The Year the World Went Mad E-Book

Mark Woolhouse

• Herausgeber: Sandstone Press
• Kategorie: Sachliteratur, Reportagen, Biografien
• Sprache: Englisch
• Veröffentlichungsjahr: 2022

Beschreibung

'An essential book.' -Matt Ridley In January 2020, leading epidemiologist Professor Mark Woolhouse learned of a new virus taking hold in China. He immediately foresaw a hard road ahead for the entire world, and emailed the Chief Medical Officer of Scotland warning that the UK should urgently begin preparations. A few days later he received a polite reply stating only that everything was under control. In this astonishing account, Mark Woolhouse shares his story as an insider, having served on advisory groups to both the Scottish and UK governments. He reveals the disregarded advice, frustration of dealing with politicians, and the missteps that led to the deaths of vulnerable people, damage to livelihoods and the disruption of education. He explains the follies of lockdown and sets out the alternatives. Finally, he warns that when the next pandemic comes, we must not dither and we must not panic; never again should we make a global crisis even worse. The Year the World Went Mad puts our recent, devastating, history in a completely new light.

Leseprobe

First published in Great Britain in 2022

Sandstone Press LtdPO Box 41Muir of OrdIV6 7YXScotland

www.sandstonepress.com

All rights reserved. No part of this publication may be reproduced, stored or transmitted in any form without the express written permission of the publisher.

Copyright © Mark Woolhouse 2022Foreword © Matt Ridley 2022Editor: Robert Davidson

The moral right of Mark Woolhouse to be recognised as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988.

ISBN: 978-1-913207-95-3ISBNe: 978-1-913207-96-0

Cover design by Heike SchüsslerEbook compilation by Iolaire, Newtonmore

This book is for my wife and daughter, the best lockdown companions I could ever wish for.

Francisca, thank you for listening. You must have heard everything in this book a hundred times before.

Nyasha, thank you for bringing joy to our lockdown lives. I’m sorry that your generation has been so badly served by mine.

CONTENTS

Foreword by Matt Ridley

Sounding the Alarm

Early Days

Models

Lockdown

Balancing Harms

Dangerous Liaisons

A Disease of Old Age

Children and Schools

Test, Test, Test

Living with the Virus

Slow-motion Replay

New Variant

The Cavalry

The Last Lockdown?

World View

Sage Science

What Should Have Happened

Disease X

Acknowledgements

Abbreviations

Bibliography

FOREWORD

Mark Woolhouse is one of the world’s most distinguished epidemiologists. His expertise has been invaluable to Scotland, Britain and the world during the Covid pandemic. His account of the first year of Covid is a remarkable story, told with great fluency and insight by somebody who was on the scientific and political inside throughout. But he is also frustrated and baffled by one big mistake that both the United Kingdom and the world made against his advice, so his book has a critical – in both senses of the word – argument to make.

That mistake was lockdown. Mark argues that however great the threat posed by the novel coronavirus, and however badly it was underestimated at first, there was always a better strategy to deal with the resulting epidemic that would have done far less economic and social harm and would almost certainly have saved more lives too. He makes his case with both passion and logic in these pages.

Why did the world go mad? Why did the blunt and brutal policy of lockdown become the one measure that almost all politicians and scientists in almost all countries agreed was unavoidable? After all, quarantining the sick and vulnerable had always been the response to outbreaks of infectious disease, never the quarantining of everybody.

The first reason was surely that for the first time in history we could. Enough commerce had migrated online that if almost everybody stayed at home, much of society could still function – especially the office workers who take decisions in government. The logistics of online retail and the technology of video conferencing had reached some sort of tipping point. Ten years earlier, we would surely not have contemplated lockdown.

A second reason was that China locked down. The attempt to eradicate the virus at source in Wuhan was brutal but largely successful. A policy that was never on the table suddenly became imaginable. There was not a little tinge of envy in some of the early comments made in the West about the power of a totalitarian regime to shut down an entire society. Neil Ferguson of Imperial College put it well: ‘It’s a communist one-party state, we said. We couldn’t get away with it in Europe, we thought . . . and then Italy did it. And we realised we could.’ Erstwhile democrats discovered in themselves a surprising love of emergency executive orders.

The reason China tried lockdown was because of SARS. An epidemic had been halted in its tracks in 2003 not just in China but in neighbouring countries and Canada by draconian action. The result was the rapid eradication of the SARS virus altogether. It went extinct except in laboratories (from whence it leaked at least five times over the next year, but that is another story). The Chinese response to Covid, echoed in South Korea and Taiwan in particular, was aimed at suppression of the virus to oblivion. By the time the virus was spreading in ski resorts, hospitals and sports stadiums in the West, that goal was always impossible. Lockdown, designed to eradicate, was not well suited to protecting the vulnerable.

