(1/9) It’s Wednesday, the sun is shining and it’s time for a short thread, with some more data supporting the hypothesis of COVID-19 resistance in the population as discussed in this previous thread. Let’s go. https://twitter.com/stevebrown2856/status/1280067312071122945?s=20

(2/9) Firstly, let’s look at the trend of regional deaths in Sweden. Sweden has lots of regions, so for simplicity let’s compare Stockholm with the rest of Sweden. Stockholm was hit at least as hard as London:

(3/9) Now let’s look at the trend by normalising against the peaks in Stockholm vs. the rest of Sweden. Well, what do you know, just like London vs. the rest of England, the trend in Stockholm starts to decline before the trend in the rest of the country:

(4/9) Reminder that Sweden has had no formal lockdown anywhere, its policy relies on voluntary social distancing and a ban on gatherings of over 50 people.

(5/9) Now let’s head back to the UK. Detailed time series of testing data are now available for individual local authorities, so we can replicate our previous analysis but at an even more granular level. Credit to @7thGalaxy2 and @Meadsjo for the inspiration on this one.

(6/9) With lots of local authorities, the trend graphs get a bit messy so let’s compare the number of positive cases on 8th April with the subsequent increase in cases to date, starting with Upper Tier Local Authorities:

(7/9) Now, that’s interesting. Any region that was hit hard and early (bottom right) didn’t see much of a further increase. One can almost see a trend in the maximum at around 7/1000 population, although testing capacity limits early on may be skewing the numbers.

(8/9) We can repeat the analysis at the Lower Tier Local Authority level, with the same result. The lack of outliers is particularly marked; the one high-ish value with perhaps 13 cases / 1000 population is Leicester where there has been a big recent random testing programme:

(9/9) So, yet again we find that regions that are hit hard also decline faster, or are unlikely to see rates rise above a certain threshold. What, other than population resistance, could explain this at such seemingly low attack rates?