3
CASE STUDY: THE UNIVERSITY OF MICHIGAN

Interviewees: Sachin Kheterpal, MD, MBA, Associate Professor of Anesthesiology and Associate Dean for Research Information Technology; Kayte Spector-Bagdady, JD, MBE, Assistant Professor of Obstetrics and Gynecology and Associate Director of Center of Bioethics and Social Sciences in Medicine; Brahmajee K. Nallamothu, MD, MPH, Professor in Division of Cardiovascular Diseases and Department of Internal Medicine and Co-Director of Precision Health

ABSTRACT

An interdisciplinary mindset and drive for a transparent approach to leveraging health data for research and precision medicine have guided the suite of activities of the University of Michigan (U-M). Activities described in this case study encompass a range of synergistic and related innovations, including the development of a data-sharing, decision-making framework; a concise and understandable informed consent pamphlet, and associated governance processes, overseen by the Human Data and Biospecimen Release Committee; and ongoing research, evaluation, and dissemination of best practices in data sharing. The leaders and faculty participating in these cross-institutional initiatives—which include the colleges of medicine, law, public health, public policy, and engineering—seek to apply policy and research advances in real time to the U-M health system, known as Michigan Medicine. This allows U-M to move toward its overarching goals of advancing critical research and supporting precision health (a population-based strat-

egy targeted to discover genetic, environmental, social, behavioral, and clinical markers to improve health outcomes).

Guided by insights from their engagement efforts as well as a commitment to beneficence, respect, autonomy, transparency, and justice, U-M faculty and staff leaders have developed and applied best practices for governing the sharing of biomedical knowledge that goes beyond current US regulatory requirements. Michigan Medicine cares for approximately four million people in its health care delivery system and engages in a wide range of research. Therefore, its data-sharing applications vary in purpose and scope. Moreover, its dual status as both a publicly supported university and a large health delivery system underpin the motivation to continuously act in ways deserving of community trust.

Collectively, the U-M’s data-sharing activities are staffed by more than 20 full-time employees and funded through a combination of federal, state, and private industry grants, university school and department operating budget funds, and foundation awards. Several interrelated activities are described in this case study to illustrate the range of operational, scientific, and bioethical considerations necessary to comprehensively guide decisions related to data use and sharing, and how insights from study participants can assist with developing organizational approaches.

BACKGROUND

Collaborators interviewed for this case study include researchers from U-M’s medical, public health, and public policy colleges who partner on many collaborative research projects and U-M Precision Health initiatives.

The foundational work supporting U-M Precision Health originated by using the participant engagement infrastructure of the NIH Clinical and Translational Sciences Award-funded Michigan Institute for Clinical and Health Research supported by the Michigan Genomics Initiative. U-M leaders invited patients and research participants to provide guidance on proposed U-M data-sharing policies. Several different themes emerged from this work, including that participants placed their trust with U-M broadly, across the various schools and colleges comprising the academic entity; they understood and supported broad data sharing for research and quality improvement within U-M, with or without explicit patient consent; and that sharing data with companies or non-academic entities raised concerns. Specifically, participants expressed discomfort with U-M sharing data with commercial companies (such as an insurance provider or a biotechnology company) without express permission. In order to address this concern, U-M ensures that data and specimens are only shared with commercial companies if the consent form explicitly discloses this. This work utilized an interdisciplinary approach to incorporate participant voices in developing a comprehensive policy approach across schools.

In addition, U-M research teams have conducted many other related qualitative, quantitative, and legal analyses of the gaps and

opportunities in this space, which have served as a foundation for their approaches and are briefly summarized in this case study. First, findings have shown that current laws and regulations do not fully protect the myriad ways that data can be generated, shared, and used in the current age, particularly from an industry perspective (Golbus et al., 2020; Price et al., 2019). Despite these regulatory limitations, focused on protections only for research participants, the vast majority of people wish to be notified if their biospecimens might be commercialized (with a minority being comfortable with such use) (Spector-Bagdady et al., 2018). This makes participant trust and institutional responsibility critical. A clear framework, context, and guidelines are necessary to ensure both timely and non-discriminatory application of machine-learning-based interventions in health care (Wiens et al., 2019). Lastly, these U-M researchers also found that participant altruism and trust in the health system and care providers were both directly associated with believing that people have an ethical obligation to allow one’s health information to be used for research (Minakshi et al., 2020).

