Covid-19 messaging at Piccadilly Circus

Blinded by science

Alasdair Palmer says there is little evidence to show that lockdowns are effective

This article is taken from the March 2021 issue of The Critic. To get the full magazine why not subscribe? Right now we’re offering three issue for just £5.

Almost everyone is now thoroughly fed up with the closure of schools and all but essential businesses, and of being told that they are not allowed to venture out of their home unless it is to do food shopping or take exercise. The measures collectively known as the lockdown — the second imposition of the restrictions in the UK in less than a year — is starting to pall.

The lockdown’s sharply negative effects on Britain’s economy, on its education, and on mental health, are becoming more obvious. Most people nevertheless accept the lockdown is necessary, because they accept that, prior to vaccinating 70 per cent of the population, it is the only effective way to restrict the spread of Covid-19.

That is what ministers emphasise at every opportunity. It is also what the Chief Medical Officer, Professor Chris Whitty, and the Chief Scientific Officer, Sir Patrick Vallance, say. So do many other scientists who are expert in infection and disease transmission. Some go further than merely stressing the effectiveness of lockdowns in combatting the virus: they say that anyone who breaks the lockdown rules is a menace to public health, and is recklessly causing the spread of a deadly disease: rule-breakers have “blood on their hands”, and deserve public shaming as well as serious punishment.

The empirical evidence on the effectiveness of lockdowns is surprisingly thin

 And yet, the empirical evidence on the effectiveness of lockdowns in diminishing the infection and death rate of Covid-19 is surprisingly thin. There is very little data that supports the claim that, at least until most people have been vaccinated, lockdowns are the only effective way of containing the spread of infection.

Of the many hundreds of scholarly articles published in scientific journals on the Covid virus since January last year, amazingly few even attempt to provide and analyse empirical data which shows that lockdowns are the most effective way of dealing with the virus. 

The most famous article which did attempt to do so was published in the journal Nature on 8 June 2020 by a team of scientists from Imperial College London. Entitled Estimating the effects of non-pharmaceutical interventions on Covid-19 in Europe, it claimed that not only were lockdowns effective, but they were the only effective way (other than a vaccine) of controlling the rate at which Covid spread and killed. The authors concluded that imposing lockdowns in the 11 European countries they studied had saved three million lives. 

Sweden did not impose the measures that are required for a lockdown: there was no stay-at-home order

That figure of three million lives saved has been frequently cited since. If it, and the reasoning behind it, are accurate, that would certainly provide very powerful evidence that lockdowns are essential for coping effectively with Covid. But both the conclusion and the reasoning that led to it are very seriously flawed.

The assumptions built into the model determined its conclusions. And one of the assumptions was that lockdowns were responsible for the drop in the number of Covid deaths that all the countries in the study saw in the spring and early summer of 2020. Sweden was one of the countries in the study. It was included partly because of the quality of its record-keeping. 

But Sweden did not impose the measures that are required for a lockdown: there was no stay-at-home order, and businesses including bars and restaurants stayed open, as did schools. Nevertheless, in the spring and early summer, Sweden, like all the other countries in the study, experienced a sharp fall in Covid infections and deaths.

You might think this fact would have persuaded the authors of the paper that perhaps locking down was not responsible for the fall in Covid cases — or at least to have caused an agonised reappraisal of the model’s assumptions. But no. It merely meant that the authors of the paper stipulated that according to the model, Sweden’s ban on gatherings of more than 100 people counted as a lockdown — which of course it was not.

The evidence doesn’t support a statistical correlation between lockdowns and a fall in infection rates

The effect was to ensure that the model was not empirical, and that the claim that lockdowns had saved three million lives was simply an assertion, not a conclusion based on evidence. In December last year, Nature published an article pointing out this and other fundamental problems with the original paper and its “lockdowns saved three million lives” claim. But by then the claim that lockdowns save millions of lives was firmly established as “backed by science”.

There are other articles in scientific journals that have made an attempt to quantify the number of lives saved by the imposition of the package of measures that together comprise “lockdown”. But they all suffer from essentially the same problem: they start by assuming that lockdowns cause the fall in deaths and infections. They then build that assumption into the model that they use to calculate what would have happened had the lockdowns not been imposed — rather than first providing evidence which establishes the truth of that connection, and only then using it to calculate the number of lives that have been saved by the imposition of lockdown measures.

There is another way of testing whether lockdowns cause a substantial fall in infection and death rates. It is to compare the infection and death rates in countries which have imposed lockdowns with the rates in those which have not. Studies have been published in the Lancet and Frontiers in Public Health which do precisely that. 

But their results do not confirm that lockdowns cause a lowering of the infection and death rate from Covid-19. On the contrary: the authors of both those papers report that they find “no association” between lockdowns and lowered infection and death rates. It’s not that the evidence they collected fails to establish that lockdowns cause a fall in infection and death rates. The evidence doesn’t even support a statistical correlation between the imposition of lockdowns and a fall in infection and death rates.

