3b	Implement robust security practices	Data repositories	Data governance
3c	Control researcher access to datasets	Data repositories Secondary researchers	Data governance
3d	Ensure data uses align with the intentions of data contributors	Data repositories Secondary researchers	Data governance Secondary data use
3e	Involve craniofacial community advisors in data repository governance	Data repositories	Data governance
4. Education for researchers
4a	Learn about craniofacial research participant values and expectations for data sharing	Craniofacial research community	Ongoing professional development
4b	Engage with craniofacial community concerns about data sharing	Craniofacial research community	Ongoing professional development
5. Remaining research gaps
5a	Better understand who is represented and who is missing from existing craniofacial research datasets	Craniofacial research community	Future empirical research Ongoing professional development
5b	Explore participant perspectives on artificial intelligence in craniofacial research	Craniofacial research community	Future empirical research Ongoing professional development

1. Research team communication with participants

Several of the working group’s recommendations related to optimizing communication between research teams and participants, both before and after participants contribute data.

1a. Communicate clearly about data sharing during the informed consent process

The working group highlighted the importance of clear communication about data sharing during the informed consent process when people decide whether to enroll in clinical research so that they know what data will be shared and in what manner. Suggestions for how to promote clear communication included using plain language, multimedia supports, and/or infographics in consent/assent forms or participant information sheets (or as supplements to institutionally required forms), as well as encouraging bidirectional conversations with prospective participants and/or their families when possible. The group recognized that data sharing may not be a primary consideration for a person—often a caregiver of a child with a craniofacial condition—who is considering enrollment in clinical research, nor would its implications be immediately apparent to those unfamiliar with data sharing for scientific studies. To address these challenges, a small number of targeted teach-back questions could be adopted by research staff to affirm understanding; these should be implemented judiciously to ensure that this practice meaningfully supports participant understanding and satisfaction. Additionally, the consent form could include space for the participant to specifically attest to their understanding of data sharing practices. Adding lengthy information to the consent process, however, should be avoided so as not to overload prospective participants and counterproductively undermine comprehension. These potential unintended consequences should be monitored.

1b. Specifically address the possibility of sharing photographs and other sensitive data

When photographs are collected as part of the research, the group further suggested that researchers could consider offering example illustrations of how photographs would be presented when shared to provide clarity for participants. If certain uses of images would not be allowed under any circumstances (e.g., using research photographs for advertising), researchers could state that these uses would be outside the scope of research data sharing without the participant’s express consent. The group also suggested that researchers might consider a tiered approach to allow participants to opt out of sharing photographs or other particularly sensitive information, such as audio/video recordings, at least in study contexts where other research data provide scientific value independent of the photographs. Importantly, researchers considering using a tiered approach should recognize that this could introduce opportunity for confusion and monitor the impact on participants’ understanding. Tiered approaches also create added responsibility for research teams to ensure that an individual’s choices are honored, which may be challenging for large studies, studies that collect data over many years, and/or those without a dedicated database manager. Researchers should carefully consider these implications before pursuing a tiered approach.

1c. Discuss the possibility that participants’ views on data sharing may change

The working group recommended that during the consent process researchers acknowledge the possibility that some participants’ views about data sharing may evolve, over time as well as across developmental stages, and clarify in what circumstances data can or cannot be withdrawn. These conversations could include acknowledging emerging technologies (e.g., artificial intelligence) that may be used to analyze data in the future but the implications of which are unknown at the time of enrollment. Some members of the working group suggested that researchers could consider whether advances in technical capabilities for data analysis might be significant enough to warrant re-consent, but they also acknowledged that planning for such a scenario would be difficult. Discussing the possibility of evolving views is particularly relevant for individuals who are enrolled as children, whose autonomy interests are expected to develop as they age. If the research team intends to re-consent minor participants when they reach the age of majority (and/or obtain or re-obtain assent at certain points before the age of majority), they should develop a plan for the re-consent process and communicate that plan to enrolling participants. Plans for re-consent should include what will happen if a participant cannot be reached to re-consent.

1d. Share direct impacts and/or findings based on data contributions with participants

The working group recommended disseminating information about ongoing and completed research to participants whose data contributed to those findings. This could be accomplished by creating and distributing plain language summaries of research findings and their implications. Because sharing information about the specific studies in which a participant’s data are used may not be feasible once data have been fully deidentified, research teams could consider reporting metrics such as overall use of datasets and general advancements in the field. Data repositories could support this approach by creating mechanisms to share information with participants and other interested parties, such as opt-in newsletters or public-facing websites.

2. Data collection and protections

The working group made three recommendations for primary researchers to consider as they collect and prepare clinical craniofacial data for sharing. These recommendations focused on maximizing the benefits and minimizing the harms of data sharing.