As Mark shows in this book, another reason was that western countries had always prepared for an influenza pandemic, which resulted in assumptions that everybody was vulnerable, that closing schools would help and that social distancing would work well. Covid was dramatically different, singling out the elderly and sparing children almost entirely, but being highly infectious in those showing few symptoms. Yet months later public announcements on the radio were still parroting the nonsense that everybody and anybody was at risk of death. An ultra-precautious mentality took hold, constantly reinforced by the deliberate policy of inducing fear and anxiety in the population to ensure compliance. This led to foolish mistakes like police harassment of hill walkers, and deadly ones like shipping hospital patients to care homes. Expert advice was badly unbalanced, with the malign effects of lockdown itself – on cancer diagnosis, mental health, loneliness and economic ruin – going largely ignored.

Ironically, another reason that lockdown became the weapon of choice was excessive optimism about timing. In March 2020, we were assured lockdown was a brief interruption to save the hospitals from being overwhelmed. Yet it was never realistic that the virus would just go away that fast. Thereafter, we were told that lockdowns would buy time till a vaccine came riding to the rescue, but this was dangerously optimistic about whether a vaccine would be developed at all, let alone fast. In the event, this miracle did occur, though we appeared to have overlooked the fact that rolling out a vaccine would take a long time, even in a very well-prepared country like Britain. But if the cavalry had not appeared, would we really have kept the country locked down for years? This was not a sustainable policy.

The final argument for lockdown was that there was no alternative. Mark demolishes this in devastating and relentless fashion, demonstrating clearly that his proposal to shield the elderly and cocoon the vulnerable by protecting those who came into contact with them, would have worked far better and done far less other harm. After all, this latter policy was what we suddenly discovered and adopted when the vaccine came along: if the elderly were to be prioritised for vaccination, why not for protection as well?

This is a book that shows what we should have done when the UK was confronted with the greatest national crisis of recent decades. Mark’s advice would be important even if these were lessons we could only have learned from hindsight. But this isn’t hindsight, this is the story of the advice that he was giving at the time as the crisis unfolded.

Matt Ridley

CHAPTER 1

SOUNDING THE ALARM

Early in the New Year of 2020 I was in my office in Edinburgh reading through media accounts of a puzzling respiratory disease – possibly a viral pneumonia – that had surfaced in the city of Wuhan in eastern China. I was concerned and tried to find out more.

On January 7th a helpful journalist sent me a copy of a report written by the Wuhan Municipal Health Committee. The report said that fifty-nine patients were suspected to have caught the mystery disease, seven of whom had become critically ill. The first cases dated back to mid-December 2019 and many were linked to the South China Seafood wholesale market in the north of Wuhan.

The same day the Chinese authorities announced that they had identified the cause of Wuhan pneumonia: it was a coronavirus.

A pandemic begins

As I’d been studying the emergence of new viruses for more than twenty years, I knew what to look out for. New human viruses usually come from animals, and most of them don’t spread well between humans. Some coronaviruses can do though, which meant they were high on the list of viruses to worry about.

The fact that there were already fifty-nine cases in a single outbreak told me that this coronavirus probably did spread from person to person, so it was potentially a pandemic virus. If it was, it was probably already too late to stop it. I knew that a respiratory virus could spread around our highly interconnected world and seed a pandemic in a matter of days and I had just learned that this one had been spreading for weeks. There were still plenty of unknowns, but I was now very worried.

On January 12th Chinese scientists published the new virus’s genome sequence – its genetic code. The genome sequence confirmed that it was a coronavirus and told us that it is closely related to the SARS coronavirus. This was more bad news. SARS is an extremely dangerous disease that killed more than seven hundred people in 2003. Only a prompt and vigorous international response had averted a full-scale pandemic. The possibility that SARS might one day re-emerge had been a concern ever since. Now we were facing a SARS-like coronavirus with unknown potential.

On January 21st the World Health Organization reported that there had been over two hundred cases and six deaths from the new virus in China, with further cases in Japan, South Korea and Thailand. We call a large but localised outbreak of disease an epidemic, but when an epidemic spreads to multiple countries across a wide region we call it a pandemic. This was not yet officially a pandemic but there could be little doubt that it was going to become one. The virus had already spread beyond China to three other countries and was most likely present in others that we didn’t know about yet.

That same day I sent an e-mail to Catherine Calderwood, the Chief Medical Officer (CMO) of Scotland. Even though no cases had yet been reported in the UK, in that e-mail I said that we needed to start preparing for an epidemic that would affect the whole country, Scotland included.

Four days later, new data were published by the World Health Organization that prompted me to write again, with even greater urgency. I explained that – based on these new data – I estimated that the novel virus was capable of infecting more than half the population, tripling the mortality rate and overwhelming our National Health Service (NHS) within two months. I acknowledged that there was a lot of uncertainty but stressed how serious this could turn out to be.

