At all large technology companies, innovation dies. Intel, Microsoft, Nokia, Intuit may claim to invent. But year after year, their ostensibly new products are indistinguishable from older ones. Could any ordinary human differentiate between phones or PCs from 2005 and 2010? Which are more glaring: the engineered security holes in Adobe's Flash plugin, or the file synchronization defects in Apple's operating system?

In time, Medidata Solutions could reach the dinosaur stage of the technology company life cycle. For now, it’s still innovating briskly.

In the past few months, the New York City firm has announced two new tools that enhance its core electronic data capture (EDC) offering and address significant, structural issues in clinical research. Medidata’s R&D agenda isn’t attacking the most intractable problems in clinical development: patient recruitment and pharmacovigilance. But the firm's road map remains the most relevant and ingenious in the industry, especially since its largest competitor's voice has been silenced inside the bureaucracy of a new corporate parent.

New Tools

Before getting to the big stuff, it’s worth noting that like any young software firm, Medidata is still sweating the details. It added a randomization tool, following in the footsteps of competitors. And it tackled a seemingly trivial but material annoyance of clinical sites: the difficulty of administering large numbers of users and passwords.

The company has attempted to address that annoyance with what amounts to a portal where sponsors can give sites a single username and password for all the trials on the Medidata platform. For sponsors with special training needs, the company expects the portal to expedite the process of helping sites actually start to find patients. The number of help desk calls should plummet as a result.

A more audacious development at Medidata is termed “targeted source document verification (SDV).” Stripped of jargon, the company has developed a software module that helps sponsors and contract research organizations (CRO) design statistically defensible plans for personally verifying some clinical trial documents—and not others.

As industry veterans know, SDV is an expensive component of one of the most costly parts of every trial: face-to-face trips to visit clinical sites. At present, most sponsors insist on 100 percent SDV. In other words, every last document gets checked. A few companies, attempting to economize, have begun to allow monitors to verify only a subset of documents. But improvised, seat-of-the-pants approaches to SDV might lead to regulatory difficulties.

SDV Management

Medidata’s vision is different. Its idea is to use software to demarcate events or time intervals, and calibrate the source document verification in each one. With clear, rigorous criteria for SDV at each stage, not to mention an audit trail, Medidata is hoping that its software-based SDV plans could be examined later and not make government officials blanche.

An aura of propriety is important. Academic researchers have never bothered with SDV; they don’t believe their relationships with regulatory officials to be in eternal flux. But the rationale for SDV in industry has been the remote but terrifying prospect of the New England Journal of Medicine or the FDA discovering record-keeping lapses.

The Medidata software allows sponsors to customize SDV plans according to any of the key parameters of the trial. As a hypothetical case, perhaps 100 percent of the case report forms might be subjected to SDV for the first ten patients at every site.

Metrics on Quality

But after that, sites that have a certain level of data quality might receive less scrutiny. Documents at sites with more problems could be given more scrutiny. As with an adaptive trial, an SDV plan would be fixed at the start of the trial, and applied without favoritism to every site.

Medidata’s customers asked for targeted SDV last year. It’s in production now, according to president Glen de Vries. He has quietly shed the chief technology officer title, but not the vitality and enthusiasm he had during the years when the entire company fit inside a conference room.

Today Medidata has 574 employees, with roughly twenty percent of them outside the U.S. (principally in Britain and Japan). De Vries says the new Medidata SDV module allows sponsors to begin to have some broad, quantifiable insights into data quality. Metrics, in other words, that sponsors can act upon. “You now have a benchmark you can use,” he says. “Is my SDV program impacting my data quality? Should I do less SDV? Should I do more?”

Enabling Action

The CRO partners who are asked to use Medidata’s SDV software don’t seem worried about a diminishing amount of work. Says de Vries: “The CRO partners are just as excited as the customers. They are looking for ways to streamline their operations.”

Like other providers of EDC, Medidata is building other systems where protocols are designed. But it has more data about the costs of clinical trials than rivals. A Medidata program called Grants Manager helps sponsors know what to pay clinical sites.

