4 ms·

If there is a publicly available picture of you that you did not consent to be published and literally nobody can currently identify it as you that is still a p

If there is a publicly available picture of you that you did not consent to be published and literally nobody can currently identify it as you that is still a privacy violation. 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.

BurningFrog 3y ago
The lack of consent makes it a consent violation, not a privacy violation.
- toofy 3y ago
  depending on jurisdiction and contexts i think it would likely make it both.
haldujai 3y ago
> literally nobody can currently identify it as you that is still a privacy violation. 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.
- JamisonM 3y ago
  "In this hypothetical it is not a privacy violation under ... PHIPA (Ontario, Canada) as consent would not be required." 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.
  - haldujai 3y ago
    Under PHIPA you can essentially use PHI for: direct patient care, research, or quality improvement. All QI does not need patient consent. Most research also does not need patient consent. 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.
    - JamisonM 3y ago
      > 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.
      - haldujai 3y ago
        Apologies, I may have misunderstood you to extend it to deidentified data. 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
        JamisonM 3y ago
        These are "uses" of data but they don't seem to in any way involve making the data public - I think if you published deidentified data to a public forum of any kind under the auspice of "educating agents" you'd get a very big fine! (You need example charts to educate certain healthcare workers, using real ones with the name blanked out is fine - publishing that on the Internet probably going to get you a violation.) 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!

BurningFrog 3y ago

The lack of consent makes it a consent violation, not a privacy violation.

toofy 3y ago
depending on jurisdiction and contexts i think it would likely make it both.

haldujai 3y ago

> literally nobody can currently identify it as you that is still a privacy violation. 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.

JamisonM 3y ago
"In this hypothetical it is not a privacy violation under ... PHIPA (Ontario, Canada) as consent would not be required." 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.
- haldujai 3y ago
  Under PHIPA you can essentially use PHI for: direct patient care, research, or quality improvement. All QI does not need patient consent. Most research also does not need patient consent. 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.
  - JamisonM 3y ago
    > 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.
    - haldujai 3y ago
      Apologies, I may have misunderstood you to extend it to deidentified data. 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
      - JamisonM 3y ago
        These are "uses" of data but they don't seem to in any way involve making the data public - I think if you published deidentified data to a public forum of any kind under the auspice of "educating agents" you'd get a very big fine! (You need example charts to educate certain healthcare workers, using real ones with the name blanked out is fine - publishing that on the Internet probably going to get you a violation.) 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!