I am writing these words at the beginning of June, but you should by now be looking back on the worst of the UK’s Covid-19 epidemic. History books will dissect every aspect of the disease and governments’ response to it, but it is already clear that there has been an unexampled disregard for the foundational pillars of the scientific method even as governments trumpet that they are “following the science”.
The Royal Society’s motto is nullius in verba — “take nobody’s word for it” — but at every stage we have failed to apply scrutiny where it is due, or even to stop and check we are on the right ladder before we carry on climbing. For the country that is the birth-place of scientific inquiry and epidemiology it is astonishing. My godfather, professor of physics at Oxford, told me that the three most scientific things you can say are, “I don’t know”, “prove it” and “I’ve changed my mind”. Let us do each in turn.
“I don’t know”: Care homes
My first thought while watching TV footage of this emerging highly-infectious disease in Wuhan and later Lombardy was, “This is mostly old people,” and reports on the impact of co-morbidities on mortality were available from the European Centre for Disease Prevention and Control by mid-February. So: how can it possibly have made sense for the UK government to guide (until 12 March) that it was “very unlikely that people in care homes will become infected with Covid-19”, even while SAGE minutes recorded scientists’ unease at this likelihood? This simply could not be the case, as the inevitable royal commission will point out.
Fatally, they didn’t just say “I don’t know”, a holding statement while parameters clarified and data accumulated. There was no application here of the precautionary principle they inflicted on the rest of us. Instead, they issued specific guidance that left care home managers powerless as Covid-positive patients were discharged from hospitals back to the homes. To protect our NHS, which only faulty modelling had suggested needed protecting, we actively seeded the disease into its most vulnerable demographic.
Until mid-April, with the escalating deaths in care homes agonisingly clear across Europe, government policy was still for patients to be discharged to care homes from hospitals without requiring negative tests. And so the toll: around half of UK Covid-19 deaths are care home residents, despite them accounting for only 0.6 per cent of our population. It is what France, Spain, Sweden, Lombardy and New York also got wrong.
What could, should the government have done? In China and South Korea, with astonishingly low death rates, care homes required two consecutive negative tests 24 hours apart before accepting patients discharged from hospital, and instituted regular temperature checks on all residents with compulsory quarantine for anybody symptomatic. In Germany, an insurance-sponsored health system creates structural overcapacity, with hospitals paid for each night a patient is retained, thereby creating an incentive for retention rather than discharge (unlike in our NHS).
Thus Germany, whose population is roughly 25 per cent bigger than ours, has suffered approximately a quarter of our Covid deaths. Clap that: NHS cultism might have done more harm than good, when common sense and scrutiny was needed rather than applause.
Care homes might have been able to weather an unsustainable, but brief, cocooning of the most vulnerable while the virus passed through the rest of society to build group immunity. Indeed, YouTube retains government chief scientific advisor Sir Patrick Vallance’s “Plan A” from mid-March – which is exactly this.
“Prove it”: Models, SAGE and scrutiny
It was the growth of evidence-based medicine, following James Lind’s 1753 publication of the results of his controlled clinical trial on his fellow “limeys” to treat scurvy, that revolutionised healthcare. The results are that we are no longer treated with leeches and attempts to rebalance our humours. Following Karl Popper’s work on falsification, we now actively try to disprove our own “null hypothesis”.
But at every stage during the pandemic, ministers have deferred to scientists who themselves deferred to the projections of models, like ancients obeying an oracle, even when data on the ground told a completely different story.
It should have led to the obvious conclusion that the policies which flowed from the models should be ditched at the double. A weather forecast is useful right up to the point when it is more useful just to look out of the window or go outside. Looking at SAGE minutes, it isn’t clear that the output of these models was ever measured against reality. Three months into the crisis, two government departments published separate estimates for the rate of viral transmission in England. One calculated from the output of a complicated model was more than three times higher than the Office for National Statistics calculation from actual tests, but guess which one made the headlines?
Statisticians on social media had a field day pointing out the chasm between modelled outcomes and reality, but it is not clear that the models on which SAGE relied (both their input parameters and mechanical dynamics) were continually refined with on-the-ground data (or simply discarded as wrong). I work in finance, where we also use models but are judged solely on results. In epidemiology it looks (at least from outside) that the models are reified with status themselves rather than regarded as tools to be discarded the moment they don’t measure up to reality.
Ministers have deferred to scientists who themselves deferred to the projections of models, even when data on the ground told a completely different story
SAGE minutes record an exercise in committee groupthink. As Dr John Lee has pointed out, “One of the key things about science is that it is fuelled by doubt, not certainty” — great science has scrutiny at its core. For all the talk of Dominic Cummings’s new broom of STEM-conscious political operatives, moving on from Oxbridge arts bluffers and their hunches and excuses, where was the “Red Team” to resist the siren call of consensus and lead to better decisions?
Where were the beloved super- forecasters with out-of-domain intellectual firepower to question whether the forecast of 500,000 UK deaths could be plausible? Spanish flu (the biggest pandemic killer in history) took half that, before antibiotics. Sure, it was possible, but was it likely? What were the odds?
