COVID-19: Thoughts on the Endgame

This note follows up on the note ‘General Orientation’ of March 26th.


Length of lockdown under Reaction 3: Late Lockdown

Most of Europe has adopted a policy of late lockdown. The first country to do so was Italy on March 9th and the last was the UK on March 23rd.

The purpose of lockdown is to ‘flatten the curve’ of infection, so that ICU capacity can match the number of cases requiring it. It does two things: stops the disease from spreading so that the number of new cases grows more slowly and then goes down; and buys time to add ICU capacity.

This is being propagated in the UK with the slogan ‘stay at home to protect the NHS’, which is an odd way of putting it, as if the prime threat is to an institution.

In fact, of course, the threat is to a group of people. The reason lockdown is being imposed on the many for whom COVID-19 is nasty but harmless, is to protect the few for whom it is deadly.

The virus is, in the words of the Swiss public health expert Jean-Dominique Michel, ‘benign in the absence of pre-existing pathology’i. Data from Italy shows 99% of victims to have been suffering from one or more conditions such as hypertension, diabetes, cardiovascular disease or cancer. Almost half of them had three or more of these conditions. Deaths correlate with age – the median age was 80 – but it looks as if this is only because older people are more likely to suffer from such conditions and have weaker immune systems. Anyone with chronic health problems of this kind, regardless of age, is vulnerable.


What will it take to be effective in stopping the spread?

Current data suggests that the incubation period is anything from 2-14 days, with an average of 5.97% of those who present with symptoms do so after 11.5 days.

During incubation they can infect others.

Once carriers present, data from China suggests that they can be infectious for between 8 and 37 days, with an average of 20 (the ‘viral carriage period’).

To eliminate the spread, lockdown would therefore have to be enforced for the incubation period of 14 days plus the viral carriage period of up to 37 days which is 51 days or just over 7 weeks.


Flattening the curve 

Introduced because of the late recognition that we are in a hole, lockdown means that we stop digging. That allows us to find out how big the hole is. If the hole is small, flattening the curve will spread acute cases over a longer period of time and so match ICU capacity. If the hole is big, it will be impossible to grow ICU capacity fast enough to match demand.

Assuming lockdown is 100% effective in stopping any further spread, the size of the hole is measured by the number of cases at lockdown. This is a function of the amount of time between the virus entering the population and lockdown.

Time is absolutely critical in this case because the spread is exponential. We are not used to dealing with exponential growth. We are used to linear growth in which change is gradual and measured in percentage increases. Exponential growth is more easily measured in doublings and what that means in terms of change is not just mathematical but something we are actually experiencing. Every day matters.

The first official case was recognized in China on 31st December 2019, meaning that the virus began its activity about 2 weeks before that. China imposed lockdown on 23rd January 2020. That means the period of first case to lockdown was 3 weeks and incubation to lockdown was 5 weeks.

The first case in the UK was recognized on 31st January, so it entered the population in mid-January. The UK imposed lockdown on 23rd March 2020. That means the period of first case to lockdown was 7 weeks and incubation to lockdown was 9 weeks.

In an exponential context, this 4 week lag in reaction time between China and the UK makes a massive difference. In those four weeks the UK has been digging itself into a huge hole, the size of which will now start to be revealed. In Italy the lag versus China was 2 weeks, and It currently has 80,000 identified cases and 8,000 deaths.

For practical purposes the precise size of the hole is not in itself critical. What matters is the size of the gap.

At the beginning of the year, the UK had about 3,400 ICU’s for adults, about 80% of which were occupied, giving spare capacity of 680. Roughly 5% of cases need intensive care. As of today, 27th March, the UK has 11,700 cases, meaning the spare capacity is almost used up.

Italy is about 2 weeks ahead of us, but almost certainly has a smaller number of cases because it introduced lockdown 5 weeks after the first case, not 7. If the UK tracks Italy, in two weeks it will need not 680 but 4,000 extra ICU’s. Because of the extra 2 week lag, UK demand will almost certainly far exceed Italy’s, where new cases seem to have peaked.

The critical variable determining whether flattening the curve will work is not the curve itself – which we can no longer control and which we know will not be flat at all but very steep – but the rate of growth in ventilator and ICU capacity.

The UK has about 8,200 ventilators. On 23th March the government placed orders for another 30,000, which suggests they are under no illusions about the potential size of the hole. Ventilators are not complicated pieces of equipment. The question is how quickly they can be made available to front-line staff and whether there are enough trained staff to use them.

Italy started with 12.5 critical care beds per 100,000 people. The UK figure is 6.6. China started with only 3.6 per 100,000, but built new ones at a speed which few if any countries can emulate. So far, no countries seem to have asked them to do the same for them. However, the example is being emulated. A new 4,000-bed hospital is being built in London’s Excel Centre.


Effects of the capacity gap

The only way of closing the capacity gap now is not to flatten the curve (because we can no longer influence that) but to grow capacity at an exponential rate: at least a couple of doublings in a couple of weeks. If that does not happen, the capacity gap will remain and that implies that some people who need treatment cannot be given it and will die. The questions then are ‘who dies?’ and ‘who decides?’.

In war, military hospitals are overwhelmed when there is a big battle. They have casualty clearing stations which use a form of triage to sort out who gets treated and who doesn’t. Usually someone does a quick assessment of the wounded, treats the lightly wounded on the spot, passes those more seriously wounded with a reasonable chance of survival on to get treatment at a hospital, and gives the most seriously wounded some pain killers and leaves them to die.

In a civil situation access to healthcare is determined by the social system. In the US it is determined by wealth, in India by caste, in most African countries by power. In Europe there is universal healthcare, so there is no default method of triage except queuing. Those who wait for too long die.