That one event happens after another event does not show that the first event caused the second

These studies do not claim that infection and death rates never fall after countries impose lockdowns. They obviously have done so, in Australia and New Zealand, for instance. Rather, these studies demonstrate that the data collected so far do not show that a fall in infection and death rates is correlated with the imposition of a lockdown.

Even if the data did show that association, it would not be enough to show lockdowns cause infection and death rates to fall. That one event happens after another event does not show that the first event caused the second. To assume it must do so is to make the same mistake as the man who found he woke up with a headache every morning after he drank gin and tonic — so he stopped drinking tonic. The history of medicine is full of cases where a causal relationship has been claimed on the basis that two events were correlated — but on closer investigation, the events turned out not to be related as cause and effect.

Critics have pointed out that there are so many differences in the way the populations of different countries react to the disease, and in the measures that their governments use to fight it, and even in the ways they measure it, that cross-country comparisons have little value in establishing anything reliable about whether lockdowns do or do not work to reduce infections. That is why a study which compares different provinces within a single country would be more helpful in the quest to establish whether or not lockdowns cause infection and death rates to fall in the countries that impose them. And one which does precisely that has recently been published.

Kasper Planeta Kepp and Christian Bjørnskov looked at North Jutland, a single province of Denmark which is remarkably homogeneous: its population has essentially the same age and health profile, the same culture and a broadly similar level of prosperity. This means that many of the problems that make it difficult to draw reliable conclusions from studies that compared two or more different countries do not apply.

The problem isn’t with the theory of lockdowns. It’s with practice

North Jutland is divided into 11 municipalities. It is a centre for Denmark’s mink farms. Last November, when there was a concern that a new strain of Covid could jump from mink to people, seven of those 11 municipalities imposed a lockdown: they issued stay-at-home orders, and closed non-essential businesses, schools and colleges. The other four municipal areas did not impose any of those measures.

Kepp and Bjørnskov compared what happened to Covid infection and death rates in the municipalities which imposed a lockdown with what happened in those that did not. They found that the reduction in infection and death rates followed an almost identical pattern in both.

It is very striking evidence that lockdowns are not effective in reducing infection and death rates, but it does not appear to have persuaded many of the scientists who study Covid. In particular, Sir Patrick Vallance and Professor Whitty continue to insist that, prior to inoculating 70 per cent of the population, lockdowns are the only effective way of combating Covid.

They may think that studies are not needed to establish so elementary a point as the effectiveness of lockdowns in controlling infectious viruses that spread in droplets in the air travelling from person to person. It is a fundamental principle governing infectious diseases spread by human contact that if you do not have contact with another individual, you cannot infect, or be infected by, them. And in this the Chief Scientific Officer and the Chief Medical Officer are surely right: it is a fundamental principle. So lockdowns must work after all, right?

Wrong. It doesn’t follow that lockdowns must work to diminish infections and deaths from Covid. The problem is not with the principle but with putting it into practice. If no one meets with anyone else, that will indeed completely prevent the spread of new infections — but a lockdown will only have that effect if there is perfect compliance. But complete isolation for everyone is obviously not a feasible option. No society can function when literally everyone goes into rigorous self-isolation. The laws governing lockdowns rightly recognise that there have to be exceptions to the rules because many thousands of people work in vital services and industries — such as the provision of food and medical care, the maintenance of law and order, maintaining the electricity, water and gas networks, and so on — who have to be able to continue to interact with others if our society is not to collapse.

Ending the lockdown does not mean letting the Covid infection rate run riot

Let us assume that lockdowns could in theory be effective in reducing Covid infection and death rates even when exceptions are made for a minimum of 15 per cent of the population. What would be the problem? Again, the problem isn’t with the theory. It’s with practice. There are clearly profound practical difficulties in getting a high level of compliance from the other 85 per cent of the population. Many people are reluctant only to leave home when they have a reason for going out that is accepted by the government. The government is aware of this. That is why it is running the advertising campaign stating Bending the rules costs lives.

What degree of compliance is needed from the 85 per cent of the population that don’t do essential jobs in order for lockdowns to be effective? I don’t think that anyone knows the answer to that question. But there is plenty of evidence to suggest that, whatever the needed degree of compliance is, we in the UK have not reached it. Could we achieve the necessary level of compliance, whatever that level of compliance turns out to be?

Again, in theory, we could — but the problem is once again with practice. Any government can in theory achieve 100 per cent compliance with any policy if it imposes sufficiently draconian punishments, backed up by a sufficiently high chance of enforcement. The government could impose the death penalty for breaking lockdown rules. I take it that no one in their right mind would want to go that far, though it might be very effective in cutting infection rates.