2a. When available, use standardized measures to facilitate interoperability among datasets

The working group emphasized that data interoperability is essential to realizing the benefits of data sharing. To ensure that participant data are optimally usable and interoperable across datasets, researchers who collect primary data should use standard measures to the extent possible. Within the craniofacial research community, this standardization could be advanced by creating working groups (including those with lived experience as well as researchers and other interested parties) to identify the most appropriate measures and make recommendations for widespread use through existing data sharing repositories such as FaceBase.¹⁵

2b. De-identify data at the earliest reasonable opportunity

The working group recommended taking steps to minimize the privacy risks posed by a potential data breach, including removing overt identifiers from study data at the earliest opportunity that is scientifically and ethically appropriate and aligned with institutional requirements. Depending on the study protocol, this may mean removing participant identifiers after participant contact is complete or deleting original audio recordings after interview transcripts have been verified. Taking these steps would help to minimize the potential harms from a data breach.

2c. Consider heightened protections for sharing particularly sensitive data

The working group, drawing in part on prior interview findings, pointed out that particularly sensitive types of data (e.g., facial images or photographs, audio/video recordings, some qualitative data) may include information of an especially personal nature. Group members further noted that there may be an increased risk of identifiability from sharing information about individuals with especially rare conditions, particularly those with multiple intersecting marginalized identities, even with direct identifiers removed. To address these concerns, the group recommended that researchers consider, with the input of those with lived experience, whether any of the data they are collecting warrants a heightened degree of sensitivity and to take additional steps to protect data in these cases. For example, if sharing the full dataset would significantly increase the likelihood of identifiability and/or harm from breach of sensitive information, researchers might share study artifacts or data summaries but maintain the full dataset with the original research team. Researchers should take special care to balance the potential harm from sharing sensitive data with the added scientific benefit of sharing a full versus limited dataset. In assessing this risk-benefit balance, it is important to understand what protections are in place in the repository where data will be shared. This evaluation should take place early in the research process as decisions about study design are made and revisited as risks and benefits evolve. Researchers should also communicate these considerations with sponsors who implement data sharing requirements to ensure awareness of potential harms to participants.

3. Data governance

The working group discussed the importance of governance within data repositories to ensure the appropriate use of craniofacial research data and proposed several recommendations.

3a. Implement clear standards for data interoperability

As noted above, interoperability between datasets is essential for scientific data sharing to produce the benefits it promises. The working group recommended that data repositories set standards for those depositing data to facilitate interoperability. For data repositories specific to craniofacial research, guidance from the clinical craniofacial research community on appropriate standards will be important. Partnership with data scientists, analysists, and others with expertise in informatics will be critical to develop and implement standards.

3b. Implement robust security practices

The working group recommended that data repositories use up-to-date security and encryption technologies. Group members suggested that repositories could consider federated or cloud-based models that restrict downloading certain data to minimize the potential for data breaches. Engaging with cybersecurity experts will be essential to ensure repositories are developed and maintained with appropriate attention to current security standards.

3c. Control researcher access to datasets

The working group recommended that craniofacial data repositories limit access to credentialed researchers with clear research goals (e.g., gleaned via review of a research proposal), and to consider requiring researchers to undergo a training process that addresses issues specific to the craniofacial community. The group also emphasized that data access committees should include a representative group of members with appropriate scientific expertise and lived experience, and that they should use a clear framework for decision-making to provide a fair and transparent process.

3d. Ensure data uses align with the intentions of data contributors

The working group emphasized the importance of ensuring that secondary research uses are not misaligned with the intention of the original data contributors, as such uses could breach the trust between participants and the researchers with whom they had a direct relationship. The group recommended that data repositories should be designed to allow access to the original consent language that participants agreed to at the time they contributed their data, as well as information about the original study’s goals. Clearly stating the terms of the original informed consent will ensure secondary researchers are aware of what participants agreed to so they can verify their research is not in tension with the goals of the original study. Repositories could require secondary researchers to attest to their agreement. Additional approaches could include requiring secondary researchers to consult with relevant experts before accessing data or inviting a member of the study team that initially collected the data to participate as a co-author on any manuscripts produced using the dataset they provided.

3e. Involve craniofacial community advisors in data repository governance

The working group recommended that community advisors, including individuals with lived experience of craniofacial conditions, should have a voice in repository governance alongside experts in fields such as clinical research and data science. Community advisors can contribute to oversight of governance structures and data access requests to identify potential harms and opportunities to improve benefits from the perspective of individuals with craniofacial conditions and their communities. The working group emphasized that the perspectives of researchers may miss important considerations from participants’ perspectives that community advisors can identify and help address in governance decisions.

4. Education for researchers

The working group emphasized the importance of disseminating information about patient-centered data protections to the craniofacial research community. The group made two specific recommendations for researcher education.

