Show Hide image

Where the science went wrong

Sage minutes show that scientific caution, rather than a strategy of “herd immunity”, drove the UK’s slow response to the Covid-19 pandemic.

 

On Tuesday, 10 March 2020 the Scientific Advisory Group for Emergencies (Sage), which advises the government on how best to respond to Covid-19, observed that there might be as many as 5,000 to 10,000 cases in the UK. It was time to consider measures to protect the elderly and vulnerable. There was, however, no need to panic. The UK was still four to five weeks behind Italy on the epidemic curve, and with timely interventions that could be pushed back by a couple of weeks. But firmer estimates of infection rates were needed. Because of the growing number of cases these were now possible and would be available the following week.

Sage did not wait another week. It convened again three days later. Professor Neil Ferguson’s team of epidemiologists from Imperial College London had the new estimates. The UK was further along the epidemic curve than assumed. The sudden change in the picture was put down to a five-seven day “lag in data provision for modelling”. The advice now was that measures to isolate the elderly and vulnerable should be “implemented soon, provided [it] can be done well and equitably”. Those “who may want to distance themselves” should be given advice on how to do so. Because these measures might not be sufficient “more intensive actions” needed to be considered to “enable the NHS to cope”.

Those measures were announced by the government on 16 March. That day Sage met again, its 16th meeting of 2020. The picture was grim. Instead of a total of 5,000 to 10,000 existing cases, this in fact was the likely number of new cases being added each day, leading to a doubling of the total every five to six days. Additional social distancing measures would need to be introduced as soon as possible if the basic objective of avoiding “critical cases exceeding NHS intensive care and other respiratory support bed capacity” was to be achieved. The Imperial group’s modelling suggested that if these were introduced, assuming public compliance, this might just be in time.

The meeting concluded with Sage agreeing “to publish a chronological set of papers and other documents which have informed the questions it has considered and its advice to date”. The participants were aware that they would soon be asked about the abrupt move away from a comparatively relaxed attitude to the spread of the virus. It was important, the minutes noted, “to demonstrate the uncertainties scientists have faced, how understanding of Covid-19 has developed over time, and the science behind the advice at each stage”. 

The material that soon began to appear on the Sage website did illuminate the assumptions behind the changing advice to the government. But there were still demands for more transparency with regard both to the membership of Sage and the minutes. In early May, the membership of the group was published. On 29 May, the minutes of its first 34 meetings, up to 7 May, were published. They show a system with several specialist sub-groups, largely populated by academics, funnelling advice to Sage, weighted more in its membership to senior figures in either the government’s own advisory network, or from the NHS and Public Health England (PHE). The government’s chief scientific adviser, Patrick Vallance, and the chief medical officer for England, Chris Whitty, were the co-chairs. The number of attendees had also grown, from between ten and 20 in the early meetings to as many as 40 when they later convened on Zoom.

In addition to general concern about why the UK has such a high death toll, two controversies fuelled the demand for full disclosure. The first was the result of reports that Dominic Cummings was part of the process and was influencing the advice. Unfortunately for this theory, Cummings was never a member of Sage and is reported to have attended only four meetings, the first on 5 March and then not again until he had returned from Durham on 14 April (although his colleague Ben Warner was a more regular presence). 

The second was the allegation that the scientists had been pushing for a cold-blooded policy of “herd immunity”, which accepted that little could be done to counter the disease until a vaccine had been developed or a sufficient number of people had been infected. It is clear from the minutes that this was never an objective. On one occasion (20 February), there was a call for more clarity on objectives but the options were then put as “flattening the peak, spreading the duration, avoiding winter”, and were about the key challenges faced by the NHS.

[See also: The real reason the UK government pursued “herd immunity” – and why it was abandoned]

Although Vallance mentioned herd immunity in a radio interview on 13 March, he insisted in a Sunday Telegraph article on 31 May that a group of people who had spent their careers treating diseases was not going to encourage people to get infected. The aim was to prevent people from becoming infected rather than to encourage infections. 

***

Sage took it for granted that the Covid-19 problem would only be solved by the population becoming immune, preferably as a result of a vaccine. That was not a policy objective but just an unfortunate statement of fact. Until then a large proportion of the population would be susceptible to infection, albeit with varying degrees of severity. There were limits on what could be achieved. The minutes of the fateful meeting on 13 March record: “Sage was unanimous that measures seeking to completely suppress spread of Covid-19 will cause a second peak. Sage advises that it is a near certainty that countries such as China, where heavy suppression is underway, will experience a second peak once measures are relaxed.”

This view continues to influence Sage’s thinking. At this stage, the problem, as we can see from events in other countries now coming out of lockdown, may not be one of a second large peak but of the virus lurking beneath the surface ready to bubble up again. This explains the caution now expressed by many Sage members as the government begins to ease restrictions. Until immunity is achieved, most likely because of the development and distribution of a vaccine, all attempts to return to normality will be constrained by a fear of the consequences.

