Title: Industry Standards for the Electronic Submission of Data to the FDA
1Industry Standards for the Electronic Submission
of Data to the FDA
- Michael A. Walega
- Biometrics Quality Technical Systems
2Agenda
- Why Data Standards
- Overview of Clinical Data Interchange Standards
Consortium (CDISC) - CDISC Working Teams
- Regulatory Implications
- What Standards to Implement?
- Summary
3Why Data Standards?
4Why Data Standards?
We may get the answers, but they dont come easy
!!! No linkage between disparate data streams
No continuity between similar data streams No
ability to employ tools without re-inventing
the wheel
5Why Data Standards?
- Within an organization
- Improve consistency and efficiency
- Enhance critical timings
- Concentrate on scientific nature of data rather
than structure of data - Between organizations
- Leverage synergies
- Sponsor-to-sponsor, vendor-to-sponsor, etc.
- Sponsor-to-Regulatory Agency
6The World Without Data Standards
Central Lab
CDMS Database(s)
Submission Database
Analysis Database
Case Report Form
Medical Imaging
Diagnostics
CDMS Database(s)
Submission Database
Analysis Database
IVRS
Data Transform
Safety
EDC/Web
CDMS Database(s)
Submission Database
Diary
Analysis Database
Lab Systems
???
Data Streams / Databases
7The World With Data Standards
Central Lab
Case Report Form
Medical Imaging
Operations Data Warehouse
Analysis Database
Submission Data Warehouse
Diagnostics
IVRS
Data Transform
Safety
EDC/Web
Diary
Lab Systems
ODM
ADaM
SDS
???
Data Streams / Databases
CDISC
8The Clinical Data Interchange Standards
Consortium
CDISC is an open, multidisciplinary, non-profit
organization committed to the development of
industry standards to support the electronic
acquisition, exchange, submission and archiving
of clinical trials data and metadata for medical
and biopharmaceutical product development. The
mission of CDISC is to lead the development of
global, vendor-neutral, platform-independent
standards to improve data quality and accelerate
product development in our industry.
9CDISC Principles
- Develop standard data models that support the
scientific nature of clinical research - Flexible, easily interpreted regulatory
submissions - Model quality and integrity, independent of
implementation strategy and platform - Global, multidisciplinary, cross-functional teams
- Maximum sharing of information with other groups
- Educational programs
- Not promote any individual vendor or organization
10Where CDISC Fits in the World of Standards
ICH
REGULATORY AUTHORITIES
Japan MHW
US FDA
EU EMEA
Pharmaceutical Industry EU USA
Japan EFPIA PhRMA JPMA
Health Care Providers Pharmacies - NCPDP
CDISC
Clinical Trials
Dictionaries MedDRA, LOINC
Models NCI, OMG
Standards HL7, XML
11CDISC Structure
CDISC Participants
Board of Directors
Alliances
Advisory Boards
Operations and Infrastructure (OIS)
Submissions Data Modeling Working Group (SDM)
Teams
Operational Data Modeling Working Group
(ODM) Teams
Testing/Applications (TAP) Working Group/Teams
Education (EDU)Working Group/Teams
12 CDISC Scope
- Data Sources
- Site CRFs
- Laboratories
- Contract
- Research
- Organizations
- Development
- Partners
- Discovery Data
- Operational
- Database
- Study Data
- Audit Trail
- Metadata
- Submission Data
- CRT Datasets
- Analysis datasets
- Metadata
ODM
SDM
ODM Operational Data Modeling SDM Submissions
Data Modeling
13Operational Data Modeling Working Group
- Mission
- To support the CDISC XML Document Type Definition
(DTD) for operational data interchange and
archiving - To stimulate adoption of the model by industry
stakeholders - To expand the usability, scope and applicability
of the model
14Operational Data Modeling Working Group
- Objectives
- Expand model support for lab data
- Address known issues with Version 1.0
- Support testing and adoption of the model,
including a proof of concept Connect-a-thon at
2001 DIA Annual Meeting - Support for data clarification and queries
- Define migration plans to XML Schemas
15Operational Data Modeling Working Group
- Objectives
- Expand protocol definition capabilities
- Increased interoperability with HL7, etc.
