When the Health and Human Services Department launched the Health Data Initiative, HHS Chief Technology Officer Todd Park expected half the work to be publishing the data and half the work to be marketing the data to application developers.

"I was completely wrong. Five percent of the work is publishing the data, 95 percent of the work is publicizing to the world of innovators that this data is out there for them," said Park, while speaking June 15 at the Government Health IT Conference in Washington, D.C.

Data promotion at HHS comes in the form of challenges, codeathons, meet-ups, conferences, even something called "health datapalooza." And while the innovation that arises from the events is good, the events are really intended to spread the word to developers that these health data sets are free and public.

"It's really not about what the challenges and codeathons physically build. It's about attracting people into healthcare. We need more people in transition, in transition into the healthcare system," said Park.

Park endorses the early release of data, versus subjecting data sets to a lengthy cleaning process--which could explain why data release only accounted for 5 percent of the effort at HHS. 

"Even the root data itself could, in a lot of cases, benefit from improvements in a lot of respects, like granularity, timeliness, etc. And one of the core principles in the Health Data Initiative is that the data suppliers and the data users actually talk to each other, so you get a sense of how the data's being used and how the data can be improved," said Park.

Park said developers can hash out problems with HHS data on its forum which is constantly monitored by someone from the department.

While the data found on healthdata.gov is helpful for community-oriented applications, some GHIT attendees said it falls short for evidence-based health applications.

"To really get down to the brass tacks of patient-centered research, it requires data beyond what's being released on healthdata.gov," said Park. "So there's a zone of data liberation that I'm calling semi-open data. There are data sets that we would never, ever, ever, ever publish as a public use file. But they are data that we could make a heck of a lot easier for researchers to access a heck of a lot faster."

Semi-open data, such as scrubbed Medicare claims data would not be added to healthdata.gov, because it could still be linked, meaning while some personally identifiable information is removed many of the variables of data are still associated. It would be possible, however, to make that semi-open data available to qualified researchers under a data use agreement.

Park said there will be more work in the next 12 months to make semi-open data available to qualified parties more quickly and inexpensively than is currently possible.

Overall, the amount of data released by HHS will likely only increase. On June 9, HHS Secretary Kathleen Sebelius asked every agency within HHS to report to her twice a year on all the data they think they have, what subsets they have made public and what additional subsets they plan to make public or more accessible in the next six months.

Related Articles:
Park: Agency data should be released more quickly, as 'beta'
NASCIO preps state CIOs for health benefit exchange IT challenges 
HHS proposes on-demand EHR 3rd party transaction disclosure rule   
HHS urges on health data app development