David Paton Profile picture
Professor of Industrial Economics Nottingham University Business School (views expressed are my own)
BLACKHAT WOMAN / women won't wheesht!! Profile picture Sharah Kasny Profile picture Anthony Profile picture 4 added to My Authors
May 23 8 tweets 3 min read
This paper in Frontiers in Public Health is the most recent & comprehensive study of Covid restrictions (NPIs). They look at 10 NPIs in 169 countries. Conclusion:
‘None of the NPIs had a substantial & consistent effect on Covid-19 deaths over time’

Some more detail.

On their graphs:
• Dots/dashes: alternative estimates of what would’ve happened if no NPI.
• If bold line below these, indicates NPI reduced deaths.
• Grey area: 95% CIs. i.e. if dotted/dashed lines within this, effect not statistically significant.
May 16 9 tweets 2 min read
Putting ethics to one side, a key assumption in this argument for taxing the vaccinated is that "it is reasonable to believe" deciding against Covid-19 vaccination causes very significant harm.

But is that actually a reasonable belief?
The authors point to two possible harms.

Potential harm 1: “externality”, i.e. increased risk of infecting others.

There's not much evidence for this now. E.g. latest ONS analysis shows *no* significant protection against infection from 2 doses: ons.gov.uk/peoplepopulati…
Apr 19 10 tweets 3 min read
There’s been quite a bit of interest in this paper claiming that if social distancing rules and lockdown had come in earlier in March 2020, thousands of Covid-19 deaths would have been avoided.

So what does the paper actually show?

The paper uses Pillar 1 positive tests to model trends in infections in the first wave.
It assumes that reductions in modelled infection growth rates can only be caused by exogenous (i.e. Govt induced) changes to social distancing and then by the lockdown.
Apr 13 5 tweets 2 min read
Latest ONS analysis indicate boosters provide no protection against infection after 90 days, relative to being unvaccinated.

In fact point estimates suggest that, if anything, after 90 days boosters may provide *lower* protection compared to many 1- & 2-dose categories …
Previous infection still provides significant protection, even if 18+ months ago (i.e. pre-alpha period).

In fact previous infection in delta period looks to give significantly greater protection even than a booster in past 3 months (95% confidence intervals don’t overlap) …
Mar 31 4 tweets 2 min read
Stanford students: "get boosted or face deportation"

Latest data for 18-29s in England indicates infection rate after 3 doses is nearly 80% higher than after 2.

Unlikely that US data will be drastically different.

"irrational" does not begin to cover this. Image newsweek.com/stanford-inter…
Mar 9 5 tweets 2 min read
Quite a few European countries are a new rise in Covid cases, starting early March. For example, England, Scotland, Germany, Netherlands & Austria.

All of these have had very different levels of restrictions & very different timing for ending rules ... …
England: no event limits since last July, masks & vaccine passports ended 27 Jan.
Scotland: event limits allowed from 17 Jan, vaccine passports ended 28 Feb, mask rules still in place.
Mar 5 7 tweets 2 min read
UKHSA reported infections by vax status continue to be striking, esp for younger ages.

Remember debates over unvaxed population: UKHSA use NIMS but some worry this underestimates unvaxed rates for most ages. Alternative (ONS) may overestimate. True values probably in between
... …
Here are the rates including both NIMS & ONS by age group for the most recent 4 weeks.
You can see impact of choice of population denominator, but boosted rates now > unvaccinated for 18-59s even if ONS population is used.
Feb 25 15 tweets 4 min read
Today is the final daily indicators update from me.
Don’t worry, I’m not going anywhere. I will still do an indicators update once a week (most likely on Fridays) & I’ll continue analysis on selected data. And opinions too of course!

However, time for a bit of a recap … …
I started posting Covid updates on 23 March 2020 focusing on Sweden & then daily updates of English deaths from 7 April.

My motivation was frustration at daily press conferences & media focusing on scary headlines based on deaths by report day rather than date of death.
Feb 10 6 tweets 2 min read
Apparently a decision on vaccinating 5-11s is imminent.
Remember for 12-15s, JCVI did not recommend vaccination based on health costs vs benefits. However, Govt offered 1 dose anyway on grounds it would lower infection rates & reduce school time lost.

How did that work out?
… ...
The Govt modelling underlying the decision for 12-15s assumed one-dose vaccine effectiveness (VE) of 55% for 6 months & concluded that could save an average of about 15 minutes of school per child. I looked at that modelling at the time:
Jan 21 10 tweets 3 min read
I offered to consider a piece submitted to the Journal, I didn't suggest that was the only way to critique.
To reiterate, I don’t think resorting to insults against fellow academics is good way of furthering debate. However, happy to give my view on @ikashnitsky 's thread ...
First, I would not dream of giving any credence to the argument that the Journal somehow needs to defend publishing Prof Allen's article. The article, from a very respected academic, went through the usual peer review process. Like all research …
Jan 21 13 tweets 3 min read
Time to review the SPI-M-O consensus statement of 15th Dec pushing for more restrictions in England.

