When in 1964 Sir John Brotherston became Chief Medical Officer for Scotland he introduced two startling innovations.
First and foremost: he placed information at the heart of health service decision-making. Second, and crucially, he also insisted that interpretation of data must remain the province not of the politicians but of the statisticians and of community health specialists. Young recruits to the fledgling Research and Intelligence Unit were told “you must be the grit in the system, the system needs grit”. Government Ministers who attempted to reinterpret data for their own ends were given a very short shrift.
This approach could be called the Brotherston Principle. It was a potential cause of considerable tension. Although the immediate benefits in terms of improved planning and resource distribution were readily apparent, some of a more nervous disposition became aware that they were for the first time vulnerable to well-informed external criticism. These tensions have persisted and are only too apparent in the response to the Coronavirus.
Until very recently anyone setting out to criticise Government responses to the current pandemic was faced with a dilemma. At times of major crisis, some argue, criticism should be reserved until after the danger has passed. But a new consensus is emerging: to reduce future deaths it is vital that current and past failures should be exposed and urgently addressed. Critical review should of course extend beyond the UK Government to the devolved administrations not only to hold those with responsibility to account, but also to provide yardsticks for comparison.
Scottish and UK responses
It is instructive to examine responses in Scotland. Of course, there is not always consensus. Inevitably, there will have been some failures which will be exposed at some point. But up to now disagreements in Scotland have largely been about the details of policy implementation and/or decisions which are essentially political – for example, proposals to discontinue jury trials or effectively suspend freedom of information requests . But, by and large, overall objectives in Scotland do appear to have been driven by proper consideration and interpretation of the data. The Chief Medical Officer for Scotland* (see below) may be a single “expert” voice at the press conferences but there seems little doubt from what she says that she is speaking collectively for a team offering a variety of expertise. And, crucially, relationships between politicians and “experts” appear to be based on a necessary mutual respect entirely in accord with the Brotherston Principle.
If you would like a metaphor for the UK response to Coronavirus look no further than the daily press briefings in Westminster. Each press conference comprises one from a revolving cast of politicians and two from a revolving cast of “experts”, an ever changing plethora of expertise, knowledge and at times it must be said ignorance. To preserve social distancing, the podiums for the three participants are some metres apart. To reach these podiums it would be best if the two “experts” on the left and right wings entered first and were then followed by the politician moving to the central podium. But no. The politician must be seen leading his team and for that reason he enters first causing a brief bottleneck at the door during which time social distancing collapses. Sadly, the politician’s desire for pre-eminence is not always mirrored by willingness to assume responsibility.
Back of an envelope
On what basis should we assess the UK government’s response? The first thing to grasp is that development of the broad outlines of policy did not need to be based on complex models of epidemics. By early February anyone armed with the sophistication of the back of an envelope, a pen and a basic understanding of exponential growth, could have used data made available by other countries to make a fair fist of predicting what was about to happen here. But, as became only too apparent, any decisions, if they were taken at all, were not driven by the data. In a nutshell, “experts” were required to follow politicians to the podium.
Armed with our pen and envelope what do we make of experts’ response? Unfortunately, they do not escape entirely without criticism. In the middle of March, a paper was released by a team from Imperial College London and the London School of Hygiene and Tropical Medicine. This paper set out the likely outcome of mitigation and suppression strategies and evidenced the need for a suppression strategy where possible. Suppression – in which the objective is to reduce the infection rate to less than 1 additional case per infected person – was belatedly implemented by the Government on 23rd March. Authors of the paper rightly have been credited with effecting a change from the previous lukewarm mitigation approach.
At this stage it was suggested that assumptions used in previous unpublished models had been modified to take account of changed assumptions about mortality and intensive care rates for the disease. Quite why these earlier studies assumed lower mortality rates than those suggested by the WHO has never been fully explained. But even if there had been a good reason, the previous approach had to be deeply flawed. Where data is seriously incomplete it is inappropriate to select and present a best guess as the probable outcome. An appropriate response would have presented a range of possible scenarios, including those adopted by WHO. Scenarios based on plausible “worst-case“ assumptions should have been at the forefront. If a large asteroid is approaching Earth and there is a 1 in 3 chance of collision you do not base your planning on the 2 in 3 chance that the asteroid will miss.