This fundamentally pessimistic outlook influenced Sage as soon it was evident in early February that the pathogen had already embarked on its global journey. Members could see little value in screening visitors at airports because there was no suitable test deployable on scale while “temperature and other forms of screening are unlikely to be of value and have high false positive and false negative rates” (22 January). Although air links provided the best predictor of where the disease would end up, travel restrictions could do no more than delay the onset of the infection and probably for only a few days. 

The restrictions would have to be 95 per cent effective to make a meaningful difference (3 February). Once the disease had reached China’s Asian neighbours, it was only a matter of time before it reached the UK.

By 23 March, the evidence suggested that cases arriving from abroad represented only 0.5 per cent of the total. Yet another report submitted that day from a national network set up to deliver large-scale sequencing and analysis of the virus’s genomes described a “large number of independent Sars-CoV-2 introductions to the UK, from multiple locations around the world.” This included Italy and other parts of Europe. On 2 June Neil Ferguson observed to a House of Lords committee that it was a surge of transmission from Spain and Italy in early March that caught out the modellers. Although those entering the UK were supposed to take protective measures, there does not seem to have been much idea to what extent, if at all, individuals had complied (23 March).

With better testing this might have been picked up. The initial policy had been to trace the contacts of those exhibiting symptoms. A reliable test was developed quite quickly but there was a problem with PHEs limited diagnostic capacity. The capacity that was available was initially largely assigned to tests for seasonal flu. This was a major factor inhibiting Sage’s own efforts to work out what was happening in the community. 

This limited capacity meant that test and trace could only work while the numbers were very low. On 18 February, it was reported that PHE could only cope with tracing only five new cases a week, each of which would lead to 160 contacts being isolated. At best, it would be possible to increase this tenfold: 50 new cases a week and 8,000 contact isolations. This was not going to be enough. Not surprisingly this led to the conclusion that once there was sustained transmission in the UK, contact tracing would no longer be useful. It was abandoned on 12 March. 

The priority was now to use the available capacity to assess the growing number of patients being taken to hospital with acute coronavirus symptoms. By contrast, Germany’s established diagnostic industry, which could scale up rapidly to meet the challenge, was one reason for its relative success in keeping track of the virus’s spread. 

On 16 March Sage “highlighted the critical importance of scaling up antibody serology and diagnostic testing” and asked PHE for a proposal to do so. Ten days later it “re-emphasised the importance of urgently ramping up testing with appropriate quality”. On 2 April, with a testing plan at last ready (it was published on 4 April), it stressed the need for a “clear public communications [strategy] to address confusions about which different tests can/cannot do”.

Sage’s pessimistic attitude influenced early consideration of social and behavioural interventions. On 13 February, noting that China had introduced stringent measures – stopping movement in and out of Wuhan and other cities in Hubei province and confining people to their homes – Sage doubted whether these would enable it to contain the epidemic. By 25 February, it was more optimistic. The evidence from Hong Kong, Wuhan and Singapore suggested that such measures (“university and school closures, home isolation, household quarantine and social distancing”) could get the critical reproduction number (the R number) to approximately 1. 

[See also: How the World Health Organisation’s failure to challenge China over coronavirus cost us dearly]

A similar effect could be achieved in the UK, helping to slow the epidemic even if it could not be halted. Two days later the advice was firmed up. While the eventual number of infections would remain a constant, the shape of the epidemic curve could be altered, and that the most significant impacts would come if such measures were introduced early and in combination.

Yet there were doubts. When it came to specific interventions the benefits were assumed to be marginal but also hard to implement and then sustain. An early discussion of travel restrictions within the UK (13 February) worried that these would have to be “draconian and fully adhered to” for an effect. The benefits of closing schools or banning large gatherings were likely to be lost if the result was that those who might otherwise have attended a big football match went to the pub or schoolchildren congregated in a park. And the measures would have to last for weeks with wider social and economic effects. This is why the first preference was to put the onus on individuals to wash their hands, clean surfaces and keep their distance, with the government intervening largely to protect the elderly and vulnerable.

In their submissions the behavioural scientists were always more optimistic on public compliance so long as communication was explicit about what was required and for what reasons. Nonetheless, the approach adopted by Sage, at least until 13 March, was to take one step at a time.

As late as 5 March, the minutes note that: “Preventing all social interaction in public places, including restaurants and bars, would have an effect, but would be very difficult to implement.” 

There was little sense of how public attitudes could quickly shift, as anxiety generated demands that the government push on rather than hold back. Once modest steps were taken, with the aim of nudging the epidemic’s peak into the summer months, the logic of the situation was likely to lead to ever more stringent measures.

***

It was the real possibility that the NHS would be overwhelmed that set the move to a full lockdown in motion. Another major area of risk had been identified on 10 March, when Sage advised that “special policy consideration be given to care homes and various types of retirement communities”. Yet thereafter there are only fleeting mentions in the minutes of the plight of care home residents. On 31 March, the NHS was urged to look at the risk of transmission in care homes but the data was poor. The issue only began to be addressed as it became evident that the statistics had been missing the high rates of transmission and deaths in care homes.

