Middle Ground, or Illusion?

A review of Mark Woolhouse’s The Year The World Went Mad

Mark Woolhouse is Professor of Infectious Disease Epidemiology at the University of Edinburgh. He was a member of two high-level government advisory bodies throughout the pandemic: the UK’s SPI-M (Scientific Pandemic Influenza Group on Modelling, a sub-committee of SAGE, the Scientific Advisory Group for Emergencies); and the C19 Advisory Group for the Scottish Government. In short, he is highly qualified and was intimately involved in advising Government at more than one level throughout the pandemic.

Given his expertise and close involvement with the response to the pandemic, it is not surprising that Woolhouse is thoroughly mainstream on some key issues. In particular, he regarded and still regards Covid-19 (or novel coronavirus as he prefers to call it) as a very serious threat. He takes the average Infection Fatality Rate (IFR) to be 1% in the time period he discusses (the first year of the pandemic, before vaccines changed the picture). This, as he also notes, made it more than ten times as deadly as flu if infected. What is more, Covid-19 is more infectious that flu, and more readily transmitted before symptoms become apparent, meaning it posed an even greater threat as compared with flu than the order of magnitude greater IFR suggests. ‘All of this told me’, he writes, reflecting on the first two months of 2020, ‘that this particular pandemic could be much more severe and much harder to control than the flu pandemic the UK had spent years preparing for.’ (p. 9). By mid March 2020 it became clear that cases were growing rapidly in the UK, more rapidly in fact than SAGE had thought. In Northern Italy, which was ahead of the UK on the pandemic curve, hospitals were under severe strain. Options narrowed. ‘At the SPI-M meeting on the morning of March 23rd I supported the committee’s recommendation of an immediate national lockdown… The UK was given a clear and simple instruction: stay at home.’ (p. 41).

More than this, not only did Woolhouse view Covid-19 as a serious threat and support the first national lockdown at the time, he was and remains critical, if not dismissive, of the Great Barrington Declaration, a well known alternative to the mainstream approach to the pandemic [for my assessment of the Great Barringtons, see here and here]. The Great Barringtons called for ‘focused protection’ of the vulnerable while allowing the majority of the population to live normal lives and develop ‘herd immunity’ through infection. As Woolhouse rightly notes, the claim by Sunetra Gupta (one of the Great Barringtons) that the virus was in far wider circulation early in the pandemic than the mainstream thought, meaning more had been infected and the population might be nearer herd immunity in Spring 2020 than typically thought, was just plain wrong (10% or less of the population infected through the first wave was the approximate figure in reality, see p. 32).

That the Great Barringtons thoroughly overestimated how many had contracted Covid-19 in the first wave meant that they thoroughly underestimated both how many would contract it, and the consequences of this, if Covid-19 were allowed to run through the population with minimal restrictions for the majority. As Woolhouse rightly says, ‘This was not a viable strategy: we would still end up with many more people needing to be hospitalised than the NHS could cope with.’ (p. 241). By that he means simply from the majority of the population at lower risk from the virus living normal lives (‘a small fraction of a very large number is still a large number’, p. 241). In addition, ‘If the virus were circulating freely it was inevitable that some infection would spill over into the vulnerable minority and cause a substantial burden of death and disease in that group too.’ (p. 241)

So far, so very conventional on Woolhouse’s part. And yet… ‘My hope — and my main motivation for writing this book — is that lockdown scepticism will become the mainstream view.’ (p. 238)… ‘If we’d trusted ourselves, our data, our systems and our science then I’ve no doubt we’d have made better choices. Lives and livelihoods could have been saved, lockdowns largely avoided and far less damage done. Instead, we were mesmerised by the once-in-a century scale of the emergency and succeeded only in making a global crisis even worse. In short, we panicked.’ (p. 257)… ‘I did not expect that elementary principles of epidemiology would be misunderstood and ignored, that tried and trusted approaches to public health would be pushed aside, that so many scientists would abandon their objectivity, or that plain common sense will be a casualty of this crisis. Yet — as I’ve explained — these things did happen, and we have all seen the result. I did not expect the world to go mad. But it did.’ (p. 257)

It is this combination of scientific expertise in a core area and soundness on some basics (rejection of the Great Barrington approach) on the one hand, and a pretty strong challenge to the way the UK managed the pandemic from a ‘lockdown sceptic’ perspective on the other, that makes Woolhouse’s book an interesting one.