I received a polite reply to my e-mails telling me that everything was under control. It wasn’t, as we were all going to find out in the coming weeks.

Warning the Scottish government’s most senior medical advisor that we were facing an unprecedented public health emergency wasn’t something I did lightly. Before sending those e-mails to Catherine Calderwood I’d consulted with two colleagues who, like me, have been studying epidemics for many years. One was Jeremy Farrar, Director of the Wellcome Trust, and the other was Neil Ferguson, Director of the Centre for Outbreak Analysis at Imperial College, London. The three of us were in complete agreement and Jeremy and Neil were already in touch with the CMO England, Chris Whitty, and the Chief Scientific Advisor (CSA), Patrick Vallance.

In that flurry of communications in January 2020, we set out how we expected the next few months to unfold.

First, the pandemic would be fuelled by mild cases but with significant mortality in vulnerable groups.

Second, there was little prospect of a vaccine against a novel coronavirus becoming widely available in less than twelve months. Third, the prospects for effective therapies were not much better.

Fourth, case isolation, infection control and contact tracing would be crucial, but capacity to deliver them could be overwhelmed if case numbers rose too high. Fifth, social distancing measures such as restrictions on public gatherings and closures of workplaces and schools would then be needed.

Time was of the essence; every day it looked more certain that a pandemic was heading our way. I continued to share information with the CMO Scotland and brought Sheila Rowan – CSA Scotland – and Anne Glover – President of the Royal Society of Edinburgh and former CSA to the European Commission – into the conversation. I was told that the Scottish government was now working ‘to address preparedness’, though not what that meant in practice. I wasn’t convinced. I’d have expected a lot more urgency if the government had been doing the same calculations that I’d been doing.

Three crucial numbers

To get a preliminary idea of how an epidemic could play out we need three numbers.

The first is called the basic reproduction number – this is a measure of how transmissible the infection is and allows us to estimate the fraction of the population who will be infected.

Next is the generation time, the interval between a person getting infected and infecting others – this sets the timescale.

Last is the infection fatality rate – this tells us how many people will die.

These three numbers drove my initial estimates of the potential scale, speed and severity of a novel coronavirus epidemic in the UK. None of these crucial numbers was known precisely in January 2020, but the most uncertain of all was the infection fatality rate. The infection fatality rate is a ratio of deaths to infections. In the early stages of the epidemic in China it was entirely possible that some deaths were being missed. However, there had to be far greater uncertainty about the number of infections. At that time, only clinical cases were being counted, not milder cases or those with few or no symptoms at all. The more of those there were, the more the infection fatality rate would be overestimated.

On January 25th the World Health Organization published an infection fatality rate estimate of almost 5% – meaning that one in twenty of those infected would die. If the true value was anywhere near that figure we were facing a catastrophe. Over the next few months, as surveillance of mild cases improved, the estimate would eventually fall to around 1%. That was bad enough, much higher than influenza which typically has an infection fatality rate less than 0.1%.

All of this told me that this pandemic could be much more severe and much harder to control than the flu pandemic the UK had spent years preparing for. That is why I wasn’t convinced by the reassurances I was receiving. I felt we already knew enough to take even stronger action. Giving advice was the easy part; getting anything to happen proved a lot more difficult.

Gearing up

There was a lot more talking than action in the weeks following that first flurry of e-mails.

The UK government’s Scientific Advisory Group for Emergencies (SAGE) met on January 22nd. I heard that it was a surprisingly low-key meeting given the gravity of the situation.

I was hurriedly appointed to a SAGE sub-committee called the Scientific Pandemic Influenza Group on Modelling (SPI-M). SPI-M met on January 27th to discuss the need for mathematical modelling of a novel coronavirus epidemic in the UK.

By the end of the month questions about the novel coronavirus had been tabled in both the UK and Scottish parliaments, so the new virus was on our politicians’ radar too.

I finally met with the CMO Scotland on February 28th and at that meeting gave an update on possible scenarios for the UK’s novel coronavirus epidemic. The main messages hadn’t changed since January, but we now had more data to support them.

The Scottish government subsequently formed its own expert committee, the Scottish Covid-19 Advisory Group, informally known as SAGE-for-Scotland. I was asked to join the group but we didn’t hold our first meeting until March 26th, three days after the UK went into lockdown, by which time the course of the epidemic in Scotland and the UK as a whole was pretty much set.

The lack of urgency was troubling. Scientists were warning in mid-January 2020 of an imminent epidemic on a scale we had not seen since Spanish flu more than a hundred years ago. This was an extraordinary event that demanded an extraordinary response and got a very ordinary one.

