Data Governance and Social Challenges to Sharing Clinical Data for Research: Problems & Solutions

Introductions / Kickoff

Workshop Goals & Scope

• NIH definition of data sharing
• Introduction to the three models for sharing clinical data for research

Speakers: Workshop Co-Chairs: Lisa Federer, Kin Wah Fung, Valerie Cotton

Session 1A: Overview Presentation Lightning Talks

Topics

• Key data governance challenges (e.g., legal, consent, policy)
• Access process
• Lessons learned

Models of Clinical Data Sharing - Lightning Talks

Model 1 ("distributed"): Academic institutions, hospitals and clinics allow shared access to clinical data in a distributed network

• Effort & Speaker: OHDSI (Hua Xu, Yale University)
• Effort & Speaker: PCORnet (Nik Koscielniak)

Model 2 ("repositories"): Researchers extract clinical data and share through established data repositories

• Effort & Speaker: DATA COUNTS (Susan Gregurick, NIH ODSS and Shannon Silkensen, NCI)
• Effort & Speaker: NHGRI AnVIL & eMERGE (Robert Carroll & Anna Lewis)

Break

Session 1B: Overview Lightning Talks continued

Model 3 ("centralized"): Centralized systems specifically designed to integrate clinical data for research use

• Effort & Speaker: All of Us (John Giannini, NIH)
• Effort & Speaker: N3C (Christopher Dillon, NCATS)
• Effort & Speaker: CMS ResDAC /VRDC (Kari Gaare, CMS)

Session 2: Panel Discussion with all Session 1 Speakers

Lessons learned from specific challenges

*** the following topics will be split between the 2 Panel Sessions on Day 1 & Day 2***

Moderated by: Lisa Federer

How each speaker/model addresses these challenges:

• Data access processes - pros/cons of each
• Balancing data sharing & data protections
• Consent or lack of consent for research/data sharing
• Hesitation to share the data
• Fostering transparency and reproducibility
• Other policy/legal or social barriers
• Enclave requirements and impacts of system interoperability
• Splitting multi-modal data from one study across repositories
• Risks/challenges when applying AI/ML cross the data lifecycle

If time: questions will be opened to the audience (in-person & virtual)

Lunch

Session 3: User Perspectives

Users/researchers describe their experiences using, accessing, combining data, or submitting data for Session 1 efforts & models

Users/researchers present

• Effort & Speaker: PCORnet/REACHnet (Model 1) (Tom Carton)
• Effort & Speaker: eMERGE/AnVIL (Model 2) (Adam Gordon)
• Effort & Speaker: N3C (Model 3) Emily Pfaff
• Effort & Speaker: All of Us (Model 3) (Julie Holm)

Q&A with the audience moderated by: Lauren Oliveira Hashiguchi (17 min)

Breakout Session Introduction

Instructions by: Kimberly Thomas

• Groups move to breakout areas, assigned based on preferences collected at registration
• Each breakout group will discuss one Clinical Data Sharing Model

Session 4A: Breakout

Logistics: Each group will designate notetakers, each group assigns someone to report out. Session will not be videocast.

Questions for each group:

• Should NLM support this model - why or why not?
• What is NLM's role in implementing or facilitating this model?

Break to Reconvene in Auditorium

Session 4B: Breakout Groups Report Out

(10 min each, assumes 6 groups)

Introduction by: Kimberly Thomas

• One person from each group reports out

Break

Session 5: Presentation + Discussion

Leveraging TEFCA's federal data exchange/sharing infrastructure to share EHR data with researchers

Speaker: JaWanna Henry, ASTP/ONC

Discussion with audience, moderated by: Kin Wah Fung

Wrap up Day 1 Recap

Workshop Co-Chairs: Lisa Federer, Kin Wah Fung, Valerie Cotton

Welcome & Goals for Day 2

Introductions: Workshop Co-Chair: Lisa Federer

Session 6:

Policies and frameworks that shape clinical data sharing strategy

• Framework & Speaker: Clinical Care Data Research Principles (Dina Paltoo, NLM) (15 min)
• Framework & Speaker: Controlled access data sharing in repositories (Cheryl Jacobs, OSP) (15 min)
• Q&A with the audience moderated by: Lisa Federer (15 min)

Break

Session 7: Continued discussion with all Session 1 Speakers

*** Continue from Day 1 Panel Session***

Moderated by: Valerie Cotton

Continue discussion from Day 1/Session 2 - consider topics raised by other sessions (including breakouts)

If time: questions will be opened to the audience (in-person & virtual)

Session 8: Data governance for participant-level linkage across multiple data sources

Data governance frameworks for sharing, using, and linking participant-level data for research

• Framework & Speaker: Frameworks for Streamlining Decisions about Data Linkage, Rebecca Rosen and Elizabeth Clerkin, NICHD ODSS (15 min)
• Framework & Speaker: A Proof-of-Concept Model to Streamline Processes for Data Access & Linkage, Heather K. Basehore, NCI (15 min)
• Q&A with the audience moderated by: Valerie Cotton (15 min)

Wrap up Day 2

NLM: Reflections and next steps.