England has finally rolled out its Test and Trace app, which we are all being exhorted to load on our smartphones. It’s good to see that the new app takes privacy seriously, using Google and Apple’s framework. But nobody seems to have worked through the implications of what the Test and Trace app can lead to, which is Lockdown by stealth.
At the start of the pandemic, countries around the world have looked at a phone app as a way to help contain the spread of Covid, essentially as an extra tool to support contact tracing. Most have primarily found it useful for signing into locations to help tracers jog people’s memory about where they have been. Others have added in the ability to use Bluetooth to add the details of who you have been close to for a significant amount of time. In England, our Secretary of State for health – Matt Hancock, who loves all things digital, latched onto the Bluetooth feature as one of the primary tools for containing Covid.
The first release of an app back in April didn’t go well, not least because the developers didn’t understand much about Bluetooth. Since then, Apple and Google have released a more comprehensive toolkit which makes it easier to estimate proximity. After initially rejecting this, the UK Government has done a technology U-turn and based its new app on this technology. That means it works on more phones, as well as keeping users safe from the eyes of central Government.
With a rise in the number of reported infections (which I’ll come to in a minute), the Government is urging everyone to download and use the app. However, that has some interesting implications, which don’t seem to have been considered. Whilst it may help Test and Trace, if everyone uses sit and complies, it could also inadvertently lead to an unintended lockdown.
To see why, we need to look at what is driving current political decisions. Basically, it is the data that positive tests for Covid are rising, as the ONS graph demonstrates:
As I write, the number of positive test has just passed the peak of mid-April. But we need to put this into perspective. We are performing a lot more tests now compared with mid-April. That’s excellent, but we need to take that into account when we look at the rise in positive test results. The chart below shows the growth of daily tests in blue. In early April we were performing around 15,000 tests per day; now it’s over 200,000. (In the following graphs I’ve used an eight day rolling average, as reporting of the results has been uneven.)
That rise may look scary, but you have to look at the bigger picture. In the figure above, the orange line charts what percentage of the total Covid tests returned a positive result. In April, it was around a quarter of all tests. Testing capacity was limited, so tests were largely done on patients in care homes and hospitals who had fairly severe symptoms, in order to confirm they had Covid. Since June, when tests have been available to anyone who thinks they have symptoms, that figure has fallen to just a few percent, as we’ve moved from confirming that someone has Covid to looking more at its incidence in the community. That change is very obvious when you look at the occupancy of Mechanical Ventilator beds in England:
At the peak of the first wave, almost 4,000 ventilator beds were pressed into use, and almost 70% of them (the orange line) were being used for treating Covid patients. In the current week, around 2,500 are being used to treat patients, but only around 60 of those beds are being used to treat Covid patients – a number which has remained relatively stable since the beginning of August.
Recapping – in April we were mostly testing the critically ill, who were already in, or heading for intensive care. Today we’re testing anyone who asks for a test. Because of the raised level of anxiety cause by local lockdown, holiday quarantines and media scaremongering, it’s a better reflection of the incidence of infection in an essentially healthy society. If we really want to compare the numbers with April, we would probably only be testing the sixty people in ventilator beds and the staff around them, which would give a figure in the low hundreds.
I don’t want to belittle the risk, but policy does not appear to put the data into context, as that shows a fairly stable position. Instead, it’s being driven by the fear or the first ONS graph showing a rise in positive tests.
It is important to base decisions on risk and not fear, but since the start of the pandemic, almost all of the messaging the Government has put out has been stoking up anxiety. The basic fact is that if you’re healthy and under 65 the risk is very low. I’m 65, and caught Covid back in March. The loss of taste and smell, which lasted a fortnight, was disturbing, but other than that it didn’t really affect my daily life. I’ve had far worse experiences with flu and even heavy colds. But when I mention that I’ve had Covid to anyone, the reaction is generally “you’re lucky – you’ve survived”. The public perception is that Covid is dangerous. The reality is that for most people, it’s not. We know that the risk for healthy people under 65 is low, but not exactly how low, as we have no idea how many people have had Covid, because we’re only testing for positive results.
That bring us back to looking at what a positive test result means.