U-M collaborators used these findings to inform creation of their data-sharing policy, centering their focus beyond simply obtaining consent to ensuring that patients and providers involved reach a collective understanding of what it means to transparently, responsibly, and ethically steward data and specimens. This feedback informed an opt-in consent system for patients agreeing to share their data within its biobank, and this process has been extended for use to the broader population of patients who access health care services through Michigan Medicine.

DESCRIPTION

The University of Michigan Medical School Office of Research also established a Human Data & Biospecimen Release Committee (Data Release Committee) to develop a decision-making framework to guide questions regarding and processes for sharing of and access to individual-level data and specimens for research (Spector-Bagdady et al., 2020). The committee consists of U-M leaders, including administrators and faculty. Collectively, committee members provide clinical, research, legal, ethical, patient-related, conflict-

of-interest, technical, and industry partnership expertise. Principles guiding the sharing and transfer of data and biospecimens are publicly viewable on the U-M Medical System website, and completion of a prescriptive checklist is required when researchers seek to share data with external for-profit organizations (UMMS, 2019; UMMS, 2018). When data are shared externally, the data use agreement stipulates that the data will only be used for research purposes and that recipients may not attempt reidentification of individuals in the dataset. The checklist differentiates among aggregated versus individual-level data as well as level of identifi-ability (de-identified, limited data set, fully identifiable).

To provide a sense of scale, the committee meets biweekly and uses the standardized checklist to review approximately three projects per meeting. The data-sharing proposals requiring review are identified via a variety of research process portals of entry: U-M’s Institutional Review Board, Office of Sponsored Research Programs, Data Office for Clinical and Translational Research, and departmental research staff. Since its establishment in 2019, the committee has reviewed more than 250 projects. Fifteen projects were rejected due to inadequate disclosure during initial consent or unacceptable terms of use from the industry partner. The interviewees for this case study, recognizing that they were undertaking novel interdisciplinary research that not only informed institutional policy but could benefit others similar institutions, also published their approach to sharing data with industry collaborators (Spector-Bagdady et al., 2020).

To create a consistent and standardized approach to evaluating data-sharing proposals, the Data Release Committee uses a rubric that considers the rationale, scope, data elements, participant consent/authorization, data recipients, and other facets of the data-sharing request. Additionally, U-M’s related data-sharing policies provide a roadmap for researchers to ensure compliance, beginning from the planning for a data-sharing request to post-approval steps by the Data Release Committee (UMMS, 2019). Notably, the rubric covers both retrospectively and prospectively collected data and does not “grandfather in” specimens collected before the new protections of the 2018 revised Common Rule which now stipulates that informed consent forms must include a statement informing

the research participant that their biospecimens (even if de-identified) may be used for commercial profit and asking whether the participant will or will not share in this commercial profit (45 CFR § 46.116[c][7]). The review is focused on transfer of individual patient data, whether identifiable or not, to for-profit or nonprofit companies, foundations, medical specialty societies, or nongovernmental agencies (not to other academic institutions).