As of the summer of 2010, Grants Manager has been upgraded to run over the internet. It now includes quantifiable ways to assess trial complexity. The software calculates a number for a trial relative to similar studies. The Tufts Center for the Study of Drug Development's Ken Getz helped Medidata develop that.

One Number

The new trial complexity score could help sponsors address one of the biggest disconnects between themselves and the scientists who help them find precious patients. Sponsors, of course, prefer ornate, baroque protocols, with multiple procedures, lab tests, patient diaries, physical exams, instrument readings, dancing unicorns, etc. etc. Those sorts of protocols are hypothesized to anticipate regulatory concerns and lead to marketing success against other drugs already on the market.

But that same complexity burdens investigator sites. Complexity-based disenchantment is one cause of investigators turning away from industry research. As such, complexity is a threat to the long-term scientific quality of industry R&D. One might think the industry would be eager to address the issue.

For sponsors, the operational decision on each trial is whether its complexity is discretionary, optional—or necessary. Until now, most of the ways to assess that have been subjective. That's made it hard for non-physicians inside sponsor companies to bravely raise their hands at a meeting and say, Do we really need everything in this protocol? Can't we simplify it?

Complexity Analysis

De Vries concedes the design of a protocol will never be a simple, one-dimensional task. But he thinks software can help. “Complexity is expressible as a single number,” de Vries says. “What it allows you to do at a glance is say, ‘Is this study going to be harder to execute than the typical study in the same therapeutic area, geography and phase? Am I giving my sites a protocol that is more complicated than my competitors?' ”

Medidata’s software may not be able to deflect the trend toward hyper-complex studies. But it helps sponsors to be aware of a) why their studies are complex and b) the financial and operational risks that accompany complex trials. (ClearTrial, out of Chicago, is engaged in a similar quest.)

In some cases, de Vries says, a sponsor may elect to proceed with a trial that is significantly more complex than other similar studies in its own portfolio, or in the industry at large. Says de Vries: “There is nothing wrong with being more complicated. There is something wrong with not knowing that piece of information and being unable to act on it.”

Medidata's Grants Manager software relies on a proprietary database of trials. “Data becomes irrelevant pretty quickly,” he says. “We have a huge set of trials, and they’re fresh.”


It’s worth noting that while all major EDC companies have software to design trials and create electronic case report forms (eCRF), Medidata is linking its tools together. Claims of pre-integrated systems are often made in the clinical trial technology niche, and can only be verified by Medidata's customers.

But in theory, Medidata’s product suite could allow someone to conceive a trial protocol, tinker with its complexity, and quickly build an EDC database. Thanks to proprietary cost data, reusable libraries of code and support for CDISC data standards, Medidata has unique abilities to unite three disparate elements of trial design—clinical science; cost; and database structure. “We set the expectations for ourselves and our customers that you are going to benefit more when you use them together,” de Vries says of the firm's tools.

According to an August, 2010 Wall Street filing, Medidata has quadrupled its customer count, to 198 from 50, since early 2007. Given that the company was not small in 2007, that suggests that a handful of top firms in the EDC space will continue to prosper. In 2010, Medidata appears likely to have $150 or $160 million in annual revenue.

A final source of momentum is a contract with the National Cancer Institute (NCI). As oncology trials are some of the most difficult, the NCI deal is a potent implicit endorsement. ClinPage wrote about the NCI strategy here. The NCI is making Medidata's technology available to affiliated academic cancer centers, although the economics of the arrangement remain murky. Medidata insists it is not donating its technology to the government. “Our business model hasn’t changed,” de Vries says, offering no specifics. “We’re not giving away [the company’s main system] Rave to anyone.”

De Vries also explained that the cancer group at the NIH enlisted another federal agency, the one administering parks, as a routine technique intended to ensure an impartial, merit-based vendor selection process. After NCI made its choice, there was a flurry of litigation to protest the choice, but de Vries suggested that’s now winding down, and the NCI-related studies are ramping up.

Editor's note: Shortly after we published this story, the NCI confirmed that it has a ten-year, $24.3 million deal with Medidata. Here's our last interview with de Vries, back in 2008.