What do we think we are doing?

The short version: Our present way of dealing with COVID-19 is insane1That’s not “insane” as a slur but “insane” as in an action no sane person would take after informed, considered reflection. Let’s not turn this into a personal dig against policy makers, though, who are under immense pressures from all directions. It’s more of a corporate insanity.. Of the only two sane approaches, both require intolerable levels of restrictions over a sustained period. In one approach these gradually scale back and COVID-19 kills a significant fraction of a year’s worth of normal deaths over about 6 months. In the other, the intolerable restrictions remain for an unknown period of more than 6 months and possibly years but COVID-19 kills almost no one. I would love to be wrong about this, but I don’t think I am. The choice before us sucks. Take your pick.

The longer explanation…

Okay let me boil this down to a simple analogy and then point out that the analogy is broken. You’re out for a lovely country walk when you unwittingly encounter a pile of dog residue2I have always loved this phrase. It communicates the essentials whilst drawing a discrete veil over the actual noxious substance in question.. So now it’s covering your shoe. You find a puddle and attempt to soak it off gently. This does not work. More vigorous engagement with the puddle has some effect. Progressively more effort results in progressively greater effect but in your heart you know that the only way water alone is going to completely deal with the problem is if you get the power hose out when you get home and give it all you’ve got.

I think this is a bit like how we’ve been trying to deal with COVID-19. We have a control dial and we’ve been trying various different levels to see what they do. At one end of the scale we have 0% where everyone just goes back to normal non-pandemic behaviour. At the other end of the scale, we have 100% where we take off and nuke the planet from orbit 3It’s the only way to be sure! I so wanted to call this setting DEFCON 1 in homage to the 1983 movie War Games but I wouldn’t want anyone to get confused and actually start firing nukes so we’ll stick with boring old “100%”.. In our shoe besmirching analogy we might imagine that any particular setting on the dial will result in a particular level of COVID-19 infection. After all this is what happens with the cleaning analogy. It is what happens to our road speed when we apply a particular pressure to the accelerator pedal in the car4Or when we “gently squeeze the brakes” as Boris Johnson so evocatively put it at one point in the crisis.. We are used to this kind of response.

But with COVID-19 this is not what happens! With COVID-19 the level of restrictions we set doesn’t affect the level of the disease but the value of the now-infamous “R”, the rate at which the infection multiplies. So the analogy is not actually correct.

We’ll come back to the control dial shortly but first we must also notice that in addition to the dial we have also been talking about an objective. We want a clean shoe or a car traveling at a particular speed. Our whole reason for adjusting the dial is to achieve that particular objective. Maybe you don’t give an immense amount of thought to the objective when it’s about cleaning your shoe or driving. But with COVID-19 it is vital that we give very serious thought not only to the short term objective (our road speed) but also our long term objective (our destination and when we want to get there and how much we want the journey to cost).

After most of a year of COVID-19 there are still only three objectives we can envisage:

  1. Build herd immunity through enough people being infected.
  2. Hold out for a safe and effective vaccine that will generate herd immunity.
  3. Flounder about semi-randomly and vaguely hope it will just go away eventually.

On the face of it, option 3 seems like madness but it has proven a favourite option for human beings over the aeons as anyone who has ever been lost5Note to young readers: the word “lost” means you do not know your exact location on the surface of Earth and you don’t know how to get somewhere safe. It’s what would happen to you if you did not have a smart phone. It can be a terrifying experience. Don’t get your elders reminiscing about their experiences of being lost, you’ll be there for ages. in a fog could tell you6As a theological aside, this is arguably what a lot of the history in the Old Testament describes in respect to humanity’s plan for relating to God.. Option 3 is the objective we are presently following.

Option 1 is rightly very unpopular because it means tens or maybe even hundreds of thousands of people will have to die before we are through the crisis. Compared with 530,841 total deaths over 2019 in England and Wales and we’re talking about a signficant fraction of year’s worth of “normal” deaths attributed to COVID-19 alone over, say, a six month period (or however long we make it take for enough people to get infected).

Option 2 seems safest. And I think we vaguely believe we are following option 2. But we are not.

