search

opensafely / vte-diagnoses-and-deaths

Study to describe the impact of COVID-19 on mortality related to venous thromboembolism in England
MIT License
1 stars 0 forks source link

Finding covid tests close to time of VTE diagnosis #7

Open matthewberesford92 opened 1 year ago

matthewberesford92 commented 1 year ago

Just been reading the OS documentation for SGSS and came across this line 'Multiple tests for the same person are treated as a single 'infection episode', no matter how far apart, and so the system will only return the earliest test (sample date and report date), and all subsequent tests are dropped.'

In our study we want to find positive tests in a time bracket anchored to the date of VTE diagnosis. We could have an issue of missing relavent covid diagnosses if someone has tested positive many months before and then tests positive close to their VTE. From my reading, the two covid tests would essentially be assigned as a single episode and dated to the time of the first.

I am sure other researchers must have overcome this issue. Any ideas @Jongmassey ?

FYI @joshuamutio

joshuamutio commented 1 year ago

It looks like this only applies to the "Earliest Specimen Dataset"; there is also an "All Tests Dataset" for SGSS which isn't restricted in the same way. The guidance says "some fields" aren't available in the All Tests Dataset, but it only goes on to list one (Case Category, e.g. LFT, PCR - which I don't believe we need to distinguish between). So, barring there being other limitations not listed in the guidance, the All Tests Dataset should be fine.