But what you can do in this study, is have a dataset with a good size (4m people from the UK!) which allows for much broader analysis than would be possible through other means.
Blood testing 4m people during a pandemic is impractical, and widescale self-reporting will provide insights not otherwise available (as this paper shows).
On at least two occasions China has swab-tested 10 million people in a single city within a week, when covid re-occurred in a city where it had been thought to be extirpated. I don't think blood testing is twice as difficult as swab testing. So, I don't think it's impractical per se. It's just impractical in the UK.
You know what's impractical? 2.7 million people dying and an unknown fraction of the survivors suffering disabilities and injuries such as diabetes, heart attacks, and cerebrovascular attacks.
That’s different to focusing resources on doing vitamin D testing for a speculative study to determine if there is a link between vitamin D and long Covid.
I’m just saying that given limited resources during this global pandemic, the best use of resources at present probably isn’t monitoring 4 million people’s vitamin D levels for this specific study.
I’m also saying that not every study has to study every possible factor - in fact this is almost impossible to do! This is a brilliant dataset of millions of people self reporting, and there are things that you can find with self reporting that you can’t find with blood testing (and visa versa). This is why other papers will do a literature review to compile together the results of multiple studies to understand the impact.