We can’t assume that those working with the earlier models failed to offer Government a range of scenarios (that would suggest a greater obsession with the elegance of the mathematical model than with its interpretation). But, if they did, providing the Government with a range of options might have been their undoing – it would have given politicians the opportunity to choose the option they liked best. And the one they liked best was to do very little. This was in accord with the dangerous, callous and wrongheaded idea that letting the disease run rampant would quickly create herd immunity. At this point, on the Brotherston principle, someone advising Government should have had the courage and necessary expertise to stand their ground. Unfortunately, there is no public evidence of such a resolute posture although of course we do not know what went on in private. But we do know that the Government’s Chief Scientific Advisor stood at his subordinate podium and echoed the herd immunity proposal associated with minimalist intervention. Perhaps the Imperial College /London School paper in March should be regarded as the resolute response, its eventual release to a wider audience the necessary act of defiance.
Sequencing the response
So what did our expert use of pen and back of an envelope tell us in early February? Based on Chinese experience it was evident that without intervention each infected person would infect on average 2.4 others (now thought to be nearer 3). The death rate was perhaps more problematic but a good working assumption was between 1% and 3 % of known cases. It was also known that deaths occurred perhaps 5 to 10 days after diagnosis. At this point the envelope-wielding analyst would see that it was only a matter of time before the first entirely indigenous case was recorded in the UK, that the disease would be rampant within less than three weeks and that the number of deaths would rapidly rise. This was the time when our armchair envelope analyst would have cancelled an overseas holiday. But critically it was the time for Government to prepare plans for reorganising hospital provision, for obtaining ventilators and personal protection equipment for all caring staff and for organising mass testing. Doing nothing was a massive failure.
The first indigenous case was recorded on 28th February although by then there were undoubtedly many, many more cases in the community. As late as 14th March with 20 known deaths and 1140 confirmed cases (true figure probably well in excess of 20,000), the Government was still apparently committed to the herd immunity strategy. Then the new model was produced and a clamour of voices insisted that the Government’s chosen course could not be sustained. But two more weeks had been wasted over and above the four weeks of inaction in February. Planning at the beginning of February would have alleviated shortages of equipment. Action at the beginning of March would have saved lives.
Horizons: how to rebuild trust
How many deaths do we expect? Some estimates suggest over 500,000 deaths for no intervention, 250,000 deaths for a mitigation strategy, a “hope” for less than 20,000 deaths on the current suppression strategy and perhaps less than 5000 deaths if there had been an immediate intervention with mass testing. These are for the next 3 or 4 months and take no account of future surges in the still non-immune population.
Is this now so much water under the bridge? Unfortunately, no. The statistical message still seems subject to massage. Take the suggested ‘green shoots’ which went out with the figures published on Monday 30th March. The problem is Monday figures have been consistently lower than expected. A likely explanation is that some hospital trusts are not fully collating figures at a weekend. Talking about green shoots, however cautiously, may be intended to boost morale and demonstrate that a strategy is working, but it comes rapidly unstuck when next day’s figures are revealed. And to give a Scottish example, as late as 20th March the Edinburgh Festivals were still uncertain whether this year’s events would take place. In justification, it was said, “There’s science and modelling and graphs coming at us from every corner and they all tell us something different”. Actually they didn’t and don’t. The only uncertainty was coming from UK politicians’ interpretations. Confidence that the festivals will return to normal in 2021 may be premature.
When should we intensify or relax current restrictions using a stop go model? This must not be a purely political decision. It must be driven by data, informed by facts. Our last recourse to the back of the envelope tells us one immutable fact. When the number of cases and deaths has substantially declined, perhaps in June or July there will still be many millions of people susceptible to the virus and a disproportionately large number of these will be vulnerable. Deciding when and how to relax restrictions must be driven primarily by the data and their interpretation and relaxation should be accompanied by a mass testing capability which will both reduce NHS staff absences and allow sensitive monitoring of the incidence of the virus.
What now? While it is essential to understand what has gone wrong the focus must now be on how to rebuild trust through truthful communication and competent decision making. Rightly there’s much emphasis on the need for effective testing and distribution of essential protective equipment. But there is also a need to overcome a fundamental problem which runs much deeper. What the performance of the Government over the last three months tells us is that the relationship between it and its expert advisers has been dysfunctional. Too much time spent on grandiloquence and justifying flawed decisions. Too little on honest appraisal of mistakes and their rectification.
