150 likes | 379 Views
caBIG Workspace Clinical Trial Management Systems. Washington, D.C. February 19, 2004. C3D Requirements (in no particular order). Rapid study implementation using Library of template CRFs based on caDSR CDEs and EVS terminology Patient centric view of clinical data via C3PR
E N D
caBIG Workspace Clinical Trial Management Systems Washington, D.C. February 19, 2004
C3D Requirements (in no particular order) • Rapid study implementation using Library of template CRFs based on caDSR CDEs and EVS terminology • Patient centric view of clinical data via C3PR • Select patient by name • Longitudinal view of patient data across studies • Eliminate duplicate data entry & maintain data integrity • Ensure data security and patient confidentiality • Web based remote data entry • Automated clinical (lab) loading process • Automated lab grading according to CTC • Automated generation/extraction for electronic submission to monitors and regulators • Robust data cleaning • Data entry validation • Post entry discrepancy management
Library of Template CRFs Built with CDEs • Forms and CDEs defined in caDSR • Template CRF layouts defined as PDF
Oracle Clinical Facts (or Why Oracle Clinical?) • Fully web-based internet architecture: Study definition & Data collection • Built in test environment – eliminate need for redundant development • Build study once • Test/train simultaneously with production data entry • Best of breed COTS solution • Used by 8 of top 10 pharma (Merck, J&J, Glaxo, Pfizer, Bristol, Roche …) • Used by 75% of CROs (Quintiles, PPD, Omnicare, ICON …) • 21 CFR part 11 compliant • Username and password protected • Complete time (and user) stamped audit trail including (soft) deletes • Verbatim data collection supported • Study, site, role, user level security • Standards enforcement via Global Library • Questions and Discrete Value Lists must be defined here before use in study
Oracle Clinical Facts - Continued • Flexible data entry from single form definition (PDF) • Electronic PDF data entry • Hand entry on printed PDFs • Batch data loading from electronic sources • Published Workflow, Image and Data Capture APIs • Interface for “Programmerless” validations, edit checks and derivations • Point of entry validation: bounds, LOVs, data type, precision, etc.. • On-line validation: multi question checks preformed on commit by form • Off-line validation: run nightly, complex validation (multi question, form, visit) • Robust discrepancy management • Discrepancy reports categorize by site, investigator and/or CRF • Data changes sensed and validations rerun to automatically resolve discrepancies • Negotiations/partnership with Oracle allow extension of HHS 80% discount to cover: • caBIG participants • Specialized Programs of Research Excellence (SPORES) • North American Brain Tumor Consortium (NABTC) • American College of Radiology Imaging Network (ACRIN)
C3D Limitations • Built around proprietary technology – Oracle, 9iAS, Forms, PL/SQL • NCICB is pushing Oracle to provide complete metadata APIs • Currently not supported on Apple Mac OS • No JInitiator for Mac • All data stored as character – limited to 200 characters • No database level support for data types • Date and Numeric type checking at data entry – can be overridden by “verbatim” data • No notion of HL7 – yet • Oracle newly released Healthcare Transaction Base (HTB) • HTB persist HL7 v3 XML messages in Oracle relational DB • NCICB is considering pilot project with Oracle to “integrate” HTB and caCORE • NCICB working with Oracle to define integration of HTB and OC
C3D Integration of caDSR Metadata with OC Global Library • Questions asked on OC Study CRFs are restricted by metadata specifications contained in the Global Library (GLIB) • NCI has its own metadata repository in the caDSR • OC has been integrate with caDSR metadata via open source scripts and procedures caDSR ObjectGLIB Component • Common Data Elements Questions • Enumerated Value Domains Discrete Value Groups (DVG) • Enumerated by reference Thesaurus DVG (external sources) • Permissible Values Discrete Value Short Values • Value Meanings Discrete Value Long Values • Form TBD Data Collection Module (DCM) • Form Module/Block TBD Question Group • New OC API added to base product to allow connection to central patient registry
Oracle Clinical 4.5 RDC CRF Matrix • Header identifies Database, User, Site & Study • Tabs identify Visits/Phases • Planned CRF pages for each visit define column headers • Unplanned pages can be added • Left column identifies Patients by position number and initials • Matrix cells indicated state of CRF page for each patient • Colors indicate discrepancy status • Red - this user’s problem • Yellow - other user’s problem • Checks indicate Verification and Approval status • Lower section of screen provides access to Summary, Discrepancy details &Audit Trail
RDC 4.5 Graphic Layout PDF Style Data Entry • It looks like paper • Less training • PDF globally accepted • PDF endorsed as submission vehicle • Available on wireless tablet
RDC 4.5 Graphic Layout (PDF) • Combs • Images • Logos • Adobe plug-in • RDC Toolbar • PDF is background • Only requires Reader • Data stored the same way
C3D CDE Compliance Review • View distinct responses collect by Study • Provides visibility of data compliance • Row 13, Response “0” is not compliant • This value is flagged by OC as a Univariate DVG discrepancy because the value is not in the list of acceptable values • Discrepancy can be resolved automatically by audited change of data (with change reason and comment)
C3D GLIB Questions linked to caDSR CDEs • Dynamic report implements link to CDE Browser details • Provides CDE Id and Version • Double click CDE Link column opens new CDE Browser with CDE Details
C3D Deployment/Development Status • Deployed into production January 2003 • ~30 existing studies with ~500 patients implemented to date – still working on back log • All new studies implemented as approved by the IRB • ~40/year expected • ~4 so far this year • 4 accruing studies (and data) migrated from legacy systems (netTrials) • 3 extramural studies in various states on implementation • NIH lab data automatically loaded daily using generalizable process • Source labs mapped to LOINC like OC lab test names • Specific lab tests placed on CRFs as required by study • Patients matched by MRN across studies and labs loaded to more than one study depending on date window (Pre-study -> Off study + 30 days) • Labs accurately graded according to Common Toxicity Criteria (v2 & 3) • Automatic extraction/generation of data for electronic submission to monitors (CTMS, CDUS) • Access to study data in unified template based reports • ~10 new CDE based CRF templates added to library
caBIG Adoption Scenarios • Scenario 1 – Central NCI hosted ASP Model • No license or maintenance fees for caBIG pilot participants • No server hardware costs • No application support costs • Protocol building services provided at no cost for caBIG pilot participants • Client hardware required - Windows PCs • Scenario 2 – Local center installation • 80% discount on license fee - $3000/named user • Annual maintenance 80% of license - $600/year/named user • Hardware & application support costs • Protocol building service costs TBD (per hour, per form, per study?)