Other things on this site...

[Blog archives] [Categories]

The British and Dutch COVID figures are both a bit off

The Dutch and British COVID figures are both a bit misleading at the moment, but for different reasons.

Both countries have had an overwhelm of their testing services. The UK's PCR testing (in December) was overloaded, getting so many cases that their results were getting delayed. But there's an extra reason now...

In January the UK Government decided that, because the rate of positive cases was so high nationally, they would no longer require someone with a positive LF test to get an official PCR test. The theoretical justification is plausible: when the rates are so high, the informational benefit you get from the expensive and labour-intensive PCR test becomes less, and the LF tests might be considered "good enough" for official stats. But - and here's where the data go strange - as soon as this announcement was made, I thought to myself, "The case numbers will go down, because instead of officially reporting their positive test-at-home test, people will put themselves in quarantine and not bother reporting it."

I think we can see this in the data. The official UK Government test statistics show a notable drop from the December peak to January, down to about 50% of what they were.

This might be a true drop-off, or it might be a drop-off in reporting, or a mixture of the two. So let's compare it to data that isn't affected by the rule change. First, the Zoe covid symptoms study data -- these show a fall from the December peak but not much of a fall, down to maybe 80%. Second, the ONS infection survey, where the fall is down to maybe 75%.

Making a rough engineer's approximation, then, I would offer a simple guess that about half of the drop in the case stats is due to an actual drop, and half of it is due to a fall-off in self-reporting. Or in other words, the official case rates since the rule-change (mid-January) can be estimated to have about a third of their cases missing. You should remember this when looking at any graphs comparing the situation internationally, because these incomplete figures are the ones that are shown.

UPDATE: Prof Christna Pagel also points this out and she's an actual expert. She stresses that the ONS survey is excellent, and should be the main source from now on for checking the UK's case rates.

The Dutch system has also been a bit overwhelmed by the sheer numbers, in January. But more recently, there was a data processing issue that meant approx 76,000 positive tests (in late Jan/early Feb) were not included in the stats. Those have now been "found" and incorporated into the official figures. It'll take a little bit of time before we can look at the graphs and sensibly see what's going on, I guess.

Overall, I'm trying to judge the situation here in NL and also back home in the UK. We had to do that in December while deciding to cancel our Christmas plans. Well - right now I've no idea who's doing better or worse, in numbers terms at least. Both countries' governments are starting to roll back restrictions, more rapidly than the scientific advisers are advising - but one of the countries has (now!) a stable government, while the other has a crisis of leadership at the top...

I should make clear that I'm not a medical expert. It's useful to blog these things sometimes, though - not least so that when I look back at all this, I'll be able to reflect on how right or wrong were my amateur's/engineer's reckonings. I've definitely mis-predicted peaks and troughs of the waves, so far, and I bet you have too...

| health |