Thread by @LawtonTri

As a researcher working with data like this, I'm aware there is a balance between the needs to plan and research, and to reassure people about what happens to their data - particularly health data which was given in the expectation of confidentiality. 1/

theguardian.com/commentisfree/…

There has been a growth - particularly during the pandemic - of massive databases seeking to include as much of the UK population as they can. And COPI rules have (temporarily) removed many of the protections people believed they had. 2/

I believe that data belongs to the people it concerns. So databases like this are best set up and run locally, with local involvement - people and organisations - to ensure data is used in a way that benefits the people it covers. 3/

Equally, the right to opt out is sacrosanct - and I am concerned that some databases do not honour this (because COPI says they don't have to). 4/

Perhaps the opt-out could be more nuanced? "I'm happy to have my data used for planning and research, but not sold" for example. Or defining the types of organisation that it should be shared with (eg for research), or levels of pseudonymisation. 5/

There's a difference between "pseudonymisation at source" and databases that ingest identifiable data and send it to NHS Digital's Privitar tool for example. And other differences in how reversible the pseudonymisation is (clearly irreversible is an extra layer of security) 6/

Local databases have other advantages too - the way we use data in real life isn't like what traditional researchers can be used to. People use data only in as much as it fits their workflow. 7/

Different workflows in different places mean that the data can look very different - despite officially its definition being the same. ie you can't look at data without considering where it came from and why it was entered. 8/

Bringing local people with you is therefore important for all sorts of reasons - and it feels like this is what's at risk of being lost here. To the point where I find myself wanting to opt out, but in a way that doesn't keep my data from being used locally. 9/

And unfortunately that option is not available - so I'm back with a binary choice - and this is going to affect a lot of people.
No answers here unfortunately! 10/

Lastly, I think transparency is key - everyone should know exactly what data (of theirs!) has been given out and to whom and why. Hopefully this prevents some of the shader options. 11/

COI: As alluded to at the top, I work on the Connected Bradford database at @CityOfResearch - I believe we do it right, and I hope this controversy helps ignite an interest in how routine data is used. n/