Interviewees described the value proposition for this work as more of an underlying motivation to act in trustworthy ways through principled treatment of data and specimens, an approach they further delineated in a 2020 New England Journal of Medicine article, “Sharing Health Data and Biospecimens with Industry — A Principle-Driven, Practical Approach” (Spector-Bagdady et al, 2020). As researchers, they are keenly aware that, given U-M’s position as a public research institution, they are accountable to the citizens of the state of Michigan, and that transparency is imperative. They further acknowledged and seek to learn from past incidents in which patient trust was damaged as a result of data breaches or intentional data sharing without patient consent. They also seek to contribute to the evidence base on engagement and data sharing while simultaneously using that evidence to shape policies and procedures. In essence, acting in good and transparent ways worthy of trust, rather than facilitating rapid data sharing and commercialization, is their guiding imperative. U-M’s concrete policies bolster this imperative, including the Policy for the Transfer of Human Data & Biospecimens to Industry and Non-Academic and Non-Governmental Entities, data-sharing agreements, and an approval process overseen by the Medical School Human Data & Biospecimen Release Committee (UMMS, 2019).

In addition to the fact that they are working at a state university, the U-M researchers have elevated the integral importance of transparency in all of their biomedical research. This is particularly evident in their work to develop a customizable template for a brief, easy-to-understand informed consent pamphlet¹ for participants in studies that use U-M’s biorepository, including the Michigan Genomics Initiative. As one interviewee put it, “If you need 16 pages to explain it, you’re not explaining it well.” The

__________________

pamphlet template uses consistent and unequivocal language to convey that all researchers in U-M will have access to a patient’s data once they consent, and, if data or specimens may be commercialized for profit, participants will be notified during the informed consent process. Further to this, if participants are not notified of potential commercialization during the informed consent process, with a few rare exceptions, U-M will not share their data or specimens with commercial companies. In some exceptional circumstances where the data or specimens are of particular value, such as for rare or orphan diseases, and it is feasible to contact the original contributor, re-consent may be sought.

An important barrier described by interviewees is the current legal and regulatory climate. At present, regulations for the protection of human participants do not cover contemporary or emerging data sharing trends, especially as sources of health data evolve to include data collected by entities outside the purview of the Health Insurance Portability and Accountability Act (HIPAA). While HIPAA provides clear guidance for sharing identified data collected by covered entities and prescriptively defines a limited data set, the 2018 revision to the Common Rule portion of the Human Subjects Research regulations may put protections for data and biospecimens collected from research participants versus patients at odds. For example, as noted above, the new Common Rule includes regulations regarding disclosure of how biospecimens will be used (45 CFR § 46.116[c][7]). But there is no such disclosure requirement under HIPAA for clinical patients. However, the U-M interviewees asserted that current regulations are “the floor, not the ceiling of protections,” and espoused the core belief that academic medical centers should establish best practices for governing the sharing of data and biospecimens with outside entities that go above and beyond current regulatory requirements. These concerns are particularly true for digital health data collected through software applications outside the classic health care provider role.

FUTURE DIRECTIONS

The U-M team hopes to continue its efforts promoting transparency, trust, and consent in its data sharing and research efforts,

and expand its research pursuits to include dialogue and participant engagement in discussions about the balance of individual risk and benefit relative to societal benefits of biomedical research. Future aspirations include better managing the limitations of written informed consent as the primary indicator of engagement and communication, reducing disparities created by demographic (especially race and ethnicity) biases in the recruitment and consent processes, and involving patients as partners on the Data Release Committee. Another area is to continue to engage researchers who use the services of U-M Precision Health to ensure their data collection and sharing practices align with their use of the biobank. By doing so, these policies have the ability to influence the broader researcher communities across campus.

In addition, the interviewees offered advice and encouragement to other organizations interested in following a similar approach. Chiefly, they urged that researchers should focus on participant engagement in what institutions should do instead of just disclosures of what institutions will do. While the human subject research regulations require many types of disclosure, they provide comparatively sparse guidance about the informed consent conversation itself, including optimal practices and essential considerations for the consent process. Finally, the team emphasized that transparency and engagement are meaningful and useful efforts for researchers to undertake, as these aspects support the creation of sustainable relationships with patients, participants, families, and communities. U-M’s research and experience indicate that greater engagement can spur broader, deeper trust that is able to be extended from a single project to an entire research enterprise.