COVID-19: playing for time - strategies to combat spread.

There's been a lot of discussion over the past few days about strategies to mitigate the spread of COVID-19, with different messages being disseminated in different countries. For example, it is well known that China and South Korea implemented a total lockdown, and the number of cases in China has decreased to such an extent that there are now more imported cases than ones developing internally. This strategy of near total population isolation has been adopted by many other countries - notably (for me) France, which yesterday announced a relatively complete lockdown due to last "at least 2 weeks". In contrast, the UK and USA appear to have been slower off the mark, with the UK instead issuing advice for only some people to stay home and providing different lengths of quarantine for different groups of people. In this post, I dissect the latest UK modelling data (Report 9 (pdf) from the Imperial College COVID-19 Response Team) that was released yesterday, and discuss some of the implications with respect to the strategies that different countries have taken.

In a public health crisis such as we are now facing globally, it is essential to provide clear and simple messages. More importantly, it is essential that these messages are transparent and based on accurate information. In situations where that information is not available, assumptions should be both clearly stated and clearly rationalised.

Imperial College modelling Report 9

The MRC Centre for Global Infectious Disease Analysis at Imperial College, London, is one of the top modelling groups in the world and a World Health Organisation Collaborating Center for Infectious Disease Modelling. The unit is headed by Neil Ferguson who has been involved in many outbreaks and has also worked closely with Chris Whitty, the UK's Chief Medical Officer. The UK advice relating to COVID-19 is based on work from this unit and, more specifically and most recently, Report 9 which was released on Monday, 16 March 2020. So what does the report say? I'm going to walk you through some of the main parts of it in the next few sections of this post, providing a bit of commentary and some of my personal thoughts along the way.

Report Introduction

The report commences with detailing two possible strategies for tackling the spread of an infection: suppression and mitigation. This is how they are described in the report (I have provided links to the references included in the report):

Suppression. Here the aim is to reduce the reproduction number (the average number of secondary cases each case generates), R, to below 1 and hence to reduce case numbers to low levels or (as for SARS or Ebola) eliminate human-to-human transmission. The main challenge of this approach is that NPIs [non-pharmaceutical interventions] (and drugs, if available) need to be maintained – at least intermittently - for as long as the virus is circulating in the human population, or until a vaccine becomes available. In the case of COVID-19, it will be at least a 12-18 months before a vaccine is available. Furthermore, there is no guarantee that initial vaccines will have high efficacy.

By "NPI," they essentially mean complete social distancing measures, which is effectively the quarantine of as many people as possible. This is the strategy that has been successfully adopted thus far in China. It works really well but the problem is it needs to be maintained otherwise the disease comes back.

Mitigation. Here the aim is to use NPIs (and vaccines or drugs, if available) not to interrupt transmission completely, but to reduce the health impact of an epidemic, akin to the strategy adopted by some US cities in 1918, and by the world more generally in the 1957, 1968 and 2009 influenza pandemics. In the 2009 pandemic, for instance, early supplies of vaccine were targeted at individuals with pre-existing medical conditions which put them at risk of more severe disease (PDF reference). In this scenario, population immunity builds up through the epidemic, leading to an eventual rapid decline in case numbers and transmission dropping to low levels.

This is a little more opaque. Essentially, the idea is to somehow control viral spread so that the impact is as low as possible. However, it still allows for propogation of the virus, and there are still therefore consequences. They say that these are similar to the measures adopted in previous epidemics, although what they don't mention so much is that there are important differences when comparing our current situation with previous pandemics. Most notably, the world has changed enormously in terms of our connectivity, particularly with the rise of the internet and changes in our travel patterns.

The authors then go on to say:

"... we focus on feasibility, with a specific focus on what the likely healthcare system impact of the two approaches would be. We present results for Great Britain (GB) and the United States (US), but they are equally applicable to most high-income countries."

Report Methods

In one of my previous posts, I tried to explain some very basic modelling concepts - but highlighted that I am not an infectious disease epidemiologist. The team at Imperial, in contrast, have extensive experience as well as access to data which I do not have. In the report, they say:

We modified an individual-based simulation model developed to support pandemic influenza planning to explore scenarios for COVID-19 in GB. The basic structure of the model remains as previously published.