It was a different story in Taiwan. The Taiwanese government introduced rigorous health checks for arrivals from Wuhan on December 31st 2019, long before most people in the UK had heard about the new virus. By the end of January, Taiwan was screening all international arrivals for signs of infection. At the same time, isolation of cases and their contacts plus quarantining of anyone deemed at high risk was made mandatory and was strictly enforced.

The UK is not Taiwan but, even so, when the alarm was first raised here we could have reacted with the same sense of seriousness and urgency. We didn’t. I can think of several reasons why.

For a start, there was a lack of global leadership. In this kind of situation, that is the role of the World Health Organization, but it failed to live up to one of its prime responsibilities. It didn’t even declare a Public Health Emergency of International Concern – a precursor to declaring a pandemic – until January 30th. It didn’t declare a pandemic until well into March. This badly undermined the case being made for early action in Scotland, the UK or anywhere else.

Another factor was complacency. Since the swine flu pandemic in 2009, the emergence of a new strain of influenza had been on the UK government’s national risk register and we had detailed and rehearsed plans for responding to it. It was assumed that those plans would work for the novel coronavirus, but this wasn’t flu and an even more vigorous response was required.

There was also a sense of déjà vu. In the early stages of the 2009 swine flu pandemic some scientists had confidently predicted a crisis on a scale far beyond what transpired, largely thanks to an initial overestimate of the infection fatality rate. Policy-makers might reasonably ask whether scientists were crying wolf again.

Those are all plausible explanations for the lack of action, but I think there was something more: sheer disbelief. We were asking officials and politicians to engage with a scenario lifted from a science fiction movie. They simply couldn’t take it in.

Looking forward

For those who could take it in, it was already apparent that 2020 was going to be a traumatic year. Many people were going to die and life was going to change for all of us. The policy-makers may have been slow off the mark but at least the scientists did respond during those early weeks.

My research group at the University of Edinburgh was ideally positioned to study this new threat because we do research on the epidemiology of emerging viruses, meaning that we study their origins, distribution and spread. Accordingly, I begged and borrowed the funding to expand and reconfigure my team to work on the novel coronavirus. The same thing was happening in thousands of research groups in universities, government laboratories and industry around the world. Scientists working on diagnostics, drugs and vaccines got to work as soon as the new virus’s genome was published in January.

Beyond scientific research, I had a good idea of what the imminent pandemic would ask of me and my team. I’d been deeply involved in the UK’s response to previous epidemics: BSE (mad cow disease) in 1996; foot-and-mouth disease in 2001; swine flu in 2009. I was familiar with the sometimes fraught relationship between science and policy. I’d worked with the media on communicating the science to a public anxious to understand what was happening. I’d experienced the pressures that come when the stakes are high and there is heated debate about how best to proceed.

I knew in January 2020 that I was about to go through all that again and more. I knew it would be the same for many of my colleagues in the UK and around the world. I hoped that science would rise to the occasion and that we scientists could make a difference. What I did not expect was that elementary principles of epidemiology – my own subject – would be misunderstood and ignored, that tried-and-trusted approaches to public health would be pushed aside, that so many scientists would abandon their objectivity, or that plain common sense would be a casualty of the crisis.

I did not expect the world to go mad, but it did.

CHAPTER 2

EARLY DAYS

Humanity is plagued by a multitude of different kinds of infection. No-one knew quite how many until my research team spent years trawling through the medical literature counting them. We found reports of infectious diseases caused by hundreds of different kinds of bacteria, of protozoa, of fungi (yes, fungi, but microscopic ones), of parasitic worms, and of viruses. This book is about a single recent addition to that grim catalogue.

Infectious diseases are responsible for a huge burden of death and disease in many parts of the world, but not so much here in the UK. We have a few thousand cases of tuberculosis a year, mainly imported. HIV/AIDS did kill thousands in the 1980s, and is still with us, but mainly affects known risk groups and nowadays can be treated. BSE (mad cow disease) caused a major scare in the 1990s but, thankfully, fewer than two hundred people died. We had only four cases of SARS in 2003.

The UK’s deadliest virus is influenza. It kills thousands – sometimes tens of thousands – every year, though most of us think of flu as an inconvenience rather than a threat to our lives. Swine flu in 2009–10 killed fewer than five hundred people.

We hadn’t faced anything as serious as novel coronavirus since Spanish flu killed an estimated two hundred thousand in 1918–19. When it came to handling a major infectious disease epidemic, the UK had no track record at all. In 2020, we would have to learn fast.

Learning fast

The new coronavirus that emerged in Wuhan in 2019 was an unknown quantity, but we already knew about other coronaviruses.

Several different coronaviruses infect mammals and birds. Some are a big problem in farm animals, particularly pigs and poultry.

