How the FDA Unlocked Millions of Files on Drug Side Effects

The Food and Drug Administration this week released software designed to let people tap into 3.6 million reports about drug side effects. On Monday, two hours after posting the code on GitHub, a website where developers collaborate, someone spotted a minor bug and suggested a way to fix it, says Sean Herron, the White House innovation fellow who worked on the project. By Tuesday, at least one independent developer had created a website to let people search the FDA’s data more easily.

The project, part of an effort called openFDA, stands in bright contrast to the way the U.S. government normally builds information technology: slowly, behind closed doors, with big checks to giant contractors. The risks of that approach were on conspicuous display last fall with the sputtering launch of OpenFDA shows that some corners of government can embrace open-source practices and be as nimble as a Silicon Valley startup.

The data on drug side effects—officially known as adverse events—has long been public but difficult to use. ”Imagine the Internet right now without a search engine,” says Taha Kass-Hout, the FDA’s chief health informatics officer. A handful of companies waded into the agency’s unwieldy bulk database to build applications, but the process was cumbersome. That’s the case for a lot of open-data initiatives that publish obscure government records but don’t do much to help citizens make sense of them.

By creating a way for different pieces of software to plug smoothly into the data—known as an application programming interface, or API—the FDA hopes private developers will find new ways to make the data useful to patients and doctors. For example, a mobile app could show which of two competing drugs has more reports of complications for patients with a particular condition.

The drug side-effect records include 3.6 million reports from 2004 to 2013 submitted by patients, doctors, and pharma companies from around the world—sometimes online, sometimes by fax, or even in paper forms delivered by mail. “How can you deal with a massive amount of information, various formats, archaic formats?” Kass-Hout says. Aside from making the records more accessible to the public, the openFDA project should make it easier to use even within the agency, he says.

The FDA is planning to do the same for other types of records, including drug recalls and changes to a medication’s safety warnings. Kass-Hout says he’s heard from interested counterparts at the Centers for Disease Control and the Centers for Medicare and Medicaid, which recently released records on Medicare payments to doctors.

The FDA kept the team working on the openFDA project “incredibly, incredibly small,” he says. Now that the code is in the wild, Kass-Hout and Herron hope the broader community of software developers will copy it and improve on it. Let’s hope other precincts of the federal bureaucracy take note.

Before it's here, it's on the Bloomberg Terminal.