4a. Learn about craniofacial research participant values and expectations for data sharing

The working group discussed the importance of sharing information about participant values and expectations for data sharing with the craniofacial research community. This recommendation was premised on the assumption that understanding which benefits are most impactful from participants’ perspectives would encourage greater alignment between research agendas and participant values. Partnerships between researchers and advocacy organizations could facilitate sharing participant testimonials focused on why people participate in clinical craniofacial research and any specific outcomes (e.g., treatment options, psychosocial support) that they hope to improve.

4b. Engage with craniofacial community concerns about data sharing

The group recommended the development of, and researcher engagement in, trainings that address the importance of data protections in connection with historical and present-day concerns that many individuals and communities have about research, particularly as they relate to craniofacial conditions, living with visible differences, and conditions associated with social stigma. These activities should be a part of researchers’ efforts to keep up to date with best practice recommendations for clinical data sharing.

5. Remaining research gaps

The working group recognized the importance of ongoing research to inform best practices for scientific data sharing and identified two sets of questions in need of additional empirical research.

5a. Better understand who is represented and who is missing from existing craniofacial research datasets

Recognizing the limitations of existing data, the group recommended further efforts to study why people may not have access to or choose not to participate in clinical craniofacial research, as well as characterizing studies’ inclusion/exclusion criteria and identifying who may be left out (e.g., individuals with medical complexity). Such research would allow for better understanding of the clinical characteristics, sociodemographics, and perspectives of those who are not included in current craniofacial research datasets, which could inform recommendations for addressing missing data. Addressing such systematic limitations of datasets is important to ensure that data sharing produces meaningful benefits for all individuals with craniofacial conditions. Additionally, better understanding barriers to participation and comfort with data sharing may reveal other strategies for improving research practices.

5b. Explore participant perspectives on the use of artificial intelligence in craniofacial research

The working group identified a need to explore participant perspectives on the value and risks of using emerging technologies like artificial intelligence and machine learning in craniofacial research. Working group members identified artificial intelligence technologies as potentially valuable for analyzing facial structure and other clinical data, noting that these technologies have potential to accelerate analysis and interpretation of data, including photographs, which could increase the potential benefits of large datasets. However, the group lacked sufficient information about the impact of these emerging technologies on study participants at this time to make practice recommendations on this topic.

Discussion

This project aimed to produce recommendations to offer guidance for the craniofacial research community as they navigate ethical challenges related to sharing clinical research data. Some of these recommendations expand on practices that are already broadly adopted or required in clinical research, such as communicating about data sharing during informed consent¹⁶ and ensuring data are interoperable and in line with FAIR Data Principles,¹⁷ providing additional nuance and suggestions for improvement in meeting participants’ needs. Other recommendations highlight future directions for the field that, in the working group’s view, would enhance ethical participant-centered research and clinical data sharing. Further, while all of these recommendations reflect important considerations, each one can stand alone as a valuable step toward advancing these goals.

A central consideration for the development of these recommendations was the prioritization of the experiences of people with lived experience of craniofacial conditions. This priority was addressed by gathering the original data that were incorporated into the working group’s deliberations,³ composing the working group, and developing the recommendations. Inclusion of those with lived experience of the condition under study is a central tenet of community-engaged research¹⁸ and is critical for developing health policy approaches that incorporate people’s real-world experiences and promote outcomes that are relevant to the community of patients and families.^19,20 It is particularly important to ensure that data sharing practices and governance structures incorporate community perspectives because all future uses of data cannot be defined at the time of individual informed consent. Data governance informed by individuals, families, and communities with lived experience therefore serves an essential role to ensure that secondary research minimizes harms and maximizes benefits to those who will be most directly affected.^21,22 Future work should continue to engage with children and families in the craniofacial community, with an emphasis on those whose perspectives are less well represented in prior research, to ensure governance practices adequately address all affected individuals and evolving real-world concerns.

There are also opportunities for further engagement with children and adolescents to learn more about how their attitudes toward data sharing evolve as they age and as real-world data use and sharing changes over time. For example, as analysis techniques continue to advance, would minors set different boundaries on data use as compared to their caregivers, and how ought any differences be accounted for in data governance? Research participants’ autonomy interests are typically protected through the informed consent process.²³ When potential participants are minors, their parents or a legally authorized representative take on the responsibility of making decisions about participation, and minors who are old enough to engage in an assent process join in decision-making.²⁴ The extent to which the approaches described in these recommendations adequately account for children’s developing autonomy interests as they are increasingly able to participate in decision-making could be further explored. Sharing data for future research raises challenging questions such as whether and how to engage with children as they mature and become more able to engage with decisions about data sharing over time. Prior work on these questions in the context of pediatric biobanking has highlighted the need to balance ethical considerations including children’s developing autonomy interests, the value of the research to society, privacy risks, and respectful engagement with participants.²⁵ As the recommendations offered here are implemented, these ethical implications should be carefully examined.

