Gov intends to monitor & implement coronavirus measures using R & the number of infections. Testing is unlikely to cover everyone with symptoms for a while yet, so how can we be confident of these two numbers—the core decision making tools in our pandemic response—being accurate?

If you are using 'science' to make-evidence based decisions, you need to ensure that you get at least two basics right: 1) the calculations and 2) the data you use.

We cannot be fully certain of R because we do not yet have capacity for universal coronavirus testing for all of those in the general public with symptoms. For the same reason, we cannot be sure of the number of infections.

I expect the government will be doing clever data modeling of the current prevalence, incidence and testing coverage data to build an internal government estimated picture of what the situation would look like nationally if everyone could be tested.

The trouble is this will always just be an estimate until you can test everyone. I wonder what % of people in general public you would need to be testing for data modeling estimations to be accurate within stastical significance? We will probably only know this retrospectively.

And what is the actual equation combining R and number of infections? E.g. will number of infections be a decimal, where highest point at peak so far equals 1, below eauals 0.X, above equals 1.X. When combined with R, what will the criteria be for each level 1 to 5?

There will be criteria for each level, but this needs to be published. The Gov say their interventions are evidence-based. Until the formula calculations, criteria & rationale are published, surely the scientific community can surely not accept this as evidence-based in practice.

#EpiTwitter #GlobalHealth @drvgallo @QM_GlobalHealth colleagues — I'd be intetested to hear your thoughts...

End of thread.