Short THREAD on log scales and why I tend not to use them.
I've been criticised at times for not plotting case numbers etc on a log scale. I want to explain why.
TLDR: essentially it's because the burden of covid ill health and on NHS is people not log people. 1/8
When you've got exponential growth (e.g. in new covid cases or hospital admissions, it will be a straight line if you put the vertical axis on a "log scale" (normally equally spaced powers of 10).
Charts show same exponential growth on a normal scale and a log scale. 2/8
Log scales are really useful for e.g:
a) seeing if growth is exponential (straight line?)
b) looking for acceleration or slow down in growth
c) comparing growth between countries at different stages of epidemic 3/8
And if I'm trying to do those things I will use log scales and (sometimes) share log scale plots.
But often I stick to normal ("linear") scales and explain that growth rates are equal or that growth is exponential etc in words instead.
Why? 4/8
Firstly, most people don't look at data charts very often.
So I try to make mine as intuitive & foolproof as possible. Most people haven't done logs since they were teenagers & don't remember how they work.
Logs are also NOT intuitive. Our brains don't work that way really. 5/8
You can label your charts brilliantly on log scales, but often people react as if they are still showing you *linear* effects - not exponential ones. This means that people might instinctively underestimate the scale of the problem.
Finally, the *impact* of exponential growth in ill health is on people. They don't come in logs.
Impact is exponentially worse the longer exponential growth continues. Acting early makes a HUGE difference.
Linear scales do a better job of showing that than log scales. 7/8
If your audience is people who have adapted to log scales then logs are fantastic. I avidly follow many people here who use them because I find them very informative.
But for a general audience who might spend 10 seconds looking at your chart, I much prefer linear. 8/8
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Quick thread on current Covid situation in England and Long Covid.
I have Thoughts about the Inquiry Report published yesterday but am still trying to organise them.
TLDR: high Covid levels remain, Long Covid remains 1/11
This wave is not over. While the number of admissions with Covid remains lower than the autumn/winter waves, it has now remained highsh for several weeks.
This means there are a lot of people out there getting sick - and having their work, plans and holidays disrupted. 2/11
Scottish wastewater data to 9 July shows a sharp decrease, suggesting that prevalence might be on its way down.
Obviously Scotland and England can have different dynamics, but it’s the best we’ve got as long as England refuses to analyse its own wastwater. 3/11
THREAD: Given tomorrow's election, I've been thinking about our nation's (poor) health, the wider determinants of health and how these have worsened and what it means for policy....
TLDR: worrying only about NHS & social care is missing the point
let's dive in... 1/25
The UK has a health problem. After steady gains in life expectancy for decades, it flatlined during the austerity years and fell for the first time this century with the Covid pandemic.
The number of people out of work for long term sickness is near record levels. 2/25
There are huge inequalities between rich & poor. Boys born in the most deprived areas can expect to die almost 10 years earlier than their peers in the least deprived areas.
Even worse, they can expect to spend 18 fewer years of their life in good health (52 vs 70 years) 3/25
As ever, I am getting lots of pushback.
Here is a compilation of the European countries I've found with recent wastewater data. Some are going up a bit, some down a bit, some are flat, none are anywhere near previous peaks.
I can't see anything here to be panic anyone. 1/3
I can't find the dashboard for Spain, but others saying it is in a wave. Perhaps it is. England has just had one - the last data we had (a couple of weeks ago from Bob Hawkins) looked as if our wave had peaked.
So, I'm not seeing reason to think things are terrible here! 2/3
Yes there are new variants growing right now. They are not growing faster than JN.1 grew in December and that wave did not end up as bad as feared.
Clearly it remains true that Covid is NOT a seasonal disease (unlike Flu and RSV)
3/3
Quick thread on the Astra Zeneca (AZ) covid vaccine since it's been in the news today.
TLDR there isn't a new "smoking gun", the AZ vax was one of first and cheapest, it saved millions of lives globally, there are better vax out there now, adapted to new variants 1/9
the AZ vaccine was one of the first approved at the end of 2020, cheaper than Pfizer, and - importantly - easier to administer in lower resource settings as it didn't require super low temperatures for storage 2/9
In most countries it was first rolled out in older adults. As it was rolled out in younger adults, a *very rare*, serious, side effect was noticed - it could cause deadly blood clots
This was spotted quickly and studied. Vax monitoring did its job. 3/9
A short thread on why this is not a scary chart and why all the evidence suggests that there is not much Covid around right now. 1/6
the above chart is recorded covid hospital admissions / reported covid cases. It is close to 100% now *because basically only hospitals can report cases since Feb 2024*
It is to do with changes in case reporting and NOT hospital testing
2/6