Lockdown — strategy, or failure of strategy?

While he supported the first lockdown, he wished we’d had more options at the time, and wanted us to plan to avoid future lockdowns (when this hadn’t happened, in his view, once again he thought we had little choice but to lock down in January 2021). The key mistake, he argues, was that in rejecting the Great Barringtons’ approach, the mainstream came to associate ‘protecting the vulnerable’ with the Great Barringtons’ herd immunity strategy. In rejecting the latter they also rejected the former, and came to rely instead on, if not positively advocate for, lockdowns, whenever cases picked up. ‘There is a middle ground’, argues Woolhouse, ‘and I believe that is where the answer lies.’ (p. 242). Very crudely, his alternative could be summarised as ‘Sweden’ +. I’ll get to this, in particular the + part, in a moment. But first, does Woolhouse accurately characterise the mainstream UK understanding and approach, the one he is criticising? My argument is that he doesn’t. What is more, he makes a number of very categorical, over confident, heavily coloured-by-hindsight, and under researched statements in failing accurately to characterise the mainstream approach. If I were to venture a suggestion as to why this is, it would be that he feels rather nose out of joint after his specific proposal (the + part in ‘Sweden’ +) was not given the serious attention he thinks it deserved.

Some analysts, in particular lockdown sceptics, define Lockdown as any Non-Pharmaceutical Intervention (NPI); that is, any restriction on normal life. On this definition, much of the world was locked down for two years straight, and many parts are still locked down. Woolhouse, more sensibly, uses a more restrictive definition: ‘a package of legally enforced measures that limit the reasons people can leave their own homes.’ On this definition, the UK was not locked down for the whole year he covers. Rather, there were three national lockdowns, starting on 23 March 2020, 5 November 2020 and 4 January 2021 (p. 51), with the first and the last being the most significant.

‘Lockdown’, he writes, ‘was never going to solve the coronavirus problem, it just deferred it to another day, and it did so at great cost.’ (p. 51). As mentioned above, at the time Woolhouse supported the first national lockdown. He didn’t support the second short-lived one in November 2020, but when the more transmissible Alpha variant arrived he again saw no alternative to the third and longer-lasting national lockdown, starting 4 January 2021, which coincided with the start of the vaccine rollout: ‘I couldn’t see any way this would end without a third lockdown. I’d thought that from December 20th when I first saw the data showing how fast the new variant was spreading, so fast that a November-like lockdown might not be enough to bring the pandemic under control.’ (p. 178). At the same time he is keen to emphasise that lockdowns were a desperate measure when all else had failed; they represented a failure of strategy rather than a strategy.

This is all very reasonable. Who could disagree? Well, Woolhouse thinks many do disagree with him. He believes lockdowns were a positive strategy for some rather than a desperate measure to avoid the overload of hospitals and buy time. Two issues strike me about this. The first is a strong form of hindsight bias he brings to this discussion. Like other insiders who have written on this time period, he conveys well his own and societal-level uncertainties and confusions early on, from the emergence of the virus at the turn of 2019 / 2020 running up to the first national lockdown in March 2020. In fact, better than some he grasps the extent to which scientists and politicians were always a step or two behind the virus, for example only realising in the last week before the March lockdown just how quickly the virus was spreading. And yet, despite this, he seems to think a comprehensive alternative to lockdown could have been formulated in this early period, and in stating this he takes a swipe at the more conventional hindsight statement that we should have locked down a week earlier, in the process turning this now common view, apparently, into belief in lockdown as strategy: ‘I do not agree with those SAGE members who have made public their regret that they did not recommend lockdown a week or so earlier than March 23rd. As far as I’m concerned, going into lockdown was a failure in itself — we needed to act even sooner to avoid having to take that step.’ (p. 50). I’m sure many would agree with him, if we allow ourselves the luxury of taking hindsight further and further back. I’m also sure that many who, like him, supported the first national lockdown at the time, did so for lack of an alternative, as an emergency break, not as he claims, a long-term strategy in itself.

