Help
Quote
The official death toll in the UK from Covid-19 – more than thirty thousand – is now the highest in Europe. It is the result of choices made at the beginning of the UK’s response to the pandemic. This is not necessarily to apportion blame: at the beginning of March, all choices seemed bad. The trolley problem, as posed by Philippa Foot in 1967, imagines the driver of a runaway tram who ‘can only steer from one narrow track onto another; five men are working on one track and one man on the other.’ This simple utilitarian problem becomes more challenging when you change the factors. Could you justify saving your mother at the expense of two people you’d never met? Intention, duty, relationships, societal consequences: all such things complicate the moral question. The epidemiological variation on the trolley problem would be something like: ‘With the most accurate models we have, what is our best course of action to minimise the statistically expected, age-adjusted, all-cause mortality?’
Sars-CoV-2 doesn’t by itself introduce moral complexity into the argument: it’s mechanically more intriguing than a tram, but just as unthinking. The dilemma, as presented to me by an epidemiologist, was whether to go for measures that might truly suppress the virus – at huge cost to the economy and to healthcare, with the very real risk of an even worse second and possible third wave of the disease later – or to accept that many deaths would occur no matter what you did. Your epidemiological duty is to minimise the expected number of life-years lost, which may mean accepting terrible suffering now to alleviate even worse suffering later. When you look at the numbers for Covid-19, the trolley problem becomes appalling, with hundreds of thousands of lives lost either way. ‘I wish I had a faith,’ she said – not a reassuring thing to hear from an epidemiologist.
The instinctive real-world response to the trolley problem isn’t to decide between one person and five, or between a loved relative and two strangers. It’s to shout and scream, pull on the brake, try to jam the wheels, do whatever you can to stop or slow the tram and get the people off the tracks. That’s what I’ve been doing in my work at the Crick Institute, as a member of a large, multinational team of scientists. We are part of an effort that extends across the world: I have never encountered such openness and generosity among scientists (we’re a competitive bunch).
The most immediately solvable problem we identified was a simple one: lack of testing. In March, during the early phase of the UK’s response, famous actors could get tested but healthcare workers couldn’t. If a healthcare worker can’t be tested, they can’t know if their mild symptoms mean they have the disease, or something trivial. It soon became clear that a substantial proportion of transmission occurs without symptoms: either because the symptoms start one or two days after an infected person becomes infectious, or because they never occur. One of the great mysteries of Covid-19 is that its severity ranges from negligible to lethal. A young, healthy person with mild or zero symptoms may easily transmit the virus to vulnerable others. You need meticulous barriers – personal protective equipment, masks especially – and a rigorous testing regime to ensure this doesn’t happen. We are short of PPE, and in March we had very little testing capacity.
On 31 March, the government blamed a shortage of chemical reagents. In reality it was a failure to prepare. On 19 March – at least a month after it should have been obvious that large-scale testing would be useful – a request was made on the prime minister’s behalf to universities and research institutes, asking them to send a particular brand of proprietary testing equipment (normally used for research) to the National Biosample Centre in Milton Keynes. By this point in the pandemic, every country in the world wanted proprietary reagents that would be compatible with these machines. You couldn’t get your hands on the resources needed. Never mind the chemicals, there was even a shortage of swabs, which are no more than glorified cotton buds. As a result, large testing laboratories have only just come online. The target of 100,000 tests per day has now been met, if you include the number of home testing kits sent out in the post with no guarantee they will be carried out, and if you take ‘per day’ to mean ‘on 30 April’. These are not the tricky antibody tests, but the comparatively simple tests for the virus itself. The chemistry has been well understood for thirty years, even if implementing it can be fiddly.
Unable and unwilling to use proprietary reagents and machines, which would have meant competing for resources with other providers, at the Crick we went for a DIY approach. We teamed up with hospitals and pathology labs, and got excellent input from Public Health Wales. Our downloadable protocols have helped other institutes to set up their own testing facilities, and we have given advice and homemade reagents to university departments and to the new large testing labs. We haven’t been especially clever; we’ve just worked hard.
