| Im struggling to understand how Covid stats get interpreted 12:19 - Dec 15 with 1017 views | nodge_blue |
This is from The Times:
For every 1,000 infections, 44 people ended up in hospital in South Africa in the first wave, 101 in the Delta wave and 38 in Omicron. He and his colleagues estimate that after correcting for vaccination status, age and previous infection, Omicron is a third less severe than the original Wuhan strain of coronavirus, which is in turn less serious than Delta.
So thats 1000 infections. Not if you get it etc. Its an infection. First wave 44 hospitalisations per 1000. Omicron 38 hospitalisations per 1000. So if its a third less severe then why isn't it 15?
How can you you allow for "other" factors when what we are looking at here is infections and hospitalisations ratio? |  |
| |  |
| Im struggling to understand how Covid stats get interpreted on 12:23 - Dec 15 with 965 views | Dennyx4 |
Are they comparing the 38 to the 101 which is nearly a third? |  | | |
| Im struggling to understand how Covid stats get interpreted on 12:25 - Dec 15 with 943 views | StokieBlue |
That doesn't make any sense to me.
"He and his colleagues estimate that after correcting for vaccination status, age and previous infection"
How can you adjust for previous vaccinations and immunity when none existed for the original Wuhan strain? That number should remain at 44 in most cases (which some age adjustment) and then the Omicron number should be adjusted upwards to take into account vaccinations and immunity otherwise you're not comparing like for like.
Given those assumptions and how the scaling would work I don't see how it could ever be less severe than the Wuhan strain, certainly not 30% less.
Being 30% less severe than Delta looks more likely given vaccination status and immunity.
SB [Post edited 15 Dec 2021 12:26]
|  | | |
| Im struggling to understand how Covid stats get interpreted on 12:25 - Dec 15 with 939 views | Ewan_Oozami |
1/3 less than 44 is about 30, not 15... |  |
| |
| Im struggling to understand how Covid stats get interpreted on 12:29 - Dec 15 with 892 views | nodge_blue |
| Im struggling to understand how Covid stats get interpreted on 12:25 - Dec 15 by Ewan_Oozami |
1/3 less than 44 is about 30, not 15... |
yes I read that wrong. I was thinking it was only a third. You're right a third less. Even then the maths don't add up? | |
| |
| Im struggling to understand how Covid stats get interpreted on 12:30 - Dec 15 with 873 views | nodge_blue |
| Im struggling to understand how Covid stats get interpreted on 12:25 - Dec 15 by StokieBlue |
That doesn't make any sense to me.
"He and his colleagues estimate that after correcting for vaccination status, age and previous infection"
How can you adjust for previous vaccinations and immunity when none existed for the original Wuhan strain? That number should remain at 44 in most cases (which some age adjustment) and then the Omicron number should be adjusted upwards to take into account vaccinations and immunity otherwise you're not comparing like for like.
Given those assumptions and how the scaling would work I don't see how it could ever be less severe than the Wuhan strain, certainly not 30% less.
Being 30% less severe than Delta looks more likely given vaccination status and immunity.
SB [Post edited 15 Dec 2021 12:26]
|
Thats what I thought. It would make sense based on Delta but not the Wuhan strain. | |
| |
| Im struggling to understand how Covid stats get interpreted on 12:31 - Dec 15 with 866 views | bluelagos |
So the 1000 infections (That led to 44 hospitalisations for wave 1 and 38 for Omicron) could well be the reported infections rather than actual infections.
Actual infections will be far more than reported ones for a number of reasons - people not getting tested as they are asympomatic, unavailability of tests etc.
So they will try and model (estimate) the actual number of infections and from that they estimate that the proportion of actual hospitalisations compared to the estimated total number of infected people.
The estimates/models of infected will also take account of community testing - where they test people on an ongoing basis - sympotamic or not - to help estimate the total numbers of infected in the whole population. |  |
| |
| Im struggling to understand how Covid stats get interpreted on 12:32 - Dec 15 with 852 views | StokieBlue |
| Im struggling to understand how Covid stats get interpreted on 12:30 - Dec 15 by nodge_blue |
Thats what I thought. It would make sense based on Delta but not the Wuhan strain. |
Worth noting that the Delta strain has killed millions of people so being 30% less severe than Delta is obviously good but doesn't mean it's harmless.
My worry is the narrative that it's "milder" is being spun in certain circles to mean "harmless".
SB | | | |
| Im struggling to understand how Covid stats get interpreted on 12:34 - Dec 15 with 830 views | nodge_blue |
| Im struggling to understand how Covid stats get interpreted on 12:31 - Dec 15 by bluelagos |
So the 1000 infections (That led to 44 hospitalisations for wave 1 and 38 for Omicron) could well be the reported infections rather than actual infections.
Actual infections will be far more than reported ones for a number of reasons - people not getting tested as they are asympomatic, unavailability of tests etc.
So they will try and model (estimate) the actual number of infections and from that they estimate that the proportion of actual hospitalisations compared to the estimated total number of infected people.
