Implementation of a Validated Statistical Computing Environment Presented

Implementation of a Validated Statistical Computing Environment Presented by Jeff Schumack, Associate Director – Drug Development Information September 16, 2005

Statistical Computing Environment (SCE) • • Data Repository of source code, input data & output Version controlled programs and output Reusable programming ability Links to Clinical Data Management System (CDMS) & Documentum 2

Prototype • Determined requirements • Performed market search • Worked collaboratively to achieve our vision • Conducted extensive pilot 3

SCE: Where Does it Fit in? Clinical Data Management Capture and Organize Data Clintrial/CRO Clinical Biostatistics Regulatory Affairs Analyze Data, Generate and Deliver Reports Submission of Reports to the FDA Documentum/Publishing SAS, Files Statistical Computing Environment Analysis, Programming and Reporting under 21 CFR Part 11 compliance 4

Interfaces Clinical Data Management Core SCE Documentum 5

Clinical Data Management Connector 6

Extraction Process Query CDMS • SAS or XML • Execute data extraction – Extraction in context – Record data and output under correct study 7

Audit Trails • • Who What When Comment if any 8

Core SCE Functionality 9

Database Connectivity Oracle Role Based Security • Require username / password • Secure – only authorized personnel allowed to connect • Level of access based on user privilege 10

External Applications • SAS system version 8. 12 (or higher) • R (Splus) version 1. 9. 1 or higher • Ultra. Edit version 10 or higher 11

Repository • Secure Oracle Database • Easily retrievable • Hierarchical Structure – Indication -> Study -> Programs (etc. ) 12

Traceable Pathway SCE provides • • • Clear pathway from “field” to report Mechanism for data extraction of analyses Dependency map of input-to-programs-to-output “Push” analyses into Documentum Audit Trails 13

Dependency Map Graphically represent “program-to-input-to-output” – Including version numbers 14

Security Manager • Individual user accounts • Password Management: LDAP • Permissions: – Group – User – Project 15

File Management • Folder types / File properties • Check out / Check in • File annotations • Source Code Editor • File Viewers • Execution and display results, logs, graphics • Audit Trails 16

Pack & Go • “Pull” files from SCE onto local disks – Indication – Study – Analysis • Useful for Archival • Collaboration with external contributors – Partners – CROs – Data Monitoring Committees 17

Study Setup - Templates • Nested Hierarchy templates – Indication – Study – Analyses • Replicate Studies using templates • Retire templates no longer in use 18

Job Management • Server based running of jobs • Define jobs / Submit jobs / Run & Record • Input / output dependencies • Table of Content user interface / comment blocks • Relative file pathways – Macro search – Common Code search • Batch execution / job sequencing • Record Outputs • Traceability – Dependency Mapping – Input data / program / macros -> main program -> Output (s) 19

Life Cycle Management • Draft / Quality Controlled / Final • Business rules • Draft from Final 20

Reporting • Audit Trail reports • Output formats – PDF, XML, RTF etc. • Combine Outputs • Interface with external reporting tools 21

Documentum Connector 22

Documentum Authentication • Username / password protected • Documentum user information verification • Documentum permissions verification 23

User Interface • Automatic population of Documentum based on lifecycle state of document in SCE • Manual population based on user triggers • User comments • Version controls 24

Audit Trails • Who • What • When • Version control information 25

Conclusion • New technology • Close to realizing HGS SCE vision • Acknowledgments: – HGS Biostats & IT teams – Waban Software – Alan Hopkins • Contact Information: – Jeff_Schumack@hgsi. com – 240 -314 -4414 26