April 8, 2020

expert reaction to the Institute for Health Metrics and Evaluation (IHME) COVID-19 projections for the UK

The Institute for Health Metrics and Evaluation (IHME) have been producing COVID-19 projections for many countries, including the UK.

Dr Simon Gubbins, Head of the Transmission Biology Group, The Pirbright Institute, said:

“The IHME model is a statistical model to estimate the number of deaths from COVID-19. The authors fit curves to the cumulative number of deaths so far and use these to project the number of deaths in future. The model does not describe the spread of SARS-CoV-2 (the virus that causes COVID-19) nor does it explicitly model the effect of social distancing on spread. It does attempt to allow for the effect of social distancing though, by assuming that its effect on the timing of the peak number of daily deaths is similar to that seen in Wuhan.

“The IHME and Imperial models were developed for different purposes. The IHME one is a statistical model that was developed to advise healthcare providers on the potential resources needed to deal with COVID-19, primarily in the USA. By contrast, the Imperial one is a complex transmission model that describes the spread of SARS-CoV-2 in the UK and was developed to assess the impact of different control measures, such as social distancing, on COVID-19 cases and deaths. Working out why the predictions are so different between the models is difficult, because the underlying models are so different. It could simply be that the IHME model is more cautious about the impact of social distancing on spread in the UK.

“As a more general point, the outputs of any model should not really be treated as a prediction of what is going to happen. Rather they represent plausible scenarios, based on knowledge at the time they were generated and assumptions made in the model (which should be as transparent as possible), that can be used to help inform decisions of policy makers.”

Prof Sir David Spiegelhalter, Chair, Winton Centre for Risk and Evidence Communication, University of Cambridge, said:

 “I am very sceptical of these IHME projections, which are based on assuming a rather simple mathematical model for the whole course of the epidemic.  I suspect they will change a lot as new data arrives – we shall see.”

Prof Sylvia Richardson, University of Cambridge & President Elect of the Royal Statistical Society & co-chair of the Royal Statistical Society Task Force on COVID-19, said:

“The IHME projections are based on very strong assumptions about the way the epidemic will progress, based mostly on using the experience in other countries to fit a smooth curve to the counts of deaths reported so far in the UK, rather than any modelling of the epidemic itself.  Methods like this are well known for being extremely sensitive, and are likely to change dramatically as new information comes in.

“Based on the counts of the deaths reported each day in the UK, they predict an eventual total of 66,314 deaths by August 4th, with an uncertainty interval of 55,022 to 79,995, and with a peak shortage of 23,745 ICU beds on April 17th.  These projections seem both extreme and too precise, especially allowing for the limitations in the daily reports.  Time will tell, but I feel these current claims should be treated with considerable scepticism.”

The IHME COVID-19 projections for the UK are posted on their webpage and regularly updated.

All our previous output on this subject can be seen at this weblink: www.sciencemediacentre.org/tag/covid-19

Declared interests

None received.