October 13, 2020

expert reaction to latest report nowcasting and forecasting COVID-19 from the MRC Biostatistics Unit at the University of Cambridge

A report, from the MRC Biostatistics Unit at the University of Cambridge, with the latest nowcasting and forecasting of COVID-19 in the UK.

Prof Sheila Bird, Formerly Programme Leader, MRC Biostatistics Unit, University of Cambridge, said:

“During August and September 2020, Birrell and de Angelis have made three important changes to their modelling.

“First, the modellers now count only deaths that occur within 60 days of an individual’s most recent SARS-CoV-2 positive test-date.  They note: “This definition reflects more realistically the burden of COVID-19”.

“Second, gradual reduction in fatality-rate of hospitalized COVID-19 patients is assumed from June 2020, estimated from the data (currently imprecisely) with age-group specific impacts on infection fatality rates.  Estimated reductions or increases will sharpen as the second wave evolves.

“Third use of serial seroprevalence data, assessed in blood donors via the Euroimmun test, was curtailed in early June because serial estimates decreased thereafter, presumably due to within-person waning of IgG antibody body prevalence when measured by Euroimmun testing.

“Today’s report, using updated methods is timely because England’s estimated daily median number of infections on 8 October was 43,000 (95% credible interval: 27,000 to 68,000), akin to the back-calculated median of 46,000 on 13 March 2020 and to last week’s estimate by the REACT-1 study.  By 23 March, pre-lockdown, the estimated median daily infection incidence had risen alarmingly to over 412,000 (95% credible interval: 338,000 to 542,000).  But, of course, in March 2020 we had severely limited swab-test and did not observe anything near even the symptomatic subset of over 100,000 daily infections.

“Currently, we have four salient advantages: a) substantially increased testing capacity, b) earlier reaction {4 days versus 10 days} to the same estimated infection incidence, c) longer double-time by a factor of at least two (if not three), and d) a Test ad Trace system that needs greater transparency about its infection control outcomes if it is to serve us well rather than fail us.

“Hence, England should not exceed an estimated daily incidence of 400,000 SARS-CoV-2 infections in wave 2.  By mid-November 2020, our report-card will be marked by this and other modelling studies for SPI-M.

“PS: The new methodology’s incidence estimates for 23 March 2020 are considerably higher than were previously estimated (147,000).”

https://www.mrc-bsu.cam.ac.uk/tackling-covid-19/nowcasting-and-forecasting-of-covid-19/ and https://joshuablake.github.io/public-RTM-reports/iframe.html

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

www.sciencemediacentre.org/tag/covid-19

Declared interests

Prof Sheila Bird: “SMB is formerly a programme leader at MRC Biostatistics Unit but is not involved in the work reported today by her former colleagues.”