Writing in the LRB dated 19 March (which went to press two weeks earlier), I assumed that well-organised democracies would follow the model of South Korea, where measures were quickly put in place to roll out testing, tracking and tracing through a moderately socially intrusive smartphone app. The wearing of masks is also widespread there – something I should have highlighted at the time. On 5 May, Angela McLean, the UK government’s deputy chief scientific adviser, endorsed this approach. ‘South Korea is really the place in the world that we can look to and say this worked,’ she said. ‘They are a fine example to us, and we should try to emulate what they’ve achieved.’ Better late than never.
South Korea suppressed its epidemic from a peak of 851 detected cases on 3 March to a long plateau at around a hundred cases a day by mid-March. Now it usually records fewer than ten new cases a day, most of them detected at Seoul’s main international airport. South Korea is reopening its economy, as are other countries that have brought their case numbers down to manageable levels: New Zealand, Australia, China, Taiwan, Vietnam. None of these countries allowed case numbers to get out of control in the way that most of Western Europe and the United States have. It remains to be seen if the South Korean approach will work to suppress a much bigger epidemic.
To suppress the epidemic you have to ensure that, on average, every person who catches the virus passes it to fewer than one other person. The lockdown measures in the UK and elsewhere have achieved this, though in an unsustainable way. When we ran our first samples at the Crick, from healthcare workers showing symptoms at the peak of the wave, we found lots of positive cases, noses and throats full of virus. Now the enemy is scattered, maybe one or two in a plate of 96 samples. If across the whole population we could find those cases, isolate them, and isolate and test all their contacts, we could stop transmission. We’d need to act fast and test thoroughly.
There appears to be a plan. Test as much as possible, and focus the tests on the people most at risk and those who look after them. For every person who catches the virus, symptomatic, pre-symptomatic or asymptomatic, isolate and test all their known contacts. Contact-tracing phone apps can help. You alert the system if you get symptoms. You can then be tested, and your contacts (determined by Bluetooth, and by old-fashioned tracing) can be isolated and tested. As I write, two different apps are under development by the NHS. Whichever ends up being chosen it should work well, provided enough people use it. But it won’t work if people want it to be perfect rather than useful, or refuse to use it because of privacy concerns. Of course it must be reasonably secure, but anyone using, say, Facebook routinely hands over more personal information than the proposed apps will ever require.
If we are determined and organised in the next phase of the pandemic – some aspects of the response so far have been good, but in other respects the first phase has been feeble and shambolic – there is a good chance this will make a difference. But for how long can we keep it up? Eventually we will need immunity to the virus. This can occur either by infection or by vaccination. The ‘strategy’ of immunity by infection – which appeared to be the UK government’s chosen route at one point – was characterised by one of my colleagues, a virologist, as ‘tantamount to surrendering before even one bullet has been fired’. Infection clearly leads to a degree of immunity in most cases, just as it leads to death in a small proportion. The problem is that coronaviruses have evolved many mechanisms to suppress immunity.
Measuring immunity is important, but it isn’t easy. The most obvious way is to look for the presence of antibodies. But antibodies to what? The virus has many components. Its main entry weapon is known as Spike. This is a large, sugar-coated protein complex that can rip a hole in the membrane of a cell to allow the virus to enter. Block Spike, and you keep the virus out. It’s easy enough to measure antibodies to Spike, but not all of them actually prevent the virus from entering cells. To find out whether the antibodies are doing their job effectively, you have to culture the virus in a high-containment facility, titrate tiny amounts of serum extracted from the test subject’s blood into the virus culture, and demonstrate that the serum blocks the virus. It’s painfully slow. We are working on ways to make these assays faster, easier and more accurate. So are many others, and for once I’m happy when another lab does something better. The procedure isn’t going to be useful for testing on a large scale: instead, we’ll have to correlate antibody tests with the neutralisation assays. Some of the newly developed commercial antibody tests will probably correlate well. Others will be useful only for epidemiologists as markers of infection. Until very recently, most of the widely available tests have been so inaccurate as to be useless.
If someone outside a high-end research lab has conducted a test for you purporting to show that you are ‘immune’, I strongly caution against assuming it means anything. Lots of people have had symptoms compatible with coronavirus. In a recent draft of a study from an excellent laboratory in New York, 99.5 per cent of people who had confirmed infection developed antibodies eventually, sometimes several weeks after the test for the virus itself. Only 38 per cent of people with likely symptoms of Covid-19 – but with no positive test – had developed antibodies. Assuming that probable infection means definite immunity is a very big mistake.
