340 likes | 750 Views
CDISC Implementation Strategies: Lessons Learned & Future Directions . MBC Biostats & Data Management Committee 12 March 2008 Kathleen Greene & A. Brooke Hinkson, BioMedical Operations, Genzyme Corporation. Agenda.
E N D
CDISC Implementation Strategies: Lessons Learned & Future Directions MBC Biostats & Data Management Committee 12 March 2008 Kathleen Greene & A. Brooke Hinkson, BioMedical Operations, Genzyme Corporation
Agenda • Take you on a journey through time to reflect on Genzyme’s CDISC implementation strategies • We will travel • Back in time to “The Past” • Through “The Present” • Into “The Future” • Questions & Comments
Introduction to SDTM • Submissions • Outsourced end stage SDTM conversion submitted to FDA • Data Operations • Submission Datasets: SDTM-like datasets • Create SDTM-like datasets from raw EDC data • Data collection: Define new CRF standards • Incorporate SDTM variables into eCRFs
July 2005 First SDTM Submission • Motive: Desire to comply with eCTD Guidance • Timeline: September 04 – April 2005 • Provided listing CRTs and SDTM datasets to FDA • SDTM datasets & define.xml never used by FDA reviewers • Outsourced: • Performed end-stage conversion (mapping & creation of SDTM datasets) • Created define.xml & annotated CRFs • Scope: 41 domains; 28 SUPPQUALS
Submissions Lessons Learned • First SDTM submission effort required significant amount of unanticipated Genzyme effort • Valuable lessons regarding implementing SDTM • Detailed knowledge of the study data • Mapping exercise required cross functional team • Interpretation of standard • There are implementation choices; no universal way all companies should implement SDTM • Implementation standards • Genzyme needs to create implementation standards and governance of the standards
SDTM Flight Attempts • Submission datasets: • Create SDTM-like datasets from raw EDC data • Data collection: • Define new CRF standards • Incorporate SDTM variables into CRFs
Why Attempts Fizzled Datasets • Initiative must be cross-functional • Change cannot be made in isolation; must have up and downstream agreement on new processes and deliverables • Did not have infrastructure to work with fully compliant SDTM datasets • Conflicts with project timelines Data Collection • Competing with other initiatives • New version of Clintrial, EDC implementations, M&A’s • Push submission requirements upstream
ODM Experiences • Electronic Submissions • define.xml: submit case report tabulation metadata to FDA • Metadata Driven Study Authoring • Begin establishing libraries of proprietary and non-proprietary eCRFs • Create vendor extensions to ODM • Generate visualizations that mirror EDCvendor’s application user interface & functionality • Import Genzyme defined ODM into vendorstudy architect tools
ODM Lessons Learned Metadata Driven Study Authoring • Make decisions regarding horizontal/vertical specifications • Successfully exchange study metadata (forms and workflow) with EDC vendors • Need infrastructure to successfully utilize tool • Limited reusability of individual study CRF builds across programs • Not just anyone should define studies using the tool • Study modeler should have strong understanding of database design and CDISC SDTM & ODM
Present Environment Caption: The scaffolding took longer to assemble than the rocket
May 2007 Second SDTM Submission • Motive: FDA requested SDTM for all domains • Jan 07 negotiated DM, AE & all SUPPQUALS • Timeline: October 2006 – March 2007 • Provided listing CRTs, CSR and CRFs to FDA in March • May provided DM, AE, SUPPQUALS, define.xml and annotated CRFs • Descriptive documentation of our mapping process • SDTM datasets and define.xml were used by FDA medical reviewer for safety review • Outsourced: • Performed end stage conversion (mapping & creation of SDTM datasets) • Created define.xml • Scope: 2 domains & 2 SUPPQUALS
Lessons Learned • FDA requesting SDTM now!! • Applied lessons learned from 1st experience to 2nd project • Weekly cross-functional meeting with vendor • Output failed WebSDM validation • Validation failures identified at Genzyme • We need to incorporate our submission requirements upstream in data collection • Not efficient implementation strategy to convert data to SDTM so late in the clinical data lifecycle • Creating extra work for stat. programming, stats. and esub • End stage conversion is expensive!!
CDISC Roadmap Purpose • To present a clear and complete picture of: • Where CDISC standards fit into the entire clinical data lifecycle • What activities must occur to integrate the standards into the processes and sub-processes within each lifecycle stage • Provide a common language and reference for further dialog, planning, design, and implementation of CDISC Standards.
Series of Initiatives Build a Complete CDISC Standards Implementation Data Flow #1: Late-Stage Conversion Provide SDTM data to FDA Submit (as SDTM) the collected data on which analysis is based Data Flow #2: Mid-Stage Conversion Data Flow #3: Standards in Collection,Processing & Storage Collect, process & store data according to standards Extend standards-based metadata-driven data flow further upstream into trial design Data Flow #4: CDISC Standards inin Trial Design
Data Flow Strategy • Meet regulatory current requests and soon-to-be requirements as soon as possible • Integrate CDISC standards more broadly and deeply into business processes • Develop clinical data based upon CDISC standards instead of converting the data to CDISC standards • Fully gain operational efficiencies from the use of standards
Metadata Repository • Currently being defined • Manage data about the data • Serves as a central hub for automation of upstream and downstream processes and tools • i.e. protocol & CRF development, SAS TLF programming • Enforces standards • Improves efficiency of process flow • Enables reusability
Data Standards Team & Governance • Data Standards Team is essential to successfully implement CDISC standards • Data Standards Team will develop, implement, maintain, educate, communicate and govern the standards globally • Standards cannot be viewed as optional • Implementation of data standards includes process changes, technology modifications and more subject matter expertise
Triage Team Charter • An interim committee to provide guidance and support to a select number of studies for mapping & programming SDTM datasets • Focus on end & mid-stage conversion activities • Will not be involved with attempts to implement standards at the protocol, CRF or database design lifecycle phases • Will be replaced by the global cross-functional governance body implemented as part of CDISC Roadmap Project
Triage Initiative *Include Clinical, Coding, IT & RA as needed • Initiative began Q4 07 will go through 2008 • Completed 2 reviews so far • Anticipate conducting 10 reviews in Q2 & Q3, with additional studies to be determined in Q4 • Currently considering expanding scope to include review of CRF and database design
Triage Lessons Learned • Process works! • Highlights importance of cross-functional communication • Need additional cross-functional resources to support initiative • Need to operationalize training for new projects going through triage reviews • Implementation questions: obtain outside guidance, when needed
Phase III: Implementation Phase II: Design Phase Triage Review CRF Standards 2008 Parallel Efforts Converge Phase I: CDISC Roadmap
Participation in Standards Activities • CDASH • HIMSS • SDTM Device Sub Team • ADaM Working Group • WebSDM User Group • FDA ODM Pilot • HL7 (Q2 2008) • CDISC User Networks (BACUN)
Future Environment • Visions is evolving • Established standards and governance • Adoption of a growing list of commercially available standards based products • Process improvements enabled by technological advances • Technological and operational infrastructure to support a metadata driven end-to-end clinical data lifecycle
Sample Future Capabilities • Ability to collect, store, analyze/report, compile/submit data to FDA according to SDTM, in conjunction with other CDISC standards • Ability to integrate other, non-CDISC, standards • ODM XML based interchanges of clinical data with vendors (i.e. EDC vendors, labs, FDA, etc.) • Metadata based protocol writing tools that establish the framework for collection, analysis & reporting at the inception of the study design
Questions email: Kathleen.Greene@genzyme.com Brooke.Hinkson@genzyme.com