The previous model was described in two papers, the first in Nature and the second in Proceedings of the National Academy of Sciences (PNAS). These are two of the most rigorous journals in the world and I have no reason at all to doubt their methodologies here. Particularly, the modelling is quite complex and appears to be described in detail. The report goes on to say:

In brief, individuals reside in areas defined by high-resolution population density data. Contacts with other individuals in the population are made within the household, at school, in the workplace and in the wider community. Census data were used to define the age and household distribution size. Data on average class sizes and staff-student ratios were used to generate a synthetic population of schools distributed proportional to local population density. Data on the distribution of workplace size was used to generate workplaces with commuting distance data used to locate workplaces appropriately across the population. Individuals are assigned to each of these locations at the start of the simulation.

My only criticism here is that they do not adequately provide information on the source data. For example, what data have they used for workplaces? Additionally, census data in the UK were last collected in 2011, have they made any modifications (or used modifications from others) to account for changes since then?

They then make various assumptions, some of which are entirely reasonable (from publicly known knowledge) and some of which are less easy to understand. For example:

We assumed an incubation period of 5.1 days.

I agree with this one. They provide two references: the first in medRXiv and the second in the New England Journal of Medicine. When I was thinking about this same issue, I based my workings on an article in the Annals of Internal Medicine which had the same conclusion.

Infectiousness is assumed to occur from 12 hours prior to the onset of symptoms for those that are symptomatic and from 4.6 days after infection in those that are asymptomatic with an infectiousness profile over time that results in a 6.5-day mean generation time.

This doesn't really make any sense to me. Twelve hours is 0.5 days. The incubation period is the time from becoming infected to developing symptoms. So they are saying that those people who are symptomatic (at an average of 5.1 days) are infectious from half a day previously - or 4.6 days. And those who are not symptomatic are also infectious at 4.6 days - so what's the difference?

Based on fits to the early growth-rate of the epidemic in Wuhan 10,11, we make a baseline assumption that R0=2.4 but examine values between 2.0 and 2.6.

This fits with known data - estimates I've seen are around 2.35. However, some estimates have been as high as 4.5 - and the reproductive rate is known to be higher earlier in epidemics before control measures have been put in place.

We assume that symptomatic individuals are 50% more infectious than asymptomatic individuals. Individual infectiousness is assumed to be variable...

On recovery from infection, individuals are assumed to be immune to re-infection in the short term.

This is a big assumption. They do qualify it:

Evidence from the FluWatch cohort study suggests that re-infection with the same strain of seasonal circulating coronavirus is highly unlikely in the same or following season (Prof Andrew Hayward, personal communication).

Andrew Hayward is a professor at UCL so we can probably safely assume that he knows what he is talking about. Still, it would be nice to have a bit more information.

Analyses of data from China as well as data from those returning on repatriation flights suggest that 40-50% of infections were not identified as cases. This may include asymptomatic infections, mild disease and a level of under-ascertainment. We therefore assume that two-thirds of cases are sufficiently symptomatic to self-isolate (if required by policy) within 1 day of symptom onset, and a mean delay from onset of symptoms to hospitalisation of 5 days.

My comments here are that this is probably reasonable - although I'm less sure about the delay to hospitalisation. The assumption that two thirds of all cases (including asymptomatic ones) is quite probably an overestimate. Based on the incubation period of 5.1 days described earlier, a mean delay to hospitalisation of 5 days means that patients are hospitalised around 10 days after infection; however, this could be a bit short if hospitalisation is only required around day 14 of the illness (which is what I've heard elsewhere).

The age-stratified proportion of infections that require hospitalisation and the infection fatality ratio (IFR) were obtained from an analysis of a subset of cases from China.

This seems reasonable - they reference a preprint article entitled Estimates of the severity of COVID-19 disease contained on medRXiv.

These estimates were corrected for non-uniform attack rates by age and when applied to the GB population result in an IFR of 0.9% with 4.4% of infections hospitalised. We assume that 30% of those that are hospitalised will require critical care (invasive mechanical ventilation or ECMO) based on early reports from COVID-19 cases in the UK, China and Italy (Professor Nicholas Hart, personal communication). Based on expert clinical opinion, we assume that 50% of those in critical care will die and an age-dependent proportion of those that do not require critical care die (calculated to match the overall IFR).

The fact that they've corrected for the UK population is really nice. Nicholas Hart is Professor of Respiratory and Critical Care Medicine at King's College London and also the Director of the Lane Fox Respiratory Service at Guy's and St Thomas’ NHS Foundation Trust, and I don't know of other evidence about the proportion of hospitalised patients requiring intensive care - this will also depend on other system factors, including the initial threshold for hospitalisation which may vary from country to country (or, even, at a more local level).

The big assumption in this section is that everyone receives optimal (hospital) care. If health systems are overwhelmed - as has happened already in Italy - the assumption will not hold true and the fatality rate of those who don't receive intensive care when it's required will be much higher. The authors of the report do go on to explain a bit more about their assumptions relating to demand for hospital places, which mitigates this concern slightly:

We calculate bed demand numbers assuming a total duration of stay in hospital of 8 days if critical care is not required and 16 days (with 10 days in ICU) if critical care is required. With 30% of hospitalised cases requiring critical care, we obtain an overall mean duration of hospitalisation of 10.4 days, slightly shorter than the duration from hospital admission to discharge observed for COVID-19 cases internationally [PDF reference - Report 8] (who will have remained in hospital longer to ensure negative tests at discharge) but inline with ]estimates for general pneumonia admissions](https://www.tandfonline.com/doi/abs/10.3111/13696998.2014.940423?journalCode=ijme20) (non-free reference).

But, similarly, I have several concerns here. Admittedly, I haven't read their previous report, so I cannot completely comment on their figures but from what I've heard, duration of intensive care is long and I'm not sure that 10 days is a sufficient time period. This impression is partly re-affirmed when they say their estimates are slightly shorter than those currently observed. What I've heard from Italy and France so far is that the durations are longer - although we're still early in the epidemic and hence the "average" duration has not been (indeed, cannot yet be) established: it may get shorter (if we get better at diagnosing earlier) but also might get longer (if we realise that earlier, more aggressive treatment helps prevent deaths).

The report then goes on to talk about the different strategies that may be put in place to deal with the epidemic. Broadly, there are five potential strategies:

1. Case isolation in the home
2. Voluntary home quarantine
3. Social distancing of those over 70 years of age
4. Social distancing of entire population
5. Closure of schools and universities

Even with these, there are many nuances that are assumed. For example, for the last ("closure of schools and universities"), it is assumed that:

Closure of all schools, 25% of universities remain open. Household contact rates for student families increase by 50% during closure. Contacts in the community increase by 25% during closure.

The final key part of the methodology is how long the measures are expected to last for. They state:

When examining mitigation strategies,we assume policies are in force for 3 months, other than social distancing of those over the age of 70 which is assumed to remain in place for one month longer. Suppression strategies are assumed to be in place for 5 months or longer.

Three to five months (or perhaps slightly longer). That's what you're facing folks - as a minimum. We need to start getting used to this situation and figure out strategies like remote presence tools for dealing with it.

Report Results

This is where it starts to get really interesting. The first figure in the report shows what the epidemic would look like: (a) in terms of the number of deaths per day per 100,000 people without any control measures in the UK and the USA; and (b) overall infections by state in the USA.

Over half a million deaths in the UK (from a population of approximately 66.5 million people). That's quite a lot. And 2.2 million in the USA. Much better than the very crude numbers you may have noticed on my own graphs (but I did say many times they were crude and not to be taken literally), but still: a lot of deaths. And also based on a lot of assumptions - including that everyone gets optimal hospital care. As I've already said, that's not necessarily going to be the case. Indeed, the report goes on to say:

For an uncontrolled epidemic, we predict critical care bed capacity would be exceeded as early as the second week in April, with an eventual peak in ICU or critical care bed demand that is over 30 times greater than the maximum supply in both countries (Figure 2).

Here's figure 2. The red line shows the number of intensive care beds available.

The interesting thing about this figure is they show the relevant decreases in the intensive care bed requirements dependent upon different strategies that they described in the methods. Essentially, in isolation no mitigation strategy is sufficient. Indeed, even when mitigation strategies are combined, there are insufficient beds available in Great Britain for the patients who will require them.

