Here’s something for a Sunday….
As I’ve been updating the numbers on my charts from Scot Gov data on Covid, I’ve noticed a really striking pattern developing where the positive numbers for one region follow closely the positive test numbers from another.
I find this quite puzzling. Obviously, in this approach taken here in graphing these measures below, I haven’t taken in to account the number of tests performed. These are just absolute numbers of positives and so they are a blunt measure in many ways. But it is interesting what we see nonetheless. I hope the government is following this too.
For reference here’s a map of the NHS regions:
NHS Greater Glasgow and Clyde population: 1.2 million
NHS Lanarkshire population: 655,000
NHS Ayrshire & Arran: 370,000
There is real similarity in the traces of GG&C, Lanarkshire and Ayrshire & Arran – the peaks happened more or less simultaneously in October and the same has been seen in January too. The traces are falling at the same rate for each region too.
Now, you might think this seems totally natural since these regions are neighbours of each other in Central/West Scotland. Maybe you would expect infection to spread in such a similar way in neighbouring regions. I agree that I would expect that. I still find the simultaneity of it all a little surprising and notable. Peaks at 24th to 26th October and then 6th – 8th January.
NHS Lothian population: 800,000
NHS Tayside population: 415,000
NHS Grampian population: 500,000
This trio 👇 is another set that I have noticed are behaving very similarly to one another. They look really different to GGC/Lanarkshire/A&A set above. There is a small local maximum at the end of October, but from the start of October to the end of December, the positive tests numbers are comparatively flat. On the 27th December all the traces take a leap and a peak is reached around 4th to 6th January. Then they all fall in parallel, faster than in GGC/Lanarkshire/A&A.
Lothian is, of course, a neighbour to Lanarkshire, but not to Tayside or Grampian. I am unclear as to why its trace would look so completely different to Lanarkshire’s (and it really does)… But Lothian, Tayside and Grampian do all sit along the East of Scotland and Lothian’s trace is quite similar to Borders and Fife (more later).
NHS Borders population: 115,000
NHS Dumfries and Galloway population: 150,000
NHS Highland population: 320,000
Looking at the next trio, Borders, D&G and Highland, they also look strikingly similar. These areas have shown astonishingly flat positive tests numbers from September right through to 27th December, again, where they all took their enormous leap to peaks between 4th and 8th January. They also now fall in parallel. These areas have very small populations and while Borders and D&G neighbour one another, Highland is in quite a different part of the country, obviously. Its peak in January is delayed by a couple of days compared to Borders and D&G. If you look at the dotted line traces, Highland and Dumfries & Galloway almost superimpose one another. This is strongly indicative of their being independent of true infections and more dependent on either testing strategy or methods or contamination or random positives. This seems particularly likely when the absolute numbers are so low, and because Highland has a population over twice as high as D&G. Overall, the spread of the population is probably also fairly similar overall in the two regions.
NHS Fife population: 370,000
NHS Forth Valley population: 300,000
Fife and Forth Valley are also neighbours. But they look more similar to one another than they do to other neighbours. These traces are also astonishingly well matched. Peaks on 6th to 8th January again, and huge leap starts in both on 27th December.
I have no explanations… only questions about these graphs! In the spring, we saw a rolling wave of infections and peaks across the country. Largely following a geographically intuitive path. We seem to have lost this in the autumn and winter and I find it very puzzling. I do not understand how infections can arrive in all areas simultaneously and that some regions follow one another so precisely, and might match one neighbour but not another.
We urgently, as has been the case since May at least, need to address the issue of false, or clinically irrelevant, positives. And if we did that, it might go some way to solving some of the mysteries of the phenomena seen in these trends. I do believe, of course, that there has to be some reflection of true infections spreading in these data.
I just wanted to share it all with you and see what answers come back from you clever people. I have some ideas, but no hard and fast answers at all. But the data never fail to surprise me when it comes to Covid.
More coming soon on positive *rates* by region.
Lots of love ❤
Stay sane 🧠 Stay strong 💪