We knew of six other coronaviruses that affect humans. Four of these human coronaviruses cause mild respiratory infections – common colds – but two are associated with more severe disease, MERS and SARS. MERS coronavirus is a problem mainly in the Middle East, where it is found in camels but occasionally spills over into humans causing a respiratory illness that is even more deadly than SARS.

SARS coronavirus and novel coronavirus are close relatives, though the new virus is even more closely related to SARS-like viruses found in bats. The International Committee for Viral Taxonomy – which adjudicates on such matters – considers all of them to be members of the same species. For that reason, the formal name for the novel coronavirus is SARS-CoV-2. I shall continue to call it novel coronavirus (though I’ll avoid the acronym nCoV). The disease caused by novel coronavirus was initially called Wuhan pneumonia, but these days it is frowned upon to name a disease after a place so this was changed to coronavirus disease 2019, abbreviated to Covid-19.

Viruses are not, strictly speaking, living organisms. They are strands of nucleic acid – the molecules that make up the genetic code – wrapped in a coat made of protein molecules.

In order to reproduce, a virus has to hijack a living cell – it could be a human cell, although viruses of one kind or another infect every kind of animal and plant – and re-purpose it for producing more viruses. A virus can only do this if it can get into the cell in the first place, which requires that one of the proteins on the surface of the virus – the key – is the right shape to attach to one of the molecules on the surface of the cell – the lock.

Novel coronavirus uses its spike (or S) protein as a key. The lock – more formally referred to as the receptor – was quickly identified as a cell surface protein called ACE-2. The receptor dictates where the virus attacks the body and therefore the kind of illness it causes. ACE-2 is found on cells in many different organs, including the lungs, heart, kidneys and intestines. The SARS coronavirus uses the ACE-2 receptor as well, so it isn’t surprising that the novel coronavirus causes a SARS-like illness.

Knowing the symptoms of any infection is central to diagnosing cases and treating patients. Our understanding of which symptoms are the most reliable indicators of a novel coronavirus infection improved during the course of 2020 and the NHS eventually settled on just three – high fever, new continuous cough and loss of sense of taste or smell – though other health agencies also list headaches, fatigue and diarrhoea.

Back in February 2020 there was a lot of debate about whether people could be infected without showing any symptoms at all, so-called asymptomatic infection. For reasons I shall come back to later on, the Chinese authorities were resistant to this idea, but we now know that asymptomatic infections are common. This makes finding cases considerably more difficult and we certainly didn’t find them all.

Most people with symptoms recover quickly but some are less fortunate and go on to develop a pneumonia-like illness. Patients admitted to hospital may need oxygen to help their breathing. A minority of cases need intensive care, and some have to be put on mechanical ventilators. Those patients are very ill and many die. For the survivors, the infection can cause significant lung damage and there is a long list of less frequent complications affecting the kidneys, heart and other organs. There can be other long-term consequences of infection too – including what is now called long covid – though this only became apparent several months into the pandemic.

Given that the symptoms of novel coronavirus infection are variable and easily confused with other illnesses we urgently needed a reliable diagnostic test. Within a matter of weeks we had several, an impressive achievement. These tests use a technique called RT-PCR to detect regions of genetic code unique to the virus.

Developing drugs and vaccines was always going to take much longer. There had been a lot of research into drugs and vaccines for the two most dangerous human coronaviruses – MERS and SARS – but without tangible success. This caused some scientists to be pessimistic about any quick breakthrough for novel coronavirus. However, vaccines had been developed for some coronaviruses of livestock, including a cattle virus that is a distant relative of novel coronavirus.

That offered some hope, but we knew from the outset that the first wave of the pandemic would have to be fought without the help of drugs or vaccines. As a first step, we needed to know more about what we were up against.

Timeline of an infection

In the early stages of the pandemic a lot of effort went into characterising the typical course of a novel coronavirus infection.

An infection typically begins when a person breathes in virus particles that then enter ACE-2-expressing cells in the thin layer of tissue – the epithelium – that lines the respiratory tract. Once the infection is established it spreads to other body tissues. For the first day or two levels of virus remain low, but they quickly build up, reaching a peak at around four to six days after infection. At that time the patient may develop symptoms and a few days later the immune system kicks in and levels of virus begin to fall. If severe illness develops it typically does so about a week after the first appearance of symptoms. Most patients that go on to die do so between one and four weeks after falling ill.

This is all important information but we also need to know when a virus infection can be transmitted to other people. Viruses get from one person to another in a number of ways: Ebola transmits through contact with bodily fluids; rotavirus through contact with faecal matter; HIV by sexual contact; Zika virus is picked up by the bite of a blood-feeding mosquito and passed to the next person by the same route.

