Quality Control For A CDISC+SDTM Based Clinical Data Warehouse

Authors

• G Wei
• J Xiao
• KK Bajwa
• SK Debrincat
• C Whalen
• M Tartakovsky

Abstract

BACKGROUND-Multiple data warehouses using data mart subsets are becoming common place in research environments. The continuous flow of data from one database to another creates the potential for errors. The OCICB has established a Clinical Data Warehouse (CDW) using CDISC+SDTM model. The clinical trial study data transfers from the Clinical Data Management System (CDMS) to the CDW via an Extract, Transform, and Load (ETL) process. We required a quality control (QC) process prior to the CDW release to the study teams. OBJECTIVES-Establish a reliable protocol to QC the CDW. METHODOLOGY- Using SAS programs, we developed a QC protocol to validate (1) the clinical data migration. Three key variables (CRF Page #, Subject ID and unique CDMS Sequence number) were extracted from CDW, sorted, and duplicate records removed. The records were compared with mapped variables from the CDMS database. SAS reports generate if inconsistencies identified. (2) A complete check of the Data structure in the CDW by each CDISC domain to ensure that the mapping fully follows the ETL specification which are generated based on SDTM data model. RESULTS- The clinical trial data for multiple protocols were validated using the methodology. It proved successful in identifying data errors and structure inconsistencies. CONCLUSION/DISCUSSION–By comparing the data in the CDW directly with data in CDMS database. The validation is a reliable QC approach for both the ETL and the mapping specifications.

Scientific Focus Area: Research Support Services

This page was last updated on Friday, March 26, 2021