Let’s look at what different control dial settings do and then we’ll explore need to do with the dial to achieve the different objectives…

So far we have experimented with something like the following three settings for the control dial7I’m allocating percentages somewhat arbitrarily. For my purposes, that doesn’t really matter as long as more alarming sounding dial settings mean larger restrictions. I could just as easily have given the dial settings names like “fluffy kitten”, “naughty puppy”, “bitey parrot”, “bitey rottweiler”, “T Rex” and “dalek”.:

  • 0%: early in the crisis when we didn’t yet know we were in a crisis. We found this gave an R value approaching 3. (I think less than 3 actually because the original spike reflected increasing levels of testing as well as increasing levels of the virus so 3 is probably an overestimate. But I’m pretty sure it was more than 2.)
  • 90%: during lockdown. This meant closing schools (with disastrous consequences for children as discovered with year end results for school leavers), furloughing non-key workers (causing a huge shock to our economy8Like a motor car, economies can handle an amazingly wide range of speeds but they respond disastrously to the kind of sudden sharp stop you get from driving into a solid wall.), requiring vulnerable people to be almost completely socially isolated (with serious consequences for mental health which will become a disaster in itself if it continues for too long). This level was intolerable and the lifting of restrictions was greeted with an immense sigh of relief by the populace when it happened in June and July. The 90% setting gave an R value of about 0.4 – so infections reduced over time9It looked a lot more like 0.9 than 0.4 but this is because the numbers we have are very skewed – at the peak in April/May we were diagnosing far fewer of the actual cases than we are now. I think in today’s terms the peak number of cases was probably something like 10 times what we actually recorded.. Because it is exponential decay, things start looking promising after a month but it would take about 6 months to get down to zero cases from where we are now10Note we’ll probably never actually eliminate the disease because it is reasonable to assume that some people will become carriers who never completely shake it..
  • 30%: where we are now. 30% means quarantine if you’ve traveled or have symptoms, face masks in public places, lots of hand sanitising, 2m distancing, quite limited socialising, no singing and working from home if possible. Even this level feels like a fairly significant imposition. But we could keep it up for quite some time. Maybe we could even get used to it as the “new normal”. Setting the dial to 30% gives an R value of around 1.7. If you started with low levels of infection, you can sit at this setting for about two months without anything feeling overly worrying but because the growth is exponential cases will suddenly seem to start exploding at some point. This is exactly where we are at present.

We have not yet found exactly what 60% might look like. But from the numbers above, I think it would be almost as restrictive as 90% and would be the magic setting that would give an R of 1. My hope is that the 60% setting is like 90% with the exception that schools remain open. 60% is a very important number because it is the dial setting where the level of infection does not change over time11Remember my percentages are somewhat arbitrary. I picked 60% for the setting that gives R=1 because it is halfway between the 30% and 90% we know about.. As we’ll see, we need this setting for both options 1 and 2.

So then, what setting is needed to achieve each of the three objectives?

Let’s start with option 3, where we flounder about. This is the only objective that ever requires setting the dial close to DEFCON 1 100%. It starts with a few months of the dial at around 30% followed by sudden alarm at the rising number of cases and a rapid and disorganised escalation to 90%. We hold 90% for several months to get infections levels low again, at which point there is a disorganised de-escalation back to 30% and the cycle begins again.

This is the worst objective. It involves more deaths than option 2 and periods of worse restriction than either option 1 or 2. And it means long term ongoing vast uncertainty. From week to week no one knows what is going to happen next. No one can make any significant plans. Economies cannot cope with this kind of sustained uncertainty. And the continuous cycling back into the 90% setting is terrible for mental health and disastrous for education. No sane person would choose this objective.

So what about objective 2, building herd immunity through a vaccine? This is a fantastic idea. If you have a safe and effective vaccine. Which we don’t. So we’ll have to hope we do get one on day and wait for it. In the meantime, we want to target as few deaths as possible because there is really no side benefit at all to letting anyone die. And we’re all agreed that it is horrible to let anyone die from COVID-19.

