To create a national learning health system, healthcare must develop data sets that are affordable, available in near real-time and at a level of granularity to produce research of real value, a trio of Philadelphia-based researchers say in a viewpoint article published in JAMA Pediatrics.
The authors--Sage Myers and Charles Branas from the University of Pennsylvania's Perelman School of Medicine, and Brendan Carr from the Thomas Jefferson University School of Medicine--laud new efforts to make federal government data more open, but say it’s not enough.
And researchers face high costs if they’re using data across multiple years and multiple states, they say.
Portals such as data.gov provide improved transparency, ease of use and access across many federal agencies, the researchers say. Yet key data is missing, they point out, on patients, encounters, care delivery and outcomes.
Plus, while healthcare data has provided insights into care variability, outcomes and cost, government healthcare data tends to be four years old. Yet non-healthcare public data is made available in one to two years.
The conflicting interests of healthcare providers, insurer and patients add urgency to create a learning health system, they say, but hamper such efforts at the same time. Researchers are hampered by restrictions to protect individuals and the proprietary interests of health systems.
Hospital-specific information often is missing from federal data sets and patient claims data often does not include specific physiologic and treatment information.
To capture a full picture of the general population, data on Medicare and Medicaid patients must be combined with that of private payers. Mandatory payment structures could require data be made available for research, they say.
A recent commentary published in Nature called for open-source competitors to “break open” patient health data from closed systems and said patients can’t be passive bystanders in the new ways data is used.