A new paper “An international comparison of the second derivative of COVID-19 deaths after implementation of social distancing measures” has received some attention over the weekend.

Perhaps because it was published by individuals from the same university, the paper has attracted comparisons with the Imperial College London study that we reviewed in our first bulletin, with some commentators even claiming that the authors of that study have dramatically revised down their fatality estimates. It is important to note that the authors of this new paper are not part of the team that advised the UK Government. Our view is that the conclusions in the paper are not valid.

The authors, from the Department of Electrical and Electronic Engineering, took data on daily fatality rates in eight countries (China, Italy, Spain, France, USA, UK, Netherlands, Germany and South Korea). They noted that the trajectory of deaths in each country after social distancing and lockdown strategies had been put in place appeared to mirror that of China. They then used the shape of deaths in China (specifically the second derivative i.e. how fast the increase in deaths was slowing) to predict the pattern of deaths in the other countries.

Their modelling suggested that the total number of deaths from COVID-19 in the UK would be between 4,700 and 7,100, and that the maximum number of deaths in a single day in the UK would be around 260. Similar conclusions were also drawn for the other countries, as set out in the table.

Despite this paper being based on data up to 24^th March, various estimates can already be shown to be incorrect. For example, we have already seen 260 daily COVID-19 deaths, a figure that the team did not expect to see until their forecast peak on 5^th April.

Whilst it is not inherently a bad idea to consider how the trajectory of a country’s COVID-19 outbreak might develop, by comparing it to other countries whose outbreaks started at an earlier date and which have implemented comparable control strategies, our view is that it is irresponsible, and potentially dangerous (as it risks undermining social distancing measures) to draw strong conclusions from such a simple analysis.

In particular, the conclusions set out are based mainly on the pattern experienced by China. This is unlikely to be a sensible comparator for most of the countries listed, as they have not introduced control measures as strong as those introduced by the Chinese Government and have indicated no intention to do so.

Finally, it is unclear to us why Imperial’s Department for Electrical and Electronic Engineering has chosen to publish a paper on COVID-19.

References

Pike, W.T and Saini, V – An international comparison of the second derivative of COVID-19 deaths after implementation of social distancing measures https://www.medrxiv.org/content/10.1101/2020.03.25.20041475v1.full.pdf

This article is one of a series provided to the public (and published with the permission of) the Covid 19 Actuaries Response Group.

It does not take an actuary to work out that on the UK’s current actual trajectory, the projections in this report will be wrong. The lessons can now be learned and digested.

It is irresponsible, and potentially dangerous,  to draw strong conclusions from such a simple analysis

Matt Fletcher – the author