Thursday, 5 November 2020

Science or logic?

Pressed Rat and Warthog have closed down their shop.
They didn't want to, 'twas all they had got.
Selling atonal apples, amplified heat,
And Pressed Rat's collection of dog legs and feet

Sadly they left, telling no one goodbye.
Pressed Rat wore red jodhpurs, Warthog a striped tie.
Between them they carried a three-legged sack,
Went straight round the corner and never came back.

The bad captain madman had ordered their fate.
He laughed and stomped off with a nautical gate.

I thought of Ginger Baker's spoken nonsense poem to music from 1968 when Captain Madman, aka Boris Johnson, did his abrupt about face and went for a month's lockdown having previously rejected calls for a two week circuit breaker intervention. A lot of shop owners will be in the same position; indeed Mrs H's favourite local clothes shop recently bit the dust. While still having its steady trade of ladies of a certain age once it was allowed to reopen in the summer, the lack of weddings and a school prom dress season meant profitability was far over the horizon.

The business community has been in despair for some time, but in my last post (27 Oct) I shared my long-held concerns, dating back to April, about whether the current approach to tackling covid is getting the balance right and my fears that the young will not only pay the price long after us oldies have gone but get pretty angry about it once they realise their fate.

Democracy Man, commenting on my last utterance, urged me to say what my strategy would be, other than (I paraphrase a bit) " I wouldn't have started from here". Well if I knew the answers and everyone else didn't that would be quite remarkable.... but there are some things I consider fairly obvious.

Firstly, there's a lot of dodgy use of statistics going on. Some of this is because many folk just don't seem to get the time lags involved. They don't understand that a negative test just means you tested negative at the time of the test - it could be a false negative, you could have been incubating the disease or you could have caught it subsequently. Testing tells us a lot but it isn't the magical solution many people have assumed it to be throughout the crisis. We heard a lot complaints that care home staff weren't being tested but, unless you test them on their way into a Premier League style bubble (to be fair some carers did lock themselves down where they worked in the main lockdown) you would have to test them at the start of every day's shift and get an immediate result, which even then would not be a 100% guarantee. We may get there but we aren't there yet.

One of the more obvious time lags is deaths lagging hospital admissions which lag new cases. Many commentators were noting recently that covid isn't even one of the top causes of fatalities. This graphic was in the Daily Mail but Sky News and others also picked up on the relevant ONS data:

The eagle eyed among you will have noted that this data related to September and so was already 2 or 3 weeks out of date when it was published. Covid deaths at the time were getting into the hundreds daily. In the spring they peaked at over a thousand. Multiply the covid deaths above by an order of magnitude and you get a very different picture. As nearly 500 deaths were reported on 4 November we seem to be getting towards the earlier peak. I have noted previously that those deaths will not all have occurred on the previous day but will have been scattered over several weeks, such are the time lags in recording and reporting deaths and their causes, so even the daily deaths figures can be misleading.

It's not just the newspapers that use dodgy stats and projections of course. Profs Chris Whitty and Patrick Vallance took some flak at Tuesday's Science and Technology Select Committee for using out of date assumptions as the starting point for their 4,000 fatalities a day number that seemed to spook the PM out of his regional tiered system into the new lockdown. This figure was subsequently branded by Theresa May as "wrong before it was even used". Whitty and Vallance also used a very selective picture of hospital loading which focused only on the 29 worst affected hospitals, not the picture over all 482 in England. Whitty bit back at the select committee that even if 4,000 was over the top we were well on track for 1,000 deaths a day and, even if MPs were happy with that number, he wasn't. And the 4,000 wasn't a prediction it was a model (hmm, I've heard that excuse before and may even have deployed it myself). Moreover, while it hadn't allowed for the tiered restrictions, they had made a difference but not enough.

Subsequently Whitty has apologised for appearing to say that cases in Liverpool were still rising across the age groups despite the tier 3 restrictions when they were not, an impression he has since corrected.

But if we take a step back for context, even hundreds of deaths a day wouldn't necessarily be regarded as any more than "it's what it is" in a major flu outbreak and wouldn't lead to calls for lockdowns. The 2017/18 flu season claimed 22,087 lives in England for example, so I would guess daily deaths may well have peaked at several hundred. Interestingly a confidence interval of around plus or minus 700 is given on the 22,087 figure, and 88% of the total were aged 65 and over. And we vaccinate against flu.

As the average age of a covid fatality in the UK is higher than the average life expectancy it isn't surprising that an increasing number of commentators have been calling for a version of  the Great Barrington declaration, arguing that it's time to shield the vulnerable and let everyone else live their lives. In this week's Sunday Times, Business Editor Oliver Shah was given a platform in the main paper, which he used to argue that the new national lockdown threatens us all saying businesses need an exit strategy:

"Boris Johnson and Sunak must plot the quickest way possible out of this shutdown and then vow never to do it again. The alternative is to risk the destruction of a generation's prospects. To flip round another slogan, we should let the NHS protect us - and let us concentrate on protecting our ailing economy."

