T-01: An Innovation: Automated Statistical Programming with CDISC Analysis Results Metadata
Director, Data Optimization
Pfizer Inc. United States
The proposed idea explores how to improve programming efficiency, quality and fast delivery to benefit our patients. The poster intends to share our experience in automated statistical programming in table and figure generations while complying with CDISC.
Statistical programming is initiated from a document called Table of Contents (ToC). In Clinical Data Interchange Standards Consortium (CDISC) standards, the Analysis Results Metadata (ARM) for Define-XML was released to support the submission of ARM as defined in Analysis Data Model (ADaM).
By examining the programming process flow of input data-->subset data-->analyzing-->reporting, we were able to automate the display (table or figure) with parameterizing all the programming process. Those parameters are conformed to ARM, thus programming being driven by ARM.
To achieve the automation, we need to apply the core principle of ADaM—analysis ready, require display template structure standardized, and fully utilize the SAS structured ODS OUTPUT dataset for inferential analysis.
After piloting this idea, we were able to achieve an optimal programming in generating tables and figures with ARM compliance initiated at ToC. It facilitates the primary programming, QC programming, review, and decrease errors. It also further facilitates the automation of ARM submission document. Production and submission are aligned. In summary, the programming process becomes highly efficient providing high quality delivery within competitive timelines.