Friday, 30 July 2021

Correcting Lockdown Sceptics

Will Jones at the Lockdown Sceptics blog has responded to my Quillette article about the corona-crazies. It's the usual stuff, but let's try one more time to put it to bed. 
The post is headlined, rather weirdly, 'Is Christopher Snowdon an anti-vaxxer?' because Jones thinks he has a savage burn to deliver.

Christopher Snowdon is plainly an anti-vaxxer, however well he tries to hide it. “Existing Covid vaccines are simply not good enough at preventing transmission and infection,” he writes. Hasn’t he read the trial results, showing 95% efficacy against infection for the Pfizer vaccine and 74% for the AstraZeneca vaccine? Or the large population study from Israel showing Pfizer’s 92% efficacy? Or the study from Public Health England showing 67% and 88% vaccine efficacy against the Delta variant for AstraZeneca and Pfizer vaccines respectively?
On what does he base his bald assertion that they are “not good enough at preventing transmission and infection”? Clearly not the science.

It's based on Public Health England's weekly vaccine surveillance report.
Jones continues...
He doesn’t appear to feel it necessary to give a single scientific reference for a claim that flies in the face of all these respectable studies, leaving the baffled reader assuming he must have picked it up in some article he read on an obscure website somewhere, presumably by a pseudo-scientific sceptic in denial.
I thought it was common knowledge. We can also look at heavily vaccinated countries such as Iceland, Malta and Israel which have recently seen new outbreaks and therefore have clearly not reached herd immunity.

As it happens, I agree with Christopher that the current vaccines are not very good at preventing infection or transmission, particularly now the Delta variant is in town. But I’m also aware that that is not the current mainstream scientific position (though it is based on recent official data and reports).

It is the current mainstream scientific opinion. The Delta variant has pushed the herd immunity threshold into the high 90s which is not going to happen and, as Professor Francois Balloux of UCL recently said, “the vast majority of the global population is expected to get infected by the virus, likely more than once over their lifetime.”

It seems, then, that Christopher is not averse to a spot of ‘crankery’ himself. But how helpful really is all this name-calling, mudslinging and smear by association? Science does not advance by consensus, by everyone agreeing, or by closing down dissenters. Christopher himself is evidently sceptical of one of the key mainstream vaccine claims – that they are highly effective against infection and transmission – so inadvertently places himself within the ambit of his own polemic. Indeed, at one point he fires a shot at the ‘smileys’, as he calls sceptics, for being sceptical of the vaccines, arguing the jabs “have been tested in clinical trials and have demonstrated their safety and effectiveness beyond reasonable doubt in recent months”. Yet he himself goes on to doubt their effectiveness!

Jones must think this is a killer argument because he spends so much time on it. However, like many smiley arguments, it is based on a simple misunderstanding. The vaccines are highly effective at preventing severe disease and death. They are not as effective in preventing transmission and infection, as the table above shows.
The science of Covid is far from settled. Snowdon takes aim at some of the more colourful and dubious characters in the sceptic camp (criticising some claims I have no wish to defend, though he is hardly kind or charitable as he does so). But he notably leaves out of his cranky panorama some of the world’s most eminent scientists who take a sceptical line on a number of the issues he raises.
Sunetra Gupta, for instance, Professor of Theoretical Epidemiology at Oxford University; Martin Kulldorff, Professor of Medicine at Harvard University; Jay Bhattacharya, Professor of Medicine at Stanford University. 
... Or where is Professor Carl Heneghan, Director of the University of Oxford’s Centre for Evidence-Based Medicine; or Sucharit Bhakdi, former Chair of Medical Microbiology at the University of Mainz; or John Ioannidis, Professor of Medicine, Health Research and Policy at Stanford University?

I don't want to get into the credentials game but it would be a stretch to describe any of these as the world's most eminent scientists. Even if they were, eminent scientists can be wrong, as Gupta was when she claimed that up to 68% of the UK's population had already been infected with Covid-19 by 19 March 2020 and that Covid's infection fatality rate was 'somewhere between 0.1% and 0.01%'.
Heneghan has also been frequently wrong during the pandemic and is one of a number of prominent sceptics to have deleted all his tweets during the second wave.  

The heart of the sceptical position on lockdowns, shared by all these scientists, is that they are not worth it.

That was the heart of the issue a year ago before the whole thing descended into pseudo-science, anti-vax quackery and conspiracy-minded wibble.

Christopher makes a number of specific claims in his piece that he seems to think are scientifically incontestable but in fact stand on very wobbly ground when given closer attention.

For instance, he asserts that the second wave in the U.K “lasted longer and killed more people”.

There were more Covid-related deaths between December 2020 and February 2021 than there had been in the first 11 months of the pandemic. At its peak, there were nearly twice as many people in hospital with the virus than there had been in March 2020. In Britain, as in most European countries, excess mortality went through the roof.

