CDASH : A PRIMER AND GUIDE TO IMPLEMENTATION

CDASH : A PRIMER AND GUIDE TO IMPLEMENTATION Mark Wheeldon NJ CDISC User Group Merck, Summit September 19, 2013 AGENDA • Introduction • What is CDASH? What it is not • Perfect libraries with CDASH • CDASH with multi-EDC vendors • CDASH aids mapping to SDTM • Practical CDASH mapping examples • Conclusions INTRODUCTION • • • Vendor neutral using CDISC since 2000 Save time & money in End-to-End Clinical Trial Universal Design with Formedix Origin™ • • • Universal Execution with Formedix Transform ™ • • • • • Data Acquisition Design & Dataset Design Origin Study and Submission Modeller™ Data Acquisition – Paper, EDC, ePRO & IVRS Conversion – ETL Submission – Annotated CRF and Define.pdf Create once work with all Train in one work in all Formedix Consult™ Consultancy Services • • Planning, preparation, implementation, multi-EDC execution Won CDISC awards – wrote much of ODM/define INTRODUCTION TO CDASH WHAT IS CDASH? WHAT IT IS NOT • Content Standard NOT a Model • Covers 22 core Case Report Forms (CRFs) - Variable Names, Definition, Question Text - Instructions to the Site, SDTM Variable Mapping • Core Designations is a killer feature - Give clinical context on why something is collected - Highly recommended, recommended/conditional, optional - Clinical Perspective on data collection • What it is not - One-to-one mapping with the SDTM - AEYN not in SDTM, AEBODSYS not in CDASH HOW CAN CDASH BE STORED? • Excel, Word, PDF Libraries - Excel for database metadata - Word for CRF mock-ups • Global proprietary EDC Library - Rave, DataLabs, Prism… - Has shortcomings ◦ Lack of full support for CDISC ◦ No alias support – automate aCRFs ◦ No discrete measurement units support ◦ Poor support for vertical structures • Optimal solution CDASH in CDISC-ODM WHAT IS CDASH ODM? Submission Clinical Context Database Content and Structure Codelists Extended ODM Structure CDISC SAVES TIME & MONEY • Libraries – Content Reuse - 70% reuse - Standards maintenance resources down by 23% - Just this use case ROI – months • Study Set-up - Specification auto-generation - Boston Scientific proved a 68% reduction - With CROs, Partners, EDC vendors … • Study Build - Automate EDC build build - 55% reduction in EDC build • Study Conduct and Analysis - 76% reduction in dataset production PERFECT LIBRARIES WITH CDASH CASE STUDY - CDASH NOT USED 342 143 MIXTURE OF EDC/CRO VENDORS STUDIES IN SAME THERAPEUTIC AREA FORMS 36 UNIQUE FORMS MASTER LIBRARY FORMS WITH CDASH FORMS WITHOUT CDASH Multiple permutations of Forms present 10 15 MEDICAL HISTORY FORMS SURGICAL HISTORY FORMS VITAL SIGNS FORMS PHYSICAL EXAM FORMS = 36 FORMS CDASH WITH MULTI-EDC VENDORS CDASH MULTI-VENDOR eCRFS AUTOGENERATED IMPLEMENTATION ISSUES CDASH WITH MULTI-EDC VENDORS • Structural Issues - CDASH vertical vs EDC horizontal - EDC systems allow only one vertical structure • Poor/Lack of support for - CDISC identifiers - lengths, character restrictions etc - Aliases – used to annotate/map to SDTM, Legacy fields - Multiple languages - Enumerated codelists • Worst case - ‘Hard coded’ CRFs required for EDC Demographics, Labs, AE etc IMPLEMENTATION ISSUES SOLUTIONS • Structural - Implement one field per question (horizontal) - Implement vertical versions - Split repeating sections on separate forms (incl/excl) • Core Libraries - Covers most use cases - Special case specific EDC eCRFs in separate library • Global aliases - Company corporate identifiers used end-to-end everywhere - Still supports EDC identifier restrictions • Code / decode instead of enumerated codelist MAPPING CDASH TO SDTM ADVANTAGES OF CDASH SELECTION OF CDISC STANDARD DOMAINS MAP TO CDASH • Interventions - CM (Concomitant Medications) - EX (Exposure) - SU (Substance Use) • Events - AE (Adverse Events) - DS (Disposition) - MH (Medical History) • Findings - EG (ECG) - IE (Inclusion / Exclusion Exceptions) - LB (Laboratory Test Results) - PE (Physical Examination) - QS (Questionnaires) - SC (Subject Characteristics) - VS (Vital Signs) • Special Purpose - DM (Demographics) - CO (Comments) CDASH ODM CDASH & SDTM ALIGNMENT CDASH CRF Label Question SDTM Controlled Terminology SDTM Variable Name CDASH Core (Highly Recommended) OPTIMIZING DOWNSTREAM DATAFLOWS 1 2 3 4 5 6 CDASH Form Non-CDASH Form OPTIMIZING DOWNSTREAM DATAFLOWS CDASH Non-CDASH PURE CDASH : VERTICAL NEEDS HELP SOMETIMES • Vital signs (VS form) • Vertical implementation suggested • Optimized for dataset production and not all easy to enter data • Must repeat on VSTEST - Cannot have repeat on time as well Form repeats for time points • Other issues - - Every variable collected at every time point even if likely to remain constant VSORRESU Units must be a pull down Error prone needs more edit checks Repeated collection of irrelevant values • Pros: Simple dataset mapping CRF - SDTM DATA MAPPING HORIZONTAL - VERTICAL CRF COLLECTION DB USUBJID TEMP TEMP_U PULSE SDTM DATASET COLLECTION DB PULSE_U USUBJID USUBJID VSTESTCD TEMP TEMP_U VSORRES PULSE VSORRESU PULSE_U ° C BPM 1001 TEMP 38.0 C 1001 38.0 ° 82 TEMP 1001 38.0 C ° 82 1001 PULSE TEMPERATURE 1002 38.5 PULSE C 79 ° BPM 1002 TEMP 38.5 C 1002 38.5 ° 79 BPM 1002 PULSE PULSE 82 79 C ° BPM BPM C ° BPM BPM CONCLUSIONS • CDASH ODM is best format to store CDASH - Union of all standards - SDTM, ODM, CT - EDC extensions and best practices • CDASH saves time and money - Clinical perspective - Promotes reuse of CRFs and end-to-end designs - Alignment with SDTM and mapping ease • Compromises have to be made - Multi-vendor and your own content - No two CDASH libraries are identical ANY QUESTIONS? ...AGENDA • Introduction • What is CDASH? What it is not • Perfect libraries with CDASH • CDASH with multi-EDC vendors • CDASH aids mapping to SDTM • Practical CDASH mapping examples •... DOWNSTREAM DATAFLOWS 1 2 3 4 5 6 CDASH Form Non -CDASH Form OPTIMIZING DOWNSTREAM DATAFLOWS CDASH Non -CDASH PURE CDASH : VERTICAL NEEDS HELP SOMETIMES • Vital signs (VS form) • Vertical implementation. .. content - Standardized CRFs = consistent data collection - Standardized CRFs = standardized datasets/mappings • Reuse of all study design components - Across end -to- end clinical trial - Across multiple