Thanks.
So... the exponent for the UK, France , Germany and the Netherlands was similar in October, so I was wrong to say that the doubling periods were different for those countries. An excellent lesson on how linear plots can easily mislead when it comes to exponential growth! But Belgium is clearly different, so something different was going on in both those cases.
We can look at the inflexion point, where steady or falling death rates change to increasing rates. In August and early September, the data is noisy. However, it looks like Germany and Britain both showed an uptick in early September, while increases in the Netherlands and Belgium were significantly later. France also has a surge in deaths around this time.
Given that deaths are a reflection of infection rates 4 weeks earlier, I think it's reasonable to conclude from this that infection rates were being influenced by local conditions (degree of mixing and restrictions) in August. There are differences between countries, as you'd expect. The timing of the uptick in the UK fatality rate is certainly consistent with Eat Out to Help Out (rise in infection rates in August).
However, all countries show an increase in that time frame, and that's unlikely to be a coincidence. Given that we know that many Covid cases were essentially imported into the UK by returning holidaymakers, I think it's likely that the same thing was happening on the continent - that would certainly account for the closeness of the timing. This was likely the main driver for the increases we saw at the end of August. Eat Out may well have accelerated the process, as it was a significant change in conditions between July and August, but ultimately is unlikely that have been the principal driver for the autumn surge.
Expanding the comparator countries I still see nothing distinctive about the UK pattern following the start of EOTHO - the pattern seems pretty consistent across the countries shown. No link to EOTHO is apparent.
Last edited: