I have an enquiring mind. Which can be trouble. Tired of unsatisfactory reporting in the various media open to me, I decided to have a look myself, and came up with these. My data source was https://coronavirus.data.gov.uk/details/testing - you can go there and download your own data. NB - the historical statistics change as information is updated, so you need to update more than just "today's" data to get an accurate image.
The first tracks the daily health stats (hospitalisation, ICU admissions and deaths) against the number of positive tests reported. NOTE the red line is plotted against the right axis, all the others the left.
In red are the positive test numbers. As you no doubt are very aware, a lot of government and media communication has been based on how many positive tests have been reported. Setting aside arguments about how many false positives there are, and the accuracy of the tests being used; there is a clear message in the this graph. From about the beginning of September (when schools went back, and Universities started to go back) there has been a massive increase in positive tests, BUT the curves for admissions and ICU have matched it albeit less steeply. For me this means whilst the focus on raw positive test numbers was inappropriate - it did indicate an underlying increase in transmission that led to ill health. I think the language about tests means many infer that a positive test means someone is ill. That is clearly not the case.
The other thing you can see is that the hospitalisation, ICU and death statistics were tracking the spring numbers to some degree when tiering and then lockdown 2 came. It wasn't just positive test results that triggered government action.
For me the interesting data is whilst testing has grown in September and October, and grown a lot - the increase in positive tests has increased much than in proportion to testing. Ignoring false positives, this shows that the infection (as measured by these tests) has dramatically increased in those months.
I have few other conclusions I want to draw or investigate. But I'd add - I would to see false positives at least approximated in this data.
Posts
