(1/9) A short thread on recent case trends in the UK and the impact of "Track and Trace" hotspot testing. Firstly, cases (by date reported) *have* gently increased since 1st July:

(2/9) If we look at cases by date of sample (for England only, in this case), we also see a rise, although this is more gentle, suggesting that there is some backlogging of cases in the daily reports:

(3/9) But that doesn't tell the full story, because testing rates have also increased; average tests / 7 days are up by 31% since 1st July, and we can see that the daily fluctuations in cases match daily fluctuations in tests performed:

(4/9) In this scenario, it makes sense to look at the *percentage of tests that are positive* as a potential warning sign. Let's add that to the chart:

(5/9) So, test positivity is broadly flat over the month. That suggests that the overall prevalence of infections is also remaining flat. So, what's driving these case numbers? Let's look at Pillar 1 (testing in hospitals and healthcare) vs Pillar 2 (testing in the community):

(6/9) We can clearly see that P1 cases are declining but P2 cases are increasing, so we're finding more cases in the community. Where? The graph below shows regional trends by date of sample:

(7/9) It may be observed that cases are increasing in specific regions, rather than overall. But don't forget that testing has also increased, so let's look at regional trends by sample date *normalised by testing rate*:

(8/9) Although cases in the North West have definitely increased (as we know), and perhaps also in the West Midlands, all other regions are either flat or declining in "real terms". The overall picture in England shows a decline over the month:

(9/9) So, in summary, more cases are being detected in the community by local hotspot testing. That is a Good Thing because it means the policy of targeted testing and restrictions is working, although it will tend to show more cases in the short term. Ends.

Addendum 1: Don't forget that PCR tests have a non-zero false positive rate, which will mean we will always find some "cases" if we test enough. ONS data suggests the false positive data (for the tests they use) is *very* low, perhaps 0.04% at the most.

Addendum 2: It is also worth bearing in mind that the testing data may include multiple tests on the same person, and may also include positives from people who are not contagious.