But the government could easily impose much stricter punishments for people who do not comply with lockdown rules. Should it? My strong conviction is that it should not, even if it turns out to be true that strictly enforced lockdowns significantly reduce the spread of Covid-19. China and North Korea are societies where the police have the kind of powers necessary to enforce uniform compliance with lockdown rules. I do not want to live in that kind of society. I hope that most other British citizens view that outcome with the same alarm that I do. 

The government embarked on a vast social experiment without knowing the outcome

A common response to suggesting that lockdowns are not worth their costs is to say that the only alternative to a lockdown is letting Covid-19 rip through our society unchecked — and that, the critics say, would lead to far higher death rates and a totally overwhelmed NHS. But the dichotomy is a false one. It is not true that the only options are either having a full lockdown or doing nothing at all to try to diminish Covid infection and death rates. Ending the lockdown does not mean letting the Covid infection rate run riot. There are a great many things that can be done to reduce infection and death rates short of a full lockdown — or any lockdown.

A paper published by the European Journal of Clinical Investigation in December claims alternatives to lockdown are at least as effective in diminishing the infection and death rate. The authors believe they can assess the effectiveness of lockdown measures such as stay-at-home orders and business closures against milder policies, such as reminding people to stay at least two metres apart, avoid gatherings of more than 50 people and wash their hands regularly. They say there is no evidence that lockdown results in a greater reduction in infection and death rates than the milder measures. They conclude that lockdowns are a misguided policy.

You do not have to accept that paper’s conclusion to believe it has not been properly established that the benefits of lockdown policies are worth their costs. There is a great deal of evidence that the costs of imposing a lockdown are enormous: on employment and the economy, on education, and on mental health, for example. That would be fine if the scientific literature had established significant benefits from lockdowns on diminishing the spread of Covid-19.

No one has ever prevented a disaster by pretending it wasn’t going to happen

Many people are upset at the effects of lockdown and there would be a revolt if it turned out the government has seriously damaged the economy, destroyed millions of jobs and blighted a generation’s educational prospects for no significant benefit. But that is what the evidence suggests has happened.

The government has embarked on a vast social experiment without knowing what the outcome will be, and without reliable evidence of any beneficial effect. It has yet to publish a full audit of the costs and benefits of lockdown on Britain.  A report by the think-tank Civitas estimates that the lockdown will cause (amongst many other costs) 20,000 additional deaths as a result of delayed treatments for cancer and other diseases. The Institute of Fiscal Studies claims that the effect of losing half a year’s schooling on the future earnings of Britain’s children will cost the Treasury about £100 billion in tax revenues. 

The lack of a comprehensive cost-benefit analysis by the government may be because the costs are mostly in the future, so we cannot see them. There is a very human tendency to discount them drastically as a consequence. But that is a very serious mistake. No one has ever prevented a disaster by pretending it wasn’t going to happen. Whatever “scientific” policymaking is, we do not have it in Britain at the moment.

Belief in the value of lockdowns in the absence of evidence for their effectiveness is astonishingly common: not just amongst ordinary folk, but amongst scientists too – who really should know better.

Nearly all of us trust scientists to get things right, which is why they can very easily mislead us. There was a good example of that last week, when Imperial College announced the results of its most recent React Study. The study measures the rate of infection with Covid in the UK, and it showed a sharp fall in the number of people infected between February 4 and 13, when researchers carried out more than 85,000 swab tests in England. Professor Paul Elliott, React’s director, stated that that the “encouraging results show that lockdown measures are effectively bringing infections down.”

His comment was duly repeated across almost all media outlets. Millions of Britons believed it.

Millions of Britons were misled. The results do not show that lockdown measures are “effectively bringing infections down”. The rate of infection fell certainly after the lockdown was imposed, but that does not show that the lockdown caused the fall – as Prof Elliott must know. Sweden has not imposed a lockdown. But Sweden’s Covid infection rate over the same period has fallen just as fast, if not faster, than the rate in the UK. More importantly, so has Sweden’s rate of Covid-related deaths. (See here for the most recent comparison).

Prof Elliott’s misleading claim was echoed by Mike Tildesley, a professor of infectious disease modelling at Warwick University. Reflecting on the fall in the UK infection rate on BBC Breakfast, Prof Tildesley confidently claimed about the lockdown that “we do know that any form of re-opening will cause the R number (the infection rate) to go up”.

But we do not know this. No evidence has yet been produced to show that the lockdown has had any significant effect on the reduction of Covid cases in the UK. It might have done so. Lots of eminent people are obviously willing to say that they believe it has – but in the absence of evidence, that does not make it true.

One can hardly blame Matt Hancock for continuing to insist on the effectiveness of the lockdown in the UK when so many of his highly-qualified advisers keep telling him that, until the vaccine starts to take effect, any fall in the infection rate must be the result of the lockdown.

But how difficult can it be for the Health Secretary to recognise that he has not yet been presented with evidence that lockdowns are effective?

Enjoying The Critic online? It's even better in print

Try five issues of Britain’s newest magazine for £10

Critic magazine cover