So if we’re plucking for option 2, we need to act right now before the death rate rises any more and set the dial to 60% and then keep it there whilst we wait for a safe and effective vaccine. No one knows how much longer this will take or whether we will ever actually develop one. But scientists do seem to be making encouraging noises. So there’s reason to hope they will find one and it will be soonish. Maybe even by the middle of next year.

A continuous 60% lockdown will be very difficult, especially for those who have to shield. Many of those most vulnerable to the virus will have to shield for the rest of theirs lives because COVID-19 is most likely to kill those who only have a year or two left anyway. But at least the economy will be able to recover over time because economies can adjust to almost anything as long as there is stability.

That leaves objective 1 to consider: building herd immunity through infections. The best way to do this is to get it over with as soon as possible with the minimum deaths and without causing the healthcare system or the economy to collapse. We coped in May with around 1000 deaths per day but we were very stretched. This seems like it might be the right upper bound for what we can cope with. So this tells us when we have to turn the dial to 60%: when the death rate reaches 1000 per day. And as immunity builds it also tells us when we can turn the dial down: when the death rate drops below 1000 per day.

It’s awful even talking about this. How can we countenance 1000 deaths a day? We should only consider this option if the alternatives are worse. But it may well be that they are. So let’s continue…

How do we start? Obviously we must isolate the must vulnerable – it is far better to build herd immunity amongst those who are very unlikely to die from the virus. So it is back to shielding for the vulnerable, starting now. But as we will see the good news is that the shielding is short term.

We then wait for the death rate to approach 1000, at which point we set the dial to 60%. This will take a bit of fine tuning so we’d best start adjusting the dial early to be on the safe side. Once R=1 we wait until the death rate starts to drop – and it will as immunity builds up in the community12Increasingly it looks like people develop at least short term immunity to the virus. Longer term we do not know but short term immunity still builds a form of herd immunity that will cause R to drop over time..

How long will it take, how many people will die and how bad is this really? I estimate that we’d have to infect up to 60% of the population before R=1 at a dial setting of 0%13It’s probably less than this because different areas of the country have differing background R values. Because growth is exponential, the areas with highest R tend to dominate when measuring average R. This means only 60% of the population in these areas needs to be infected. Other areas will need less than 60% infection rates and over any given amount of time they will get less because their R value is lower. Sorry that is technical and would take a lot of text to explain in more detail.. In the very worst case scenario we would lose about 1% of the people who get infected. 1% of 60% of 67 million (the whole UK population) is 402,000 people. But we are shielding the most vulnerable, we probably don’t need to infect as many as 60%, millions of people have already had it14A recent study suggested 3.4 million in the UK have antibodies. We know that many people fight the infection with T cells, not antibodies so 3.4 million is a lower bound. If the virus kills one in 20 then the 42,000ish who have died suggest an upper bound of 8.4 million. and the 1% number was before medical advances like dexamethasone. We could realistically hope for a number as low as one tenth of this, i.e. about 40,000. If we’re more pessimistic we might say the factors above only reduce the numbers to a quarter, i.e. 100,000.

That sounds terrible. And, frankly, any individual death is terrible. But over the country as a whole it is quite a small number compared to the half million who already die each year in just England and Wales. And from an individual risk perspective, without any shielding people are about as likely to die from a case of COVID-19 as they would be to die in the next year anyway. With shielding we are all a lot less likely to die from COVID-19 than our existing risk of dying in the next year. This is an individual risk level we are able to live with generally.

And how long will it take? At the present R level it would take around 13 weeks to get to 1000 deaths a day. Then at 1000 deaths a day, it will take something another 100 days for 100,000 people to die. That’s seven months during which we enjoy just over three months of having the dial at 30% and just under four months of having the dial move to 60% and then steadily progress down to 0%.

Conclusion

Where does this leave us? Critically both option 1 and option 2 require us to set the dial to 60%. That means some kind of sustained and significant lockdown. Option 1 involves an unbearable loss of something like 100,000 people. Option 2 involves and unbearable sustained lockdown of 60% for an unknown period, perhaps years, perhaps forever. We face a horrible choice.

But both are far better than option 3, floundering. This option just doesn’t make sense. We should avoid it.