This is already the case in the NHS, where people with critical illnesses die when waiting for transplant organs or the availability of equipment, and it will probably be the case now. One alternative is for civilian doctors to make the decision in the way that military ones at casualty clearing stations do. In Italy doctors are already making live or die decisions based on age and state of health. The other is for the government to issue guidelines.

The latter is unlikely because it introduces agency, giving some people, rather than a system, power over the life and death of others. That would be morally and politically controversial, perhaps even unacceptable, and raise the question of how decisions would be made. A utilitarian might use life-years saved, or a formula such as (life-quality x life years of the population), but no such measure exists.


Getting out of the hole

A strategy involves achieving a determinate goal with limited resources against opposition in an uncertain environment.

In this case all four conditions apply, with the twist that the opposition is simply an RNA sequence programmed to reproduce itself. However, it does behave in a quasi-strategic way, deploying disguise, deception, maintenance of the aim, agility and resilience. Its big initial advantages are in its own speed of deployment and the total lack of intelligence on the part of its opponent – us. We are closing the intelligence gap and developing offensive weapons in the form of a vaccine, but at the moment it still has the advantage and has used speed to build a very strong position and put us firmly on the defensive. It still has the initiative.

We have deployed four counter-strategies in response:

  1. Immediate test and trace: S. E. Asia
  2. Rapid lockdown: China
  3. Late lockdown: Europe
  4. Herd immunity

Number 1 is the most successful. There are few deaths and almost no social and economic disruption. It requires preparation and dedicated resources lacking in most of the world and is no longer an option for others.

Number 2 also works but at higher cost on both scores. The fast reaction reduces both deaths and the economic cost that would arise with delay.

Number 4 is risky, but seems to be working in Sweden, the only country really putting it into practice. Economic activity is muted rather than at a standstill, and casualties are limited because the few to whom the virus is a threat are well protected. As of 26th March, Sweden had 2,800 identified cases and has suffered 66 deaths. There may be specific local reasons why it is a feasible option there. Whether it works in the long run depends on two assumptions being true: that infection confers immunity and that the virus does not mutate into something more virulent.

Both could be false: getting a cold (some versions of which are caused by other coronavirus variants) does not prevent you from getting another one, and the virus already has many different strains because mutation is very fast. In 1918, one mutation proved to be deadly. It was spread because severely ill soldiers were taken out of the trenches and moved around, whilst those with mild symptoms stayed where they were, effectively isolating them from others. Today, the opposite is happening – the severely ill are being isolated, and if there is a second wave it is probably going to be a mild variant. But nothing is certain.

Other European countries have adopted number 3 more by default than design. It is by far the weakest of the four.

Because the delayed reaction creates so many cases, the potential capacity gap means that all effort has to placed on closing that, and lockdown is a necessary pre-condition of doing so. Whereas strategy 1 has the single goal of eliminating the virus and so minimizes casualties and avoids disruption, strategy 3 opens up the likelihood of heavy casualties and huge economic cost at the same time and eventually forces a choice between them. This is the worst of all possible worlds, except for total denial (i.e. no strategy at all).

It is now too late to flatten the curve, so the only way of closing the gap is to increase treatment capacity exponentially. Expect that to form the main effort over the coming week.

Whether that is possible remains to be seen, but if it is, it will be a near-run thing, and it will be hard to avoid some form of triage, whether explicit, or, as is more likely, implicit.

It is also impossible to avoid massive economic damage. This is not just about money. An economic shutdown destroys lives. People lose jobs and security, their futures are put in jeopardy, and small business owners could lose everything. The longer it goes on the more serious the effects will be, and they will change not just in quantity but quality. If supply chains break down, poor families could struggle to feed themselves. As small businesses fold, suicides could increase. As people are incarcerated in their homes, relationships will come under strain. Domestic violence is already on the increase. At the extreme, death lurks in this direction as well.

So the pressure to lift lockdown will grow daily. At some point this strategy will be forced to change to some form of targeted response, which places a bet on herd immunity and introduces massive testing so that future outbreaks can be handled with a version of counter-strategy number 1. Look out for signs of a pivot in this direction. As it stands, it leads nowhere. There is no exit strategy.

The exit will be a forced one, produced by the collision of two conflicting goals: the lives of the threatened few and the wellbeing of the many. China kept up lockdown for 8 weeks, with some relaxation of its stringency along the way. In Europe that might not be enough to reduce infection to zero. The Imperial college paper recommended 20 weeks. Today Scotland’s Chief Medical Officer told the BBC that the modelling that had been done ‘always said that we need at least 13 weeks of some sort of measures in order to really get this virus away from spreading among people’.

But the pressure to lift lockdown will be such that some compromise between the goals will probably be sought. In China, total lockdown was eased after just over 2 weeks to allow what they called ‘orderly production’. They were down to zero new cases after 6.5 weeks.

In estimating economic damage it therefore looks as if 8 weeks could reasonably be set as an upper limit for total lockdown, and 13-14 weeks as an upper limit for less economically damaging restrictions. 3-4 weeks would be an absolute minimum to limit deaths, but unless there is a massive surge in ICU capacity it would soon become apparent that it is not working and if that happens lockdown could be relaxed or abandoned. Large scale testing and a shift to a targeted strategy is the main thing that could shorten the need for restrictions on the general population.

Goal conflict is a classic source of strategic failure because the closer you get to achieving one goal the further away you get from achieving the other. The open question now is only how big that failure will be.


Stephen Bungay, 27th March 2020

Director, Ashridge Strategic Management Centre


i Jean-Dominique Michel, ‘COVID-19: The Game is Over?!’, 24th March 2020

Back to top