listen to the drip, drip of
By Agnes Shanley, Editor in Chief
We are drowning in data. Every year, Berkeley researchers tell us, we generate 30% more information, in paper, electronic and other forms. Two years ago, we had already stacked up a 5-exabyte Everest of data, equivalent to 500,000 Libraries of Congress, or 800 megabytes for every person on earth. I know because its all in my email inbox.
Privately, we all struggle with this issue each day. But few industries suffer more from this data deluge than pharmaceuticals. Despite significant investments in information technology (IT), knowledge-based pharma remains knowledge poor in its day-to-day operations, at every step of the value chain, from discovery through distribution.
In this issues cover story, we uncover the waste on the drug development side with a benchmarking survey on data management. As it reveals, many gifted and well-paid scientists and engineers spend 15% of their time trolling through servers and file cabinets for documents they need. Sometimes they never find them, triggering rework, redundant tests, and the loss of untold millions of dollars each year.
In his memorably-titled book, Rembrandts in the Attic, Kevin Rivette, founder of Aurigin Systems, describes the problem of IT and knowledge management, and what underutilized intellectual property costs companies. At a surprising number of firms, R&D teams are literally re-inventing the wheel, duplicating technology that the company already owns the rights to, whose patent is buried in some obscure and forgotten file.
But smaller Rembrandts are also being lost every day a best practice that a quality team at a sister facility in another country developed, critical calibration data, specifications for a key replacement part, information on the best way to operate a specific dryer.
Many of these documents are still on paper, and, at all too many facilities, theyre difficult to find. Shouldnt we be beyond paper by now? As FDA emphasizes process understanding in its approach to regulation, paper-based systems put companies under enormous risk.
But even electronic databases at most pharmaceutical companies arent much better, since all they do is move the flat, paper format onto a computer. In a word, theyre still document-centric. Firms may offer better ways to search for these documents but, they often fail to provide a context for data, and ways to link seemingly unrelated pieces of information for example, maintenance data for a specific piece of equipment, incoming quality and vendor data for a specific raw material, and final product lot data. Because problematic patterns arent recognized, theres no proactive early warning system for operators, scientists and engineers when an issue reemerges.
Software programs have proliferated, promising to ease the burden of sifting through this information. More often than not, though, they merely overlay an expensive patchwork of Band-Aids over a much deeper problem. Meanwhile, every day at a facility near you, researchers, engineers and operators riffle through paper-based documentation or electronic documents maintained in separate silos.
Whats needed is a fundamental connection between databases. Relational databases allow users to make connections between different data points, but they dont always work well with the types of data that drug companies deal with. (Read on p. 44 how Novartis solved this problem, and advanced its PAT efforts).
Initiatives such as PAT and the move to personalized medicine only shine a searchlight on pharmas data management problems. On-line quality measurement will increase the amount of quality data, hybrid drug-devices will result in multiple streams of compliance data, while enhanced efforts in genomics and proteomics will add to the challenge.
E-mail to tell us how youre coping with the data explosion. If your message bounces back, please dont give up. Im only cleaning out my inbox. Again.