Beyond his strong form of hindsight bias, a second issue brings out quite starkly his mis-characterisation of mainstream views on lockdowns. Pursuing his distinction between lockdown as an emergency measure representing a failure of strategy (the basis on which he supported them) and lockdowns as strategy (his claim about mainstream views on the issue), Woolhouse makes the following bold claim: ‘The world was given an intervention that only made sense in the context of eradication as the preferred means to control a disease that was clearly here to stay.’ (p. 53, my emphasis) In other words, in his view the mainstream supported lockdowns as a strategy on an impossible premise. He repeats the point later on: ‘Lockdown only made sense in the context of eradication, but when the World Health Organization was forced to abandon its eradication strategy it did not change course and lockdown quickly became the international norm.’ (p. 255)

If the UK Government and the World Health Organisation actually thought as he claims they did, they would have been proposing lockdown forever. But of course they weren’t proposing that. Indeed, it’s worth noting that while the WHO might have endorsed eradication for a while, the UK government never did: when the first lockdown ended, the UK Government started relaxing controls, reaching a high point with Chancellor Sunak’s ‘eat out to help out’ policy in August 2020. It is far more reasonable to view the UK Government’s strategy as muddling through until vaccines became available (that is not to say other policies weren’t also tried, such as Test and Trace, and a Tier System of controls short of lockdowns through the autumn).

Does Woolhouse not see this? He does, kind of, but he is very reluctant to give it any credit. In fact he believes that if this were the strategy, it was ‘grossly irresponsible’, because it would amount to planning for the ‘reasonable best case’ of vaccine success. (p. 187) So unwilling is he to see that this might actually have been the strategy, and a partly reasonable and successful one, that he says regulatory bodies and scientists who proposed vaccinating vulnerable groups first were inconsistent and had performed a U turn: the regulator’s decision, he claimed, ‘was supported by many public health experts who had dismissed the idea that we should take special measures to protect the most vulnerable in other ways. That seemed inconsistent to me, but we got there in the end.’… ‘Once vaccines became available, the attitude towards protecting the vulnerable completely changed — you could call it a U-turn.’ (p. 186; p. 246). This is nonsense. Everyone was concerned to protect the vulnerable. The question was how to do this. As noted above, Woolhouse dismissed the Great Barringtons’ approach to protecting the vulnerable. The difference between Woolhouse and the mainstream is that the latter was impressed by neither the Great Barringtons’ proposals nor Woolhouse’s (we’ll get to his soon). What they did support was a strategy that actually worked — vaccination. There was no U-turn.

The fact that vaccines were around the corner in November 2020, and known to be by Government, may also explain the muddle and prevarication around the November lockdown, but also the determination to go ahead. For Woolhouse, ‘it was apparent to everyone that the latest lockdown — just like the previous one in March — would not solve the problem, it would just defer it… by the Christmas holidays the situation would not be much better than it was when the lockdown began.’ (p. 169). Well, except in the obvious sense that stopping cases rising saved lives because we deferred the problem until vaccines were ready to be injected into people’s arms. In case you think I am being unreasonable here, I should point out that Woolhouse himself believes the November lockdown reduced the number of cases (at a time we didn’t have vaccines): ‘The November 2020 lockdown itself turned out to be more successful than many had expected… The incidence of cases had fallen by almost half as we entered the final week of lockdown.’ (p. 171) Sometimes, a problem deferred is a problem at least partly solved.