- Extend model to support real-time data
- Increased support for multimedia data types
- Accomplishments
- v1.0 examples available at CDISC website
- Clinical data models and study metadata are close
to full functionality - Expect v1.1 summer 2001, v2.0 2Q2002
16Laboratory Data Team
- Mission
- To define requirements for expanding the ODM XML
DTD to improve laboratory data interchange - To test the CDISC data model with complex real
laboratory data to assure its functionality - To explore other opportunities to standardize
aspects of laboratory data processing
17Lab Team Objectives
- Propose additions and revisions to the ODM Group
for the XML data model to optimally support
laboratory data - Circulate the revised ODM data model for public
review and comment by laboratory stakeholders - Develop a testing plan for the data model
- Evaluate the data model with multiple testing
scenarios involving real data
18Submission Data Modeling Working Group
- Mission
- Define standard metadata models that describe the
structure, usage, content and attributes of Case
Report Tabulations (CRT) and other data sets
submitted to Regulatory Authorities - Models are intended to help FDA reviewers view
data and replicate analyses, tables, graphs, and
listings without requiring complex programming or
data transformations.
19Submission Data Modeling Working Group
- Overall Strategy
- Follow lead of FDA Submission Guidelines
- Regulatory Reviewer(s) primary customer(s)
- Define basic safety metadata standards
- Guide data set organization -- not rigid
structures - 80 of domains and 80 of variables
- Use representative examples, not hard rules
- Allow flexibility for science and sponsor
differences - Begin with 12 safety domains identified in FDA
e-submissions guidance - Post standards openly and encourage ongoing input
by all
20Submission Data Modeling Working Group
- Metadata Strategy
- Organize data sets in folders follow FDA
Guidelines - Define structure of analysis for each data set
- Patient-level, Visit-level, Incident-level,
Item-level - Add common variables to all data sets
- Core, selection, etc.
- Classify variables by
- Origin (where its from) Source, Derived,
Mapped - Role (how it s used) - Key, Selection, Review,
Support - Attributes (what it means) - Labels, Types, and
Codes - Allow sufficient latitude for adding other
variables and domains where scientifically
appropriate
21Submission Data Modeling Working Group
- Documentation Strategy
- Describe the information needed to understand the
domain - Clarify and address anticipated questions
- Provide philosophy behind the domain structure
- Document decisions and use as training material
- Rationale and criteria for the decisions
- Document variables that could collect, code or
predefined
22Submissions Data Standards Team
- Objectives
- Recommend standard vocabularies, codes
dictionaries and other conventions for
representing data - Revise and support the 12 safety domain models
- Publish sample data scenarios as examples for
applying the models - Publish additional documentation on the
assumptions and use of the model - Define new models for PK Data
- Define new domain models for additional safety
and efficacy data sets - Increase interoperability with the ODM XML Model
23Submission Data Standards Team
- Accomplishments
- Published v1.0 of 12 safety domain metadata
models in October 2000 - Publish v1.1 of safety domain metadata models in
July 2001 - Completed sample data scenarios
- Provided models and sample data scenarios to FDA
to aid in preparation of a new CRT Guidance - Defined content for additional model
documentation - Supported analysis of computer-readable metadata
for FDA
24Variables Common to All Data Sets
25Example Vital Signs
- Developed to accommodate planned, actual, elapsed
time collection - Each parameter has a separate variable name
- Could be expanded by adding additional variables
26Example Vital Signs
- Domain name VITALS
- Description Vital Signs
- Location VITALS.XPT
- Structure 1 Record per subject (per
