OSHA has proposed a set of set of changes to current tracking of workplace injuries and illnesses.
Currently information about workplace injuries and illnesses must be recorded, but only on paper. Further most of this information is never reported — OSHA only receives detailed information when it conducts an investigation, and receives summary records from only a small percentage of employers who are selected to participate in the annual survey. (Additionally BLS receives a sample of this information in order to produce specific statistics for its “Survey of Occupational Injuries and Illnesses”
OSHA proposes three changes. The first change would require establishment to regularly submit the information that they are already required to collect and maintain (quarterly submission of detailed information for larger establishment, and annual submission of summary information from any establishment with more than twenty employees that is already required to maintain these records) . The second change makes this process digital — submissions would be electronic, instead of on paper. And the third change would be to make the data collected public — searchable, and downloadable in machine-actionable (.csv) form.
These proposed changes raise an interesting and important combination of questions about how to promote government (and industry) transparency while protecting individual privacy. My colleagues at the Berkman Center, David O’Brien, Alexandra Woods, and I have submitted an extensive comment on these changes with some proposed recommendations. This comment is made on behalf of the Privacy Tools for Research Project, of which we are a part, and has benefitted from extensive commentary by the other project collaborators.
To summarize (quoting from the conclusions of the comment):
We argue that workplace injury and illness records should be made more widely available because releasing these data has substantial potential individual, research, policy, and economic benefits. However, OSHA has a responsibility to apply best practices to manage data privacy and mitigate potential harms to individuals that might arise from data release.
The complexity, detail, richness, and emerging uses for data create significant uncertainties about the ability of traditional ‘anonymization’ and redaction methods and standards alone to protect the confidentiality of individuals. Generally, one size does not fit all, and tiered modes of access – including public access to privacy-protected data and vetted access to the full data collected – should be provided.
Such access requires thoughtful analysis with expert consultation to evaluate the sensitivity of the data collected and risks of re-identification and to design useful and safe release mechanisms.
I invite you to read the full comment here: