Why the UK has changed its coronavirus strategy (and why there’s hope)
The UK government dramatically changed its strategy for dealing with coronavirus this week. The key reasons behind this can be found in a report put together for the government’s advisory group by scientists from Imperial College London. I spent yesterday studying the report so that you don’t have to (and because science nerd + social distancing, I guess it was inevitable). I’m a Cambridge grad in Biology and Management, now working in public health. My mum is an A&E consultant and dad is a GP in Birmingham. Thought I’d share a non-technical summary to try and help non-scientists understand what’s going on.
Read on to find out what the report said, why it’s changed the UK strategy and, crucially, what we can expect to happen next. If you want to know why there’s light at the end of the tunnel, scroll straight down to “Is this going to be the new normal forever?” (spoiler: it’s not)
A very tiny bit of epidemiology (bear with me)
To understand the reasoning behind global government responses to the coronavirus, we need to understand one key thing. Every virus has a ‘reproductive number’ (R0) — the number of new cases of a disease directly caused by one infectious person in an entirely susceptible population. It’s basically an indicator of how easily a disease transmits between people. For coronavirus, it is estimated that R0 = 2.4 so every infected person will pass the disease onto 2.4 other people if society is functioning as usual. If R0 is more than 1, an epidemic grows until most of the population has been infected and you achieve herd immunity. If R0 is less than 1, the number of cases shrinks over time and the disease disappears from the population.
A government can use various policy tools to reduce R0 at this stage in an epidemic. Deploying a couple of them will make the number of cases grow more slowly, and using lots of them will make it start shrinking:
· Isolating infected people for 7 days [‘case isolation’]
· Quarantining the other people in their household for 14 days [‘household quarantine’]
· Limiting physical interaction between people [‘social distancing’]
· Shutting schools & universities [‘school closures’]
· Shutting workplaces
Adding all of the above together = ‘lockdown’
There’s also another key policy available: isolating people most vulnerable to the disease e.g. those over 75 or with underlying conditions. This doesn’t have much impact on the transmission of the disease but is very important for reducing deaths. I won’t mention this again but it’s included as a given in all of the scenarios described below.
UK strategy
Until Monday, the UK’s strategy for the COVID-19 epidemic was mitigation — getting R0 close to, but still more than, 1. This would slow but not stop the disease and allow it to infect up to 80% of the population, which scientists believe would enable us to develop herd immunity. The aim of mitigation would be to ‘flatten the curve’ so that peak demand on intensive treatment unit (ITU) beds in UK hospitals would not exceed the NHS’s capacity — see the diagram we all know and love below. ITU beds are the only ones with the ventilators needed to treat the most seriously ill coronavirus patients.
The mitigation measure introduced last Thursday (case isolation) was the first of several designed to ‘flatten the curve’ in the UK. The government would have decided when to layer on additional measures based on:
- Measures of actual disease spread — the best measure being the number of patients testing positive in ITUs, as all patients in ITUs will be tested whereas there will be lower testing coverage of patients who are less seriously ill.
- Predictions generated by detailed mathematical models like the one this report discusses. These take in huge numbers of inputs, both:
a. about the virus (e.g. R0, average time from infection to ITU admission)
b. about the population (e.g. school class sizes and population age distribution)
N.B. we have to make plenty of guesses and assumptions about these inputs given how little we currently know about the virus. We are much more confident in the info we have about the population as it’s based on tons of data and has been verified by using it to model outbreaks of previous diseases, where we can check that its predictions fit with what actually happened.
Until a few days ago, the model predicted that these mitigation measures would keep demand for ITU beds below ITU capacity. Then, two updates to the available data caused the UK government to change its strategy:
- Previously, they thought that around 16% of the cases admitted to hospital would also require admission to ITU. Latest evidence from Italy and the UK suggests it’s more like 30%. This number is an important input if your overall aim is to make sure that ITU demand is lower than ITU capacity.
- They also received more certain confirmation that the NHS has a maximum achievable ‘surge’ ICU capacity of around 5,000 beds, not more.
Changing both these inputs also changes the conclusion about whether it’s possible to stick with mitigation without totally overwhelming the NHS: it’s not. Even the best mitigation strategy of case isolation + household quarantine would have seen 260,000 deaths assuming a perfectly in-control NHS. But peak demand for ITU beds would also have been 12x capacity. In countries where the health system is totally overwhelmed like this, e.g. Italy, the death rate seems to be about 10x higher. Based on this, continuing with the mitigation strategy would probably have resulted in several million deaths in the UK.
So it has become absolutely necessary to move to a more aggressive strategy of suppression — getting R0 below 1 so that the number of new cases starts decreasing towards 0.