Shah has written many strong columns in the newspaper but I'm afraid this wasn't one of them. Firstly, if Johnson and Sunak knew how to plot a way through this, let alone specify a realistic exit strategy, they would long since have done it, as would their equivalents in many other countries. There isn't a global pandemic playbook to crib from. We thought we had one, but it had been written for influenza. Secondly, a lockdown of 2 or 4 weeks duration will certainly not eliminate the virus, it simply pushes the curve of cases and fatalities back a few weeks. So the numbers will rise again. The inevitable result is that, until or unless vaccines make an impact, we face periodic "reset" lockdowns if the virus is not to run rampant. Few politicians are being honest enough to own up to this but it seems inevitable to me. 

This didn't stop Sir Vince Cable arguing that the circuit breaker concept is "crude, lazy and defies common sense".

Others have argued against lockdowns on purely medical grounds. The Daily Mail has a tame oncologist from a London teaching hospital, Prof Angus Dalgleish, who argued that if lockdown was a drug it wouldn't get approved because they do "more harm than good".  Now I have some sympathy with Dalgleish, given the fact that he probably hasn't been able to carry out as many operations as normal. Indeed the scandal (for I believe it is such) of why so many cancer operations and other elective surgery such as joint replacements have been deferred is a bit of a puzzle to me. Most hospitals had just about got back to where they would normally be and some have really got it sorted: Croydon is now doing 20% more operations than pre-covid by effectively creating a separate hospital within a hospital combined with Premier League style testing and isolation. Dalgleish notes than less than 0.01% of covid deaths are people under 45 and claimed that, at his hospital, only two or three dozen patients have tested positive for covid since London was put in tier 2, compared with half their admissions in the previous peak. But he isn't in one of the country's hottest spots.

Plenty of people have argued that lockdowns don't work at all, some misquoting the World Health Organisation, which nevertheless has noted they build up a lot of problems. WHO's David Nabarro told the Spectator:

“lockdowns just have one consequence that you must never, ever belittle, and that is making poor people an awful lot poorer.” 


“The only time we believe a lockdown is justified is to buy you time to reorganize, regroup, rebalance your resources, protect your health workers who are exhausted, but by and large, we’d rather not do it.”

so they should "not be used as the primary means of control of this virus". Which isn't at all saying "lockdowns don't work".

Moreover, there is a big logical problem with the Dalgleish and Shah arguments: they are effectively decreeing that the NHS can cope. Fine if it does, but if not it isn't only covid deaths that would soar. Elective NHS procedures would be put on hold and non-covid deaths would also soar as a result. In some cases there would be time lags but the total of excess deaths would be very much higher. That isn't so much science as logic.

There is of course an argument (which I have certainly flirted with) that those excess deaths are "worth it" for the broader societal gains. The government admitted on Tuesday that the Treasury had not carried out any assessment of the effect of the new lockdown on jobs. It is very difficult - and for some controversial - to try to equate deaths and other disbenefits but the government does that all the time, explicitly or implicity, so I find this revelation truly astounding. We've heard a lot about Prof Neil Ferguson's fairly basic (and fairly crummy) model, which seems only to count covid deaths. There must be some reputable modellers trying to make a broader assessment, whether or not this has been commissioned by the government.  I find it staggering that we haven't heard from them after so many months. Those who assert that letting the virus run wild would be disastrously damaging to the economy as well as health may well be correct. But if it were so overwhelmingly the case it shouldn't be difficult to produce an analysis that demonstrates it.

I accept that drawing conclusions from any models which value health as well as economic impacts is problematic. Several decades ago I spent a fair chunk of my career working on the development and application of risk assessment techniques for major hazard processes and transport. At the time I researched the difficulties of balancing costs of preventing accidents that might happen with resulting risk to life*. Though fraught this stuff is done all the time. Most people have heard of the NICE (National Institute for Health and Care Excellence) approval process for drugs to be deployed in the NHS. That scheme values a quality adjusted life-year at £20,000 to £30,000. Many would appreciate, if not necessarily accept, that road safety schemes are approved on the basis of cost and risk, effectively putting a price on saving a life (though actually putting a price on reducing the probability of accidents that could happen and might cost lives) . So of course the government should be attempting to balance all of the various costs and disbenefits in the strategy it adopts to contain covid. It should also try, difficult though that is, to have an adult discussion with the electorate over the choices it is making and the uncertainties it is grappling with.

As it is well established that people are generally not good at intuitively assessing the probability of these type of risks I am not surprised the government has ducked that challenge. I read that less than two-thirds of a sample of MPs were able to correctly put a probability on not getting a head from two tosses of a coin which was depressing but was still a higher proportion savvy MPs than I would have guessed (though maybe the group included a lot of gamblers).