position/qualifier) - Purpose CRT
- Keys USUBJID, VISIT (, POSITION)
- Sort Consistent with structure
27Example Vital Signs
28Example Vital Signs
29SDS Future Objectives
- Expand model details
- Variable definitions
- Vocabularies, Code Lists and Dictionaries
- Models for other domains
- PK data
- Study Domain, Investigator/Site Domain
- Social History
- QOL, therapeutic efficacy data, Hospital
Admission/Discharge/Pharmacoeconomics - Connections/Interoperability with ODM
- Future technical formats (e.g., XML Metadata)
30Analysis Data Modeling (ADaM) Team
- Mission
- To provide metadata models (One PROC Away) and
examples of analysis data sets that are used to
generate the statistical results for a regulatory
submission
31Analysis Data Modeling (ADaM) Team
- Objectives
- Limit scope to examples for primary and secondary
efficacy analyses - Provide samples for different analysis types
- Develop documentation for analysis datasets
- Accomplishments
- Developed draft strawman analysis dataset for
change from baseline for SDS labs model - Developed draft principles statement
32FDA/CDISC Submissions Data Standards Metadata
Meeting
- Held in February 2001
- Meeting Goals
- Increase FDA understanding of CDISC models
- Achieve buy-in and feedback on CDISC models
- Achieve buy-in for incremental approach
- Identify requirements for incorporating CDISC
models into rolling FDA guidelines - Discuss plans for moving forward
- How CDISC can help FDA raise internal awareness
of standards - Joint FDA/CDISC meeting planned for late summer
2001 - Next steps for guidance documents
33Future of E-submissions at FDA
Ref Randy Levin, FDA/CDER Update on Electronic
Initiatives, Guidance and Regulations, Presented
at DIA Clinical Data Management Meeting,
13Mar2001.
34Future of E-submissions at FDA
- XML
- eXtensible Markup Language
- Standardized method for rendering complex
structured metadata/data in text format - Sponsored by World Wide Web Consortium
- Nearly universally adapted by internet community
and software industry for data exchange - Flexible
- Non-proprietary ASCII format
- Extensive support by vendors
- Good match to hierarchical patient data
- Adaptable to data warehousing
35Future of E-submissions at FDA
- XML example
- PatientId 123456789
- EnrollDate 2001Jan01
- Gender Male
- PatientInits HAL
- DateOfBirth 2000Jan01
- RandomDate 2001Jan02
- P001
-
36Regulatory Implications
- FDA Involvement
- Liaisons from electronic submissions groups
within CBER / CDER actively involved - Safety domains internally reviewed
- Guidance re submission of CRT data has been
drafted internal review ongoing - ICH Involvement
- ICH liaisons have been involved
- Data standardization not addressed till 2004
- Supports CDISC initiatives
37Impact on Pharma
- Submissions to Regulatory Agencies
- North America
- Metadata / Data
- CBER/CDER ready for CDISC SDS submission
- Volunteers???
- Impact in early 2002 (SDS), late 2002 (ADaM)
- Rest of World (ICH)
- Data normally not required as part of submission
- No formal guidance re data standards till 2004
(or later?) - Will follow FDA (CDISC) lead
38What to Standardize?
- Regulatory Submissions Data
- Short-term
- Safety / Background Characteristics
- 12 domains identified in FDA NDA e-submissions
guidance - Occur in majority of Phase I-III clinical trials
- Well-defined
- Initially via FDA guidance
- More completely through CDISC SDS model
- Long-term
- Other data domains (Efficacy, QOL, Diary)
39What to Standardize?
- Analysis data
- Long-term
- Highly variable
- Follows development of regulatory submissions
data standards - Safety / Background Characteristics will be first
(CDISC ADaM)
40In Summary...
- Data standards will soon be a reality
- e-Sub Guidance to address standards for
submission of data to FDA - CDISC data models offer an excellent baseline
41Contact Information
- CDISC website www.cdisc.org
- ODM and SDM Working Groups Wayne Kubick
(wkubick_at_enteract.com) - ADaM Team Art Devault (adevault_at_pcyc.com)
- NJ ASA Presentation michael.walega_at_covance.com