I've taken the various bits of analysis and visualisation I've done with
#RStats
in relation to COVID-19 data, tidied the code up a bit, updated it using the latest data and put it all in one place:
I still can't get over how stark the divide in COVID-19 cases between Northern Ireland and the Republic of Ireland is. It's been like this for *months*.
I've seen a few people recently shouting about how new ONS data shows that the "true" death toll from COVID in England & Wales is only 17,371, rather than the ONS figure of 157,816.
This is obviously nonsense, but evidently still needs debunking, so here goes...
So this is not good.
Both new COVID cases and hospital admissions are rising again in England.
Just how bad this is remains to be seen. But trying to work out how worried we should be (if at all) requires some understanding of *why* it's happening, and honestly๐คท
/THREAD
This is one of the stupidest things I have ever read in the name of science.
No, red wine does not ward off COVID.
Not being poor (which is strongly correlated with red wine consumption) does though.
I accidentally made a map comparing rainfall and sunshine hours across the UK in 2021.
Conclusive proof that Yorkshire>Lancashire (among other things).
High-res version here:
Since
@ONS
now publish experimental inflation indices showing how different population groups experience inflation rates differently (thanks to
@BootstrapCook
) we can see how inflation rates differ by income, and it isn't pretty.
I try and share the R code that produces all of the graphs that I post on here alongside each graph.
I was wary at first, but it's turned out to be an *excellent* decision.
Here's why and why I think you should do the same:
Reasons to be (cautiously) optimistic:
COVID cases in Scotland have been rising steadily for 6+ weeks, but the number of patients in hospital with COVID has barely risen over that time.
This is very different to how things looked at a similar stage in October.
Hong Kong is currently experiencing a horrifying COVID-19 outbreak. The mortality rate is approaching the UK's peak in January 2021.
Why, if Omicron is so much milder?
Most of that mildness is really due to vaccination and Hong Kong has hugely failed to vaccinate older people.
Playing around with ONS' data on prices and this is something I hadn't realised before.
The variation in the price of a pint across England has increased *massively* in the last decade or so.
More good news for Scotland - cases have been falling for a couple of weeks and now the number of people in hospital with a positive COVID diagnosis looks to have started to fall too.
Not counting deaths with pre-existing conditions is saying to people with these conditions that their lives don't matter.
The argument being made is that only the lives of people with no pre-existing health conditions count.
People with pre-existing conditions are real people. Their lives matter. They are your parents, your grandparents, your friends, your children. Very possibly you.
It seems extremely bizarre to claim that this means the person died from asthma and not from COVID. Perhaps they wouldn't have died if they didn't have asthma, but they *definitely* wouldn't have died if they didn't have COVID.
This is obviously repugnant. It's the even nastier cousin of the 'they were old and would have died soon anyway' argument that we've heard trotted out many times, in spite of the fact that the average years of lost life per COVID death is ~10.
But I think it says a lot about the kind of person making these arguments that rather than go and look for the data to answer a question, their first resort of just to send an FOI request demanding somebody provide it to them on a plate.
#30DayMapChallenge
Day 7 - Raster
I normally make a new map each day, but I'm quite proud of this one that I made a few months ago - annual sunshine hours vs. rainfall across the UK on a bivariate map.
#RStats
code here:
If you are seriously making the argument that only 17,371 people have *really* died from COVID then you are saying that the ~140,000 people who died with pre-existing conditions were expendible and didn't matter.
After a year off (because, you know, stuff happened), the Health Survey for England ran again in 2021 and we now have some data from it via
@NHSDigital
The increase in the proportion of young people who don't drink alcohol is *astonishing*.
Recent analysis from AP found that 99.2% of COVID deaths in the US in May were in those not fully vaccinated.
In England, PHE data shows the figure is more like 57%.
This sounds bad, but actually, this is *good* news for the UK ๐งต
This also looks like good news - the recent increase in people in English hospitals with positive COVID tests is mostly driven by patients who are being treated for something else, *not* COVID.
Once you get past your mid-50s, chances are you have a long-term health condition. By the time you reach your 70s, you have to be extremely lucky not to.
Data from ICNARC shows that the vast majority of people being admitted to critical care were happily living without any support prior to admission.
These are not people who were already at death's door.
I've been looking at vaccination rates in the USA and England a bit more.
The US has vaccinated 10m+ teenagers, but has lower vaccination rates in older age groups compared to England, where vaccine rollout was much more strictly tied to age.