One particularly important challenge for clinical craniofacial research is ensuring appropriate privacy protections when sharing facial images or photographs of individuals with visible facial differences. Medical researchers²⁶ and journal editors²⁷ alike have acknowledged how difficult it to sufficiently anonymize facial images and highlighted the importance of individual choice in this context. In a previous interview study, individuals with craniofacial differences and their caregivers expressed varied perspectives on their comfort with having research photographs shared beyond the original study and what would constitute adequate deidentification.³ A prior survey of parents of children with and without congenital conditions similarly reported variability in preferences for privacy protections of facial photographs in research but found overall that parents of children with congenital conditions, as compared to those without, were more comfortable with the collection of facial photographs for research.²⁸ These perspectives may reflect a recognition of the scientific and social value that these visual data sources can provide. This raises critical ethical questions about the extent to which the value of future research for society can justify risks to individual participants.²³ Some scholars have suggested that the potential value of large datasets to the communities they stand to benefit may outweigh risks of identifiability, provided that the community deems the protections in place to be adequate and trustworthy.²⁹ Continued empirical and ethical exploration of this issue will be important to assess the likelihood of re-identification, impact of resulting harms, and overall balance of risks and benefits on both an individual and community level.

Further enhancing the social value of data sharing, defined as its potential to improve health and well-being,²³ is another important consideration reflected in several working group recommendations. A review of ethical guidelines for international data sharing found that social value is supported by factors including the quality of the data, protocols used for analysis, and accessibility and interoperability of datasets.³⁰ Balancing the potential benefits and risks on both a community and individual level will require careful consideration of justice implications to ensure that risks are not disproportionately experienced by groups that are already disadvantaged. As clinical data sharing for craniofacial research moves forward and analyses increasingly incorporate technologies like artificial intelligence,³¹ it will be important to take into account these varied perspectives and for researchers to approach sharing of visual data sources thoughtfully. For example, researchers can develop their approaches by building on existing frameworks for the trustworthy use of artificial intelligence in patient-centered research.³²

Finally, an essential consideration that spans across our recommendations is the role of prior trust among research participants in data sharing decisions and the ways that data sharing can affect trust across both clinical and research realms. Clinical research participants may have prior clinical relationships with members of the primary research teams who initially enrolled them in research. This prior clinical trust can influence trust in research and enrollment decisions, particularly when the clinical and research teams are based at the same institution.³³ When data are shared with unknown secondary researchers, participants rely on the primary research team’s assurances that future research will be conducted in a trustworthy fashion that aligns with the expectations that were set at the time of enrollment. Breaches of this trust could cause relational harm between the participant and broader research enterprise, the primary research team, and the clinical team. Consequently, trustworthiness must be a central feature across the spectrum of data collection, sharing, and governance practices.³⁴

The recommendations described here reflect the working group’s deliberations, but there may be additional topics that were not addressed or nuances that would benefit from further detail. For example, as others have recommended,³⁵ engagement with ethical issues in information technology will be essential for ongoing data governance. The working group drew on prior empirical interviews to frame its recommendations rather than on an existing ethical framework, which centered participant experiences but may have limited consideration of issues that did not arise in either the prior interviews or the working group activities. While the team sought to include a diverse range of perspectives in the development of recommendations, there may be divergent viewpoints that are not reflected in this work, for example from families or researchers outside the U.S. Many, though not all, working group members had previously worked together to some degree and this may have limited deliberations.

The working group welcomes refinements to these recommendations and expects that these will continue to evolve in response to changes in technological capabilities, the funding landscape, and societal research priorities. Formal consensus development methods, such as a Delphi approach,³⁶ could be a valuable next step to refine the recommendations produced through the deliberative approach described here. Further, future work should evaluate the dissemination and implementation of these recommendations, including reach to and acceptance by key audiences such as community partner organizations and research oversight bodies, as well as any unforeseen impacts that may warrant reconsideration. As these recommendations are implemented, it will be important to evaluate their impact on the ethical conduct of research, including risks and benefits to participants as well as broader public health ethics considerations such as community benefit, trust, and accountability that are critical for research involving health data.³⁷ Ongoing empirical and normative work will enhance understanding of whether and how data sharing practices guided by these recommendations enhance the rights and welfare of participants and the broader craniofacial community.

Conclusion

A multidisciplinary working group developed a set of recommendations for ethical clinical data sharing in craniofacial research, based on empirical data and on the group’s collective clinical, research, and personal experiences. The resulting summary offers guidance for the craniofacial research community to learn from and build upon as they collectively work to advance participant-centered clinical data sharing practices and scientific outcomes for individuals with craniofacial conditions, their families, and their communities.