1. DAF Phase 3-Data Access for Research Frequently Asked Questions DRAFT VERSION 4-5-2016

2. DAF Phase 3 Focused, Time Bound Purpose Standards based access (using FHIR and DAF implementations guides) to existing data (starting with the common clinical data set-CCDS) to meet needs of patient centered outcomes research and researchers – Six capabilities defined as functional requirements for three primary actors (Data Source, Data Mart, Query Composer and Query Responder) based on PCORNet model functionality – Will need new FHIR resources for C2 – C6 (want to propose for September 2016) Agile development and demonstration of proof of concept through CDRN pilots Identify policy issues to be resolved and refined into technical requirements (e.g., privacy) Defer tackling new requirements and focus on today’s best practices Avoid building new solutions to problems being addressed in other forums (e.g., privacy, provenance, information model) Future coordination with other FHIR based initiatives once demonstrated PoC

3. DAF Phase 3 Privacy, Security and Related Issues Privacy Policy and Technical Requirements – Learn what CDRNs are doing now: How IRB-defined policies for privacy are maintained/synchronized among all components – FHIR based authorizations, access controls and consents methods are in rapid development – use what is available and swap out when stable Consent – We only want to capture how this is done today, not suggest a standard or implementation mechanism – Note: The OCPO “Patient Choice Technical Project” is working on basic consent, consent for research, and granular consent. Public virtual meeting discussing this project will be held April 12, 2016 at 2:30 pm (ET) https://siframework1.webex.com/siframework1/onstage/g.php?MTID=ec46e9e11c0cd3f07 1359ea5c11d8bec9 https://siframework1.webex.com/siframework1/onstage/g.php?MTID=ec46e9e11c0cd3f07 1359ea5c11d8bec9 Security – Capture what mechanisms are being used today: How security policies, user identification, authentication, and authorization are maintained/synchronized among all components – No technical solution suggested – Advice pilots of Smart on FHIR/HEART projects and evolving OAuth profiles

4. DAF Phase 3 Privacy, Security & Related Issues CONTINUED Privacy and Other HIPAA Considerations – Capture what mechanisms are being used today: How security risks (to confidentiality, availability, consistency) are analyzed & managed among all components – No technical solution suggested PHI – Intent is not to include PHI in Data Mart metadata (not same as patient data metadata) – Type of Data in Data Mart Metadata No PHI or PII – Type of Data in the data mart PHI / PII / LDS / De-identified etc. Provenance and Auditing – Capture into the DAF capabilities C1 thru C6 the existing metadata How audit data are aggregated, e.g., to detect & report security-relevant events among components How patients are informed of data disclosures – Compare with FHIR Provenance and FHIR Auditing – Provide advice to both HL7 and Pilots on how they map

5. DAF Phase 3 Patient Matching & Pseudonymization Patient Matching between the Data Source and Data Marts – IDs are produced and provided by Data Source – Data Marts re-use the ids for communication back with the Data Source – Data Marts can have other ids for their own purposes De-identification Service – It is up to the CDRN and its Data Sources to use /not use the service – See Patient Matching above on how it is supposed to work PCORNet Data Marts based on CDM contain PHI – Some of the 18 PHI (age, DOB, encounter dates) – Minimum necessary when not aggregated result – However, PATID is pseudonymized (mapped to local ID and used as link across DM (potentially network) tables – Mapping and pseudo-identifiers outside CDM and DAF 3 scope – Other DM use of PHI matter of local policy – do not want to lock in design that makes use of individually identifiable data

6. DAF Phase 3 Coordination w/ other “related ONC initiatives” Intent is agile development and pilot implementation to prove FHIR methods for supporting researchers – then coordinate with other initiatives and FHIR profiles eCQM, SDC, etc. – Each initiative has different scope and objectives but may eventually share underlying data and FHIR based standards – FHIR and related development and DSTU are rapidly evolving – Until all are proven and relatively stable, coordination is likely to confuse and slow progress toward actual demonstration of proof of concept and industry uptake HL7 FHIR Resource Coordination – DAF will support the coordination and forwarding the necessary materials working with FMG starting in April as things mature

7. DAF Phase 3 Data Models & Mappings For Data Marts Biased towards PCORNet CDM because that is the most commonly accepted model in PCORNet; Pilots are welcome to use other models – Based on focus to initially support PCORI and assumption that over the next several years most clinical data source systems, e.g., EHRs, will support FHIR resources, profiles and APIs, DAF 3 chose to map DAF FHIR profiles to CDM, widely used by PCORNet DAF will provide FHIR to PCORnet CDM mappings DAF will assist development of other mappings if required by the Pilots Presuming and after successful demonstration of this mapping in the ETL and downstream use/value of FHIR in query and response, we would expect mappings from FHIR to other models to support PCORI as well as other secondary uses (e.g., eCQM)

8. DAF Phase 3 C1: Specific Points C1: Standardize data extraction mechanism from clinical data sources to populate data marts Types of Data Sources – Most predominantly EHR – Others include HIE’s, DWs, Other Data Sources, which satisfy Data Source Functional Requirements for extraction Incremental vs Complete Extraction – Maybe today it is complete, but Data Sources and Data Marts should remember to architect to support a partial list of patients. Metadata about patient data extracted and needed in C2 but not Metadata about the DM which is part of C2 When to run extraction (change to on demand) Source supports FHIR API (for CCDS) and transformed to CDM to be loaded into DM Pseudoidentifier (see slide on Patient Matching & Pseudonymization) FHIR mapping should be extensible in future to other DM data models

9. DAF Phase 3 C2: Specific Points C1: Standardize metadata about data marts, CDRN’s, PPRN’s and data sources LPP has developed table with organizational hierarchy in FHIR style – DM Environment – DM Provenance – DM Governance – DM Data Quality – Data Dictionary (Data Types – DM metadata informed by CDM Harvest Table (v3) See Privacy and C1 discussions of PHI and pseudonymization