The estimates/models of infected will also take account of community testing - where they test people on an ongoing basis - sympotamic or not - to help estimate the total numbers of infected in the whole population. |
Ok. I can see that. Although I would have thought it was virtually impossible to guess actual infections in the first wave given the lack of testing.
But your answer makes more sense. | |
| |
Login to get fewer ads
| Im struggling to understand how Covid stats get interpreted on 12:38 - Dec 15 with 800 views | nodge_blue |
| Im struggling to understand how Covid stats get interpreted on 12:32 - Dec 15 by StokieBlue |
Worth noting that the Delta strain has killed millions of people so being 30% less severe than Delta is obviously good but doesn't mean it's harmless.
My worry is the narrative that it's "milder" is being spun in certain circles to mean "harmless".
SB |
Yes its clearly not harmless. And the milder nature gets negated by the increased transmissibility. | |
| |
| Im struggling to understand how Covid stats get interpreted on 12:49 - Dec 15 with 727 views | bluelagos |
| Im struggling to understand how Covid stats get interpreted on 12:34 - Dec 15 by nodge_blue |
Ok. I can see that. Although I would have thought it was virtually impossible to guess actual infections in the first wave given the lack of testing.
But your answer makes more sense. |
Sampling - you actually don't need that big a sample size as you may think. Opinion polls are based on responses of around 1k for a voting population of around 25million.
The 10k poll of the exit polls is incredibly accurate - again 25m population.
I don't know how big the ONS sample sizes are - but I do know these bods are all statisticians and will use robust methods for estimating infection rates.
And these rates will then be used by the modellers to build their models and to extrapolate forward what they think will happen. They know what they are doing, whilst accepting a level of uncertainty in their projections is inevitable.
People wanting 100% accuracy or certainty are going to be disappointed. People wanting the best possible forecasts should accept these guys know their sh1t. | |
| |
| Im struggling to understand how Covid stats get interpreted on 13:01 - Dec 15 with 668 views | homer_123 |
| Im struggling to understand how Covid stats get interpreted on 12:49 - Dec 15 by bluelagos |
Sampling - you actually don't need that big a sample size as you may think. Opinion polls are based on responses of around 1k for a voting population of around 25million.
The 10k poll of the exit polls is incredibly accurate - again 25m population.
I don't know how big the ONS sample sizes are - but I do know these bods are all statisticians and will use robust methods for estimating infection rates.
And these rates will then be used by the modellers to build their models and to extrapolate forward what they think will happen. They know what they are doing, whilst accepting a level of uncertainty in their projections is inevitable.
People wanting 100% accuracy or certainty are going to be disappointed. People wanting the best possible forecasts should accept these guys know their sh1t. |
Big big big differences in sample size requirements depending on your confidence level requirements and your margin for error.
Then....depending on what you are doing your main overall sample may not be representative of differing sub and demographic groups. |  |
| |
| Im struggling to understand how Covid stats get interpreted on 13:07 - Dec 15 with 639 views | bluelagos |
| Im struggling to understand how Covid stats get interpreted on 13:01 - Dec 15 by homer_123 |
Big big big differences in sample size requirements depending on your confidence level requirements and your margin for error.
Then....depending on what you are doing your main overall sample may not be representative of differing sub and demographic groups. |
Absolutely - and one big bias in infection rates from testing - is the bias of "self selection" of those ordering and using test kits - namely those who ask for tests are more likely to be elderly / vulnerable etc.
That is why raw data needs to be adjusted and can't simply be used/quoted as someone tried to do on here a few weeks back. These guys know their sh1t :-) | |
| |
| Im struggling to understand how Covid stats get interpreted on 14:27 - Dec 15 with 552 views | WD19 |
| Im struggling to understand how Covid stats get interpreted on 13:01 - Dec 15 by homer_123 |
Big big big differences in sample size requirements depending on your confidence level requirements and your margin for error.
Then....depending on what you are doing your main overall sample may not be representative of differing sub and demographic groups. |
For those that are interested, the ONS CIS study looks like a bit like this:
"Since August 2020, we expanded the survey to invite a random sample of households from the AddressBase. Fieldwork increased in England, and coverage of the study was extended to include Wales, Northern Ireland and Scotland. Survey fieldwork in Wales began on 29 June 2020 and we started reporting headline figures for Wales on 7 August 2020. Survey fieldwork in Northern Ireland began on 26 July 2020 and we started reporting headline figures for Northern Ireland on 25 September 2020. Survey fieldwork in Scotland began on 21 September 2020 and we started reporting headline figures for Scotland on 31 October 2020.
Ultimately, the swab target is to achieve approximately 150,000 individuals with swab test results at least every fortnight from October 2020 onwards in England, approximately 9,000 in Wales, approximately 5,000 in Northern Ireland and approximately 15,000 in Scotland (approximately 179,000 total across the UK). The blood target is to achieve up to 125,000 people with blood test results every month in England, and up to 7,500, 5,500 and 12,000 per month in Wales, Northern Ireland and Scotland respectively (approximately 150,000 in total across the UK). The absolute numbers reflect the relative size of the underlying populations."
And yes, as BL says, they know what they are doing. |  | | |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