Other mischaracterisations and overly categorical statements

In addition to mischaracterising the mainstream approach to lockdowns, Woolhouse makes a number of other statements in which he mischaracterises his opponents’ arguments and / or makes overly categorical statements. These include the claim that consideration was not given to the harms of lockdowns and different views about transmission in schools. For the sake of (relative) brevity, let’s discuss just one of his claims, one beloved of lockdown sceptics, the idea that Government set out to scare people into complying with lockdowns by exaggerating the risks. Woolhouse doesn’t hold back, writing this about March 2020: ‘the actual risk to more than half the population was extremely low. Government advisors were concerned that this might reduce acceptance of lockdown. A solution to the problem was suggested by the SAGE subgroup SPI-B, which advises on behavioural science… a crucial section reads…: A substantial number of people still do not feel sufficiently personally threatened; it could be that they are reassured by the low death rate in their demographic group… The perceived level of personal threat needs to be increased among those who are complacent, using hard-hitting emotional messaging.’ Woolhouse goes on: ‘SPI-B’s advice gave them [Government] a charter to exaggerate the risks to ensure compliance. Michael Gove — Minister for the Cabinet Office — took full advantage, declaring that “we are all at risk” and “the virus does not discriminate”.’ (pp. 107–8)

As far as Woolhouse is concerned, this messaging from Government was not a one-off or an accident. Rather, ‘the risks to the majority of the population from novel coronavirus were consistently and deliberately overstated by both the government and sections of the media in order to bolster support for lockdown.’ (p. 194, my emphasis)

‘Consistently and deliberately’ — this is a strong claim to make. But the fact of the matter is that Woolhouse is plain wrong, as the only sources he references show. The quote from the SPI-B paper includes no specific (numerical) mention of levels of risk. There is the claim in it that some people weren’t sufficiently scared, given the levels of risk. This is quite possibly true: in mid to late March, given the lag between cases and deaths, awareness within the general population of absolute risks by age group was still developing. Experts and government already knew that there was a steep age gradient, and at this point in time rough numbers, which turned out to be pretty accurate, were thought by age to be, for example, 0.5% IFR aged 55; approximately 1% early 60s. Did the general population know that? Were they sufficiently worried about it if they were aware? SPI-B drew upon a survey that showed the public were most certainly aware of increasing risk with age. But as to awareness of the absolute level of risk by age, this is less clear. The survey showed, to continue with the example of the late middle-aged group, that of those between 55 and 64, 5% thought infection would be personally “life-threatening” (and thus 95% thought it wouldn’t). Whether that is an over-estimate or an under-estimate, as compared with the approx. 1% death rate in that age range, is in part a subjective judgement, dependent on what “life-threatening” means.

What is clear is that SPI-B did not push the argument that everyone was at equal risk. A cursory glance at the SPI-B paper Woolhouse references reveals that it is littered with information about the steep rise in risk with age. On this point, and its connection to the need for general controls, SPI-B’s message is the wholly conventional one, the one Woolhouse himself makes in his critique of the Great Barrington Declaration — the need to limit all social interaction so as to limit the circulation of the virus to reduce the chance of the most vulnerable catching it.

What of Michael Gove, Woolhouse’s only other reference for his claim that Government pushed the message that we are all at equal risk? This is equally wrong on Woolhouse’s part. Gove was very clearly referring to the risk of infection, not the outcomes of infection. He referenced Prime Minister Boris Johnson and Health Secretary Matt Hancock in support of his argument. At this point in time, 27 March 2020, Johnson was not seriously ill, and as far as I am aware Hancock never became seriously ill. In fact 27th March was the day it was announced that Johnson had tested positive. Gove’s point was that everyone was at roughly equal risk of becoming infected, and that was pretty much true.

