search

opensafely / emis-qa

MIT License
0 stars 0 forks source link

Track the strange movements in patient ids #3

Closed sebbacon closed 3 years ago

sebbacon commented 3 years ago

See #2 for general background info

Total numbers

On the 10th March we had approx 34 million registered patients. On the 15th March there were about 300,000 fewer registered patients (precise numbers here). We've been advised by EMIS CMO that 300k is too much for "normal churn".

Strange patient id drift

During the same period, over 1.1m patients were in the 10/3 data but are not in the 15/3 data, and over 860k patients were in the 15/3 data but not the 10/3 data.

sebbacon commented 3 years ago

This is in EMIS' court now. We could run some kind of analysis on practice, age etc associated with these odd patients, but they're going to take a look when the current build finishes (expected today) (slack)

sebbacon commented 3 years ago

From EMIS (link):

we did some more digging into patient ids set changing... there seems to be lot of infrastructure changes happening at the moment particularising around historical "deleted" patients being re processes of the past month basically..

those changes saw lots of historical movement being implemented as deletes, so the core / ingest team suggests that it is finished now and we should monitor for the next few re re frefreshes

as previously stated we do not think any of it impacts your paper outcomes though as has mostly to do with patines deleting and migrations

for example there is an org that had 9000 historical deleted patients 'deleted' on 3rd of March

image

sebbacon commented 3 years ago

In conclusion: as far as well can tell this is "something to do with implementation of mergers that you shouldn't worry about".

I think we've reached the end of the line on this one