Why didn’t the government just aim for suppression from the outset? There are several likely reasons:
· Wanting to reduce negative impact on the economy
· Wanting to avoid societal disruption
· The endgame is not clear — if you suppress an epidemic back to 0 cases, you then still have most of a population susceptible to it and vulnerable to a second wave. I will talk about this more in a bit.
Why might you argue that they should have aimed for suppression from the outset?
· There are still very many unknowns about coronavirus so ‘playing it safe’ at this stage gives us more time to research and plan
· Even using their old, incorrect assumptions (that ITU capacity was higher and demand would be lower), if 80% of the 66 million population had got the virus at 0.5% mortality rate, you still would have been looking at 260,000 deaths. The only way to bring this number down is through suppression measures.
So we are now aiming for suppression to avoid millions of deaths. The mitigation measure we already had in place (7-day case isolation) is not a bad start — but the model in this report shows it’s totally insufficient. On Monday, they brought in two additional measures: 14-day household quarantine and general social distancing.
If we just keep these three measures, we’ll see around 90,000 deaths plus ITU demand peaking at 2x capacity. This is shown by the orange line in the figure from the report, below (B is a zoomed in version of A). But we can also see that adding school closures (the green line) keeps ITU demand below capacity (the red line). It also reduces deaths to around 24,000. It’s very likely that the government will be adding school closures as a suppression measure in the next few weeks (update: Boris Johnson has confirmed that school closures are “imminent” as of Wednesday afternoon).
Why haven’t they shut schools and universities already? Various reasons have been given for why schools remain open. I am unsure which are considered the most important:
1) Lots of NHS staff have school-aged children. If schools are closed, some of them might have to take days off work to look after their kids. This may reduce ITU capacity or lead to challenges elsewhere in the health system which ultimately increase the case fatality rate.
2) We are still very uncertain how much children transmit coronavirus compared to adults, given that they have much milder symptoms. If they aren’t big transmitters, then shutting schools could make little difference. But we don’t know yet.
3) Shutting schools causes general societal disruption and economic losses.
4) If children off school are sent to their grandparents because their parents are at work, it puts their grandparents at risk.
One measure which doesn’t appear on any of the graphs, despite having been modelled, is a complete lockdown i.e. case isolation + household quarantine + social distancing + school closures + workplace closures. We have already seen this enforced in various countries, including Italy. The report also says (unsurprisingly) that a lockdown would have the largest impact on lives saved. But, for some reason, the associated numbers for deaths and ITU demand weren’t published so we don’t know how much better they would be. I’d guess this indicates that the government are trying to avoid a lockdown if possible.
We can expect ITU admissions for coronavirus over the next few months to follow the pattern shown by either the orange line or the green line in the figure above, depending whether or not school closures are added to existing measures. If/when they are introduced, ITU admissions will begin declining around three weeks later.
What happens next?
What happens next is uncharted territory. China is the first country which just began to lift suppression measures a few days ago. The figure below shows the same graph as above but extended further into the future. If we model the suppression measures continuing for 5 months then being lifted, we see the epidemic re-emerge on a huge scale in the second half of 2020. Very bad. That’s because suppression takes you back to very low case numbers but you still have a whole population susceptible to infection later.
So we can’t just go into lockdown for 5 months and then go back to normal, as a second wave of the epidemic will undoubtedly occur. As one potential solution to this problem, the report offers an ‘adaptive policy’ where you repeatedly turn suppression policies on and off while keeping caseload below ITU capacity:
Below is what that looks like in terms of caseload. The blue line is turning suppression policies on (at the top) or off (at the bottom). The orange line is caseload.
Every time you relax suppression, more people contract the disease and some people die, but society functions more normally for a short time. Depending which exact combination of suppression measures you choose, you need to spend about 2/3 of time in suppression and can relax it for around 1/3 of the time without overwhelming ITUs.
Is this going to be the new normal forever?
No. We won’t be living in lockdown forever, but it could be the new normal for a while (maybe 5 months based on this model). There are a few different potential options for getting out of lockdown and it will become clear which of these are most feasible as we learn more about the virus:
1) Turn on/off suppression measures until we have a vaccine, probably in around 18–24 months. This assumes we can develop a vaccine. It might not be possible to reliably produce one if coronavirus keeps moving the goalposts by mutating like the flu and developing new strains.
2) If we can’t develop a vaccine, turn on/off suppression repeatedly until enough people have been infected that we achieve herd immunity. This is a very unappealing option as it would take years of major societal disruption.