Respiratory viruses – such as influenza, SARS and the novel coronavirus – are transmitted when we exhale, cough, sneeze or vocalise. For this to happen the virus must be replicating in the upper respiratory tract – the nose, throat and pharynx. Infected cells burst, releasing virus particles and the whole cycle begins again.

Respiratory viruses have always worried epidemiologists on the look-out for pandemic threats because it is so hard to stop them spreading. Transmission happens when people come into close contact, which they do all the time. In the absence of a vaccine, to stop a respiratory virus from spreading you have to stop people behaving as people normally do.

There is surprisingly little data on the transmission of respiratory viruses between humans but – working with Bryan Charleston and his team at the Pirbright Laboratory in Surrey – I have studied the transmission of animal viruses for many years. In early 2020 it just so happened that we were completing some experimental studies of influenza virus transmission in pigs.

Influenza virus is not a coronavirus, and pigs are not people, but I thought our findings might be relevant to novel coronavirus in humans too, if only to suggest what we should be looking out for.

First, even under experimentally controlled conditions, there is a great deal of variability between pigs: some are highly infectious for a prolonged period, others only briefly if at all.

Second, the amount of virus in a nasal swab is a good indicator of how infectious a pig is.

Third, levels of virus decline after a few days and the pigs cease to be infectious, even if we can still detect virus fragments.

Finally, the pigs can transmit the virus perfectly well without showing any symptoms at all.

All these features did turn out to apply to novel coronavirus infections in humans. The last one on my list – transmission without showing symptoms – is particularly important. It was quickly established that even asymptomatic cases had detectable levels of novel coronavirus in the upper respiratory tract, which meant they were likely to be infectious. We now know that asymptomatic cases are about one-third as infectious as symptomatic cases. We also know that symptomatic cases are infectious for one or two days before symptoms appear – the pre-symptomatic phase. This makes sense as we can detect high levels of virus during that phase of the infection.

These are important facts to know. A pandemic will be extremely difficult to control if it is helped on its way by large numbers of people who are infectious but are not yet showing – and may never show – symptoms.

Throughout January and most of February 2020 the majority of novel coronavirus cases were still in China. The Chinese National Health Commission and the China Centre for Disease Control were managing the epidemic but information was slow to emerge, and often did so in the form of official pronouncements rather than scientific reports, which made it hard to evaluate crucial evidence.

Even well into March we did not have a good estimate of the infection fatality rate, though the consensus was beginning to settle on a value of around 1%. We still did not know how many asymptomatic infections there were.

Early pandemic response

Meanwhile, China had taken drastic steps to contain the virus. On January 23rd Wuhan – a megacity of eleven million people – was cut off from the rest of China and put into an extraordinarily strict lockdown. Such an intervention was unprecedented in modern times and public health experts – myself included – were sceptical that it would work. It did and it didn’t. It worked in terms of bringing the epidemic under control in Wuhan within a few weeks and slowing the spread across China, but it failed to contain the virus within China.

By the end of February, cases had been reported from forty-eight countries. This did not deter a World Health Organization mission to China from concluding that ‘China’s bold approach . . . has changed the course of a rapidly escalating and deadly epidemic’. It hadn’t, but those misleading words were to pave the way for more lockdowns.

It wasn’t until March 11th that the World Health Organization declared a pandemic, though epidemiologists had been urging them to do so for weeks. By that date, more than one hundred and twenty thousand cases and four thousand deaths had been reported from over a hundred countries from all around the world. It was patently obvious we were in the midst of a pandemic.

Most countries, including the UK, were already taking action and, for them, the World Health Organization’s long-delayed declaration was pretty much irrelevant. The more consequential part of the announcement of the pandemic was the Director General’s call for ‘urgent and aggressive action’ to bring the virus under control, using China as example of success. He did this despite the profound doubts of many public health experts – again including me – that China’s lockdown strategy was the best approach to use in the rest of the world.

The UK had reported its first two cases of novel coronavirus on January 31st, both imported from China. By the end of February, the count had increased to twenty-three cases but there had not been any deaths.

The UK government’s pandemic response strategy in early March was badged as Contain, Delay, Research, Mitigate.

The Contain phase revolved around using case finding and contact tracing to prevent infection becoming established in the community.

Delay meant slowing the spread in the community using interventions designed to reduce the transmission rate – these could include social distancing measures that reduce the number of contacts between people.

Research covered the development of diagnostics, drugs and vaccines.

Mitigation was about patient care and NHS capacity. Preparing the NHS was a major focus of government planning, culminating in the rapid construction of Nightingale hospitals in England and Louisa Jordan Hospitals in Scotland.

The UK government announced on March 12th that it was moving from the Contain phase to the Delay phase and was abandoning testing in the community so that our still limited testing capacity could be used in hospitals. Though it was controversial even at the time, it was a defensible decision, for two reasons.