SAGE minutes record an unmistakeable shift in tone and advice when this forecast was presented. Its sheer awfulness created such gravity that advisers (who can only get it wrong by not being cautious enough) would never escape its event horizon. Why weren’t Oxford’s team, who specialise in zoonotic viruses and who looked at the same data as Neil Ferguson’s modelling-led team but came to wildly different conclusions, on SAGE’s panel to provide an alternative view?
Why were there no economists on SAGE? Economics is not the bloodless pursuit of money but the science of decision-making under uncertainty where resources are finite; could they really have brought nothing to the party? Only now are the economists pointing out that the recession is likely to take many times the lives taken by the virus being listened to by the government.
There is much more to science than models, and the most accurate analysts throughout were those who relied on other pillars of science, which does not only proceed from models after all: it also has, inter alia, experiments, defaults (Popper’s “null hypotheses”) and controls.
In mid-March, Stanford’s Nobel laureate Michael Levitt (biophysicist and professor of structural biology) discussed the “natural experiment” of the Diamond Princess cruise ship, a petridish disproportionately filled with the most susceptible age and health groups. Even here, despite the virus spreading uncontrolled onboard for at least two weeks, infection only reached a minority of passengers and crew. Levitt concluded that we must have high levels of immunity that can clear the virus.
Three months later, the “dark matter” of innate immunity and T-cell immunity from exposure to previous coronaviruses backs up this view. And using very simple mathematics (not “15,000 lines of uncommented code”) Levitt demonstrated that the virus’s spread had never been exponential but rather was running out of steam in a predictable pattern from day one. Who listened?
I posted evidence of this same clear deceleration in Covid-19’s growth rate for the UK and other European countries towards the end of March. Back then, the media were full of alarm at the “exponential growth” about to overwhelm our health system, even while the data itself clearly showed we were already near the tipping point of the bell-curve (meaning the disease is on the wane). We were already past the point where lockdown could have made much difference.
The eccentric biostatistician Knut Wittkowski came at things from a different angle: the “null hypothesis” — a default, in layman terms. In the absence of evidence to the contrary, he assumed that Covid-19 was a normal viral respiratory disease, and at the end of March wrote a compelling but neglected paper showing how the emerging data backed up his view that “respiratory diseases [including Covid-19] . . . remain only about two months in any given population”. His fate was to be another Cassandra. Nowhere in the SAGE minutes is there any evidence of somebody meekly raising their hand to ask the most obvious question of all: “Is this just ‘another one of those’?”
“I’ve changed my mind”: Lockdown
We also have “controls”: not all countries behaved the same. Neither Sweden nor Japan locked down. So, if the lockdown hypothesis were true, Stockholm would be a morgue and Greater Tokyo (population 38 million) a necropolis. The Free Swedes pointed out all along that lockdown would be much easier to get into than out of: no kidding. The moment these “controls” proved that the hypothesis of lockdown was wrong (evident by around mid-April), it should have been abandoned without delay.
We were assured going into lockdown that it wouldn’t be for a day longer than required, but somehow allowed the default to change such that the suspension of our most basic civil liberties could be put on a 28-day review cycle. By mid-April it was clear that lockdown made no difference to the spread of the virus, though it will take years to unpick why some countries did better than others. But there was no challenge to lockdown either from the cabinet or the opposition; the scrutiny provided by our legislature was absent when it was most required. And most of the press competed to cheer on government policy and scold such public sceptics as there were.
The social costs of lockdown are extraordinary: the burden of proof for every day it was maintained should have been more extraordinary still. Instead, we developed mass Stockholm Syndrome (ironic, when schools and pubs in Stockholm itself were still open). In the fog of war and with data still unclear, most of us can understand why Boris Johnson declared the first lockdown: to give the NHS time to build capacity and to gauge the wave that approached.
But by mid-April (albeit with Johnson ill), the data was clear: there was no danger at all of defences being overwhelmed. Dominic Raab should have had the courage (and the authority?) to lift lockdown immediately. It should never have been down to a cabinet, let alone the so-called coronavirus quad of ministers or a poorly prime minister, to decide when rights we have taken for granted since Magna Carta should be restored: their suspension should have needed justifying daily, with the bar set high.
Even if we are uninterested in blame allocation (for now), we should be profoundly concerned at the precedent: the legislation was fatally flawed by the omission of a dead man’s switch, such that civil liberties would be restored automatically the moment their suspension was not overwhelmingly and objectively justified.
We needed checks and balances at every point. Presented with accurate data from the ground (rather than output from models) the government should have been not just enabled, but compelled, to say: the evidence has changed, we’ve changed our mind: no harm, no foul. Some backbenchers made this point, but couldn’t get us out of the eel-trap.
I have no idea whether there will be either a second wave or a totally different epidemic in the next few years. But we urgently need to repair the infrastructure not just of our healthcare and procurement systems but also of the decision-making processes underpinning our scientific advice, our policy response and our legislative safeguards, ready for the next crisis that comes along. It should start with ensuring that the right questions will be asked, and that scrutiny is maintained and sustained when the going gets tough. We did not have to get this wrong the way we did: we must not repeat the fatal steps that brought us to where we are now.
Enjoying The Critic online? It's even better in print
Try five issues of Britain’s newest magazine for £10Subscribe