July 1, 2020

expert reaction to latest nowcast and forecast and regional R estimates from the MRC Biostatistics Unit at the University of Cambridge

The MRC Biostatistics Unit, at the University of Cambridge, have released their latest nowcast and forecast and regional R value estimates for COVID-19 in the UK.

Dr Daniel Lawson, Lecturer in Statistical Science, University of Bristol, said:

“This model attempts to describe COVID-19 transmission ‘right now’, from data that is collected with a time lag.  The model has been previously tested to pandemic influenza and seems to fit the data.

“Nowcasting is an important problem, and is particularly difficult when interventions are changing.  This model does not directly model the subtleties of easing lockdown.  That means the nowcast will be systematically wrong if interventions, or people’s response to them, have changed since the last collection of the data.  These results should be considered with caution since the data lags 1-2 weeks behind and interventions are rapidly changing.  The real number of infections expected from one initial infection could be significantly higher than these estimates if contact rates have increased recently.

“Intervention efficacy in the UK cannot be precisely known until after they are tried, though we do at least know ‘now’ which interventions are in place.  Learning the effect of interventions on COVID-19 from other countries’ experiences is a critical ‘next step’.”

Prof Keith Neal, Emeritus Professor of the Epidemiology of Infectious Diseases, University of Nottingham, said:

“The model is estimating the number of new infections to be higher than more direct surveillance systems such as the ONS survey and the Zoe-Covid symptoms app.  The difference is now narrowing though.  Their estimate of the number of new cases also has a widespread range, the highest estimate is nearly four times larger than the lowest estimate of new infections.

“The data is again reported as ‘Midlands’ although this is two regions, East and West Midlands.  The spike in cases seen in Leicester would have been added into the figures for the combination of two regions.

“If the number of new infections in the Midlands is relatively low, the rest of this area excluding Leicester of about 9 million people must have few cases if the model is correct.

“The estimate of R becomes increasingly less accurate as the total number of cases falls.  This applies even more as R is estimated for smaller areas.”

https://www.mrc-bsu.cam.ac.uk/now-casting/

All our previous output on this subject can be seen at this weblink:

www.sciencemediacentre.org/tag/covid-19

Declared interests

None received.