Data sharing

[tinatin manjavidze]

Dear all,

Not sure if I'm writing on a correct page, but I'm completely sure that a lot of people here had the same issue/problem.
How do you think, is it possible to upload the health related information file to GitHub? .dta or any other type. If file contains sensitive information about health but without personal information, of course. Around 10 GB

Sorry if I'm asking not on the right place, but since this forum is highly relevant for me (working in stata) and gained a lot of information from the posts, hopefully I did not make a mistake

Thank you in advance,

[Leonardo Guizzetti]

I guess the main questions I would first ask are
1) are you legally allowed to share this data in whole or in part?
2) do health and privacy protection laws in your location require you to make any special measures to protect the data?
3) is explicit patient consent or ethics approval required?
4) what are your legal rights and obligations under any agreements with the body that produces/distributes this data?
5) can this anonymized data become personally identifiable?

If those questions provide a clear path, then the most general way would be to create a plaintext, fixed-width data file with a separate import script which reads in the data and applies formats, labels, etc. One might be needed as well for SAS or SPSS if those will be common among potential users. Of course, the data should be compressed to minimize file size and ideally encrypted to ensure safe use of the data.

[tinatin manjavidze]

Thank yo soooo much. If the answer for last question is yes, does it mean that then I can't upload it? Since the dataset contains 500 variables for each case, its possible to identify the case.

[Leonardo Guizzetti]

I would at least be consulting someone more senior than myself, and possibly a lawyer just to check on the legal aspects. I obviously can't give you specific advice on this aspect.

[tinatin manjavidze]

Thank you, you were really helpful

[Maarten Buis]

You can, but it is probably unethical. So you probably shouldn't.

Moreover, you will probably get into trouble with your university and in many jurisdiction you will get into trouble with the law. Also very good reasons not to do so.

In your university there is probably an ethical committee (e.g. an ethical review board) that can help you with the ethical aspects, then there is probably a legal department that can help you with the legal aspects, your library may or may not have expertise in handling the distribution of data. You can talk to all those, but expect the outcome of those discussions to be that distributing that data is just not feasible.

[tinatin manjavidze]

Thank you, Maarten