Two paths to controlling COVID with vaccinations

 

As vaccination is rolled out across advanced economies, the main danger has become mutations of the virus, or variants. We all know about the B.1.1.7 variant that emerged in the UK in September and helped generate the rapid rise in cases in December. We also know about the ‘South African’ variant (B.1.351), which appears to reduce the effectiveness of all vaccines to some degree. But these are just two of the better known variants, which seem to be emerging all the time (see also here).

COVID variants are the reason that so many countries have now severely restricted travel (‘almost closed borders’) into their countries in recent weeks. One of these variants could severely reduce the effectiveness of a country’s vaccination programme. It is likely that scientists would be able to change their vaccines to deal with these variants, but that will still mean at least another year before everyone is re-vaccinated and therefore means another year of lockdowns. We also do not know whether scientists could ever win a race between vaccine development and the ability of the virus to mutate.

Virus variants arise when there is a large amount of the virus about. One of the many failings of Western countries during this pandemic is to believe that you could safely vaccinate in an environment where domestic cases were high. That is exactly the environment that encourages mutations that are better at avoiding vaccines. If Western countries had followed an elimination strategy after the initial outbreaks in Spring last year, it would have been possible to keep borders open between these countries and the chances of producing a variant that can bypass vaccines within these countries would have been eliminated. In addition we would have had a small fraction of the deaths we have seen, with much less disruption to the economy.

Western countries now face a choice of how they handle COVID as vaccines are rolled out. Although no doubt reality will be far more complex than this, I think we can illustrate two types of outcomes available to them by thinking about two possible paths. I will call these two alternative paths ‘elimination’ and ‘living with COVID’. As with any attempt to look into the future, it assumes an existing technology that could change, and it involves some assessments that may prove wide of the mark (or maybe just wrong).

Elimination

This path tries to use a combination of almost closed borders, vaccination and other measures to enforce social distancing (including lockdowns) to get cases down to very low levels. At these very low levels, measures to enforce social distancing can be relaxed and an efficient test, trace and isolate (TTI) programme can keep R<1. During the summer, zero-COVID is achieved. The advantages of elimination are (see also here)

1. Once elimination is achieved, there is no chance that COVID variants that can bypass vaccines will emerge at home. While elimination is being achieved those chances are greatly reduced. This gives scientists a chance to develop new vaccines that can deal with known overseas variants while keeping everyone safe and COVID-free.
   
   

2. Countries that have achieved elimination can very easily open their borders to other countries that have done the same. Obviously this advantage is not that great if no other Western countries eliminate the virus.
   
   

3. Once elimination is achieved, the domestic economy can fully recover, there will be no more deaths from COVID, and there will be no more new cases of long COVID.

There is an important caveat to these three advantages. No ‘almost closed border’ is going to be foolproof, as both Australia and New Zealand have seen. When failure occurs, if TTI cannot eliminate the new cases, a short sharp lockdown is required to ensure the country returns to zero-COVID.

Living with COVID

This alternative arises when governments do not make any attempt at elimination, and two other outcomes prevail

1. Because of widespread vaccination together with TTI, countries can ensure R is very slightly below one in the summer months (If R is a lot less than one, elimination may occur naturally.)
   
   

2. However because COVID cases are still around by the end of summer, as winter approaches R increases to be above one, and cases increase. However because of vaccination cases never increase by enough to threaten health services, and once spring appears (or possibly before that) cases start to decrease again. (If cases rise in winter sufficiently to overwhelm the NHS, a further lockdown will be required.)
   
   

In this situation COVID becomes more like flu (it clearly isn’t like flu without vaccination), causing many deaths in winter but largely disappearing in summer. We learn to live with COVID.

Opening up borders now becomes a more messy affair. Countries would have to take informed guesses about whether other countries COVID cases were ‘safe’ (meaning they contained no dangerous variants) or not. However if most countries do not eliminate COVID, it will be easier to open borders to many countries than under elimination, once those countries have a similar number of cases that don’t include dangerous variants

There are two risks under this strategy. The positive risk is that vaccination turns out to be able to keep R<1 for most of the time, so COVID gradually dies out. This may happen because people feel more confident about being vaccinated as time goes on, and vaccines get better. Under this outcome, we don’t have to live with COVID for very long.

The negative risk is that during winter months when cases are high a mutation may develop that could bypass existing vaccines (partially or completely). This would be a very dangerous outcome, because it is likely to be some time before this variant is detected, by which time its numbers have multiplied substantially. (An advantage of the elimination strategy is that all outbreaks are treated as if they were dangerous, and are obviously much easier to see.) It is not clear to me how such a situation could be controlled without at least a year of lockdowns while new vaccines are developed and/or rolled out. The same danger arises from a failure in the ‘almost closed border’ that lets in a case with a COVID variant that can bypass vaccines.

Compared to the elimination strategy, there are two obvious advantages in living with COVID. The first is that lockdown can be relaxed or ended earlier. The second is that the country becomes less vulnerable to failures in its ‘almost closed borders’. The certain disadvantage compared to elimination is more deaths, and there is also the risk of allowing the development of variants that can bypass existing vaccines.

East versus West

At present the elimination strategy is being adopted by a group of countries which is small in number but does include China as well as Australia, New Zealand and Taiwan. There is no indication at the moment that any other countries are likely to join this group. It is also unclear whether these countries will continue with elimination once vaccination is widespread. They may be doing similar analysis to this (although no doubt better) to help them decide.

One prudent policy for the existing elimination countries is to wait and see how ‘living with the virus’ works out. If Western countries do learn to live with that strategy, the negative risks I point out above do not in practice occur, and they succeed in opening borders to each other at least, then there will be a strong incentive for the elimination countries to abandon that strategy. Equally, however, if it looks like Western countries are constantly playing catch-up with new domestically generated virus variants that bypass existing vaccines we may begin to see some Western countries begin to adopt elimination.

The UK government’s strategy for ending lockdown

Of the four criteria the UK government is going to use in deciding how quickly to ease its lockdown, steadily reducing cases is not among them. Instead we have “Infection rates do not risk a surge in hospital admissions” which is a much weaker criteria. It is actually exactly the criteria you would have under the ‘living with COVID’ strategy. It is therefore clear if unsurprising that the UK government has no intention of switching to an elimination strategy.

Much has been made of the government’s new step by step, data driven patient approach. What is less often said is that such an approach is essential to avoiding another resurgence in cases when most people have not yet been vaccinated. (Perhaps Tory MPs wanting to open everything quickly are only there to make the government look good?) Opening schools all at once in March is a risk. While it is good that secondary school children will be required to wear masks at all times, in France children aged six and over are told to do this. There is probably a lot more that could be done for schools in terms of ventilation, bubble sizes and so on. We also need to prepare for the possibility that schools going back may raise R to almost one, which would effectively invalidate the government's approach.

What is also not widely appreciated is that SAGE forecasts, like those released yesterday, are going to play a critical role in deciding how quickly lockdown is relaxed. Only in this context does the need to ensure “Infection rates do not risk a surge in hospital admissions” make sense. What SAGE needs to forecast is what level of cases in the summer, with lockdown at an end, is sufficient to ensure that the increase in cases over autumn and winter does not overwhelm the NHS. It is a very difficult forecast to make because we have no experience of what happens when most people are vaccinated, and if the government is wise it will base its calculations (on when to ease lockdown) on a worse case scenario. Unfortunately if it did that it would be acting out of character.

You also have to ask whether, if you intend to follow data rather than dates, it is wise to then announce a complete set of dates attached to a detailed list of the restrictions that will be lifted all the way to the end of lockdown? This gives a mixed message at the very least. In reality, given that we are far from clear how effective each element of the lockdown is, there may be a danger that the whole timetable is built on sand. But in political terms, this timetable is going to have a strong influence on expectations (where caveats are forgotten or just ignored by print media). The net result may be that disappointing those expectations will be a large political cost that politicians will not want to incur. A timetable this precise may have the effect of taking decisions away from the scientists when the data turns out to be worse than expected.  

Did the scientists advising the government on COVID-19 make serious mistakes?

The government has now said it will begin to wind down the lockdown with an extensive test, trace and isolate (TTI) regime in place. Containment, having been abandoned on March 12th, is back but this time the UK government is serious about it. It has a much greater chance of success not just because there is more testing capability available, and there will be an app to help with tracing, but also because it will take place with a degree of social distancing.

We can only hope that this works, and that government actions match their words. But what seems very puzzling to me is why the scientists advising the government seem to have been preoccupied with moderated herd immunity as the only way to deal with this virus, and largely neglected the possibility that containment could be made to work.

Managed herd immunity plus behavioural worries

This is not just a UK issue. Herd immunity, together with limited and advisory social distancing, is the strategy that is being tried in Sweden, following the advice of their chief epidemiologist. It is controversial there too, with many other scientists arguing against this strategy, and with deaths in Sweden well above those in other Nordic countries.

The logic of the argument for herd immunity is set out in a very interesting debate on Channel 4 on 13th March. It was between John Edmunds, an epidemiologist and member of SAGE, and Tomas Pueyo. (Although I saw this at the time, thanks to @cubic_logic for reminding me of it.) Both seem to agree that a lockdown of some sort is required to avoid the health service being overwhelmed, with the disagreement being how quickly that is required, but at the end it seems both are in agreement about the principle of herd immunity.

Why not impose a lockdown until a vaccine is discovered. You can also achieve herd immunity with a vaccine, without hundreds of thousands of people dying. If there was no cost to a lockdown, and it could be successfully imposed for a year or more, the optimal strategy is clearly to impose the lockdown as quickly as possible and retain it until the vaccine arrives. Obviously a lockdown of that length would be very costly, but the reason Edmunds gives in this debate for not doing this is that people would stop observing it. This is the ‘behavioural’ element to the story that has been talked about.

The behavioural element is crucial for the following reason. Even if it is not successful, containment could allow the following strategy. You start with containment (as we did), switch to total lockdown when containment is obviously failing, and once that lockdown brings the number of cases right down abandon it and revert to containment. This on/off for the economy could get us to the point where a vaccine is produced in large quantities. Compared to herd immunity we would almost certainly cause far less deaths without too greater economic cost.

To reject this outcome requires a pretty firm belief that people would become so tired of lockdowns by the Autumn that whatever the government did there would be an uncontrolled epidemic in the winter with a worse outcome for the NHS than managed herd immunity could achieve. Where that belief comes from I don’t know, and neither will we know until after the pandemic ends because the government has refused to publish the scientific advice it received until then. You would think that this behavioural assumption would depend in part on the economic support individuals were receiving, but as far as I know no Treasury officials or other economists were involved. Equally no sociologists were involved.

It is here that politics and science interact. Speaking to reuters’ journalists, John Edmunds is reported to have said:

“We had milder interventions in place because no one thought it would be acceptable politically to shut the country down.” He added: “We didn’t model it because it didn’t seem to be on the agenda. And Imperial didn’t look at it either.”

This is dangerous territory for any scientist. In the blink of an eye a scientist believing something is not on the agenda can be a politician saying that no one told us about this alternative.

Once those advising the government believed that lockdowns were off the agenda or too fragile because of the behavioural response then the government’s strategy begins to make more sense. You try containment, because containment would be the best outcome, but you suspect that without some social distancing and other measures it is unlikely to work, but you don’t want to impose those for behavioural reasons. You then move to managed herd immunity, where you want to impose as little restrictions as you can get away with, again for behavioural reasons, such that the NHS stays above the water line.

The strategy in action

On 9th March, during the containment phase, according to reuters:

“Johnson held out against stringent measures, saying he was following the advice of the government’s scientists. He asserted on March 9: “We are doing everything we can to combat this outbreak, based on the very latest scientific and medical advice.” Indeed, the government’s Scientific Advisory Group for Emergencies, SAGE, had recommended that day, with no dissension recorded in its summary, that the UK reject a China-style lockdown. SAGE decided that “implementing a subset of measures would be ideal,” according to a record of its conclusions. Tougher measures could create a “large second epidemic wave once the measures were lifted,” SAGE said.”

Whether this was a unanimous view is disputed by Lawrence Freedman, who says some members did favour something closer to a Wuhan lockdown.

Thus on 10-13th March the Cheltenham racing festival went ahead. There is now evidence of a concentration of COVID cases in the area of Cheltenham racecourse. Other sporting events were allowed to go ahead. While this seems negligent if your concern is reducing deaths, if you think herd immunity is the only option and you want to delay even a partial lockdown for as long as possible it makes sense. On 12th March containment ended, and in his first COVID press conference the PM said “many more families will lose loved ones before their time.” Testing is now limited to hospitals.

Yet at the same time the Imperial team produced a new analysis suggesting that further measures were needed to avoid overwhelming the NHS. This was ‘the science changing’ according to the BBC. In reality this was still the strategy of managed herd immunity. Freedman reports:

“Now, promoted by the Imperial study, the advice from the operations subgroup of the Scientific Pandemic Influenza group on Modelling (SPI-M-O) on 16 March was notably different in tone from a week earlier. The measures at first envisaged – case-by-case isolation, household isolation and social distancing of vulnerable groups – was now “unlikely to prevent critical care facilities being overwhelmed”. Everything now had to be tried, including “general social distancing and school closures”, which offered the best chance of disease control.”

But was this really just the science changing? The reuters article says:

“What allowed Britain to alter course, said Edmunds, was a lockdown in Italy that “opened up the policy space” coupled with new data. First came a paper by Edmunds’ own London School team that examined intermittent lockdowns, sent to the modelling committee on March 11 and validated by Edinburgh University. Ferguson’s revised Imperial research followed. Woolhouse, the Edinburgh professor, confirmed the sequence. Edmunds said these new studies together had demonstrated that if the British government imposed a lengthy period of tougher measures, perhaps relaxed periodically, then the size of the epidemic could be substantially reduced.”

What changed was the politics, which had “opened up” the alternative strategy of intermittent lockdowns that I described at the beginning of this post. It was inevitable that the politics would change. I noted before Chris Giles commenting that the herd immunity strategy was time inconsistent: once deaths started increasing and people saw the pandemic overwhelming other countries politicians were bound to change strategy. But Johnson remained very reluctant to do so. As an article in the New England Journal of Medicine says about the measures introduced on 16th March:

“People should work from home if possible, but that was largely up to employers to decide. Vague. Anyone over 70 was advised to avoid “nonessential social contact.” Vague again. The government was “moving emphatically away from” mass gatherings, but again these were still not (and are still not) banned. British understatement was in full swing — citizens, businesses, and nursing homes were asked to read between the lines and go beyond explicit government policy. The prime minister’s father announced that he would be going to the pub if he chose to, since they needed the customers.”

We had to wait until 23rd March for the current lockdown to be imposed, that replaced advice with something much stronger. The French President threatened to close its borders with the UK if this was not done. As I noted in last week’s blog post, the 10 days delay between the containment phase ending and full lockdown probably cost the majority of lives lost in the UK. I don’t think the difference between weak advice (16th March) and the currrent lockdown (23rd March) can be pinned on the scientists.

What mistakes did scientists make

How much of all this the fault of the scientists advising the government. They certainly made mistakes. Three stand out. The first was NERVTAG keeping the threat level at moderate in mid-February, with Edmunds the only one objecting. This failed to convey the urgency of the situation to the government. The second, mentioned above, was scientists sticking to an agenda that seemed to rule out a lockdown, or other scientists pretending they knew with any certainty that lockdowns could not be sustained for long. Cummings taking part in the meeting didn’t help here. People who say his presence had no impact on the other people in the room, when his actions could potentially hurt many of those people, have no imagination.

The third, and by far the most important, was the failure to realise that containment if taken seriously was an alternative to herd immunity or total lockdowns. East Asian countries like South Korea suggest that alternative, if done properly and with supporting measures, can work for some time. You test people who may have the virus, and if they do you trace their contacts over the period they have been infectious and isolate both them and their contacts. The South Korean experience suggests that you can even bring down very high numbers of cases using this strategy. Professor Ferguson now agrees this is the country to learn from.

As long as a vaccine comes along within two years, this is a strategy that can get you through the period with no vaccine with far less cases and deaths than with herd immunity. This option comes up in the Channel 4 discussion noted above, but Edmunds dismisses it as having been tried and failed in the UK. In the UK this phase failed to stop the number of daily cases increasing.

One of the lessons from the countries with successful TTI strategies is that you close down travel from infected areas very quickly, or impose compulsory isolation for incoming travellers. (One of those countries, Taiwan, has an epidemiologist as its deputy President, which just shows that not all epidemiologists, or even a majority? believe in herd immunity.) The UK failed to do this during our containment phase. In addition, having mobile phone based tracing apps is extremely helpful in tracing who people have been in contact with, and the UK did not have that during its containment phase. But more generally, mild forms of social distancing can be used to help TTI work. Stopping large gatherings like Cheltenham races is an obvious example. We can put the same point another way. The UK did nothing to help the containment phase work. Was this due to politics or the absence of advice from SAGE?

If TTI fails, as it has in Singapore for example, then governments quickly need to impose a lockdown. But this is not a reversion to a herd immunity strategy, but just a measure required to get cases back down to a level where TTI can work again. Even if TTI has to be accompanied by occasional lockdowns, it will still involve many less deaths than herd immunity or ‘running the NHS hot’ does, simply because of the mathematics.

Do these failures by scientists advising the government, if they were not politically influenced, get politicians off the hook? Absolutely not. Most politicians, on being advised that the only sure way of responding to a new virus was to let hundreds of thousands of people die (as they were told in the UK), would surely want to be absolutely sure that there really wasn’t any alternative. They would ask basic questions, like are other countries doing the same, or are there examples of countries that have chosen a different and less deadly approach. Scientists would not have kept what China and South Korea are doing from ministers, and they are unlikely to have said that these countries' experience was of zero relevance.

If our government had done that, I’m pretty sure we would have had containment with more support in terms of controlling international travel, banning large gatherings and so on. Herd immunity wouldn't have been on the agenda, and a lockdown would have been imposed earlier. Probably most of those who have died would still be with us. The reason we tried containment in a half hearted way without social distancing support or border controls were because politicians were too content with a strategy that would kill huge numbers of people.

We are where we are because our current Prime Minister and some in his cabinet, even in the face of predictions of hundreds of thousands of deaths, was too content to do nothing, take it on the chin, and too fearful of curtailing economic freedoms to interrogate the advice they were given. It was the same slowness to act, rather than anything coming from the scientists said, that led to inadequate PPE provision, or to waiting until mid-March to appeal for ventilators or make requests to labs for assistance with testing.

What happens next

There still seems to be divergent views within government about how quickly the lockdown should be relaxed, where by relaxed I mean allowing more people to get COVID-19. There shouldn't be any debate. The number of cases need to be brought down as quickly as possible to a level where the TTI regime can be pretty sure to work. The current lockdown has to continue until that point is reached, and in addition some sort of quarantine system has to operate on anyone coming into the country.

At that point the TTI system can begin to operate. (Earlier testing should iron out any teething problems.) Only when this system has been operating for enough time to see that it is clearly working can we begin to relax lockdown. But these relaxations in lockdown need to be done step by step, and in each case we need to ensure that TTI can still work successfully.