Woolhouse’s alternative

Early in the Pandemic Woolhouse and his team developed a strategy to manage the pandemic without lockdowns — ‘Segmenting and shielding of the most vulnerable members of the population as elements of an exit strategy from Covid-19 lockdown’. Recall that Woolhouse, unlike supporters of the Great Barrington Declaration, thought that Covid-19 was a far more serious risk than flu. He agrees with the mainstream assessment that the average Infection Fatality Rate across the population was around 1% in the UK before vaccines. Related to this, given the number of deaths in the first few months, it was also clear to him that only around 10% of the population had acquired immunity though infection during the first wave (Spring 2020). So how does his proposal differ from the Great Barringtons’ one of ‘focused protection’, which he rejects? There are two sides to this: a superior (in his view) approach to shielding; and the recognition of the need for the rest of society to live with restrictions if not full lockdown, to keep cases down. The first part is the + part, as I am calling it, the second the ‘Sweden’ part. As he notes, because Sweden, the actual country rather than ‘Sweden’ (its approach to Covid), maintained general restrictions throughout, it, like the UK, got nowhere near to herd immunity before vaccines. In other words Sweden itself did not follow the Great Barringtons’ approach (p. 213). But given that Woolhouse’s claim to novelty rests more on his Segmentation and Shielding proposal, in what follows I focus on the + part rather than the ‘Sweden’ part.

Woolhouse’s shielding proposal covered both the vulnerable themselves and those with immediate supporting roles for them. In the community it also included those who simply lived with the vulnerable. Age was the biggest single predictor of serious outcomes. The IFR passed the 1% point at ages above the early 60s. This seems to suggest that around 15 million people in the UK were in a category that Woolhouse himself says warranted enhanced protection (it’s a subjective judgement, of course). Add to that those caring or living with people 60 and over and the numbers requiring shielding seem impossibly high. But this is where segmentation comes in to it. While 1% is the average, many people of a given age will in fact be at higher risk, and many at lower risk, once other risk factors in addition to age are considered.

Woolhouse complains that his peers rejected his proposal, when he put it forward in Spring 2020. He thinks it wasn’t given a fair hearing (and/or, self-critically, ‘we simply didn’t make the case well enough’, p. 98). Yet on the basis of what was known at the time, how could it have worked? Woolhouse writes: ‘I was told on numerous occasions that implementing Covid-safe measures to protect the vulnerable in the community was just too difficult. Yes, the scale is daunting — we are talking about several million people — but surely no more daunting than locking down the entire country, and we were all too willing to do that.’ (p. 98). But hang on, where has this several million people come from? How has the 15 million with an IFR over 1% based on age (with carers and those living with to be added) been reduced to several million? This is the product of a study, QCOVID, published in October 2020. Woolhouse writes: ‘According to QCOVID, 91% of the deaths occurred in the 15% of the population at greatest risk. What has happened is that we have removed the healthy elderly from the top 15% and replaced them with younger individuals with co-morbidities and other risk factors’ (p. 91). Or even more extremely, ‘we could identify 5% of the population who were over fifty times more likely to die from novel coronavirus infection than the other 95%.’ (p. 235). 5% of the UK population is several million, the number he mentioned earlier in the text.

Given that the data emerged later, his colleagues could be forgiven for not wanting to base policy on it in Spring 2020. But Woolhouse’s hindsight aside, what of the merits of the proposal once data was available? There are three clear problems it seems to me:

1. No matter how the groups are sliced and diced, the average IFR across the whole population has to aggregate to 1%. The larger the shielding group, the more unwieldy and impossible the task of shielding. On the other hand, the smaller the group the larger the non-shielding group, and, with a still non-trivial IFR, the more people would easily be exposed to the virus with significant number still experiencing severe outcomes (a small proportion of a very large number is still a large number). As Woolhouse himself said in his Spring 2020 paper: ‘A smaller vulnerable population may be logistically easier to protect, and perhaps more likely to comply, but is likely to incur a smaller proportion of the severe disease burden. At the same time, a consequence of protecting a smaller proportion of the population and relaxing restrictions for a larger proportion is that overall transmission rates are higher. The implication is that S&S [Segmentation & Shielding] will be much more difficult to implement successfully if the proportion of the population designated vulnerable is too small.’ Yet this latter scenario sounds just like the 5% / 95% split. A factor 50 difference in group average IFR implied the 95% would have had an average IFR of around 0.29% as compared with around 14% for the 5%. 0.29% may not sound too bad, but it is around six times the risk posed by seasonal flu, and Covid is more contagious.

