Six months....impossible

Originally posted by madamswer: ↑

Yes, @poyndexter, I have been crunching those numbers.

I have been keenly trying to obtain this sort of data, because it is the only way that we can get a feel for the true contagious nature of this virus, and how many unreported cases there might be (without which, we cannot determine the mortality rate).

So I am glad Worldometer did all the work.

Why the rate of testing is important is because, assuming it is sufficiently random and reaches a number of occurrences which is statistically representative, the cases it picks up it offers a good proxy for the real infection rates (which would included cases that would otherwise not have been reported).

Of course, as you rightly point out, how tests are targeted (and the time-weighting of the of tests conducted) are limiting parameters on the veracity of the data, but in the absence of other pointers, dividing number of cases divided by number of tests conducted, it is the best of bad indicators of true infection rates.

So it is worth analysing the numbers, albeit on a basis that requires considerable qualification.

And like so many things about this virus, the data - especially across countries - provides a lot of conflicting signals.

For example, taking the countries that account for 95% of the total number of detected cases globally, we see that the "infection rates" (recall: Number of Cases divided by Number of Tests) varies between countries to such a large extent - the range is 1% (Russia, if that datum can believed, or 2% for Australia, which is believable, I like to think) to 56% (Phillipines, again, an unreliable outlier, with only 50 tests conducted per million citizens, compared to the average of around 6,000 tests per citizen), and a 12% median - as to make the data almost meaningless.

View attachment 2087197

And its not as if there is any statistical bias to the number of tests conducted which might explain the wide variance between countries.

Because the number of cases detected per test is totally uncorrelated to number of tests conducted per capita.:

View attachment 2087206

Expand

I'm not sure how easy it is to model, or work through with excel, but the way I'm thinking about it is in terms of the R0 value. Clearly this is the crucial variable.

My inclination is that the R0 for an asymptomatic patient is likely to be low. Pre lockdown this might have been ~1.5, and post lockdown it's likely dropped to ~0.5.

The r0 value for a symptomatic mobile patient pre Covid concern might have been as high as 10 or even 15. This likely dropped to 5 once the world got wind of avoiding people with a fever/respiratory symptoms. This then dropped to maybe 2 with a rapid test and trace campaign*, and now down to less than 1, with these measures and a total lockdown. If we look at the time frame for these other measures, it helps explain how Aus 'flattened the curve' early. It isn't only the lockdown intervention which needs to be examined.

We also need to partly exclude/reduce the R0 value for the people who became known covid positives, either before arrival in Aus, or during their 14 day isolation.

*I think the introduction of 'trace and test' at a low initial number of positive cases, is the key reason Australia has kept the numbers low.