Like a boxcar of iron ore on its journey to becoming a steel ingot, clinical data undergo a series of routine transformations. Even with the most meticulously organized or technologically adept organization, it has proven difficult for the industry to achieve anything close to efficiency in the long process that converts a fact collected at a clinical site at the beginning of a trial into a study report or a statistical table reviewed by the FDA at the end.

Could that be changing? Statistical powerhouse SAS and electronic data capture (EDC) leader Medidata Solutions have partnered to automate the messages flowing between their systems and, they say, eliminate or expedite some of the cumbersome manual loading, transfer and mapping of data.

Other EDC firms have partnered with SAS before. So the arrangement is hardly unprecedented.

But this time SAS and Medidata say they have a shared vision that goes beyond the notion of “integration” of clinical data management and biostatistical analysis. The two companies are promising a more seamless progression of data from sites through data management and into the hands of biostatistical wizards who analyze what happened in the trial.

‘Patient to P-Value’

“We’re very excited about this,” says Jason Burke, worldwide director of the SAS health and life sciences global practice. “We think this is moving down a path that enables people to do things they have not been able to do before. You should not have to care if the information lives in an EDC environment or an analytical environment. It should just be available to you.”

“This integration offers customers an opportunity to revolutionize clinical development by speeding data transparency from patient to p-value,” Burke noted in a press release. “SAS and Medidata can further push legacy clinical data management systems (CDMS) toward obsolescence.”

As an example, Burke told ClinPage that the partnership could facilitate looking at data across trials. Using legacy CDMS, that is not an impossible task but it is far more intricate than most outside the industry would suspect. At present, it might take SAS programmers weeks or months to figure out how to look across several trials. That might be a matter of hours in the new world. “What we’re describing here is an end-to-end environment that is a natural part of the way people are working with information, so that they can see the pedigree of data and how it has evolved over time,” says Burke.

Leveraging Standards

How? Both SAS and Medidata have focused on data standards from the Clinical Data Interchange Standards Consortium (CDISC). And both of them can voice considerable passion for metadata, the data about the data. It’s not enough for two systems to share the number 120, they point out, without having a common label for that figure. Their systems will have a single reference set of metadata, although the raw data will remain in separate, secured locations. 

In technical terms, the companies have built a way to use graphical, nonprogrammer techniques to connect data in Rave, Medidata’s EDC application, with SAS Drug Development, a collaboration environment for data analysis.

Time Misspent

Burke relays the plight of a senior SAS programmer at a top ten pharma who recently confided that she spends the bulk of her time doing what amounts to housekeeping. She doesn’t have time to write much of the SAS code she is expertly able to create; instead she painstakingly moves data from one system to another, trial by trial. He hopes the SAS-Medidata collaboration will allow her and others in the industry to spend more time on high-value tasks, and less on moving or mapping data from one system to another.

Needless to say, SAS has seen to it that the information’s progression from one system to another is in accordance with 21 CFR Part 11, and that the chain of custody on any particular bit or byte is easy to examine. As such SAS sees SAS Drug Development’s new Medidata-enabled functionality as something that can further facilitate collaboration.

Adaptive Angle

Burke believes that the Medidata partnership could facilitate adaptive designs. How else can decisions be made, he suggests, if the data have not traversed the divide between the EDC platform and the analytical toolbox?

Ironically, Burke explicitly says the partnership might initially result in less SAS code being written, simply because more data mapping and data transfers will happen without the intervention of an expert or team of programmers

In an internal demonstration of the technology, Burke reports, data were prepared for analysis without anyone writing a single line of code. “That is a phenomenal shift,” says Burke. “Both systems now share the same information about the data. You don’t have to figure out what the table names are or the variable names are or how it should be structured.”

Life-Cycle View

There is similar excitement at Medidata. Glen de Vries, chief technology officer, concedes that the company has been engaging in a number of technology alliances. But the relationship with SAS is something that allows the customer to span a longer section of the clinical data life cycle than ever before, connecting the sites at the start to the analysis and reporting tasks at the end.

“We think we are raising the bar on integration by which systems like this will be judged,” de Vries says. “To us, this is really the second half of the complete picture that you need for a customer to have that end-to-end capacity.”

As at SAS, Medidata vows that studies can be defined without writing a single line of code. “Because both systems are sharing metadata and understand what the other system’s data means, we’re going to continue to build avenues by which information can pass between the system,” says de Vries. “Frankly, you don’t want your data manager wasting their time thinking about something that should be as trivial as handing data off between two systems. You want them thinking about data quality.”

Achieving Scale

In the future, de Vries says, the sharing of metadata will facilitate adaptive trials. “If you have an adaptive study and all of a sudden there is a new kind of data and new type of metadata, and you have to educate every system, there is a cost.” With the approach of SAS and Medidata, he says, the introduction of a new data element could be done once—and populated throughout the system. “You need to have this backbone for information sharing that is going to be able to react to adaptive decisions,” he notes.  “There is no need to define metadata twice.”

Indeed, that backbone already speeds the data on its way from other Medidata partners such as trial-design firm Fast Track Systems and patient diary concern invivodata, with which metadata is also shared.  Says de Vries: “If you take a Fast Track protocol and load it into Rave, those metadata definitions are going to be the ones that wind up in SAS Drug Development.” Diary data from invivodata handhelds, of course, will be able to flow into SAS Drug Development just as seamlessly.

De Vries says there will be a small number of ways that others in the industry will be able to duplicate the efficiencies his firm and SAS will deliver. Says de Vries: “There are basic principles in computer science that dictate [that] the data model you use for analysis is different than the data model you use for transactional purposes. Metadata sharing and the seamless ability to reuse in one place and to scale more data through because of that reuse is the only sensible approach. You can take systems that don’t really work that way and make them look like they work that way, with smoke and mirrors, but they will never really achieve scale.”