August 7, 2020

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

The MRC Biostatistics Unit at the University of Cambridge have released their latest nowcasting and forecasting for COVID-19.

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

“The Cambridge team warns that their reported estimates – essentially by back-calculation from COVID-19 hospitalisations and deaths – for the number of new infections daily are relatively insensitive to changes in transmission of SARS-CoV-2 in the past two weeks.

“Hence, this week’s report, which estimates 3210 new infections daily (credible interval: 1710 – 5820) in England, does not differ much from the previous week’s central estimate.

“What is new is the second caveat on interpretation, which is absent from previous reports:

“The central estimates for incidence in each region are less than 800 infections per day, a substantial proportion of which will be asymptomatic (my italics).”

“What does “substantial proportion” mean?  Guidance from the authors is lacking but a range from 30% to 70% asymptomatic covers all bases.

“At 30% asymptomatic (approximating assumption in modelling studies in March 2020), we’d expect 2250 new symptomatic infections daily; and 15700 weekly (credible interval: 8400 to 28500).  In the week of 23-29 July, 4966 persons in England were swab-test positive, of whom 4642 were transferred to Test & Trace, 30% only of the weekly central estimate for new symptomatic infections and 55% of the weekly lower bound.

“At 70% asymptomatic (approximating the estimate by REACT-1 surveillance), we’d expect 960 new symptomatic infections daily; and 6700 weekly (credible interval: 3600 to 12200).  The 4642 symptomatic individuals transferred to Test & Trace in the week of 23-29 July would thereby represent 70% of the weekly central estimate for new symptomatic infections and overshoots the lower bound by nearly 30%.

“I conclude that “substantial proportion” is unlikely to mean that as many as 70% of new infections are asymptomatic.  “Substantial proportion” needs to be considerably higher than 30% of new infections being asymptomatic for Test & Trace not to be deeply mired.

“Even at 50% asymptomatic, we’d expect 11200 weekly new symptomatic infections (credible interval: 6000 to 20400), the recent performance of Test & Trace would be far from world-beating unless under 50% delivery on central estimate (and under 80% delivery on lower bound) is the performance of duty that England expects, and accepts, in pandemic 2020.

“And so, what is meant by “substantial proportion”? The proportion may be age-dependent. If so, we need to know . . . and Test & Trace is where the answer should come from!

“If the proportion is not age-dependent but indeed is so high that 50% of new infections  are asymptomatic, then Test & Trace must surely now change tack: adopt the recommendations by the Royal Statistical Society’s COVID-19 Taskforce on assessing not only how many developed symptoms and tested positive during or soon after quarantine but also the extent and correlates of asymptomatic infection in T&T’s two quarantined high-risk groups (and their adherence to “stay home”). T&T’s two high-risk groups are:

1. Members of the household of symptomatic cases

2. External close contacts of symptomatic cases.”

https://www.mrc-bsu.cam.ac.uk/tackling-covid-19/nowcasting-and-forecasting-of-covid-19/

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

Prof Sheila Bird: “I am not involved in the work discussed but it is led by my former colleagues at MRC Biostatistics Unit. I am a member of RSS’s COVID-19 Taskforce which made three key recommendations on gleaning intelligence from Test, Trace & Isolate.”