The data, however, tells a more complex story. Looking at the weekly total deaths, it’s clear that the second peak is much lower than the first (see below). The higher Covid mortality in winter compared to spring can be seen (blue bars) to be due to more of the non-excess deaths in winter being classed as Covid, whereas in spring there were many excess deaths beyond those defined as Covid (green bars). This discrepancy will likely have a number of factors behind it, but you have to think that how many people were being tested and treated as a Covid case has to play a large part, given that a Covid death is defined as a death from any cause within 28 days of a positive Covid test.


Note also that because it was winter the baseline was higher, meaning in percentage terms the excess peak was less than half the spring peak. This helps to put the winter epidemic into perspective.

The reason 'more of the non-excess deaths in winter [were] classed as Covid' is that they were caused by Covid. There were fewer non-Covid, non-excess deaths in winter because lockdowns, masks and social distancing got rid of the flu and the restrictions may have prevented some other deaths (what happened to the 'lockdown deaths' by the way?) 
It's debatable whether the peak really was higher in April - three of the four deadliest days of the pandemic were in January - although the peak in excess deaths certainly came in the spring. But I'm not talking about the peak. I'm talking about the overall number of Covid deaths. Even a cursory look at the graph above (which ends in February for some reason) tells you that there were more Covid deaths in the second wave than in the first.

Jones repeats the old smiley canard that 'a Covid death is defined as a death from any cause within 28 days of a positive Covid test'. Assuming he is not being deliberately misleading, it is kind of pathetic that he still doesn't know that the ONS data is based on death certificates. It is not based on the rough and ready estimates published on the dashboard which rely on deaths recorded within 28 days of a positive test. If you look at the death certificate figures, there are 24,000 more deaths than if you use the 28 days measure, and we know that around 90% of these deaths had Covid as the underlying cause.
It is very likely that some Covid deaths in the first wave were not identified as such due to a lack of testing, but this does not change the basic conclusion. There have been over 150,000 Covid deaths in England during the pandemic (according to death certificates). Only around 40,000 of them occurred in the first wave.
Just to remind you, the Lockdown Sceptics blog repeatedly claimed that there would be no second wave and that the pandemic ended last summer.

The below average deaths since the end of winter have also meant that 2021 is now a below-average year for age-adjusted mortality (so far), the low mortality of the spring and summer having already cancelled out the high mortality of January and February.

Look at the graph he uses to illustrate this! Does it look like the 'low mortality of the spring and summer' has 'cancelled out the high mortality of January and February'?! There have been several weeks when there have been fewer deaths than average, as you might expect after a virus has wiped out 150,000 people, but those dips are far smaller than the peaks in January and February. 

Jones links to one of his own crackpot blog posts (titled 'What pandemic?') to support his claim that '2021 is now a below-average year for age-adjusted mortality'. This is based on data from the Institute and Faculty of Actuaries. Read this thread from an actuary to see why Jones' interpretation is wrong.

Christopher is very critical of sceptics for questioning the reliability of the PCR test and the definitions of Covid case and death based on it. But this was a very live topic in autumn 2020, with a number of top scientists including Professor Carl Heneghan wading into the fray, as this correspondence in the BMJ records. The Lancet published a piece from NHS scientists in December stating that the operational false positive rate of PCR testing was estimated to be “somewhere between 0·8% and 4·0%”. 
The 'correspondence' is a Rapid Response on the BMJ (otherwise known as a reader's comment) and is worth clicking on as a reminder of some of the garbage being spouted by smileys in late 2020. The author claims that Covid is no more deadly than seasonal flu, that 'deaths are currently running at normal levels' (in mid-November) and that 'there is no sound evidence of any second wave'. It was written by this person. Needless to say, their Twitter output is a sewer of conspiracy theories, quackery and anti-vaxxing.

As for the Lancet article (a comment piece), I am very familiar with this because smileys frequently link to it despite it being from last September. It is only popular with 'sceptics' because it says 'preliminary estimates show [the false positive rate] could be somewhere between 0·8% and 4·0%.' It cites a SAGE document from early June 2020 for this claim. Smileys have clearly never read it because the document also says: 

DHSC figures [3] show that 100,664 tests were carried out on 31 May 2020 (Pillar 1 and 2 RT-PCR tests). 1,570 of those tests were positive for SARS-CoV-2 (1.6%). The majority of people tested on that day did not have SARS-CoV-2 (98.4% of tests are negative). When only a small proportion of people being tested have the virus, the operational false positive rate becomes very important. Clearly the false positive rate cannot exceed 1.6% on that day, and is likely to be much lower.