The first thing we see from the second graph is that only a small percentage of people being tested are coming out positive. That number will be an overestimate of who is actually infected in that sample, as it includes three different groups:
- People who have Covid
- People who may have had it in the past, but are still shedding viral RNA, and
- False positives.
All of these people with a positive result will be told to self-isolate for 14 days, which is the strictest form of lockdown, as they’re not allowed to leave home.
Currently, we’re testing around 200,000 people every day, which is generating 5,000 positive results, who will be told to self-isolate for 14 days. On any one day, that means around 70,000 self-isolating. Everyone who signed up for a test should already be self-isolating, and if it takes another two days to get the confirmation that they’re OK, that adds another 600,000 people self-isolating every day who are just waiting for a negative result, giving a total of 670,000 people confined to home.
If we increase testing capacity to a million tests per day, that bumps the number of people in self-isolation to around 3.3 million people on any one day, of whom, very few will actually have it.
The myth of false positives
If we look at the chart of total tests against positive results, we can see that only around 2% of people are testing positive, so out of every 100,000 people tested, around 98,000 are Covid free. They may have had it in the past, but we don’t know. These are people who will be self-isolating from the point they feel symptoms, or, more likely, have no symptoms, but are told that they’ve have been in contact with someone who has. From the point that they are told to stay at home, manage to get a test and then wait for the result currently takes an average of three days, during which they are locked down.
We’ve been hearing a lot about the burden of false positive test results recently, with claims that they are causing thousands of people to go into self-isolation. According to a recent article in the Lancet, “the current rate of operational false-positive swab tests in the UK is unknown; preliminary estimates show it could be somewhere between 0.8% and 4.0%”. Matt Hancock has been taking the lower figure, with the media latched onto this to predict massive numbers of people being unnecessarily locked down as a result.
It’s not difficult to work out an approximate value, and I don’t understand why people haven’t done this. The FPR should remain pretty static over time, as it’s nothing to do with Covid, it’s a facet of the testing process. If you calculate it as a percent of the overall tests each day, it can never be bigger than the total number of positive tests. That gives us an upper limit. If we subtract that number from the total positive results for a day, the result will be the actual number of valid Covid infections. The graph below shows that calculation for a range of values of FPR from 0.8%, reducing down to 0.2%.
The red horizontal axis marks the point where the false positives would account for every positive test. Any line which falls below that has to be too high a value of FPR. The light blue line, which represents an FPR of 0.4% is therefore the highest value of FPR that is possible, but that would imply virtually no cases in July, i.e. less that 0.1% of those tested having Covid. It’s more likely that the real FPR is around 0.2%. In which case, only one tenth of each day’s positive results would be false. That’s low enough to ignore, which means we can concentrate on the real issue – which is keeping people at home during testing delays.
As you can see, it’s not difficult to work out these numbers – all you need is some primary school arithmetic and an understanding of Excel. But that seems to be something our leaders lack. I hate to compare anyone’s level of intelligence to that of Donald Trump, but the statements being made by Matt Hancock and Dido Harding fall into exactly that category. Baroness Harding was brought in by Matt Hancock to run the Test and Trace program after its first round of failures. She was dubbed the “Queen of Carnage” during her time as CEO at TalkTalk, when four million customers had their data stolen. When asked by an enquiry whether that data had been encrypted, she replied “The awful truth is, I don’t know”.
Last week, when she was asked by the House of Commons Science and Technology why testing was not keeping up with demand, she replied that she “doesn’t believe anybody was expecting to see the really sizeable increase in demand” for coronavirus tests. To paraphrase the words of a former fictional lady peer “To be uninformed once, Mrs Harding, may be regarded as a misfortune; to be uninformed twice, looks like carelessness”. If you stoke up anxiety, as this Government has done, then tell everyone that they can have a free test, it does not take a genius to predict that demand will rocket. It appears that Matt appointed Dido because of shared interests – they’re both keen equestrians. He might have done better to get off his high horse and appoint a couple of racecourse bookies, as they would almost certainly have a far better understanding of numbers and how to assess risk.
