How Britain (narrowly) avoided lockdown last Christmas

Exactly a year ago today, the cabinet met to decide whether or not to lock down to tackle Omicron. At the time, published Sage documents had outlined a range of 600 to 6,000 daily deaths unless more action was taken. Recalling the anniversary, I had an interesting exchange with Graham Medley, who chaired the SPI-M modelling committee that fed into Sage. ‘We obviously made sure that the people we were talking to did understand,’ he told me.

That raises the prospect that a subset of people may have been briefed that Sage was discarding real-world South African data on the mildness of Omicron – so its ‘scenarios’ could bear no resemblance to reality. As proved to be the case. But this did mean that Britain was very nearly locked down for a fourth time on the strength of ‘scenarios’ that even the authors may not have regarded as credible. 

‘Who knew or understood what, and when, is likely to be the main point of the inquiry,’ Professor Medley added. But we can put together a timeline now. The decision to lock down was made by cabinet members – many of whom had resorted to asking for their own advice as they did not trust Sage. This even extended to the Prime Minister, who said later that he spotted Omicron’s fall in Gauteng, the South African epicentre, and used that to challenge his advisers’ assumption that the ‘doubling time of two days’ meant it would keep growing without lockdown.

The whole episode offers an interesting example of how the government works in a crisis, raising questions of accountability among modellers whose work can decide whether a country is locked down or not, the nature of caution (is lockdown cautious or reckless?) and the flaws in a Sage system that still exists now, unreformed, ready to spring back into action when needed. Given how many of the key players then are still in place now, we can’t expect much scrutiny over the mistakes made. 

So here is how things unfolded.

Monday 22 November: variant detected

• Omicron first detected in South Africa: then known as B.1.1.529. Four days later it’s named a variant of concern by the WHO and named ‘Omicron’. Using Greek letters is to avoid a virus being named after a place of discovery: ‘the Kent variant’ became ‘Alpha’, etc.

Thursday 25 November: travel disruption

• HMG announces that six countries in southern Africa have been added to the UK’s ‘red list’ from midday on 26 November, requiring returning travellers to enter hotel quarantine on arrival in Britain.

Saturday 27 November: cases found in Britain

• First cases detected in the UK, in Chelmsford and Nottingham.
• Boris Johnson makes a statement: ‘We need to buy time for our scientists to understand exactly what we are dealing with.’ (‘Buying time’ is usually a euphemism for restrictions.)
  • Four more countries are added to the red list.
  • Face coverings are made compulsory in shops and on public transport.
  • Suspected Omicron cases are required to isolate for ten days – regardless of vaccination status.
  • Moves to accelerate boosters and make them available to younger people by reducing the gap between second and third doses (confirmed by JCVI two days later).
• Chris Whitty also stressed that ‘it is easier for the virus to achieve the ability to escape for infection than it is to achieve escape for severe disease’.

Sunday 28 November: South Africa flags ‘extremely mild’ Omicron

• On BBC1, Andrew Marr interviews Angelique Coetzee – the doctor who first spotted Omicron in South Africa – who says patients with the strain have experienced ‘extremely mild symptoms’.

Saturday 4 December: South African data – a potential game-changer?

• South African authorities published data saying those in hospital were there for an average of ‘2.8 days… compared to an average length of stay of 8.5 days for the past 18 months’. This has game-changing implications for hospital capacity: if patients are out in less than three days, this means hospitals can cope with higher peaks and are at far less risk of being overwhelmed.
• The report also found:
  • Fewer people hospitalised with Omicron have ended up in intensive care: 8 per cent, compared to 25 per cent for Delta.
  • Fewer patients need oxygen: only about a third. A big change, the study says, because until now pretty much every Covid patient has needed oxygen.
  • Most Omicron patients were not admitted for Covid: they registered due to the hospital policy of testing everyone admitted.

But how much could the UK read into the South African data? At the time, it was unclear whether Omicron’s milder effect was due to South African health features. For example, SA had seen a large Beta wave, which may have provided better immunity against Omicron than the Alpha and Delta waves experienced in the UK (where Beta never took off). It was summer in South Africa and winter in the UK. Covid always finds it easier to spread in winter with more people indoors etc. With different demographics and infection history, it was not obvious that what held for Gauteng would hold for the UK. But nor was it obvious that these traits would not apply to the UK. This would become the central dilemma for UK policymakers.