First, we couldn’t deploy – in the community or anywhere else – testing capacity we hadn’t got. That problem had its roots back in January when – had the government taken on board the seriousness of the situation – we could have started to build testing capacity immediately.

Second, it was already too late, novel coronavirus was firmly established in the UK by mid-March and containment was no longer feasible. When the UK government made the switch from Contain to Delay they said that their decisions going forward would be based on careful modelling. So, it’s time to talk about models.

CHAPTER 3

MODELS

Scientists use mathematical models to study complex biological processes at every conceivable scale from single molecules to ecosystems. An epidemic has fewer moving parts than an entire ecosystem but it’s still extremely difficult to predict how it will play out, even if you’re an expert in public health. Models can help.

Epidemic maths

An epidemic is the consequence of infection being transmitted from person to person. The crucial characteristic of this process is that the more infections there are in the population the greater the risk that an uninfected member of that population gets infected. The technical term for this characteristic is positive feedback and it is why epidemics grow exponentially.

Exponential growth is multiplicative (for example, 1, 2, 4, 8, 16, . . .) rather than linear (1, 2, 3, 4, 5, . . .) – it will be a recurring theme in this book. An epidemic can’t grow exponentially for ever – sooner or later it has to slow down because there are fewer people left to infect – but it can do so in the early stages, as we would soon see with novel coronavirus.

Numbers of cases of non-communicable diseases like stroke or diabetes do not behave this way. For those diseases we sometimes find clusters of cases within families or in a local area but there is no person-to-person spread, so no positive feedback and no exponential growth.

Positive feedback makes it hard to say what will happen if you put in place an intervention designed to reduce the rate of transmission, such as lockdown. If, for example, a lockdown halves the transmission rate (by halving the number of people we come in contact with) it’s not obvious what that would do to the size of the epidemic. We’d expect it to be smaller, but would it be half the size, or more, or less? Models are good for answering that kind of question. (The answer, by the way, is that sometimes halving the transmission rate has little impact on epidemic size, sometimes it has a big impact and sometimes it stops the epidemic from taking off altogether – it depends where you’re starting from. Epidemics are complicated.)

This is much more than an academic nicety. If those commenting or advising on the response to novel coronavirus do not understand the dynamics of epidemics – and many clearly did not – then their comments or advice can be misleading. Throughout 2020 we saw one example of this after another in public discussions of the R number, herd immunity, elimination, travel bans and the second wave. The resulting confusion clouded those important and necessary debates about what to do next and increased the likelihood of getting crucial decisions wrong.

My favourite example of epidemiological models outperforming expert opinion involves the work of a pioneer in the field, Roy Anderson of Imperial College. Roy and his colleagues modelled the future scale of the HIV/AIDS epidemic in the late 1980s when it was a new and still relatively rare disease. Their predictions of millions of deaths globally were ridiculed by public health ‘experts’ who failed to grasp that – unlike more familiar infections such as flu – this would be a long, drawn-out epidemic and difficult to stop. As the world now knows, Roy’s models were right and the critics were wrong.

SPI-M

For all these reasons, SPI-M – the modelling sub-committee of SAGE – played a pivotal role throughout the UK coronavirus epidemic. SPI-M was co-chaired by Graham Medley – who had invited me to join the committee – and Angela McLean, Deputy CSA. I knew Graham and Angela from when we all worked together in Roy Anderson’s team at Imperial College in the 1980s. SPI-M is a large committee and the meetings were always lively. The discussions were well-informed and uninhibited, ideas were put forward and challenged, every session was a crash course in our rapidly expanding knowledge of coronavirus epidemiology.

In happier circumstances, working with SPI-M would have been a joy, but the meetings during 2020 always had an undercurrent of deeply felt concern. The whole point of the modelling was to help us glimpse the future, and none of us liked what we saw there.

Since the models were informing literally life-or-death decisions it was vital that the outputs and the recommendations coming from SPI-M were as robust as possible, and that any uncertainty was reported alongside the headline results. To achieve this, SPI-M uses ‘ensemble’ modelling.

Ensemble modelling works by consolidating the outputs of multiple, independent models rather than reporting only the output of a single ‘best’ model. If the models agree, we are more confident in our conclusions. If they disagree, we look for the reasons why and, in the process, may learn something about questions we still need to resolve. Ensemble modelling is a tried-and-trusted approach to modelling complex problems where there are many uncertainties. Climate change modelling works the same way.

SPI-M was a tremendous resource for the UK government in 2020, an assemblage of top scientists in the field, with a structure ready and waiting to deliver modelling outputs in real time as needed, working under the direction of the Department of Health and Social Care (DHSC).

That said, SPI-M frequently found itself at the centre of the storm. The models were (quite rightly) intensely scrutinised and strongly challenged. When, for reasons we’ll come to later, trust in the models declined, there was plenty of negative publicity, not least in the November 2020 BBC2 television documentary Lockdown 1.0 – Following the Science?