This had major consequences not only for the death toll but the epidemic’s curve. By the end of 31 March, just over a week into full lockdown, there was reasonable confidence that this had immediately pushed the R number in the general community below 1, and possibly as low as 0.6. This continued to be the assessment. The problem was that Covid-19 was continuing to spread within hospitals. These “nosocomial” cases began to represent an increasing proportion of the total. Only on 14 April was it observed that, in this respect, care homes were a concern but the data was still poor. 

On 23 April, just before the major increase in testing capacity, “preliminary calculations” (this was far from a firm prediction) suggested that “the level of incidence could fall to ~4,000 cases a day by 4 May 4 and ~1,000 cases… by 11 May”. But at that meeting, it was also observed that “a small but significant proportion of deaths relate to deaths in care homes, rather than in hospitals”. A dedicated testing strategy was therefore required to reduce the spread in care homes. 

Charlotte Watts, the chief scientific adviser at the Department for International Development, was asked to lead a working group on the issue. This was not exactly her departmental responsibility, although her academic background was in epidemiology. To get the numbers of new cases down it became an increasing priority throughout May to reduce the levels of transmission within both the health and social care sectors. Until this was done, the disease could spread back into the community and the situation would be too precarious to start easing restrictions.

Discussion of how to get out of lockdown began in the weeks after it had been imposed. On 2 April, it was observed that no changes could be introduced without better data. At any rate, for the moment, “lifting measures too early could lead to a second wave of exponential growth of the epidemic”. Contact tracing would need to be added to testing. The search for contacts would have to begin as soon as an individual experienced Covid-19 symptoms – at least until a 24-hour test was widely available. The system would then only work if at least 80 per cent of contacts agreed to self isolate. 

On 4 May, a paper from the modelling group SPI-M-O did not see enormous risk in the early easing measures proposed, and which have now been introduced. It was much more cautious once “businesses involving close, sustained contact with many people” were opened. This could lead to “levels of infection as high as those seen in social care”. The discussion the next day at Sage noted that what mattered was not simply the level of R but also the incidence of cases, which was why the high levels of infection among health and social care workers needed to be addressed. Even with an effective test and trace programme, should there be a major relaxation of measures, R was likely to be pushed well above 1.

[See also: Why a true “exit strategy” from the coronavirus lockdown is not possible]

***

The transparency surrounding Sage’s deliberations is welcome. As can be the case when demands for transparency are met there is often a surprising lack of interest in exploring what has been disclosed. The published minutes can help us draw some conclusions about why the UK experience of the virus has been so painful. 

First they show that as the epidemic took hold the government was largely following Sage’s advice. It could and should have questioned the advice more, not least because other countries were acting after 9 March with greater urgency. But the “science” encouraged the UK to be relatively slow in imposing stringent measures. 

Second, that is not a sufficient explanation for the high level of fatalities in Britain. The problem was not so much that individual outbreaks were more severe than elsewhere – compare, for example, New York’s suffering to London’s – but because the outbreaks were much more uniformly spread across the country. It may be that much of this was the result of families returning from their half-term breaks and skiing holidays in Spain and Italy. 

Third, this would have been better understood at the time if the UK had been able to test at volume. Sage’s scientific approach was very data-driven, which would have been fine had the data been readily available and timely. Data suggests the UK is now at the fore in its testing programme but, until late April, it had only patchy information on the incidence and impact of Covid-19. 

Fourth, too little attention was given to what was going on in care homes where deaths now account for about a third of the total. The reason it is taking so long to see off the disease is that it established itself in care homes as well as hospitals. 

Lastly, there is still no clarity about acquired immunity. The only safe assumption is that the virus will lurk around, likely to return if we lower our guard. Until a vaccine is found, and if we want to live our lives more fully, we will have to learn to cope with the risks.

The final meeting of Sage for which minutes have been published was the 34th. It took place on 7 May, just after Neil Ferguson resigned from Sage because of a breach of social distancing rules. He had been a constant presence at the meetings since the first on 22 January. 

The 7 May meeting opened with members noting his important contribution to Sage’s work and the need to continue to draw on his team at Imperial College. This was followed by a discussion of the pressures on the group and how they might be mitigated. Too many commissions and requests for advice were coming in and these needed to be filtered so that members could concentrate on urgent scientific questions. There was a need for pastoral support. Ian Boyd, who had recently joined Sage to serve as “an independent challenge function”, offered to help on this matter. (He is chair of the Research Integrity Office.) 

So it was that a group of experts who had been advising the government on how best to look after the physical and mental health of the nation for four months during a pandemic were now worrying about their own. The minutes recorded the need to “help the resilience of participants of Sage who will continue to work under intense pressure on the Covid-19 response for many more months”.

Lawrence Freedman is emeritus professor of war studies at King’s College London

This article appears in the 12 June 2020 issue of the New Statesman, A world in revolt