But what happens if we implement all controls and aim for complete suppression? Well, things are definitely better. We already know from China and South Korea that this will control disease spread and allow hospitals to manage, and that cases may even drop. But the measures need to be strict. And they also have to be implemented for a really long time - any let up lets the virus come back, and the death rate surges again. This is shown in the third figure included in the report which includes two panels. The first panel shows the overall picture, the second is zoomed in to the bottom portion of the graph so that it's easier to see what's happening in relation to the number of intensive care beds available; both panels show what would happen in Great Britain if various mitigation measures are taken for a 5 month period (shown as shaded/blue in the figure) beginning towards the end of March 2020 (i.e. about now):

The current UK government strategy is roughly approximate to the orange line. That is, the advice is for older people to isolate themselves, and cases or potential cases (including family contacts) to also isolate. What happens is that we see an increase over the next few weeks but then the increase starts to slow down and the intensive care bed capacity is only slightly exceeded in the beginning of May before a secondary rise occurs over the summer. I would guess this is probably related to some aspect of the model that accounts for increased mixing during the summer months when people want to get out of the house, but I can't be sure.

The green line represents a policy more akin to that which was previously adopted in China, South Korea and Italy, and is now being adopted in France as well as many other countries around the world. Indeed, this is the only policy that can lead to an R of less than 1. As the authors of the report say:

Our projections show that to be able to reduce R to close to 1 or below, a combination of case isolation, social distancing of the entire population and either household quarantine or school and university closure are required

The problem, of course, comes following the "blue" period when measures/controls are relaxed. There is a huge rebound in terms of the number of cases - and this is particularly true for the policy of (almost) complete lockdown. Hence, presumably why the UK government has chosen the strategy they have.

What have we learned so far?

There are many take away messages from this report so far. For instance:

• The world has changed. The virus is "out there" and there's no putting the genie back in the bottle - much as every strain of flu in the last 100 years is descended from the 1918 influenza epidemic.

• We need to learn patience. The delay from infection to onset of symptoms forces this upon us, as once measures are taken we need to wait for quite some time to see what impact they have.

• Complete social-distancing is the only measure that works. But it only works as long as it can be enforced for - as soon as it stops, the danger will be back.

So, what are our governments doing about it in terms of strategy, and which strategy is best? The two examples I will use are France and the United Kingdom - they are the countries I know best; they are roughly in the same situation in terms of the number of people currently affected; and they have opted for different strategies: these are described first below. However, other countries are similar to one or the other - the USA, for example, is so far following a similar strategy to the UK, whereas most of therest of the world has been taking measures similar to France.

Strategy: France

France has opted for almost complete lockdown:

Les déplacements sont interdits sauf dans les cas suivants et uniquement à condition d'être munis d'une attestation pour :

• Se déplacer de son domicile à son lieu de travail dès lors que le télétravail n’est pas possible ;

• Faire ses achats de première nécessité dans les commerces de proximité autorisés ;

• Se rendre auprès d’un professionnel de santé ;

• Se déplacer pour la garde de ses enfants ou pour aider les personnes vulnérables à la stricte condition de respecter les gestes barrières ;

• Faire de l’exercice physique uniquement à titre individuel, autour du domicile et sans aucun rassemblement.

Translated, that says:

Leaving your house is forbidden except for the following cases and only when carrying a document of authorisation:

• Going from home to work when remote-working is not possible;

• Essential shopping in the closest authorised shops;

• Visiting a medical professional;

• Going out to take care of children or for helping vulnerable people with the strict condition that social distancing and barrier measures are respected;

• Taking individual exercise, local to where you live and without meeting anyone else.

Fines will range from 38 to 135 euros. An FAQ (pdf) has been produced providing a little bit more detail.

In terms of the economic impact, the goverment has stated that they will help business - making available 300 billion Euros for that. There will be help for many things like tax and utility bills

Strategy: United Kingdom

The United Kingdom has, thus far, taken a different approach. They are advising "social distancing" but in a more targetted fashion. The guidance is shown in the following figure:

Social distancing/isolation is recommended for certain sectors of society, but not for all and, at present, schools and universities remain open although people are encouraged to find ways to not move around. However, no economic measures have (yet) been announced... In fact, as I've been writing this post, there is a statement and press conference going which I think is escalating some of the advice - but it is still not equivalent to France. However, some economic relief packages have been described, including the release of 330 billion pounds. I think it is likely that more information as well as further restrictions will come in the days to come.

