CDISC Standards End-to-End: Enabling QbD in Data Management Sam Hume <shume@cdisc.org> 1
Shared Health and Research Electronic Library (SHARE) A global electronic repository for developing, integrating and accessing CDISC standards metadata in electronic format. SHARE should dramatically improve the quality, reusability and integration across CDISC standards and controlled terminologies, and improve interoperability with healthcare. SHARE is a requirement for standards-based automation, which is key to ROI from standards implementation 2
How SHARE Impacts QbD Reduces variation Reduces the interpretation conducted by each company Reduce manual programming work needed to apply the standards Increase the rigor of the standards definitions More complete metadata by design Meta-model enforces content and relationships Additional metadata needed for end-to-end model More of the IGs captured as metadata End-to-end process perspective Define-XML up front 3
How SHARE Impacts QbD Reduced cost of poor quality Increased opportunities for automation Reduced re-work needed Higher levels of re-use Tighter controls on standards governance processes Change and version control Increased end-to-end perspective from the start Quality rules Reduces waste All standards metadata in one place (all standards end-to-end) Search and visualization tools 4
Today, numerous standards & documents in various formats - waste 5
SHARE Makes the CDISC Standards Metadata Available from One Place V1.1 V1.2/3.1.2; V1.3/3.1.3, V1.4/3.2 in progress V3.2, plus ISO 21090 datatype 2013-12-20 package, all 5 sets 6
SHARE ISO 11179-based Metamodel Organizes and Constrains the Metadata 7
SHARE Metamodel in Action 1. Concepts for systolic & diastolic blood pressure 3. The values are part of a superset terminology 2. Implemented as terminology values 6. Root terminologies are used to represent variables 7. Variables are included in domains 8. Variables are mapped to one another 4. And subset terminology 5. Which is a version of a root terminology 8
SHARE Relationships Improve Quality 9
SHARE - Describes the Relationships in the Standards Metadata Relationship Relationship Relationship Diagram Dave Iberson-Hurst
SHARE Relationships Improve Quality From CDASH 1.1 no terminology named SCCD Updated with correct terminology From CDASH 1.1 UG no SDTM variable named EXREFID Reference left in, but no related variable 11
Submission and Archival Analysis and Reporting Trial Execution Trial Design and Setup Reducing Variation: End-to-End Data Process View Author Protocol Design Trial Design CRFs Annotate CRFs Setup EDC Trial Trial Registry Generate MP Data Sets Capture CRF Data Capture Lab Data Generate CSR Generate TLF Generate Tabulations and Listings Generate Analysis Datasets Generate SDTM Generate Submission Metadata Generate SR-CRFs Create Submission Data Archive 12
Today, Variation Caused by Lack of Complete Metadata and Tools Scalability, Traceability, Portability, and Maintainability Challenges Author Protocol Design Trial Design CRFs Word Excel EDC Designer Trial Registry Extra processes, code, metadata, and data are required to drive the end-toend workflow. Long-term maintenance of these extras is difficult. Annotate CRFs Adobe Pro Setup EDC Trial EDC Designer 13
CDISC Standards End-to-End 14
SHARE Meta-model: Relationships in Action CDASH variable NCIt Concept (Variable) Part of DM domain Part of CDASH Standard BRIDG & ISO 21090 Mapping Maps to SDTM NCI EVS codelist and codes NCIt Concept (CT code) NCIt Concept (CT codelist code) 15
SHARE Rules Improve Quality Validation in place and/or upon submission Section level indicators Attribute level indicators Hover over notes Hard & soft (warning) failures 16
SHARE Process Improves QUALITY Approve or reject Impact analysis report Workflow to request review Reviewer: Data steward 17
SHARE Machine-Readable Metadata Supports Automation Downloadable exports 18
SHARE MDR Framework Other External Resources (e.g. value Sets) Versioned Standards Operational Collections Research Concepts Integrated BRIDG / ISO 21090 Templates A Versioned Standard is comprised of a set of Operational Collections and associated variables and rules for a specific use case (e.g., SDTM, CDASH) An Operational Collection is a grouping of Research Concepts. There may be additional rules between objects within the Operational Collection. An Operational Collection may be analogous with a SDTM Domain, a CRF, an ADaM data set, or a similar level of operational structure. Basic elements that are the foundation for the content described in the SHARE repository. A Research Concept defines one or more related pieces of clinical data, which is developed using an Integrated BRIDG / ISO 21090 Template. Integrated BRIDG / ISO 21090 Templates use BRIDG classes and relationships to fashion small re-usable patterns that are commonly needed in clinical research BRIDG ISO 21090 data types BRIDG is the foundation model for SHARE that provides classes, attributes and associations use to creates core building blocks. 19
SHARE Benefits Improves the quality and completeness of the standards by enforcing business rules Improves integration towards using the CDISC standards end-to-end, from protocol through analysis and reporting Improves automation through machine-readable standards Improves standards re-usability through improved search Improves standards management through version control, impact analysis, and a complete audit trail 20
R1 Q1 2014 SHARE R1: Initial Capability First version of ishare Implemented using SOA s Semantics Manager Meta-model based on ISO11179, BRIDG and ISO21090/HL7 Datatypes Cloud-based hosted at Amazon Exploring out-of-the-box functionality to develop, maintain and access CDISC standards. Load of Initial set of CDISC standards into SHARE SDTM 1.2 (SDTMIG 3.1.2) CDASH 1.1 BRIDG 3.2 and ISO21090 All CDISC Terminologies New versions (e.g. SDTMIG 3.1.3, 3.2) now being added. 21
R1 Q1 2014 R1: Machine-Readable Standards and Basic Functions Export machine-readable standards ODM v1.3.2 Define-XML v1.0 & v2.0 CSV / Tab delimited / Excel Version control & impact analysis Workflows (e.g. new requests, metadata governance) Reporting (e.g. governance metrics) 22
R2 Q4 2014 R2: SHARE Concept Model Implementation of concept models in SHARE eshare content download site Single, trusted source for all CDISC data standards Load additional CDISC standards content (e.g. SEND) Supports Therapeutic Area standards development Improved Controlled Terminology process integration Value Level Metadata content RDF/OWL export format 23
eshare: Accessing SHARE Content Export formats include: ODM v1.3.2 Define-XML v1.0 & v2.0 CSV / Tab delimited / Excel RDF/OWL Website for easy download of SHARE machinereadable content 24
Thanks! Q & A 25
Strength through collaboration. 26