There are four ‘seasonal’ coronaviruses – 229E, OC43, NL63 and HKU1 – that cause mild disease in nearly everyone, only occasionally causing pneumonia. They can be given to healthy volunteers to study the immune response. They cause the ‘common cold’, and in experimentally infected humans they give rise to an antibody response. That response wanes after a few months, and the same people can be experimentally reinfected, though they tend to get milder symptoms the second time round. It is thought that adults get reinfected on average about once every five years. Sars-CoV-2 causes mild disease in most cases, and gives rise to antibody responses in nearly all cases. We don’t know how long these responses will last, but it is likely that people who suffer only mild disease will be susceptible to reinfection after a few months or years. Humanity has never developed ‘herd immunity’ to any coronavirus, and it’s unlikely that Sars-CoV-2 infection will be any different. If we did nothing, a likely possibility is that Covid-19 would become a recurring plague. We don’t know yet. It may have seemed like an aeon, but we have been aware of this virus for only a few months.
We need a vaccine. The good news is that the virus is not mutating in such a way as to make immunisation especially difficult. One extremely crude vaccine consists of killed virus. This induces very decent immunity in experimental models. More sophisticated vaccines may well be better, and there are more than ninety in various stages of development. Vaccines that induce neutralising antibodies to the virus are strongly predicted to work: maybe not perfectly, and maybe not without complications, but I don’t know a single immunologist or virologist who thinks a vaccine is impossible.
Some are worried that because so many countries are instituting effective control measures it will be hard to demonstrate the effectiveness of a vaccine. However lackadaisical the UK response may have been, it has at least followed a vaguely scientific path. The situation in the US is different. The wealthiest country in the world, with the best bioscience and the largest testing capacity, is in the grip of the worst of the pandemic. It isn’t difficult to work out why this is. The Trump administration’s failures do mean, however, that it won’t be hard to generate the numbers required to show vaccine efficacy in the US. How quickly a vaccine can be rolled out is another matter. I have a bet with a colleague at the Scripps Institute. He thinks it will be four years. I think it will be 18 months, not least because the Bill and Melinda Gates Foundation and other large donors are backing initiatives to put promising vaccine candidates into production before there is conclusive proof they work. But I’d choose this hopeful pragmatism over the fatalism of the trolley problem.
Sars-CoV-2 doesn’t by itself introduce moral complexity into the argument: it’s mechanically more intriguing than a tram, but just as unthinking. The dilemma, as presented to me by an epidemiologist, was whether to go for measures that might truly suppress the virus – at huge cost to the economy and to healthcare, with the very real risk of an even worse second and possible third wave of the disease later – or to accept that many deaths would occur no matter what you did. Your epidemiological duty is to minimise the expected number of life-years lost, which may mean accepting terrible suffering now to alleviate even worse suffering later. When you look at the numbers for Covid-19, the trolley problem becomes appalling, with hundreds of thousands of lives lost either way. ‘I wish I had a faith,’ she said – not a reassuring thing to hear from an epidemiologist.
The instinctive real-world response to the trolley problem isn’t to decide between one person and five, or between a loved relative and two strangers. It’s to shout and scream, pull on the brake, try to jam the wheels, do whatever you can to stop or slow the tram and get the people off the tracks. That’s what I’ve been doing in my work at the Crick Institute, as a member of a large, multinational team of scientists. We are part of an effort that extends across the world: I have never encountered such openness and generosity among scientists (we’re a competitive bunch).
The most immediately solvable problem we identified was a simple one: lack of testing. In March, during the early phase of the UK’s response, famous actors could get tested but healthcare workers couldn’t. If a healthcare worker can’t be tested, they can’t know if their mild symptoms mean they have the disease, or something trivial. It soon became clear that a substantial proportion of transmission occurs without symptoms: either because the symptoms start one or two days after an infected person becomes infectious, or because they never occur. One of the great mysteries of Covid-19 is that its severity ranges from negligible to lethal. A young, healthy person with mild or zero symptoms may easily transmit the virus to vulnerable others. You need meticulous barriers – personal protective equipment, masks especially – and a rigorous testing regime to ensure this doesn’t happen. We are short of PPE, and in March we had very little testing capacity.