How has the failure of the Immensa lab in Wolverhampton affected hospital admissions in the areas that saw the most incorrect test results returned?
The government line is "not at all".
I beg to disagree.
๐งต
Something like a third of the population have high blood pressure (hypertension), ~5 million people in the UK have diabetes, just under a third are obese etc. etc.
I posted this graph yesterday as part of the
#30DayChartChallenge
and a few people have asked some reasonable questions about why I chose to present this data in this way.
So I thought I'd write a thread to explain my thought process...
When it comes to COVID, we already know that the vast majority of deaths where COVID is mentioned on the death certificate list COVID itself as the underlying cause.
If we look at cases, and specifically the rate at which they are changing relative to 7 days previously, then it's really striking how *every* age group has suddenly changed direction at the same time.
Dear rude people of the internet,
Before you hit send on that oh so clever tweet saying โthis analysis is shit, you should look at x insteadโ, how about finding out if there is actually any data on x to analyse first?
Yours, someone who is tired of your shit x
So this, I think, strongly suggests one of two possibilities:
- some population-level shift in behaviour has suddenly increased the number of people getting COVID
- BA.2
And prevalence of these conditions is strongly linked to age. Slightly different definition (though similar), but here's a great study from Scotland on the prevalence of long-term health conditions by age.
COVID is a respiratory virus. Many of the people who get severely ill have trouble breathing and need ventilation. It is hopefully obvious that having asthma will potentially exacerbate these difficulties and will therefore be listed as a contributory cause if they sadly die.
So that is a *lot* of people with pre-existing conditions. If any of them were unlucky enough to have caught COVID, get severely ill and die, all it takes is for these conditions to have *some* impact for it to end up as a contributory cause on the death certificate.
So much of what we observe in the COVID data is driven by age dynamics - cases rising and falling in different age groups at different times.
Here's the full tapestry of the pandemic in England (for cases at least). So many different stories written into this plot.
I was a bit sceptical at first, but I'm increasingly convinced that the gender gap that has emerged in COVID cases in 20-39 year-olds in England is a direct consequence of the football.
And it's too early to expect any effects from the semi-final or final to show in this data...
The claim comes from an FOI request that some genius sent to the ONS asking them for the number of deaths where *only* COVID was listed on the death certificate without any other pre-existing conditions listed.
But this new claim is that *any* deaths where there is a pre-existing condition listed on the death certificate alongside COVID aren't really COVID deaths. Because obviously these people are sick and were just about to die anyway before they caught COVID.
To illustrate how stupid this whole argument is, let's take an example - asthma. The ICD-10 code for asthma is J45, so it's on the list of pre-existing conditions.
But what does it mean to die with both COVID and a pre-existing condition on the death certificate?
Death certificates list an "underlying cause" and can list several "contributory causes".
Omicron has really taken hold in the US, how worried should we be?
The US hasn't seen the nice decoupling of cases from admissions and deaths that we have in the UK. Partly because of lower testing (I believe), but also lower and less well targeted vaccination uptake.
So that's a long-winded way of saying that I don't know what's going on, and I haven't seen anyone else come up with a compelling theory yet. I'm sceptical about it being waning vaccine effectiveness though.
I'm a little surprised at the number of high profile data visualisation people on here that I've seen dunking on this chart from the NYT.
We've seen (and I've made) a million boring line/area charts of COVID cases. It's refreshing to see somebody trying something new.
Well, no.
The number of people being admitted to hospital with COVID, or testing positive in a hospital setting, has started rising *at the same time* in all age groups.
Here is some *much* better news.
Rates of COVID-19 vaccination have picked up *massively* in the last week.
Yesterday, we reported vaccinating over a quarter of a million people. In a day. That's pretty amazing.
This pushes us away from any explanations that might be age-specific, for example the (entirely plausible) idea that the protective benefits of boosters are starting to wear off. Because the booster rollout was very staggered with age.
Here's another way of looking at recent patterns in COVID case rates.
Cases still falling, but much more slowly, in all age groups except for primary school kids and their parents age bracket where cases are rising again.
@ledredman
@HalseyJane
Sadly the kind of people making these arguments are also the kind of people who will just claim that all excess deaths are "lockdown deaths" caused by the response to COVID. It's like reasoning with treacle.
Comparing COVID admissions now with 12 months ago, it's impressive how well vaccines have kept adults protected from severe disease.
For under 18s, it's a rather different picture, both during Delta and now Omicron waves.