But responsibility for this fundamental failure does not rest only with the UK Government. The Brotherston Principle demands that Government fully respects its expert advisors, it also demands of those advisers that they draw on the best expertise, work together to achieve a best consensus, and speak out strongly if the Government chooses a course of action which flies in the face of reason or squanders precious resource of time or people on justifying failure.
The experts must be the grit in the system because the machine will not function without grit.
Featured image: Marco Verch, Professional Photographer and Speaker from Cologne CC BY 2.0
Secondary image of Rishi Sunak at testing lab in Leeds courtesy of HM Treasury via Flickr CC BY-NC-ND 2.0
*Dr Catherine Calderwood, CMO, resigned late on April 5 after she twice broke her own/Scottish Government advice/rules by visiting her second home in Earlsferry, Fife, some 45 miles from Edinburgh (with her family) within 60 hours. Her resignation had earlier been rejected by the First Minister who admitted a “terrible message” had been sent: see the Scottish Sun Guardian and the Herald
Further reading:
Germany’s devolved logic: Guardian
How Britain’s coronavirus strategy unravelled: Guardian
How the UK got coronavirus testing wrong Free to read Financial Times
German virus response shines unforgiving light on Britain: Financial Times
Fearchar says
Thank you.
Kirstie says
Hi James- a very pertinent analysis. Although “statistics and scientific method” was not my best subject, I found the UK government’s initial stance bizarre. Even if there was some merit in the herd immunity approach (although I couldn’t see it), how on earth would society, the NHS, and the economy cope with such enormous numbers of I’ll people?
Gordon says
Interesting piece James. Without taking on board the WHO advice to test,test,test there will be no full picture of extent. This needs to happen if the points you make can then be used to inform strategy.
Lisa says
Back of the envelope calculations based on the Chinese experience? You have more confidence in the Chinese Communist Party than I do! I am very concerned at the idea that UK and other European epidemiologists might be basing their modelling on Chinese data.
Alex. says
Interesting indeed. However a basic question must be that of how good is the data. Is the widely differing figures for each country indicative of very different success rates in coping with the virus or is it a data/statistical artefact? For example why does Iceland with a far higher infection rate than the UK have a dramatically lower death rate? Why should Sweden with a similar (albeit slightly lower) infection rate to the UK have a much lower death rate, despite a much less draconian lockdown approach? I have doubts about the quality of the data and perhaps even also as to what constitutes a Covid 19 death. There is also the problem that by relying on models the quality of the output will depend on the quality of the input. If, as is believed to be the case by some in the medical profession that this virus is seasonal then we could be looking at much cheerier news in the next few weeks.
One could also ask why we are seeing an upsurge in non Covid 19 deaths in Scotland – could this be a direct consequence of the lockdown with surgeries being close and much normal hospital work being abandoned? This does point to another issue for the politicians as too narrow a focus on Covid 19 risks failing to take into account longer term issues which are potentially developing. Not least is the economy where prolonged lockdown will have devestating long term effects, not least in our ability to afford a quality health service; I do worry from listening to the First Minister that the Scottish Government doesn’t have sufficiently wide focus and this could be to our long term detriment.
As regards Governments being guided by the experts in Scotland we have an interesting example. It is very clear by her actions that our Chief Medical Officer as was, the person that the FM considered to be the very best available, believed that there was no risk to anyone in travelling in a sealed environment ( her car) to a different property occupied by the same group of people.; on the face of it that seems an eminently reasonable position to take. However was that view ever expressed by her to the Government and if so was it overruled? Expressed or otherwise clearly the medical view was overridden by the political calculation. In either event clearly the gross hypocrisy of her position required instant dismissal, although rather inexplicably that did not happen.
Interesting times.
David Gow says
Thanks for a v thoughtful response…on the point you raise in para 2 this has indeed been reported and commented upon…sick folk and/or family are worried about overloading the NHS so are not going to A&E though Jason Leith and Calderwood;’s interim successor are urging them to do so here…DG (co-editor)