Twenty six patients. Eleven patients. Three patients.

Despite the billions spent on clinical research, regulatory decisions often boil down to adverse events in a handful of patients. There are widely divergent views on whether the small numbers are statistically sufficient. But no one would argue that casting a wider net could be good for science, industry and the public.

So it was auspicious to see that Phase Forward‘s safety software group, formerly known as Lincoln Technologies, has begun working on a project with FDA and the Pentagon. The effort is using the Waltham, Mass., company’s Empirica software to dip into a dataset of 12 million-patients, those in the U.S. Army and their dependents.

Big File

“The program will be managed by the Pharmacovigilance Center of the Office of the Surgeon General of the Army, which oversees medical benefits programs for Army personnel,” the December, 2008, Phase Forward press release notes. Chan Russell, president of the company’s safety group, makes it clear that some aspects of the work began as long ago as 2004 and were scaled up slowly and somewhat quietly given the Army’s preference for a low profile.

All the same, the effort appears relevant to pharmacovigilance challenges faced by industry, which is obliged to look for clues in larger and larger databases partly to be able to keep up with what academic scientists and the legal community are learning about drug safety. Says Russell of the Army project: “We are getting their data up where they can see it in a useful way. We are trying to give [the Pentagon] tools where they can quickly look at a drug-event combination and quickly drill down on individual patients, what the adverse event really is, are they plausibly related to the usage of the drug.”

In accord with HIPAA, Phase Forward’s safety experts never see any identifiable data. They’re on the prowl for trends. “We’re dealing with aggregates,” Russell explains. The military data are surprisingly clean, not having been munged up by the sorts of events that can complicate commercial databases—corporate mergers, for example. “The Dept. of Defense is really taking care of a bunch of people and trying to use safety information to optimize what they’re doing fundamentally around patient care,” he adds.

Deidentified Data

In broad strokes, Russell says Phase Forward will help FDA and the Pentagon on the technical or IT side of things, and let government researchers and physicians do the heavy scientific thinking. They’ll be sifting and sorting the data to find safety signals that might not otherwise appear in ordinary clinical practice, and trying to apply those lessons rapidly. As part of the data mining, Phase Forward says it is considering the development of a new health-care oriented application to help analyze large volumes of deidentified data, but it doesn’t have a timetable for the release of such a product.

“It’s really a third kind of data,” says Russell, who says pharmaceutical industry interest in large health care databases is nothing new. In other words, the military data are not immaculately controlled clinical trial data, nor less-than-pristine medical claims data.

Russell is apparently allowed to say that the Army data are in electronic form—not on paper. Half of the patients in the database are women. He adds: “It is both inpatient and outpatient. It is people who go to military treatment facilities and what amounts to claims data. You have dependents and retirees. We get prescriptions.” If the patients see external sub specialist physicians, that’s in there, too. Says Russell: “As databases go, it is pretty broad.”

Future Tools

Is the project part of the 2007 FDAAA legislation, in which Congress attempted to give the FDA a bit more of an iron-fortified budget? One might presume so, but Phase Forward can’t be certain. “We don’t know what line item of their budget it came from,” Russell says.

It’s not a classified matter that, under the FDAAA law, the FDA is preparing to be able to analyze data from 100 million patients. That effort may be one of the more crucial ones for the medium- and long-term future of both the industry and the FDA. In the event of another global drug safety controversy—another Vioxx, perhaps—the agency could benefit from the ability to precisely assess population-level risks. Doing so early, before CNN or anti-industry academics make their own estimates, would be nice.