On 31 March, the government blamed a shortage of chemical reagents. In reality it was a failure to prepare. On 19 March – at least a month after it should have been obvious that large-scale testing would be useful – a request was made on the prime minister’s behalf to universities and research institutes, asking them to send a particular brand of proprietary testing equipment (normally used for research) to the National Biosample Centre in Milton Keynes. By this point in the pandemic, every country in the world wanted proprietary reagents that would be compatible with these machines. You couldn’t get your hands on the resources needed. Never mind the chemicals, there was even a shortage of swabs, which are no more than glorified cotton buds. As a result, large testing laboratories have only just come online. The target of 100,000 tests per day has now been met, if you include the number of home testing kits sent out in the post with no guarantee they will be carried out, and if you take ‘per day’ to mean ‘on 30 April’. These are not the tricky antibody tests, but the comparatively simple tests for the virus itself. The chemistry has been well understood for thirty years, even if implementing it can be fiddly.
Unable and unwilling to use proprietary reagents and machines, which would have meant competing for resources with other providers, at the Crick we went for a DIY approach. We teamed up with hospitals and pathology labs, and got excellent input from Public Health Wales. Our downloadable protocols have helped other institutes to set up their own testing facilities, and we have given advice and homemade reagents to university departments and to the new large testing labs. We haven’t been especially clever; we’ve just worked hard.
Writing in the LRB dated 19 March (which went to press two weeks earlier), I assumed that well-organised democracies would follow the model of South Korea, where measures were quickly put in place to roll out testing, tracking and tracing through a moderately socially intrusive smartphone app. The wearing of masks is also widespread there – something I should have highlighted at the time. On 5 May, Angela McLean, the UK government’s deputy chief scientific adviser, endorsed this approach. ‘South Korea is really the place in the world that we can look to and say this worked,’ she said. ‘They are a fine example to us, and we should try to emulate what they’ve achieved.’ Better late than never.
South Korea suppressed its epidemic from a peak of 851 detected cases on 3 March to a long plateau at around a hundred cases a day by mid-March. Now it usually records fewer than ten new cases a day, most of them detected at Seoul’s main international airport. South Korea is reopening its economy, as are other countries that have brought their case numbers down to manageable levels: New Zealand, Australia, China, Taiwan, Vietnam. None of these countries allowed case numbers to get out of control in the way that most of Western Europe and the United States have. It remains to be seen if the South Korean approach will work to suppress a much bigger epidemic.
To suppress the epidemic you have to ensure that, on average, every person who catches the virus passes it to fewer than one other person. The lockdown measures in the UK and elsewhere have achieved this, though in an unsustainable way. When we ran our first samples at the Crick, from healthcare workers showing symptoms at the peak of the wave, we found lots of positive cases, noses and throats full of virus. Now the enemy is scattered, maybe one or two in a plate of 96 samples. If across the whole population we could find those cases, isolate them, and isolate and test all their contacts, we could stop transmission. We’d need to act fast and test thoroughly.
There appears to be a plan. Test as much as possible, and focus the tests on the people most at risk and those who look after them. For every person who catches the virus, symptomatic, pre-symptomatic or asymptomatic, isolate and test all their known contacts. Contact-tracing phone apps can help. You alert the system if you get symptoms. You can then be tested, and your contacts (determined by Bluetooth, and by old-fashioned tracing) can be isolated and tested. As I write, two different apps are under development by the NHS. Whichever ends up being chosen it should work well, provided enough people use it. But it won’t work if people want it to be perfect rather than useful, or refuse to use it because of privacy concerns. Of course it must be reasonably secure, but anyone using, say, Facebook routinely hands over more personal information than the proposed apps will ever require.