Epilogue

You may wish to know which of these two bad options is the one I think is the lesser of two evils. In my present view, I think option 1 is better. i.e. let the virus itself build herd immunity.

Here are my reasons:

  • First, it is a matter of justice. The people dying are largely those who have less life ahead of them. The victims worst affected by option 2 are the children whose education and future economy are so heavily impacted. If we have to trade off the well-being of older adults with that of children, I think we should privilege the children.
  • First and a half, it is also a matter of justice for those who have to shield. 7 months of shielding is better than having to do so indefinitely.
  • Second, the individual risks are lower than we think. Almost all of us are far more likely to survive COVID-19 than otherwise with option 1.
  • Third, option 2 may well not work anyway. People on the whole have not been good at keeping to the restrictions on a long term basis. The best we may be able to hope for is a short term 60% lockdown.
  • Finally, the damage done in option 1 is not a consequence of anyone’s action whereas it is in option 2. In option 1, the damage is done by a virus that is happening to us. No human choice is in the causal chain that leads to the damage. This is a form of the trolley problem, which is well worth exploring in the present crisis. Good people disagree on their response to the trolley problem so it would make perfect sense if you disagreed with my reasoning here.

   [ + ]

1. That’s not “insane” as a slur but “insane” as in an action no sane person would take after informed, considered reflection. Let’s not turn this into a personal dig against policy makers, though, who are under immense pressures from all directions. It’s more of a corporate insanity.
2. I have always loved this phrase. It communicates the essentials whilst drawing a discrete veil over the actual noxious substance in question.
3. It’s the only way to be sure! I so wanted to call this setting DEFCON 1 in homage to the 1983 movie War Games but I wouldn’t want anyone to get confused and actually start firing nukes so we’ll stick with boring old “100%”.
4. Or when we “gently squeeze the brakes” as Boris Johnson so evocatively put it at one point in the crisis.
5. Note to young readers: the word “lost” means you do not know your exact location on the surface of Earth and you don’t know how to get somewhere safe. It’s what would happen to you if you did not have a smart phone. It can be a terrifying experience. Don’t get your elders reminiscing about their experiences of being lost, you’ll be there for ages.
6. As a theological aside, this is arguably what a lot of the history in the Old Testament describes in respect to humanity’s plan for relating to God.
7. I’m allocating percentages somewhat arbitrarily. For my purposes, that doesn’t really matter as long as more alarming sounding dial settings mean larger restrictions. I could just as easily have given the dial settings names like “fluffy kitten”, “naughty puppy”, “bitey parrot”, “bitey rottweiler”, “T Rex” and “dalek”.
8. Like a motor car, economies can handle an amazingly wide range of speeds but they respond disastrously to the kind of sudden sharp stop you get from driving into a solid wall.
9. It looked a lot more like 0.9 than 0.4 but this is because the numbers we have are very skewed – at the peak in April/May we were diagnosing far fewer of the actual cases than we are now. I think in today’s terms the peak number of cases was probably something like 10 times what we actually recorded.
10. Note we’ll probably never actually eliminate the disease because it is reasonable to assume that some people will become carriers who never completely shake it.
11. Remember my percentages are somewhat arbitrary. I picked 60% for the setting that gives R=1 because it is halfway between the 30% and 90% we know about.
12. Increasingly it looks like people develop at least short term immunity to the virus. Longer term we do not know but short term immunity still builds a form of herd immunity that will cause R to drop over time.
13. It’s probably less than this because different areas of the country have differing background R values. Because growth is exponential, the areas with highest R tend to dominate when measuring average R. This means only 60% of the population in these areas needs to be infected. Other areas will need less than 60% infection rates and over any given amount of time they will get less because their R value is lower. Sorry that is technical and would take a lot of text to explain in more detail.
14. A recent study suggested 3.4 million in the UK have antibodies. We know that many people fight the infection with T cells, not antibodies so 3.4 million is a lower bound. If the virus kills one in 20 then the 42,000ish who have died suggest an upper bound of 8.4 million.

Leave a Reply

Your email address will not be published. Required fields are marked *