1. Infections:
SPI-M best case peak = 600k infections/day
Worst case = 1.2m
Mid-point = 900k

Actual peak (from today's ONS infection survey): 416k
... ...
2. Hospitalisations:
SPI-M best case peak = 3,000 admissions/day
Worst case = 10,000.
Mid-point = 6,500.

Actual peak (7-day ave): 2,014
Jan 21 8 tweets 2 min read
The latest UKHSA vaccine surveillance gives infection numbers by 2 & 3 vaccines.
Here are rates for 2-dose vs unvaccinated for 18-59s (most affected by NHS sackings) latter using both NIMS & ONS pop estimates

Which group do you think the Govt is planning to sack?

Remember there is uncertainty about the unvaccinated population. UKHSA argue for using NIMS population estimates, but some worry this underestimates unvaxed rates. In contrast, ONS is likely to overestimate unvaxed rates …
Jan 19 5 tweets 1 min read
With the good news today, worth being clear that the rationale for ending Plan B should NOT be that cases are falling.

Sooner or later infections will rise again, perhaps by a lot. The key point is that masks, WFH, vaccine passports etc. are not a proportionate response ... …
The only possible case for such extreme* measures like Plan B might be if:

1. We are facing an urgent crisis like the imminent prospect of health services being overwhelmed

AND ...

(*if you think they are not extreme, you have lost perspective)
Jan 19 6 tweets 2 min read
Latest ONS estimates on likelihood of testing positive by vaccination & reinfection status:
• No significant reduction in likelihood of testing positive after 2 AZ or Moderna doses or 2 Pfizer doses > 180 days.
•Very small reduction (16%) for 2 Pfizer doses > 90 days.
... ...
Remember NHS staff with 2 doses will not be sacked despite the ONS finding no evidence of any reduction in likelihood of testing positive in most cases.
Jan 7 6 tweets 2 min read
The latest @UKHSA figures on infections by vaccination status make for startling reading.

Hospitalisation & death rates still much lower in vaccinated, but UKHSA now report higher rates of infection among vaccinated for *every* age group, even those with high booster rates.

Remember UKHSA argue in favour of using NIMS population estimates. However, for some age groups, this may underestimate unvaccinated rates. The alternative is ONS population which has the opposite problem & the true rates values are probably somewhere in between …
Jan 6 9 tweets 2 min read
The Govt has published its evidence paper, purportedly backing up its decision to make secondary children in England wear masks.

The document is … interesting!

First the paper highlights the wide range of evidence highlighting the significant costs of masks in schools in terms of “negative impacts on teaching, learning and wider health & wellbeing” e.g. …
Jan 2 4 tweets 1 min read
The Imperial modelling indicated that with R0 = 2.4, there would be 510k GB Covid deaths by autumn 2020 in a do nothing scenario & that optimal mitigation could reduce that by 50%, i.e. 260k deaths averted by autumn 2020 ... ...
They then estimate suppression (strict lockdown) would reduce deaths by somewhere between 80 to 98% (& spread over 2 years).
These estimates of deaths averted by lockdown proved not just to be wrong but catastrophically wrong ...
Jan 2 6 tweets 2 min read
Why would anyone give the slightest credence to the discredited Imperial predictions from back in spring 2020?
... ...
Remember infections were already decreasing before all 3 English national lockdowns. This means Covid deaths would not have increased to anywhere close to the Imperial predictions
onlinelibrary.wiley.com/doi/full/10.11… ...
Jan 1 4 tweets 1 min read
Governments face a clear choice:
• No restrictions on people's freedom, hospitality etc & see infections go up & down including really big outbreaks.
• Lots of restrictions on people's freedom, hospitality etc & see infections go up & down including really big outbreaks.
... ...
This really shouldn't be hard given what we now know. Yet even in England, Ministers still talk about whether infections rising might mean we need more restrictions.

Do they not have access to the data from Scotland, Wales, NI, RoI, France etc?
Dec 31, 2021 4 tweets 1 min read
Since introducing vaccine passports/Covid certification, none of the UK Govts have provided any evidence this is an effective policy.

Now UKHSA report close to zero effectiveness of double vaccination against infection after 20 weeks ... ...
Remember the current law says someone double vaccinated can go to the football or a nightclub even if they have Covid.
Someone unvaccinated (even if previously infected & with a significant level of immunity) has to prove they are not infected ...
Dec 30, 2021 4 tweets 2 min read
Warwick modellers doubling down in their latest pre-print here.
Their 95% confidence interval for deaths/day whether or not there are post-New Year restrictions looks to be about 1.75k to 4.75 ...
medrxiv.org/content/10.110… ...
And that is assuming Omicron is 50% less severe than Delta!
1.75k deaths /day (which we can think of as their absolute best case scenario would be *odd* even as a worst case scenario. More details in this earlier thread ...