Photo / Slide Animation of the Fun Kind
Software for Clinical Development

Lilly: Implementing SAS Drug Development

Notes from the 23rd DIA CDM Conference

Susan Duke of GSK chaired a session on Tuesday, March 18th, called "Statistical Computing Environments: Collaboration through Metadata, People, Process and Software." During this session, Christopher Colangelo, Associate Senior Statistician, presented Eli Lilly's effort to implement SAS Drug Development (SDD) as a centralized but globally available repository for clinical data.

But first, and for those unfamiliar with this relatively new concept, a Statistical Computing Environment (SCE) is a combination of hardware, software and processes, that enable the collection, management, analysis, and transparency of operational data within a life sciences environment performed in compliance with government regulations and operating best practices. From a technical perspective, it is also an extension of a relatively new software category called "Clinical Data Repository" or CDR.

You can learn more about both SCE and CDR via the following posts on this web site:

At Lilly, the selection of SDD arose from the recognition that internal staff in Data Management, Data Delivery, Statistics, Medical Communications and external partners such as CROs needed an environment that would allow for greater collaboration within a secure environment.

From an operational standpoint, the system had to:

  • Control the storage of all clinical data regardless of source
  • Support the creation, editing and execution of statistical programs
  • Allow reports to be generated and distributed to all stakeholders
  • Enable browsing and retrieval of data and related metadata
  • Facilitate internal and external collaboration globally

The implementation was done in a phased manner starting with a production pilot. The initial experience allowed the project team to control risk by surfacing and addressing issues that would have to be solved prior to a general roll-out of the system. It also enabled the creation of a small team of experts who could then be called on to help with system training and adoption to the larger user community. Last, but not least, the phased approach allowed Lilly to avoid "burning its bridges" by making sure that the new system would perform as expected prior to the old system being shut down.

The first two phases of the implementation, each lasting roughly 6 months and focused on the US, Canada and India, began in the 3rd quarter of 2007 and will be completed in the 2nd quarter of 2008. As of March 2008, the system has almost 250 users. Europe and Australia will be brought on line in the 3rd quarter of 2008 and Japan by the end of the year. It is expected that the system will have about 600 users by October 2008.

With the recognition that the new system would be mission-critical, the project team and the Sponsor made sure that several critical success factors were in play. These are:

  • Alignment and advocacy from upper management
  • A strong partnership between Lilly and SAS Institute (the software vendor)
  • Earmarked resources from IT, Data Management and Statistics
  • Unprecedented collaboration between the "business" and IT

Although the project is still in progress, Lilly has already recognized several benefits from the implementation including:

  • Tighter security & control than the LAN being replaced
  • Ability to integrate existing systems and enable others
  • Centrally-located clinical data
  • Fewer systems/platforms to maintain/support
  • Streamlined approvals / E-approvals
  • Streamlined code validation
  • Streamlined data flow
  • Reduced CPU time for execution of some SAS jobs
  • Scheduled job execution

To his credit, Mr. Colangelo also pointed out several system adoption issues that will need to be addressed in the future. Some of the "soft" issues were:

  • a reluctance to do program development directly in SDD
  • concerns about performance in a server-based (rather than PC-based) SAS environment
  • enthusiasm vs. dismay having to work in a non-Unix environment
  • excitement vs. reluctance going to a new environment

Some of the "hard" issues were:

  • the inability to develop or execute programs other than SAS
  • the lack of workflow/process tools/methods

While it is too early to tell how well management and user expectations will be met by the system once in use by all stakeholders, the project team should be commended for taking a methodical but deliberate approach to implementation and for giving proper weight to technical, business, process and cultural issues.


The comments to this entry are closed.