Constructing a Data Warehouse for Pharmacokinetic Data

The construction of a warehouse using data from DuPont Pharmaceutical’s Clinical Pharmacokinetics Group with the SAS/Warehouse Administrator product is described herein. Data sources incorporated into the pharmacokinetic warehouse existed in multiple formats (Oracle, Excel, SAS data sets, etc) and were represented by numerous individual files, rather then few large files. In most cases, the data density enjoys low granularity. The pharmacokinetic data warehouse draws data from many sources represented by multiple departments and spans long drug development cycles with varying degrees of detail. Additionally, data originated from both internal legacy systems or was obtained from an external contract research organization (CRO). Remote systems have been easily identified within the warehouse and views rather than distinct copies have been incorporated as data sources. The SAS/Warehouse Administrator product was essential to this effort as it organized the metadata from the various operational data definitions (ODD), assisted in the cleaning / scrubbing of the data via user-written code, and facilitated the creation of multi-dimensional databases (MDDB), datamarts and infomarts which were front-ended by exploitation tools including SAS/IntrNet based and SAS/EIS derived applications. More importantly, the warehouse is under the central control of a single administrator and refresh is scheduled in a manner consistent with the needs of the department using the Windows AT scheduler (Windows NT 4.0). The current data warehouse will assist in multiple aspects of electronic transmission of regulatory data: data (SAS XPORT), reports, and computer assisted new drug applications (CANDA). The following SAS systems and products discussed included Base SAS, SAS/GRAPH, SAS/Warehouse Administrator v1.3, SAS/SHARE, SAS/CONNECT, SAS/MDDB, SAS/IntrNet v1.2, and PH.Kinetics v2.00.