Statistics and how they are compiled

[LookslovelyinSpringtime]

Can someone please explain to me where the government statistics are coming from and how they are compiled?
How do they know how many people have been infected with Omicron?
My brother has recently been in hospital with Covid. He was tested when he was admitted. He asked which variant he had. The hospital didn’t know!
Those who take a LFT do not know the variant. Those taking a PCR do not know.
Many people with Covid test and isolate without informing anyone that they have Covid. Others don’t know they have it because they are asymptomatic or have mild symptoms.
The statistics must be guess work or completely wrong.

They send samples to a lab and look at the make up of the virus.

If they do that for 1000 random samples and find 500 Omicron, they can extrapolate the figures to give a guide figure for the population. Something like that probably.

Science & maths. Calculations, not guesses.

@MerryChristmas21

Science & maths. Calculations, not guesses.

Can you explain how this works? You haven’t answered the question.

They randomly sequence a percentage of samples.

Judging by the percentage of samples that come back as a variant they can then estimate the amount of that variant that is circulating.

Where do the samples come from? If not from hospitals, is it PCR tests? Which percentage do they test? We aren’t told any of this!

As explained, not all tests can be checked for variant, but enough are checked to work out a percentage. I think it’s currently around 90% or more omicron, so it’s more than likely your brother had that.

Yes, they get the samples from PCR tests, these can be from home tests, walk in/ drive by test centres, or hospitals.

I’m not sure what percentage of samples get sent for sequencing, it probably varies based on capacity. I believe we are one of the top countries for sequencing though.

They may well have sequenced your brother's sample but will have removed the connection to your brother. So the hospital might not know the strain of each individual patient, but overall they will be able to say that e.g. 90% of patients with positive tests at the hospital had Omicron. It's a lot of admin and additional data protection red tape if you keep the link to an identifiable person, and wouldn't (I don't think) make any difference to a patient's treatment/outcome so no reason to do it.

And earlier in the pandemic I seem to recall that we were sequencing c.1/3 of samples which is incredibly high and would give me a lot of trust in the numbers that are produced.

You can probably work it out

An incredibly quick google reveals

End of June 2021 uk had sequenced 600,000

So find out how many tests done up till that time and you have your ratio

So it's there if you look for it

Does this help, OP?

In short - we sample a % of positive PCR tests (up to 20%, as it depends on volumes at the labs with the facilities to do so). From that we can make reasonably statistically robust estimates.

www.bbc.co.uk/news/health-59460252

Are positive LFT results taken into account when gathering data or is it just PCR tests?
I've registered a positive lft but just wondered if it's not classed as a confirmed case until a pcr comes back.
(Nothing to do with variants but I thought someone might know )

@HumourReplacementTherapy

Are positive LFT results taken into account when gathering data or is it just PCR tests? I've registered a positive lft but just wondered if it's not classed as a confirmed case until a pcr comes back. (Nothing to do with variants but I thought someone might know )

A registered positive LFT is counted yes. If you then take a positive PCR, that is counted and the LFT removed (so you only count once).

Positive PCR samples are sent for genomic sequencing.
The sequencing identifies what variant the infection was.
Genomic sequencing takes 3-7 days to get results.
What % of PCR samples got sequenced, varies by national lab capacity, which depends mostly on staff availability -- but also information needs. No great value in sequencing lots in low periods.
It's been as high as 49% of all samples at times.
In July 2021 was as low as 5%.
30-35% in late November 2021.

As long as the sampling is reasonably unbiased, 10% is enough to make confident extrapolations about general prevalence.

UK is world leading on sequencing. Am pretty sure No one does more sequencing (higher % of positive samples) than UK. China might beat us, I suppose.

We aren’t told any of this!

Sure I have heard a few people mention it on Radio 4 and explain how they work it out. I don't think it get's mentioned a great deal because its quite boring sorry

UK-COG has been putting out press releases since March 2020. Not their fault if media didn't pick up on the PRs very much until September 2020.

Thanks @PatriciaHolm

Thanks this helps me understand it.