SPI-M had to contend with one challenge right away; it was set up to tackle the wrong disease. That’s apparent from its full name: the Scientific Pandemic Influenza Group on Modelling. The committee was created in the wake of the 2009 swine flu pandemic to model the much-anticipated next influenza pandemic (something we may experience in coming years).

For the most part, the influenza expertise in SPI-M was an asset; many of the epidemiological drivers – particularly human behaviour – are similar for novel coronavirus and for flu. Nonetheless, adjustments were needed and some of the models had to be re-written from scratch.

Even then, the new, bespoke coronavirus models betrayed their influenza pedigree. A good example is that the models included schools but not care homes. Schools were the major driver of the 2009 swine flu epidemic in the UK but they were not the major driver of novel coronavirus – this is one of many features of its epidemiology that are more like SARS than flu. Care homes, on the other hand, were crucial. A large proportion of deaths due to novel coronavirus occurred in care homes. As we were to learn, the course of the epidemic in care homes was distinct from that in the wider community – that needed to be modelled explicitly.

Another missing feature was shielding of the vulnerable. Shielding featured prominently in the government’s response from March 2020 onwards, but most of the modelling of this intervention was done by independent researchers. Shielding surely had some effect on transmission to people most likely to die – that was what it was intended to do, after all. It was also likely that many people at risk were taking precautions of their own – independent of government advice. The influenza models didn’t capture any of this.

A striking omission from the original influenza models was lockdown. A full lockdown was not part of the UK’s planning for an influenza pandemic. The models included options for some elements of social distancing – notably travel restrictions and closing schools and workplaces – but not an instruction for most of the population to stay at home. Over the past ten years, SPI-M had built up an evidence base for the interventions we expected to use to control a pandemic. Now we were contemplating doing something different. We’d done our homework, but we’d prepared for the wrong exam.

The influenza legacy was a weakness, but the novel coronavirus models were still useful tools. They were flexible enough to answer the many different questions being asked by government, such as when cases would peak, how much intensive care unit capacity would be needed or what difference more testing would make.

I told a House of Commons Select Committee in November 2020 that decisions on how to respond to an epidemic like novel coronavirus should not be taken without input from models. As I’ve explained, they provide insights into how the epidemic might unfold that couldn’t be obtained in any other way, particularly in the early stages.

That said, I wouldn’t want decision-making to be over-reliant on models either. In a Good Practice Guide for modellers I co-authored in 2011 we stressed that models should only be one of the lines of evidence informing policy-makers, never the only one.

In March 2020, however, you could easily get the impression that the UK government’s mantra of ‘following the science’ boiled down to following the models. That’s how it looked and that’s how the media presented it.

The R number

One of the most prominent outputs of the models was an estimate of the R number. The formal term for R is the ‘case reproduction number’ and it has a simple definition: R is the average number of cases generated by a single case. The single case is referred to as the index case, and the cases that it generates are known as secondary cases.

The R number is a measure of how well the infection is spreading in the population. R greater than one means that, on average, an index case generates more than one secondary case and the number of cases will grow. R less than one means that, on average, each case generates less than one case and the number of cases will decline. R equals one means that the incidence of new cases stays the same. The phrase ‘bringing the epidemic under control’ implies reducing the R number from above one to below one.

For a given infection in a given population, the maximum possible value of R is called the basic reproduction number, written as R0 and pronounced ‘R nought’. For novel coronavirus in the UK in March 2020 R0 was about three. At that stage of the epidemic, the case reproduction number, R, was equivalent to the basic reproduction number, R0, but once steps were being taken to reduce transmission rates the R number fell and it would fluctuate between 0.7 and 1.5 for the rest of the year.

SPI-M put a lot of effort into those estimates of the value of R. The outputs of these calculations were reported by DHSC as the government’s official weekly R number. The R estimate was never a single value, always a range, the size of the range reflecting the degree of uncertainty in the true value. There is inevitably some uncertainty because, in the absence of detailed and systematic tracing data telling us directly how many secondary cases are generated by an index case, we can only measure R indirectly, estimating it as best we can from the epidemiological data available.

Though R is important to infectious disease epidemiologists, it isn’t that useful as an operational public health tool.

R doesn’t tell you about the number of infections in the population – that’s the prevalence – or the number of new infections – that’s the incidence.

R only refers to infections, so it doesn’t tell you about the numbers of hospitalisations and deaths, which matter most to the NHS and government.

Nor does R help anyone understand their individual risk, which matters most to you and me.

R does tell you about the trajectory of the epidemic but not how fast the epidemic is growing or shrinking. You get that from the doubling time, which I’ve long argued is a more useful number to communicate than R.