The Health Data Liberation meetup group is devoted to helping providers, insurers, and institutions to adopt open data practices in the field of healthcare.
The 8 Principles of Open Health Data guides our thinking. Here’s the complete text:
In order to fully realize the potentially profound benefits of digital health data, open access to de-identified non-aggregate health data is essential. We propose eight principles of open health data, modelled on “8 Principles of Open Government Data,” to guide management of, access to, and governance of this data.
- Open access to de-identified health data is essential to optimizing the health of populations and individuals.
- Data must be de-identified in order to address individual concerns about privacy and the economic and social consequences of disclosure, but as a society, we should work toward eliminating negative consequences of disclosure of health information.
- De-identified data should be available to any user for any purpose, with minimal barriers to access. Rigorous de-identification of data results in some loss of fidelity and precision. Researchers who present credentials and institutional review board approval and plans for secure data storage and destruction should be provided a less rigorously de-identified data set.
- Participation in health care funded with public money, including health care based on publicly funded research (essentially all health care), should stipulate participation in open health data.
- A custodial entity that serves and protects both society’s and the individual’s interests in the data should manage the data repository.
- Health data must be as complete as possible—including prenatal through postmortem clinical data, behavioral and genomic data, data about environment and occupation—to fully understand the connections between health and its social and medical determinants. Given the enormous obstacles to open data posed by electronic medical records at this time, a subset of data relevant for public and population health purposes should be defined to make data acquisition feasible. However, the ultimate goal should be to make the record as complete as possible.
- Data must be added to the database in a timely way (e.g., not after death) in order to produce the most significant benefit. The record should be updated frequently, but need not be updated in real time.
- Data must be collected and provided as much as possible using national and international health data community standards, but non-structured data should be collected where possible or necessary.
Rebecca Wurtz, MD, MPH Northwestern University Chicago IL
Daniel X. O’Neil Smart Chicago Collaborative Chicago IL
