How so? Because of the "easily" qualification? Is this ironic? If it can't be traced to you it is by definition perfect privacy.
You might insist that people knock before entering your room to respect your privacy. No identification is relevant here; your housemates know who and where you are already.
Someone might ask you what you did this weekend, and you might not want to reveal that you spent the weekend on a hobby you're embarrassed or sensitive about. Maybe you spent the weekend writing your novel, but you aren't ready for people to know about it yet. If you tell a white lie that you spent the weekend camping, you're doing so to preserve your privacy. Again, no identification is relevant here.
If something can be linked back to you, then you lose control of the information and your privacy is compromised. But that's an implication of the definition, not the definition itself.
It is a violation because of the lack of consent, but it is also a violation because now you have to live with the fact that maybe some other information can link you to that picture in the future. There is no such thing as perfect privacy if any link can be created in the future, and the fear of the creation of that link going forward is itself a problem for the subject.
In this hypothetical it is not a privacy violation under HIPAA (US) or PHIPA (Ontario, Canada) as consent would not be required.
Whether such a level of deidentification is possible is a different question and I agree it’s best to avoid the situation altogether.
> if any link can be created in the future
An example of this is CT/MRI scans of the head that were shared in open databases. The technology now exists to make a 3D rendering of the patients face which was not anticipated 15 years ago when these datasets were made.
That is not categorically true, PHIPA requires research ethics board to review research disclosures to determine if consent is required. So what is "too much" information that might violate privacy under Ontario law is a moving target, as our ability to violate privacy with less and less information progresses the standard should be more and more restrictive.
Research use of course does not mean that the data needs to be public, as in my example, and raw data is often not made public. In your example it was, if that same database were created today I am confident the ethics board enlisted to review the disclosure would disallow it.
Research requires REB approval. Institutional policy usually requires a REB to determine that the proposed initiative is not research but this varies. It is trivial to get QI approved.
The act of deidentification is considered a "use" and therefore would require an REB approval or be structured as QI. However, once deidentified the data is no longer protected under PHIPA and can be used for anything.
There is some nuance to this, such as you can't have a planned use of the deidentified data and not have disclosed it in the initial QI or research plan, but you can subsequently use the data as long as it was not planned at the time of initial application.
Without getting lost in the weeds, there is little protection on deidentified data once it exists.
That's what I said I think.. the law is about making it exist and the standard for being allowed to make it is dictated by the REB - so that's a moving target, you can't say specifically what would be allowed today will be allowed 5 years from now.
There are easier non-REB exceptions to consent that can facilitate deidentification and subsequent use that are commonly used:
PHIPA sets out a limited set of acceptable uses of personal health information without consent, including, for example, the following purposes:
• planning or delivering programs or services
• risk management, error management or activities to improve or maintain the quality of care or any related program or service
• educating agents to provide health care
These are just standard legislative exceptions saying "you can use the data you have to run the operation without fear of getting in trouble" not publication rules.
ETA: You obviously know about the legislation, if you know of instances of orgs using these rules to publish data.. blow that damn whistle!
Some institutions and systems have internal policies that add exceed PHIPA. For example ConnectingOntario which is the closest thing to a provincial EMR only allows access for direct patient care (i.e. not even REB approved research). Some academic hospitals also add approval requirements for the disclosure of deid data to other institutions or on credentialed access repositories.
I started my medical career in Ontario and this is done all the time for things like publishing case reports, online (or non-institutional) lectures and case banks (e.g. on the publicly accessible Radiopaedia).
Not sure what you mean by whistleblowing, this is completely legal in both Ontario and the US and in my opinion entirely ethical (for the most part).
I’m not saying it’s a good idea or ethical to use these exceptions to consent for posting marketing materials on TikTok. The CPSO which regulates physicians may still find the action unprofessional/unethical for reasons not related to violating the law, but their standard is different (“behavior unbecoming of the profession” and “behavior that degrades patient trust in the practice of physicians”).
On a side note I think I’m getting old but what does “ETA” mean (other than estimated time to arrival)?
Thank you for the information.