3) My favourite option: suppress the disease to low enough levels that you can isolate all confirmed cases. Lift suppression measures so that freedom of movement is restored, but test every single suspected case and aggressively trace and isolate anyone they have contacted. This could enable society to go back to mostly normal functioning while preventing the epidemic from growing out of control. South Korea have been doing this spectacularly well since their epidemic started and it is a genuinely feasible and scalable option using existing technology. If we can overcome privacy concerns, ‘big data’ like smartphone location data can be used to inform anyone who has been in contact with a confirmed or suspected case that they need to self-isolate. Check out this article for a flavour of what it looks like. Tech companies have been recording our lives in intricate detail for some time — this is an example of where it can be used for good. Details like the following are posted on public websites: “Patient №12 had booked seats E13 and E14 for a 5:30 p.m. showing of the South Korean film, “The Man Standing Next.” Before grabbing a 12:40 p.m. train, patient №17 dined at a soft-tofu restaurant in Seoul. Patient №21 drove her car to attend a weekday evening church service.” The information is kept anonymous — so you know if it’s talking about you but nobody else does. People who haven’t been in contact with a case can go about their usual business. My bet is that this is the way we are going, until we can hopefully develop a vaccine. If the test, trace, isolate approach doesn’t work perfectly and we begin to see the early stages of another epidemic, then societal suppression measures will need to be reintroduced temporarily to bring it back under control.
There is one unknown that could change everything: we don’t yet know what proportion of infections are asymptomatic (where the person has no symptoms and doesn’t realise they have the disease). This is because there has thankfully been no epidemic of coronavirus which has spread completely throughout a population. If lots of infections turn out to be asymptomatic, then the death rate (number of deaths divided by number of cases) would be lower than we currently estimate because ‘number of cases’ is bigger than we realise. Some numbers to illustrate:
We have seen 8,000 deaths and 200,000 confirmed cases to date worldwide (N.B. true case number is probably a lot higher but not everyone has been tested).
((8,000 ÷ 200,000 = 4% fatality rate))
But we can imagine a (highly optimistic and unlikely) scenario where 95% of cases are actually asymptomatic and going untested, so there are really 4 million infections to date.
((8,000 ÷ 4,000,000 = 0.2% fatality rate; this is closer to the 0.1% mortality rate of a typical flu strain))
We won’t know the answer to this question until scientists have finished perfecting a new antibody test which is currently in development. This will let us test whether someone has had coronavirus at some point in the past, even if they no longer have it. If we find that lots of infections are asymptomatic, this could be a gamechanger down the line.
[Update on 20th March: the antibody test could also move us towards a 4th option for getting out of lockdown, as long as people who’ve recovered from the virus are no longer likely to transmit it (which we still need to find out). If we can identify who has had the virus already, those people may be able to move around normally. This will be particularly useful at first for care workers and health workers who need to interact with vulnerable groups, but could likely be extended to other members of society later on.]
So what do we do?
The most important thing to do now is not to panic. Countries are adopting suppression measures at increasingly early points in their own epidemic timelines thanks to the lessons learned from uncontrolled outbreaks in places like Wuhan and Italy. Every day, we learn more about coronavirus and how to protect ourselves from it. Government crisis strategies are working to ensure the continued functioning of essential supply chains like food and waste management.
But we can’t be complacent. The current public health threat is the most serious we have faced in generations. The pressure on the NHS will be immense in the coming months. So we must take suppression measures extremely seriously. Evidence increasingly suggests that you can transmit the virus before you develop any symptoms yourself, so it is vital that we all follow government guidelines strictly to protect vulnerable members of our families and communities. But just because we must be physically separated doesn’t mean we can’t look out for one another. We can check in (virtually) on friends, relatives and strangers - offer support and connection as we navigate the uncertainty together. This won’t be easy but it can be done.
If people find this post useful, I might try and share some more bits and pieces later on. You can find me on Twitter @FionaConlon4
P.S. If you think you might have coronavirus symptoms, do follow the instructions at NHS 111 online. Don’t go to a pharmacy/surgery/hospital and don’t ring an ambulance unless it’s a life-threatening emergency. Let’s keep each other safe and save these services for the patients in greatest need.
Citation of the report I’ve been talking about (link here):
Ferguson, N.M., Laydon, D., Nedjati-Gilani, G., Imai, N., Ainslie, K., Baguelin, M., Bhatia, S., Boonyasiri, A., Cucunubá, Z., Cuomo-Dannenburg, G., et al. (2020). Impact of non-pharmaceutical interventions (NPIs) to reduce COVID- 19 mortality and healthcare demand. 9th report from the WHO Collaborating Centre for Infectious Disease Modelling, MRC Centre for Global Infectious Disease Analysis, J-IDEA, Imperial College London.