I also accept that the picture is still very confusing, data is very incomplete and many aspects of the "science" are as yet poorly understood, so even the most robust of plans would be subject to rapid revision when the facts change. I don't know how Shah expects a clear and definitive exit strategy for business in those circumstances.

So even if the government had a plan it would not necessarily survive contact with the evolving situation. Indeed, Johnson has probably had several plans and on at least two occasions they have rapidly been binned. The first occasion was in March when he really didn't want to lockdown (and herd immunity may or may not have been part of the plan) and the second was last week when the commitment to a regional tiered approach got ditched. Mike Tyson famously said "everyone has a plan until they get punched in the mouth". The punch Boris Johnson has taken to the mouth twice now is the spectre of people lying on  trolleys and the floor in the corridors of NHS hospitals. He has presumably judged that those pictures would be intolerable to the public. Indeed, they may actually be intolerable to him personally. Hence the NHS is not there to save us, as Shah would have it, the priority come what may is to "save the NHS". What Shah nor anyone else writing similar anti-lockdown columns hasn't told us is what they would do when the hospitals get overwhelmed across the board.

Although a further lockdown buys time, the idea that a short circuit breaker gives the opportunity to fix things, as advanced by the Wales First Minister Mark Drakeford (he was talking about test and trace) is risible. Whether or not we are in a lockdown doesn't change anything much in terms of the challenge to improve how those processes operate. And when Angela Merkel proclaims "very, very difficult months ahead" for Germany despite its stronger testing capability and three times as many hospital beds per head of population compared with the UK it's obvious that greater testing and hospital capacity isn't the whole answer.

So I feel it's not so much an understanding of science that is lacking much of the time, especially since "science" hasn't got anything like a full understanding of covid yet. Most of what is selectively presented to us isn't based on science: it's empirical data. No element of scientific methodology has gone into its capture or analysis. There are no theories followed by controlled experiments to validate the theories, other than in the development and production of vaccines and tests on drugs. People - some of them admittedly very knowledgeable and experienced - pore over the data and jump to conclusions. So there isn't much science going on but the main problem is the lack of application of sound logic. 

Nevertheless it was scientific understanding that convinced me we mustn't let the virus rip and run its course after I picked up a copy of New Scientist, which gave two reasons. In addition to the one we hear all the time, about avoiding overwhelming hospitals, the other reason is that the more the virus spreads the more opportunity it has to evolve. It quoted one American scientist as saying "we are fortunate that the virus is not mutating fast". A recent analysis of more than 18,000 genomes from around the world found very low levels of genetic diversity. This is fortunate because it increases the chance of rapid and successful deployment of vaccines. The study concluded that these viruses were all so similar that a single vaccine should protect against them all.

New Scientist also rubbished the Great Barrington declaration, which argues for shielding the vulnerable while everyone else "should immediately be allowed to resume life as normal" allowing herd immunity to build up. For SARS-CoV-2 the herd immunity threshold would be 60 to 70% of the population. This depends on the typical R number, the average number of people that each infectious person infects, assumed to be about 3.5 for covid. For measles the numbers are 15 and 95%. There are two problems. Firstly herd immunity from a virus has only ever been achieved in humans through vaccination. Secondly, even if the death rate is less than 1% letting the virus run free would hospitalise and kill millions. If I apply those numbers to the UK population of 67 million some 45 million people would need to have been infected, potentially killing 450,000 people. (See: I just got Neil Ferguson's headline number without needing a "model". God knows what they pay him for....) To be fair to Vallance, he was saying much of this about herd immunity back in March.

Moreover, herd immunity can only be built if the immune response totally prevents individuals from picking up and transmitting the virus. That sometimes happens but often doesn't: a lot of the time an immune response stops us from falling ill but doesn't prevent onward transmission. The same is true of vaccines. We don't yet know whether natural immunity or the vaccines under development will halt transmission. Until we do assuming herd immunity is "unscientific and irresponsible".

A further nail in the coffin of the herd immunity argument came soon after when it was reported that covid-19 antibodies fall away quite rapidly after recovery. This isn't game over for natural immunity as the role of T-cells wasn't part of the relevant study. There is hope that vaccination responses may be stronger and early studies on older people look promising.

Nevertheless I was convinced that we can't at this stage allow most things to return to normal, we must buy time and opportunity for vaccine deployment. If the vaccines don't work we may have a diferent problem to face but we aren't there yet.