2. The QCOVID breakdown of risk factors is very detailed. Only by digging deeply into all possibilities are they able to split the groups into 5% and 95% with a fiftyfold difference in average risk between them. In addition to a long list of medical conditions, it is noteworthy that included within the risk factors are social (deprivation) and racial characteristics. Is Woolhouse proposing that these risk factors should have been a part of the implementation?

3. That final word immediately above — implementation. You won’t find any meaningful discussion of that in Woolhouse’s book. In fact you won’t find anything remotely close to a concrete plan of action. Before we even think about the real world social logistics, this point holds for the most basic point about the QCOVID study. It is a research study assessing risk factors. It doesn’t attempt to assess how the information might be used as a part of an alternative to lockdowns. Is Woolhouse suggesting that the very detailed analysis of numerous medical risk factors contained within the study should have been used to generate an overall risk profile for each member of the entire UK population, then a 5% / 95% plan enacted? Of course it wouldn’t have been 5% / 95% because of all the carers and those living with them, so…

Woolhouse doesn’t really get to first base with his plan. None of this is to say that in general he isn’t on to something. More protection for the more vulnerable and more thought to what could have been done to reduce spread in the general population short of lockdown is a good general starting point when thinking about what to do in the future. But on the other hand it is not hard to see why using lockdown as an emergency brake while placing great hope in the vaccine programme (the approach criticised as ‘grossly irresponsible’ by Woolhouse) seemed the more reasonable course of action to policy makers as they grappled with the unfolding pandemic in real time.

Sweden +

Could the UK have pursued some kind of “Sweden” +, even if not the particular plan promoted by Woolhouse? Perhaps. It is a counterfactual we will never be sure of the answer to. Yes, many people voluntarily socially distanced to a great degree before the major lockdowns (lockdowns one and three, March 2020 and January 2021). On the other hand, the imminence of lockdowns was heavily signposted by government, leading to pre-adoption perhaps. Without the warnings, the projections and the signposting of lockdowns, would the voluntary measures have happened to the same extent?

One of the biggest difficulties with the idea that the UK could have copied the approach of other countries (or other countries copied the UK’s approach) is that each country faced specific problems and a specific pattern of spread. Woolhouse acknowledges the issue: ‘There are good arguments that the UK was predisposed to suffer a severe epidemic whatever we did. Here are some suggestions: infection was seeded quickly and widely thanks to our high volume of international travel; our crowded cities and our lifestyles promoted the spread of the virus; and our ageing population and generally poor population health meant that more of us were at risk of developing severe disease.’ (pp. 208–9). To that could be added years of NHS underfunding and a chaotic Boris Johnson led government.

The UK did learn from the first wave. The NHS in particular was better organised for the winter wave of 2021. There were also improved treatment protocols. And of course most of all we had vaccines by early 2021. The conventional analysis that shifting everything a week earlier in March 2020 would have saved many thousands of lives is true based on the mathematical properties of exponential growth. A different kind of government would have paid people to isolate and would have enacted some fairly basic ventilation measures in schools and other workplaces. How much differences these and other measures would have made given Woolhouse’s observations about the nature of the UK is hard to say. What does seem clear to me is that Woolhouse’s colleagues were right to be sceptical about his Segmentation and Shielding proposal. This is not to say it and related ideas shouldn’t be interrogated in more detail in thinking ahead. After all, as Woolhouse somewhat depressingly concludes, Covid-19 will not be the last pandemic.