This is the obvious point that sceptics keep missing. The false positive rate cannot be higher than the positive rate. The world has been engaged in mass testing for over a year now. We do not need to rely on a preliminary assessment from last June. 
Why would the 'sceptics' rely on a document that is over a year old when they could look at more up-to-date evidence? It is because no credible scientists believes that the false positive rate is anywhere near 0.8%, let alone 4%.


We know the specificity of our test must be very close to 100% as the low number of positive tests in our study over the summer of 2020 means that specificity would be very high even if all positives were false.  For example, in the six-week period from 31 July to 10 September 2020, 159 of the 208,730 total samples tested positive. Even if all these positives were false, specificity would still be 99.92%.

Case closed.

Christopher argues: “The UK had a positivity rate of just 0.2% as recently as two months ago [May]. The false positive rate cannot possibly be higher than the positivity rate, but this simple logic continues to elude the Covid-sceptical community.” This simple logic is too simple, however, as it doesn’t grapple with the fact that the operational false positive rate can vary, including with the volume of tests.

Jones links to a blog post by "pathological" Claire Craig to support this false claim. Truly desperate stuff. 

Christopher claims U.K. infections fell in January because of the lockdown, but fails to engage with the fact that Sweden’s fell as well without a lockdown.

They did indeed. And then they rose again, as I say in the article. Lockdown sceptics are fond of saying that infections fall before lockdowns begin. This is usually untrue, but even when it has happened (as it did before England's most recent lockdown) there is no guarantee that rates would have continued to fall if a lockdown hadn't been enacted. Sweden is a good example of a post-Christmas decline turning out to be a false dawn. 

Jones tries to wriggle out of this by not showing the infection data and only showing the mortality data. He claims that too few people were vaccinated in Sweden for vaccines to have reduced the death toll in the third wave. I disagree, but whatever the reason for the lower mortality after February, Sweden had a high infection rate throughout winter and spring as well as a high rate of hospitalisation. The charts below shows case numbers and intensive care admissions for Covid.

Sweden and the UK are different countries and one cannot be treated as a control group for another, but comparing their outcomes in winter and spring is a priori evidence that lockdowns work. As a major travel hub with many big cities, the UK has been hit harder by Covid from the start. If the UK had followed Sweden's trajectory, it would have been disastrous.

Inevitably, he then brings up Florida and makes the usual comparison with California as if a state on the other side of the continent was the only available comparison. He talks about 'lockdown states' as if it were a binary choice between being in lockdown and being “open”. Most states haven't been in lockdown for a year and in reality there is a patchwork of local, city and county rules, Florida included. 
I am so bored of hearing about Florida that I cannot be bothered to discuss it again. I will just repeat that it is quite possible to keep R at or below 1 for a period of time without lockdowns. No one has every claimed otherwise. It is a non sequitur to say that this proves that lockdowns don't work. It doesn't even test the proposition. 
Lockdowns are only needed when the rate of infections gets so high that it threatens to overwhelm the health service and kill very large numbers of people. I remind you that Florida now has the highest infection rate in the USA, if not the world.

Jones repeats the smiley mantra that 'in England infections have peaked and declined prior to lockdown on all three occasions'. This is not true. Infections fell somewhat a few days before the first and third lockdowns but, crucially, not the second. The first and third lockdowns were preceded by heavy restrictions which reduced R. All schools, pubs, restaurants, theatres, gyms, etc. were closed several days before the first lockdown. Much of England was already in de facto lockdown long before the winter lockdown began, the hospitality industry was closed in most of the country and heavily restricted in the rest. All the schools were closed for Christmas. Many workplaces were also closed for Christmas. It is not surprising that the infection rate fell, but there is no guarantee that the decline would have continued after the Christmas holidays.

The second lockdown was not preceded by other measures and the infection rate only fell after it began. It beggars belief that anyone can look at the curves in autumn, with the pronounced dip in November, and believe that this is the natural waxing and waning of an epidemic. What possible reason can there be for a respiratory virus to suddenly decline in November when it was rising in October and rose again in December? 

The clincher is that there was a similar trend in Wales (below) but the dip started and ended two weeks earlier. In England, cases peaked on 11 November (five days after the lockdown began). In Wales they peaked on 29 October, six days after the Welsh lockdown began. 
It's an almost perfect natural experiment involving two neighbouring countries. It is very difficult to deny that the drop in infections was caused by the lockdowns but of course Jones does. He cites a study which is wrong for the reasons I explain here.

I realise that I am wasting my time responding to these people. Anyone who is still arguing about false positives 18 months into the pandemic is beyond help. The Lockdown Sceptics blog has been a major source of disinformation since it was launched last year and has likely led to the deaths of some of its readers. If you look at the comments section, you will find an absolute madhouse of paranoia and ignorance. Nothing short of intubation will bring them to their senses. Maybe not even that.

No comments: