Keir Starmer is wrong. Despite what he said in PMQS on Thursday. Tracking and tracing of COVID-19 was not critical over the last 10 weeks. It made perfect mathematical sense to stop testing during the middle phase of the epidemic and here’s why.
“Tracing is critical” Sir Keir said. No, Sir Keir, it’s not critical under all circumstances. The fact is, in peak pandemic, tracing would make almost no difference.
The government is coming under criticism for tracking and tracing cases early on, and then abandoning this, only to be looking at restarting again in the near future. This isn’t a policy failure, it’s smart, and it’s entirely in-line with the mathematics. It is probably also still somewhat too early to return to that approach, although there will come a time when it is highly appropriate.
Why? The answer lies in the immutable mathematics of the pandemic.
We need to recognise two key facts:
- That a proper “trace and test” system will be small scale by its nature.
- It’s crucial to deploy that system at points in a pandemic where it actually makes a difference.
Let’s look at those two points:
A Test and trace system will be small scale by it’s nature
The reality is that close contact tracing is going to be enormously labour intensive. It’s not simply about doing a vast number of tests, it’s about doing those tests on the right people. It’s about identifying everyone that an infected person has been in contact with in the previous 14 days and testing all of them. Doing 100 targeted tests in that way is clearly a vastly more difficult and labour intensive undertaking than doing 1000 tests on people at random.
Therefore, the reality is, that no system of that nature is going to be able to really test vast numbers of the right people. Sure, you can tests lots of people, but testing lots of the wrong people, or even the right people at the wrong time is pointless.
The timing is crucial. It’s not good enough to simply contact everyone who has been in contact with an infected person in the last 14 days and test them all at once as soon as possible after the diagnosis of the initial person.
You will get a whole bunch of negative tests this way. This is because the window of time during which those people will have a positive test will be different depending whether they had contact with the infected person just yesterday, or 14 days ago, or on multiple times through the previous 14 days. And they need testing at times which reflect the incubation time.
For example, there’s no point testing someone who only came into contact with the infected person yesterday. They won’t be positive yet. You need to wait and test them in a few days time. And probably several times over the next 14 days to spot if they become positive.
You also have to keep in mind that if you test someone who came into contact with the infected person 14 days ago, they might well test negative, but they might have already gone from a negative test, to a positive test, infected a bunch of other people, and gone back to negative, all in the last 14 days. Thus, for that person, even if you test them and they are negative, you probably still need to trace and test all of their contacts. Whereas, for the person who only had contact with the infected person just this morning, you don’t need to – but you do need to tell them to self isolate for the next 14 days.
It should be apparent that doing this properly is incredibly detailed work, which requires highly trained public health workers who understand exactly what they are doing, communicate well with each individual involved to extract the relevant information, and then have superlative detective skills to round up all these contacts.
Yes, electronic apps on phones will help, but they are an only a tool to help assist this work if it’s being done properly.
And there’s very little point at all in doing this work with 90% effectiveness. If 10% of cases are slipping through your trace and test system, you will achieve very little at all, due to the exponential growth potential of the virus.
This is why pointing at other countries and saying “look they’ve done lots more tests than us” means virtually nothing without really understanding what those tests are intended to achieve.
You should also consider this: There is a world of difference running this kind of thing while a country is lock-down compared to when it is functioning normally.
While all the pubs, bars and cafes are shut, yes, you might get away with a trace and test system which is sub-par and still find it is workable. However, when you suddenly find you have to trace and test everyone who went to a busy pub on a Friday night a week previously – the level of complexity rises rapidly.
The conclusion is this: Real trace and test systems are, by their nature, going to be small scale affairs if you are seriously looking at trying to identify and isolate every last case. You won’t be able to apply when there are large numbers of cases circulating.
It’s crucial to deploy trace and test at points in the pandemic where it makes a difference
Consider this graph:
As discussed above, our test and trace capacity allows us to eliminate a small number of highly selected cases. This is represented by the red line.
The first point is an obvious one to make, which is that, as a proportion of the total number of cases, that trace and test capacity obviously becomes a near meaningless drop in the ocean if it is deployed near the pandemic peak (notice the blue line demonstrating the cases in excess to capacity of the trace and test capacity).
On the other hand, that red bar is a much bigger proportion (virtually all) of the total number of cases when deployed very early on.
However the real point of this isn’t the effect on the total number of cases (the Y axis of the graph), it’s the effect on the timing of the graph (the X axis).