NHS England have just published their weekly breakdown of patients in hospital 'with' or 'for' COVID.
In London, both numbers have really shot up in the last 2 weeks, but two thirds of patients in acute hospitals who have COVID are there *because* of COVID.
The areas in Great Britain which currently have the highest rates of COVID-19 cases are generally areas where fewer people work from home.
Lots of other factors at play, but the prevalence of red (high cases, low WFH) and turquoise (low cases, high WFH) is clear.
There is always a bit of a lag on genomic surveillance, but as of 26th February, the BA.2 variant of Omicron was dominant almost everywhere in England.
Our new paper looking at alcohol, drug and suicide mortality during the first 2 years of the pandemic in the USA and the UK nations has just been published in
@RSPH_PUHE
Lots of attention on these deaths early in the pandemic, so what actually happened?
This graph looks pretty remarkable - in mid September, COVID case rates suddenly became highest in the least deprived areas, a sharp reversal of the prevous trend.
So what's going on here?
๐งต
And pretty much at the same time in all regions, although there is a little bit of variation - admissions have already exceeded their Omicron peak in the South East and South West in the oldest age groups.
Three things that I think will be interesting to look out for tomorrow:
- New data from NHS England on the with/for COVID breakdown of admissions
- New data from NHS England with amore detailed age breakdown of admissions
- A new
@UKHSA
surveillance report
Why is it so hard to work out what's going on with this new wave of COVID cases in the UK?
The answer lies in the fact that a *lot* of things have changed in comparison to previous waves, and it's *really* hard to parse out the specific impact of each contributing factor...
I've seen a few sources talking about the new COVID-19 variant being more transmissible among younger children.
I've no reason to doubt this, but it's hard to look at this graph and infer that young children have been driving the recent rise in case numbers.
Genuinely amazed at the number of tedious COVID chumps who seem to think that pointing out that the average age of a COVID death being close to life expectancy demonstrates anything other than their own ignorance of how life expectancy works.
Still lots of reasons to be very cautious about interpreting this (day of the week effects, changes in ability to get tests, changes in test processing times etc etc) but this looks pretty good I'd say.
To put this another way:
When nobody is vaccinated, 100% of deaths are in unvaccinated people.
When everyone is vaccinated, 100% of deaths are in vaccinated people.
There's some amazing data hidden away in ONS's user requested data section.
Last year they quietly published death data by age, deprivation decile and cause, which means we can look at age patterns in alcohol deaths by IMD, with sadly predictably grim results:
I've been meaning to make a plot like this for a while, after seeing
@jburnmurdoch
do something similar.
This is COVID case rates by age, relative to their peak level last winter, coloured by booster coverage.
As the lines turn green, they go down ๐
COVID vaccination rates by age in the US (in blue) compared to England (in red) - double jabbed in darker colours.
The US has better coverage in the youngest age groups, but much worse coverage in older ages.
As COVID risks vary so much with age, this is bad news for the US.
But nor am I convinced that the timing of relaxing restrictions entirely fits with what we're currently seeing. And there is the oddity that admissions started rising *before* cases, even though admissions should lag cases.
Every single person I know working in Public Health, be that PHE/W/S, local PH teams or other organisations, has been working incredibly hard in this past year and achieved some *incredible* things.
If your worldview involves them being the baddies, then youโre very wrong.
There really is a lot going on in the recent COVID case numbers in England.
Cases back to being highest (and rising fastest) in primary age children.
Cases starting to rise again in their parents' age group.
Falls stalling in other ages.
Let's do some graphs...
One for the 'but they are all incidental admissions crowd here'.
In every English region, the majority of patients in hospital with COVID are being treated *for* COVID. In many areas, it's a substantial majority.
And the numbers are rising *everywhere*.
This is great.
Everything you ever wanted to know (and plenty you probably didn't know you ought to know) about choosing and using fonts from the always excellent
@lisacmuth
So I think both of these explanations are *plausible*, but I'm not totally convinced about the timing of either of them. BA.2 has been growing since early January, so that doesn't really fit (IMO) with a sudden, simultaneous change in cases and admissions.
The number of patients in English hospitals being treated for COVID keeps on falling and is now smaller than the number of patients 'with' COVID but being primarily treated for something else.
BA.2 clearly has a growth advantage over OG Omicron, but UKHSA analysis doesn't suggest it leads to more severe outcomes, or is any better at overcoming vaccinations.