Wednesday 8 December: Plan B in England and Vallance’s briefing

– Plan B introduced in England. Face masks are compulsory in most indoor public venues, staff are asked to work from home if they can, ‘Covid passes’ (requiring proof of vaccination or a negative test) needed for large venues. This fell short of ‘Step 3’ post-lockdown, a ‘rule of six’ for indoor social gatherings and the closure of nightclubs.

– The Vallance briefing to cabinet. Draft Sage advice is shown to cabinet members and Patrick Vallance, the UK’s chief scientific officer, which presents a view that makes the case for interventions. One senior cabinet member, speaking to The Spectator, summarises it as saying that Omicron’s faster rate of spread more than compensates for any lower severity – so action is still needed.

Vallance explained it by saying the rate of growth is exponential, but severity is a linear measure. They think Omicron will be somewhere between half to the same severity as Delta. But just assume it’s half what Delta is: that still doesn’t help you, because of the growth rate. If you’re doubling every two days, then all you purchase is two days extra. You’re back to where you started with half the severity in terms of hospitalisations.

Were ministers briefed on the possibility of a far shorter hospital stay?  The same minister summarises the Vallance briefing as saying: ‘The rate of growth in South Africa is exponential. It’s doubling every 2.5 days. Hospitalisations, always: ten days, or more.’ The average hospital stay in South Africa was 2.8 days, not ten days. But this 2.8-day figure was never printed in any Sage minute – in spite of later recognition that ‘halving the length of stay would permit double the admissions’. Ministers regularly asked about hospital stay duration, and were usually told it was too early to tell.  These ministers spotted a common refrain from those in Sage: any bad news from South Africa (such as faster transmission) was modelled and reported in an instant. But any good news – even hugely good news – was either ignored, or placed in the ‘too early to say’ category.

A Sage paper from 22 December made it possible for Simon Cook, a Spectator data reporter, to reverse engineer NHS data and show that UK hospital stay had reduced.

Friday 10 December: South African transparency

• South African data continues to flood in – the country’s largely privatised healthcare system is (unlike the NHS) big on data and transparency. The below from Pretoria shows the far lower death ratio for Omicron, taken from this weekly study.

Saturday 11 December: first UK Omicron impact model produced

• A London School of Hygiene and Tropical Medicine paper highlighted a range of scenarios which showed between 2,140 and 7,190 daily hospital admissions depending on the range of optimism. ‘Under the most optimistic scenario (low immune escape of Omicron and high effectiveness of boosters), a wave of infection is projected which could lead to a peak of over 2,000 daily hospital admissions’.
• ‘Due to a lack of data, we assume Omicron has the same severity as Delta.’
• No mention was made that the assumption of same-severity was massive and scientifically groundless given that, in South Africa, all figures pointed to a far milder outcome. An economic analyst at JP Morgan thought this all a bit fishy and began to run his own numbers…

Sunday 12 December: Boris warns of Omicron ‘tidal wave’

• Covid alert level is raised to 4: ‘Transmission is rising exponentially, social distancing continues’.
• Johnson makes a statement: ‘No one should be in any doubt: there is a tidal wave of Omicron coming… At this point our scientists cannot say that Omicron is less severe.’
• He also announces ‘Get Boosted Now’ – a major acceleration of the vaccine programme with a target to give a booster to every adult by the New Year, a month earlier than previously planned. 
• Other NHS appointments are deprioritised to accelerate boosters. 

Monday 13 December: The first JP Morgan study – ‘This would be manageable without further restrictions’

• JP Morgan research note is published, flagging flaws in LSHTM paper. It says LSHTM ‘assumes that the virulence of the Omicron variant is the same as the virulence of the Delta variant. But, evidence from South Africa suggests that Omicron infections are milder than Delta infections… If, in addition, the duration of the stay in hospital declined from eight days to three days, then bed occupancy by Covid-19 patients at the end of January 2022 would be 33.6 per cent of the peak seen in January 2021. This would be manageable without further restrictions.’
• Rishi Sunak and Jacob Rees Mogg (both former financiers) receive the JP Morgan note on the day it is released and start circulating it to others. It ends up being passed around cabinet like a samizdat paper. 

Monday 13 December: Johnson says it’s time to ditch the idea that Omicron is milder

• In a video statement to the BBC, he said: ‘The idea that this is somehow a milder version of the virus, I think that’s something we need to set on one side and just recognise the sheer pace at which it accelerates through the population.’