Which strategy is better?

This is a difficult question to answer, and it partly depends on what the national priorities are.

The difficulty with imposing complete confinement on the entire population is that it may work initially, but over time people will become bored and frustrated - inevitably they will look for ways to break the rules. In contrast, by not imposing strict rules immediately (and leaving it to individual "buy-in") as is the UK strategy, the government is asking a lot of a lot of people. There is no guarantee that people will cooperate - as was seen in Paris on Sunday, a day after bars, cafés, restaurants and other non-essential public spaces were closed, when people spent the day congregating in the parks and other open spaces.

One way of dealing with this is to impose restrictions to control the spread of the virus, but then ease these off when they start to have an effect. This was something that was considered by Imperial College modellers, and is shown in the next image:

Here, they've modelled the weekly number number of patients requiring intensive care in orange, with implementation of a really strict policy of social-distancing that is relaxed if intensive care cases fall below a certain threshold, and reinstated if the number of intensive care patients passes another, higher threshold. It is not entirely clear from the report which policies are included here (the description in the main report text differs slightly from the caption - which just says "other polices"[sic]), but the principle is what's important to understand.

Looking to the future - my interpretation of the data

There are two issues when considering the future. The first is whether the data on which we're basing our strategy (i.e. in this case, the modelling report described above) is any good. The second is about the different strategies that are being (or could potentially be) taken by different governments.

Is the report any good?

I raised a number of concerns when I discussed the methods above. Most notably, there are a lot of assumptions that are made, they aren't all well justified, and I'm not sure I agree with some of them. However, I agree with a majority of the data they have based their models on and, in any case, the data are specific to Great Britain whereas the important thing we should be taking away is the principles. Thus, while there may be differences in numbers (notably, I am concerned the death rate is higher than currently estimated), the trends and impact of the mitigation/suppression strategies is extremely useful.

In other words, the big picture is excellent while some of the finer details may not be quite so good.

Are goverment strategies any good?

It might be argued that the priorities of France and the United Kingdom differ: one appears to be trying to maintain economic health, but potentially at the cost of population health, while the other is thinking first of the population's health and leaving economic questions to be sorted out later. However, that is a rather simplistic dichotomisation: with any decision, there are likely to be immediate consequences and consequences that are delayed. As the modelling described above shows, confinement could lead to a rebound epidemic later on if it's then lifted or people ignore it. But, of course, letting your population die is political suicide for any government.

But what have we learnt, what does the future hold, and which country is doing better in terms of their strategy? My opinion is that it's best to come down rather hard at a certain point near the beginning and then ease off again. In today's super-connected world - where news information from other countries can be available (almost) immediately - I strongly believe that means isolation is required now for all members of the population in all countries, as I've written about before. This provides multiple benefits:

• It stops people dying.

• It gives us time to find out more about this disease - there is much we do not currently know.

• It gives us time to try to develop treatments and, more importantly, a vaccine.

• It gives us time to develop better models for the future - to figure out which measures really work to slow down spread of the virus, and how we can live our lives.

I therefore remain concerned about the strategy of the UK and the USA. I believe that complete suppression measures should be implemented. Alternatively, they need to be providing more information about their thinking: how they envisage the epidemic progressing, and what evidence they have.

Final thoughts

This has been a long post but I make no apologies: now many of us are stuck at home, you should have time to read it! I have tried to dissect some of the current epidemiological evidence about COVID-19 and how it is influencing government strategy. My conclusions remain unchanged from previously: that a strategy of lockdown and isolation is currently required by all. This is also one of the main conclusions of the report:

We therefore conclude that epidemic suppression is the only viable strategy at the current time. The social and economic effects of the measures which are needed to achieve this policy goal will be profound. Many countries have adopted such measures already, but even those countries at an earlier stage of their epidemic(such as the UK) will need to do so imminently.

Epidemic suppression strategies will save lives, and will buy us precious time. We need to work together to combat the COVID-19 virus: it is a global threat, it needs a global response, and we need to do this collaboratively, openly, together.

Please stay safe, and please stay isolated.