So returning to Democracy Man's challenge, to keep control of the virus the only thing that matters is keeping R below 1. Exactly how doesn't really matter other than to the specific people the restrictions affect. Unless the virus can be eliminated some people are going to catch it and it will continue to circulate in the community, even if the numbers of cases are small. So whatever scope (some call it a budget, or headroom) is available while keeping R below 1 should be used to provide essential services, education and keep open all parts of the economy where people don't gather in close proximity for more than very brief periods of time. Keep going all the things that really matter for the long term and all the things that pose low risk. Everything else might have to be put on hold.

Whether the controls are national or regional is partly a matter of preference. We now have a semi-scientific experiment with Scotland going regional, England national and Wales having a shorter circuit breaker. I have my opinions on these but we should get some useful data by December.

Keeping schools open is a total no brainer. When children are at so little risk I was very impatient about the schools being closed for so long and I am relieved to see that, this time round, only the teachers unions (surprise, surprise) want to see them closed. I am intrigued at how any teacher, but particularly any science teacher could claim "the science" backs the idea that schools need to close. The already overwhelming argument was reinforced by a study published this week that said adults living in houses with children at school are at no higher risk. 

Universities are a tougher call. It was always obvious that the virus would spread among the student population in a 2020 covid version of fresher's flu. Indeed I mischievously chuckled at the start of the uni term about a national experiment with herd immunity among 18 to 21 year olds, keeping them in their uni accommodation and away from relatives until December when they had all caught the virus. But actually I think the government's decision was simply that otherwise lots of universities would need bailing out. Large lectures can be given on line and face to face tutorials in smaller groups can be run so I'd still leave universities open to operate as best they can.

I entirely understand why businesses that have spent time and money making themselves supposedly "covid safe" - I assume there is a whole new army of safety elves engaged in those assessments - are furious that they have now been closed down anyway. For the most part this makes little sense in terms of the individual establishments, especially shops where there are low people densities. Having a Black Friday sale scrum has been avoided in England (unless it is rescheduled into December) but Wales might have to take a view on that. Restaurants and pubs, where you sit opposite someone and talk at them, are much more questionable and would be the first type of business I would close. 

The most dangerous thing is gathering in houses, which is why all the UK national governments have retained controls over who you let into your home.

I accept that the rules are not just framed on the basis of what specific activities are considered low enough risk. It seems to me that the British public only has two modes: frightened to death and mainly compliant in lockdown, anything goes otherwise. So maybe there have to be periodic lockdowns to get the message through.

I realise this isn't very different from what most governments are doing. But I feel Johnson needs to channel his Churchillian tendency and tell everyone that, until we can start to put the first generation vaccines into play - which may only be a few weeks now - we just need to "keep buggering on".

But of course, on my version of the planet golf courses wouldn't close. Mrs H is even more irritated by this than me, calling it the "politics of envy and resentment". I think she's probably right, because it sure as eggs are eggs ain't based on science. Anyway, what's the point of being "world king" if I can't declare golf courses open?

So yes, my approach is only a tweak on what is being done and what most countries are doing. Because the only other option is to live in a country like China or South Korea with a very different approach to civil liberties. Or live at the arse end of the world - New Zealand.

* Don't ask - I might dig out my publications on the subject! For now I'll just say that not much appears to have changed in 35 years


Pressed Rat and Warthog is on Cream's Wheels of Fire Studio album, 1968. But listen to the equally apposite Born Under A Bad Sign though....and of course White Room.

Flu data from Public Health England's annual surveillance reports available at

Oliver Shah's column "Stem the tide. Protect the economy. Save jobs" was in the Sunday Times on 1 November

Circuit breaker idea is crude, lazy and defies common sense. Vince Cable in Daily Mail 19 Oct 2020

Prof Angus Dalgleish's piece If lockdown was a drug it wouldn't be approved does more harm than good was in the Daily Mail on 29 October

Bruce Y Lee explained how the WHO Special Envoy of Covid-19 David Nabarro is being misquoted and taken out of context in the WHO Warning about covid-19 coronavirus lockdowns,, 13 October

Angela Merkel was quoted in the Guardian's piece Global Report: Merkel says Germany faces "difficult month ahead" in covid fight, 26 October

Germany has 8 hospital beds per 1,000 people, UK 2.5. For a full list of countries see Wikipedia

New Scientist magazine on 27 October included an editorial "Now is not the time" (to let our guard down), a super piece titled "The evolving virus - the coronavirus has mutated very little, but as more people are treated or vaccinated it could face pressure to evolve" and a comment column "Absurd about the herd. Assuming that herd immunity will result from letting most people get covid-19 is unscientific and irresponsible". Both the latter pieces by Graham Lawton.

Coronavirus: Antibodies fall rapidly after covid infection dashing hopes of herd immunity, Sky News reported on the REACT-2 study by Imperial College on 27 Oct,

Sir Patrick Vallance tells Sky News about 60% of people will need to become infected in order for the UK to enjoy "herd immunity".  Sky News 15 March

No comments:

Post a Comment