Tuesday 14 December: More good news from South Africa

• A Discovery Health/South African Medical Research council presentation confirms that ‘risk of severe disease and hospitalisation [is] significantly lower in Omicron infection compared to prior variants’ and two Pfizer doses provide ‘significant protection’ against Omicron.
• Some 100 Tory MPs voted against Covid passes for nightclubs and large venues (which would require patrons to show proof of vaccination or a negative test).
• Sunak arrives in California for a business trip.

Wednesday 15 December: Whitty – ‘all the things we do know are bad’

• Whitty said: ‘This is a really serious threat at the moment. How big a threat – there are several things we don’t know, but all the things that we do know are bad.’ He added: ‘I just think there’s a danger people have over-interpreted this to say, this is not a problem and what are we worrying about?’

• Dr Jenny Harries, head of the UK Health Security Agency, proclaims Omicron to be ‘probably the most significant threat’ since the start of the pandemic.
• Cases in Gauteng had started to fall, South African government data showed, countering the Sage theory of a ‘doubling time’ of two days. This was widely reported in the UK.
• SPI-M-O modelling said there would be 3,000 to 10,000 daily hospital admissions and 600 to 6,000 deaths a day without further restrictions. ‘To prevent such a wave of hospitalisations, more stringent measures would need to be implemented before 2022.’
• Unison, a trade Union, calls for schools to close. 
• Sunak, in California, gets word that a press conference is planned for Friday 17 December. Fearing the announcement of a lockdown, he books a flight back to London and urges the PM to hold a cabinet meeting and make a collective decision which would provide political cover to avoid another lockdown.

Thursday 16 December: Sage suggests restrictions

• Sage meeting (minutes published 18 Dec):
  • ‘The number of Omicron infections in the UK has continued to increase very rapidly with the doubling time in England currently around two days… a significant package of non-pharmaceutical interventions [ie, lockdown measures] would be required to slow growth in infections.’
  • ‘In Gauteng, which is ahead of the UK in the epidemic trajectory, high levels of infection are now leading to significant numbers of hospitalisation’. No mention that cases in Gauteng had peaked, as was by then widely known (and reported in the Mail Online a day earlier). Johnson later says he spotted this himself.
  • ‘Without intervention beyond those measures already in place (Plan B), modelling indicates a peak of at least 3,000 hospital admissions per day in England.’

Friday 17 December: Tories in trouble

• By-election result shows that the Tories lost the North Shropshire by-election.

Saturday 18 December: Neil Ferguson says ‘Omicron poses a major, imminent threat to public health’

• Neil Ferguson is back on manoeuvres. The FT splashes on an Imperial College London report, saying ‘Omicron no less severe, experts say’. This was now contradicted by a weeks’ worth of South African data.
• Ferguson’s group reports ‘no evidence of Omicron having lower severity than Delta, judged by either the proportion of people testing positive who report symptoms, or by the proportion of cases seeking hospital care after infection’.
• Real-world data continues to contradict Ferguson and Sage, but it’s still left to banks to point this out. A research note from Jeffreys Equity Research finds ‘deaths due to Omicron are also lower across all age groups… a 2-10x reduction for age 5-9 and 75-79, and a 5x reduction for age >80’

• Brexit minister Lord Frost resigns over the introduction of Covid passes, writing to Johnson: ‘We also need to learn to live with Covid and I know that is your instinct too… I hope we can get back on track soon and not be tempted by the kind of coercive measures we have seen elsewhere.’
• Professor Medley, chair of Sage modelling, says on Twitter that ministers are only shown ‘scenarios’ that would require action. 

Sunday 19 December: a new LSHTM model

• LSHTM updates a new model, released on SPI-M-O minutes – which is still nowhere near reality... Below is how it compares to actual outcomes.

• Raghib Ali’s minority report. Then education secretary Nadhim Zahawi requests an independent review from Dr Raghib Ali, a Cambridge academic with a large Twitter following who has been doing a lot of private work on Covid. Ali advised Zahawi that ‘the most likely scenario was going to be the best-case scenario, or better’. Ali’s advice to Zahawi is understood to have been similar to his 9.30 p.m. Twitter thread of that day. Zahawi sent this to Johnson ahead of a cabinet meeting. Ali also sent the advice to the Cabinet Office and the Times.

Monday 20 December: cabinet decides against lockdown

• Cabinet meeting to discuss options, which lasts more than two hours. They decide against lockdown – with only Michael Gove advocating for it.

After the Cabinet meeting, the heat quickly died out of the debate. The picture from South Africa became clearer by the day and it was soon evident that the UK was not following the trajectory feared by Sage. This point was made daily at The Spectator’s Data Hub: Michael Simmons, a data reporter at The Spectator, had used scanning technology to grab the Sage graphs and draw the actual figures against them. But there was little reflection from any Sage scientists on why their figures went so wrong – or how close Britain came to being locked down on a false premise.

