With stories circulating that some think we need a new national lock down because R may be over 1, we need to go over old ground on these inaccurate numbers. Sage updates us on R, a measure of how many people someone with CV 19 will infect, and on the growth rate in infections.
The latest SAGE Report says the R figure is now in the range 0.9 to 1.1, a 22% spread. The Report admits R “cannot be measured directly so there is always uncertainty”. They tell us different groups work R out in different ways. Some use hospital admissions and death rates data. This used to be the main way which I criticised in the past. They now concede this data may have a lag of 2-3 weeks in it. There are also the issues over how reliable the death rate figures are as some of the CV 19 ascribed deaths are people who had had the disease well before death and had other serious medical problems. Some use contact pattern surveys of people’s behaviour. This relies on people providing accurate returns, and leaves open big judgements about how it relates to the spread of the disease. The third identified system is the one that should produce more accurate results being based on the consistent and regular testing of a sample of the population. This should in particular give more accurate figures for growth or decline in the disease which would be a more useful figure than an estimated R.
SAGE blends the results from all these different methods , arguing they should draw on all of them as “there is uncertainty in all the data surveys so estimates can vary between different models”. You would have thought instead of this consensus blended approach they would identify the most accurate ways of calculating relevant figures and create consistent and accurate data to do so. They give us these ranges, and then add qualifications. They point out where the incidence is now small the data may be more unreliable.
Because they are combining results from a range of ways of computing R, all with their problems, they present it as a range. They assert that “The most likely true values are somewhere towards the middle of these ranges”. Why? What if the sample testing result was at one of the extremes? Shouldn’t this be taken more seriously as a better indicator of growth rates and therefore of R? They also stress local areas can have flare ups which are not representative of the surrounding region or local government area.
People deciding to lock down places and areas need clear and reliable data that there is a real problem with a surge in the virus and its spread. These generalised stories based on national R estimates are not the way to settle whether the economy can recover or whether we can have some of our lost freedoms back.