Why does this all matter? It’s because the current policy could lead to an unintentional and damaging lockdown. We’re hearing politicians talking about “eliminating” Covid, as if there is some magic way to make it go away. They seem to think we can all huddle away until a vaccine arrives, which, if you believe the PR, is sometime soon. As epidemiologist Mark Woolhouse told New Scientist in early April: “I do not think waiting for a vaccine should be dignified with the word ‘strategy’. It’s not a strategy, it’s a hope. Current Government thinking appears to be that the UK can put up the “No entry to Covid” signs and get back to normal. The problem with siege strategies like this is that when you eventually come out, the red Indians, wolves, zombies, etc. are still waiting outside. At which point it’s Groundhog Day for the pandemic.
Which leads us back to the NHS Test and Trace app and how it may hinder, rather than help. It works in two ways. It uses Bluetooth to scan for other phones running the app and then logs how many of those you remain in close proximity to and for how long. Depending on how many interactions you have over the course of a day and how long they last, an algorithm assigns a risk to each contact. If a user tests positive, they can share this information with a central server run by the Department for Health and Social Care, which will inform all of the users who fall into that high risk category to self-isolate.
Users can also scan a QR code at shops and restaurants, which will be recorded on the phone. It appears that these are not cross-checked against known locations which have been visited by someone with Covid, but are used purely as an aide-memoire to help contact tracers.
We don’t know how many high risk contacts a user will generate each day. If you use your phone at work or use mass transit, that number could be high. That will bump up the number of people being told to self-isolate and be tested, which further increases the demand for tests. If the QR scans also allow more people to be traced, we see a further increase.
Although the Government is claiming that test capacity will hit 500,000 tests a day by the end of October, industry reaction is that it will be late. Given that we are also told that demand is currently four times capacity, it is unlikely that the test capacity will catch up with demand this year. The effect of this is critical, as any overload delays the time to get a negative result back, which is what people need in order to get back to work.
As we saw above, with three-day self-isolation, at the current level of testing, around 680,000 will be self-isolating at any one time. If test results are delayed for six or seven days, that number doubles. Moving to the target of testing 1,000,000 people per day, that would lead to around 7 million people self-isolating at any one time, almost all of whom are not infected.
As if that’s not bad enough, all of these can then be asked to go through the process again. It’s one of the little ironies of the system, that the Test and Trace app keeps on working regardless. As soon as you’re cleared to go back to work, it will start looking out for new infection opportunities. The more you go out, the more likely you’ll be told to isolate again. By the third time they’re told to stay at home, most people they will probably take the Dominic Cummings or Margaret Ferrier approach and disregard it, at which point the whole edifice crumbles. It is difficult to believe the amount of damage that these two people have done to the country by their self-centred actions.
What is missing in the current policy is any attempt to find out who has had Covid and who can safely catch it. That last bit may sound contentious, but we should be looking seriously at how to live with Covid. As an example, rather than confining students to their rooms, it might be worth encouraging them to do the things that students normally do to get infected. It’s unlikely that’s any more dangerous than the normal consequences of fresher’s week and would mean that once they have all had it, it’s safe for them to go home at Christmas. The alternative, of telling them to isolate before going home to parents and grandparents, sounds suspiciously like the disastrous policy of freeing up hospital beds in April by sending Covid patients back into care homes.
Instead of knee-jerk reactions based on data they don’t understand, we need policy makers to looking at the real problem, which means developing a strategy that will allow those most at risk to resume their normal lives Around 2.2 million people are currently shielding in the UK because of underlying, severe health problems, with another 8.8 million over 70 who are advised to stay at home. Nothing in the current strategy does anything other than condemn them to stay in their homes until a successful vaccine arrives, which may not be in their lifetime. In other words, we are writing off one in every six people in this country because the current policy is stoking up fear.
Covid is unusual, as it is quite specific regarding who suffers major symptoms. With over nine months experience of its effect, we now have a pretty good idea of who is at risk and who has minimal risk. Yet our strategies don’t acknowledge that – our leaders fail to understand the meaning of data and put their heads in the sand as apps and anxiety drive us inexorably back to another lockdown, even if it’s not called that. It’s time to understand the data and the risks and look at how we can live with Covid. Otherwise everyone’s life is going to be blighted.
Read more of my articles on Covid here.