Wednesday 22 December: more evidence piles up for mildness

• A Scottish study finds Omicron to have a two-thirds reduction in hospitalisation. 
• Imperial study (report 50) finds that Omicron is a milder strain after all, presenting a 40 to 45 per cent reduced risk of severe hospitalisation, according to UK data.

Thursday 23 December: Sage still quiet on far shorter hospital stays

• A Sage memo sounded as if the average hospital stay was still a mystery, saying: ‘A reduced length of stay would allow more capacity within hospitals to manage this, and, to a first approximation, this would scale linearly with the change, i.e. halving the length of stay would permit double the admission.’ In fact, Sage data from the day before allowed this figure to be reverse engineered by The Spectator.

Tuesday 28 December: Oxford scientist echoes South African data

• Sir John Bell confirms that the hospital stay duration for Omicron is closer to three days, as seen in South Africa. But this was an interview, with the all-important statistic still not published by the NHS, UKHSA or Sage.

Thursday 30 December: Warwick model severe outcomes

• Warwick University, one of the modellers used by Sage, releases its own figures dealing with scenarios ranging from 10 per cent to 100 per cent severity compared to Delta. 

Some overall points…

Boris Johnson’s position. His allies say he was never going to lock down for Omicron, and that he had confronted the scientists, who told him he was wrong. This chimes with reports of Johnson using the phrase ‘the evidence just isn’t there’ in multiple meetings. Others say Johnson resisted earlier lockdowns (to the chagrin of Dominic Cummings) but ended up taking the line of political least resistance and following public opinion, which was usually strongly behind lockdowns. So the cabinet meeting was intended to create a force against lockdown, countering the Sage apparatus. 

Rishi Sunak’s position. In an interview with The Spectator in the summer, he details how he cut short a trip to California to return and urge the Prime Minister to hold a cabinet meeting that would give the political support (or cover) needed to reject lockdown. Although he liaised with ministers beforehand, he spoke very little in the actual meeting – saying that the PM had ‘already heard enough’ from him.

Michael Gove’s position. He was one of the few at that cabinet meeting who did advocate for a lockdown (or a significant strengthening of measures), believing that delays in the past had led to worse outcomes overall – and that the same would prove true this time. This remains one of the biggest debates about lockdown, with ICL’s Neil Ferguson claiming that a ten-day delay in the first lockdown cost 20,000 lives. But others have claimed that his figures were cooked and that the virus was falling before lockdown due to the public taking evasive action: an unforced response that was the basis of Swedish policy, but didn’t factor in UK modelling which assumed that, without restrictions, people would behave as normal. Gove had been appointed levelling up secretary by then, so was less involved in the discussion.

Sage’s position. Designed as a group of some 200 experts to provide technical advice to the government, Sage changed when its membership was made public. This allowed some of its more vocal members – such as Ferguson and John Edmunds – to be seen as the voices of ‘the science’, given that ministers had said they were following ‘the science’. A Sage ‘consensus’ document would be drawn, a political process giving power to whoever held the pen.

Rishi Sunak has questioned how far the Sage summaries reflected the degree of consensus; ‘The Sage people didn’t realise for a very long time that there was a Treasury person on all their calls. A lovely lady. She was great because it meant that she was sitting there, listening to their discussions.’ He says she would tell him: “Well, actually, it turns out that lots of people disagreed with that conclusion,” or “Here are the reasons that they were not sure about it.” So at least I would be able to go into these meetings better armed.’

What we don’t know. The January 2021 lockdown was extended for months on the basis of sage ‘scenarios’ – were they similarly flawed? I have been unable to find the data or assumptions behind those studies so don’t know (for example) if they assumed no behavioural response (i.e. that only lockdown would make people hunker down). In which case, how to calculate the avoidable economic and social damage? This matters not to find the guilty men, but ensure a higher quality of decision making next time.

The secret red team? A point I haven’t been able to stand up. Francois Balloux, a professor at UCL, has also said he was part of an unofficial ‘borderline red team’ group. This suggests the government was seeking more non-Sage advice than was known at the time. He hasn’t elaborated, but I have heard talk of a shadow team that met weekly. Who convened it and who it reported to, I have no idea (yet). I would say ‘we’ll wait for the inquiry’, but that could take years: journalism can complete the picture far more quickly. If I get any more info, I’ll update this post.