The difference in impact you can make on the X axis is very clear. Look at the regions labeled “1” and “2” . Deploying your trace and test tactic early on can significantly delay the time to peak, where as,when you move into the exponential growth phase of the graph, it makes very little difference to this at all. (Green line 1 is much longer than green line 2).
In fact, this visual demonstration vastly under-estimates this difference, because you can’t easily fit the massive exponential growth phase accurately onto a graph, but it gives you the visual idea, that the first green line is clearly longer than the second. You just have to bear in mind that the graph under-estimates that difference.
This is why the British Government chose to undertake trace and test early on, and abandoned it as the pandemic progressed.
It made perfect sense, in February and March, to try and push the peak of the pandemic back into the summer, and hence, using that period of time denoted by that green line labelled “1” to delay things was sensible, but there was no sense in expending time, money and effort on trace and test policies in the midst of the epidemic peak. It simply makes no beneficial difference.
Why are the Government looking at re-introducing “trace and test” now?
Why then, are the government now looking at introducing trace and test going forward?
To understand this we need to understand something about the nature of entropy. Entropy has a specific meaning within physics, but is also a general principle of the world around us: Things tend towards states of stability. It is very difficult to maintain any system in a highly unstable state. It will always want to collapse into its most stable configuration. Consider the diagram below
You can probably spot instantly that the ball on the left is in an unstable configuration, while the ball on the right is in a stable configuration. A slight nudge on the ball on the left will cause it to roll down the hill, and it will continue to accelerate from an initial push, while that on the right, if given a slight nudge it will tend to slow, and eventually come back to its current position.
What is true of the ball is true, in an analogous sense, of viral epidemics. Essentially all viral epidemics are about the virus reaching a stable state within its host population.
If you have a population which is mostly not immune to the virus, then it is very much like the ball on the left – unstable. The smallest “nudge” by increasing the of number of infections will cascade into exponential growth. An initial impetus of infection leads to ever increasing acceleration of viral growth.
However, in a population which is immune, you can give much bigger “nudges” and not get the same effect. More like nudging the ball on the right. The virus may infect a few people, spread a bit, but as it is constantly bumping into people who are already immune, the cascade effect of exponential growth is no longer possible. It will tend to self-limit to smaller, regional outbreaks.
The fact is, as the UK comes down off the slope side of its epidemic peak, we are now in a much more stable configuration with respect to COVID-19 than we were beforehand. How stable, depends a lot on exactly how many people have already had COVID-19 and how much herd immunity there is. and how long that herd immunity lasts. (Which is admittedly something of an unknown.) In this sense, the more infections there have already been, the better. According to the University of Manchester over 25% of the UK population may well have already had COVID-19.
Now, at this point, it again, makes sense to start tracking and tracing COVID-19. Under these circumstances, we aren’t anticipating another huge peak, so limiting small numbers of cases could make an outbreak of what might have been 100 cases into an outbreak of just 10. That’s clearly worth doing because we are back to the scenario of being able to make a difference to the number of individual lives that can be saved. If you can do tracing and testing at a a rate which can “keep up” with the epidemic then that’s worth doing. And if you are a a situation of stability, then your tracing and testing can keep up with the outbreak.
Why are the Government not rushing back to trace and test?
While government are again looking at introducing an intensive trace and test system, they don’t appear to be rushing to do so. This was what Sir Keir Starmer was criticizing in PMQs.
The reason for this is because there is a fair degree of uncertainty about how much herd immunity we now have in the UK population.
We have a fairly unusual “hybrid” situation with COVID-19. Some countries slammed the door shut on COVID-19 before it ever really took off. They have populations who are massively susceptible to COVID-19 going forward. They are still very much in the situation like the ball on the left hand side diagram. Some countries like Sweden have not really gone the full lock down route. They are much nearer the ball in the right hand side diagram, metaphorically speaking.
The UK picture is a mixture of the two. If you imagine New Zealand as being like the Ball on the left, and Sweden as being the ball on the right, the UK is somewhere in between. The exact curvature of the landscape on which our “ball” is rested is not really clear.
This is because it’s very difficult to tell to what degree our numbers of current cases have dropped off because of the natural peak of the first wave and herd immunity, and to what degree this is due to the lock-down suppressing infections reaching their natural peak.
If it is primarily the latter, then our state with regards COVID-19 is still unstable. Our “ball” is still sat on an unstable and bumpy landscape. We are prone to another large outbreak. Now, if that’s true, it actually may well not be smart to suppress that at this point, because all this will do is delay that second peak (remember back to the green lines). If we suppress the second peak towards autumn and winter, the impact of it will be far greater than allowing it to come through the summer.
Unfortunately a really good test for immunity is still not widely available. There is a reason why Prof Chris Whitty has maintained since very early on that we need immunity testing, and why he has always seemed rather ambivalent about testing for active infections. We are seeing that reason come to the fore now.
As we don’t have that test widely available, the way we will tell the difference is by closely monitoring the number of cases that we see as lock down lifts.
Effectively lifting lock-down gradually is like giving the “ball” a series of little “nudges” and seeing what happens, to try and “feel out” what kind of landscape the ball is sitting on. Is it on a hill, or in a valley? We will watch how the ball moves, that is, how the case numbers respond, to try and assess this.
And here, things are rather counter intuitive:
If case numbers stay low then we can assume it is because there is widespread herd immunity: Which means it is sensible to go ahead and establish trace and test procedures.
If numbers seem to pick up very rapidly as lock down lifts, we can assume there is still a marked “second wave” to come. If that is the case, then, unfortunately, there is a strong argument for letting that wave move through the country over the course of the summer months. If we use test and trace and heavy lock-down measures to restrain it, we run a severe risk of simply pushing it back to the winter of 2020, which could be far worse.
However, this presents the government with the awkward position of seeing numbers rise, and yet appearing to “do nothing”. Which may prove politically untenable, all the more so, if Sir Keir is breathing down Boris’s neck about testing.
This is the real danger in the current climate: I believe Sir Keir has good intentions, but pressing the government to make moves which seem like a good idea when in fact they are not scientifically sound, could ultimately cost lives. Sir Keir is astute enough to realise this, and this is why he has, for the most part, been very careful in his approach during this crisis, and not gone after the government like a rabid dog over every policy, as many on his side have wanted him to do. He has, for the most part, shown commendable restraint in his approach to this crisis. He ought to make sure he is well apprised of the dilemmas involved in trace and test.
The two views on COVID-19
It’s worth closing this article by just reminding readers of one of the great answered questions about COVID-19.
Most theorising about what is the best way to deal with COVID-19 hangs on a central difference of opinion on the nature of the problem we have before us which emerged early on in this epidemic:
Virologists in many other parts of the world, looked at the virology of COVID-19 and said “It’s a coronavirus, SARS and MERS were coronaviruses, it will behave like SARS and MERS.”
Epidemiologists in the UK looked at the behaviour of COVID-19 and said “It’s an airborne droplet spread illness, often mild or asymptomatic (unlike SARS and MERS), but occasionally serious. It’s therefore more like a flu. It’s epidemiology will follow that of flu”.
Both sides have got at least part of this wrong:
Those in the first group probably thought COVID-19 would be limited by restrictive measures. They were basically wrong. SARS and MERS topped out at a few thousand cases and died out. At five million confirmed infections world wide and counting, COVID-19 has not proved to be containable in the same manner.
The UK approach to think of it as flu led to some some errors of judgement about who was at risk. The initial assumptions was that it would be those who are most at risk of flu (mostly those with respiratory diseases). This turned out to be at least partially wrong. It turns out that things which are risk factors for heart disease are probably more relevant for COVID-19 risk: Old age, Obesity, diabetes, ethnicity, male sex. (Which is not to say there is no increased risk from having pre-existing respiratory diseases)
However, the big question remains about who is right about the long term future of COVID-19.
If the approach of many other parts of the world is right, then, at some point, COVID-19 will still die out like SARS and MERS.
If the UK view is right, then we are looking at COVID-19 eventually becoming a disease like measles, chicken pox or flu, where you got occasional pockets of disease break out, and it never really goes away.
An awful lot of what seems confusing about government strategy is to do with hedging our bets between those two outcomes, because no one really knows, although the UK experts seem to lean towards the second view.
So if it all seems confusing at times, it’s because it is. There cannot be “clear advice” on a situation which is far from clear. However, it doesn’t follow that “if the situation is far from clear then we should play safe and stay in lock down and trace and test everybody”, because if the UK epidemiologist’s view is right, then that will actually make things worse in the long run.
That, is the true answer to Sir Keir Starmer’s question at PMQs. The problem is, there’ simply no way to reduce it to a three part soundbite slogan for the Prime Minister to quote from a podium – and thus Sir Keir, and the wider population, will likely go on being confused.