If we are determined and organised in the next phase of the pandemic – some aspects of the response so far have been good, but in other respects the first phase has been feeble and shambolic – there is a good chance this will make a difference. But for how long can we keep it up? Eventually we will need immunity to the virus. This can occur either by infection or by vaccination. The ‘strategy’ of immunity by infection – which appeared to be the UK government’s chosen route at one point – was characterised by one of my colleagues, a virologist, as ‘tantamount to surrendering before even one bullet has been fired’. Infection clearly leads to a degree of immunity in most cases, just as it leads to death in a small proportion. The problem is that coronaviruses have evolved many mechanisms to suppress immunity.
Measuring immunity is important, but it isn’t easy. The most obvious way is to look for the presence of antibodies. But antibodies to what? The virus has many components. Its main entry weapon is known as Spike. This is a large, sugar-coated protein complex that can rip a hole in the membrane of a cell to allow the virus to enter. Block Spike, and you keep the virus out. It’s easy enough to measure antibodies to Spike, but not all of them actually prevent the virus from entering cells. To find out whether the antibodies are doing their job effectively, you have to culture the virus in a high-containment facility, titrate tiny amounts of serum extracted from the test subject’s blood into the virus culture, and demonstrate that the serum blocks the virus. It’s painfully slow. We are working on ways to make these assays faster, easier and more accurate. So are many others, and for once I’m happy when another lab does something better. The procedure isn’t going to be useful for testing on a large scale: instead, we’ll have to correlate antibody tests with the neutralisation assays. Some of the newly developed commercial antibody tests will probably correlate well. Others will be useful only for epidemiologists as markers of infection. Until very recently, most of the widely available tests have been so inaccurate as to be useless.
If someone outside a high-end research lab has conducted a test for you purporting to show that you are ‘immune’, I strongly caution against assuming it means anything. Lots of people have had symptoms compatible with coronavirus. In a recent draft of a study from an excellent laboratory in New York, 99.5 per cent of people who had confirmed infection developed antibodies eventually, sometimes several weeks after the test for the virus itself. Only 38 per cent of people with likely symptoms of Covid-19 – but with no positive test – had developed antibodies. Assuming that probable infection means definite immunity is a very big mistake.
There are four ‘seasonal’ coronaviruses – 229E, OC43, NL63 and HKU1 – that cause mild disease in nearly everyone, only occasionally causing pneumonia. They can be given to healthy volunteers to study the immune response. They cause the ‘common cold’, and in experimentally infected humans they give rise to an antibody response. That response wanes after a few months, and the same people can be experimentally reinfected, though they tend to get milder symptoms the second time round. It is thought that adults get reinfected on average about once every five years. Sars-CoV-2 causes mild disease in most cases, and gives rise to antibody responses in nearly all cases. We don’t know how long these responses will last, but it is likely that people who suffer only mild disease will be susceptible to reinfection after a few months or years. Humanity has never developed ‘herd immunity’ to any coronavirus, and it’s unlikely that Sars-CoV-2 infection will be any different. If we did nothing, a likely possibility is that Covid-19 would become a recurring plague. We don’t know yet. It may have seemed like an aeon, but we have been aware of this virus for only a few months.
We need a vaccine. The good news is that the virus is not mutating in such a way as to make immunisation especially difficult. One extremely crude vaccine consists of killed virus. This induces very decent immunity in experimental models. More sophisticated vaccines may well be better, and there are more than ninety in various stages of development. Vaccines that induce neutralising antibodies to the virus are strongly predicted to work: maybe not perfectly, and maybe not without complications, but I don’t know a single immunologist or virologist who thinks a vaccine is impossible.
Some are worried that because so many countries are instituting effective control measures it will be hard to demonstrate the effectiveness of a vaccine. However lackadaisical the UK response may have been, it has at least followed a vaguely scientific path. The situation in the US is different. The wealthiest country in the world, with the best bioscience and the largest testing capacity, is in the grip of the worst of the pandemic. It isn’t difficult to work out why this is. The Trump administration’s failures do mean, however, that it won’t be hard to generate the numbers required to show vaccine efficacy in the US. How quickly a vaccine can be rolled out is another matter. I have a bet with a colleague at the Scripps Institute. He thinks it will be four years. I think it will be 18 months, not least because the Bill and Melinda Gates Foundation and other large donors are backing initiatives to put promising vaccine candidates into production before there is conclusive proof they work. But I’d choose this hopeful pragmatism over the fatalism of the trolley problem.
