Just before the start of the firebreak lockdown, I wrote an article for Nation.Cymru explaining the expected path of cases and deaths during the lockdown. I’m pleased to say that my predictions turned out to be pretty accurate.
Now that we’re 11 days past the end of lockdown, we are just about in a position to look back and see if it worked.
And the answer, as government politicians are keen to say, is: yes, it did. It brought down cases; they’re right. But I fear that the message which came over at the end of the lockdown, which felt like “that’s it folks, back to normal life” (although it wasn’t quite that – although many clearly interpreted it as such) could end up undoing all of our good work faster that we’d anticipated.
So, let’s take a look at what happened. In the chart, I’ve labelled key dates:
- Lockdown starts (Friday 23rd October)
- Cases peak (Friday 30th October)
- Schools return (Tuesday 3rd November)
- Lockdown ends (Monday 9th November)
- Cases at their lowest (Wednesday 11h November)
The red lines on the graph show the daily figures. They are very erratic, so it’s important not to place too much emphasis on the figures from day to day. However, by taking a 7-day average, we can more clearly see the trend. Key dates 2 and 5 are based on that trend – the green line.
It took seven days from lockdown for cases to peak. From then, they fell fairly quickly and encouragingly, for 12 days. That’s four days short of the length of the lockdown. The fall bottomed out on Wednesday 11th November.
It’s inconceivable that the start of this ‘bottoming out’ was due to the end of the lockdown, on the Monday just two days earlier, so we have to look for other factors. There are two possibilities:
1. Irregularities in the data
We know these have happened, as was seen with the 444 cases recorded on Tuesday 10th November (this was explained by PHW as a delay in some of the test data). But these irregularities are largely evened out by taking an average. There have been a number days recently with what I would consider suspicious data – i.e. unlikely from a statistical point of view, so I don’t discount this possibility.
2. The return of schools on 3rd November
Schools returned eight days before the lowest point in cases. This corresponds almost exactly to the time from lockdown to the peak, and seems to me to be the most likely explanation for the date of the trough.
This leaves us, as I write, 11 days after the end of the lockdown, and just beginning to see the effect of the end of lockdown coming through into the data. Because a 7-day average line on a graph cannot include data points beyond the current day, it lags behind by three days. So we are currently only seeing the data to eight days after the end of lockdown in that average.
If I am right, and I really hope I am not, we might see a steeper increase emerging in the next few days.
The worst-case scenario is that the lockdown has bought us a month: two weeks of lockdown, followed by two weeks to return to where we were. In any case, I am not sure that the Welsh government’s aim of no more lockdowns before Christmas will be sustainable. With three local authorities back in double figures on Friday (Swansea, Cardiff and Rhondda Cynon Taf), it’s clear that there are still big problems, and these can’t be ignored without severely stressing our health service.
Of course, nothing like this had been tried before, and lessons can be learnt from this. Perhaps a “what if” would be useful here: “what if schools hadn’t re-opened until the end of the lockdown?”
To try to answer that question, we can look back and see what the rate of decrease was, around its maximum. The week of 2nd to 9th November was the period of greatest fall; from 1299.4 to 889.1 – a fall of 410.3, or 31.6% in 7 days, from which we can estimate that the first week of lockdown cut cases by around a third.
In a 6-day period (the amount we lost by schools going back – Monday 2nd, at least around here, was an Inset Day), that’s 27.1% (6/7 of 31.6%). So that could have seen us trough at a little under 600, instead of 806.3. If we’d had a three week, 23 day lockdown, with schools closed, cases might have been brought down by around 59% (to somewhere around 470). In any scenario though, the time to double before lockdown was around 16 days, so it still doesn’t buy a huge amount of time.
My own feeling, at the moment, is that without further measures before Christmas, the NHS could start struggling again. It’s still five weeks away. That’s two more doublings.
What we can’t be sure of is how fast cases will rise now, as the effect of the end of lockdown comes through in the figures in the next few days. If it’s low enough for an enhanced track and trace to cope with, that could buy us more time. And with an R not much above 1 (time to double over two weeks – compare March, when time to double was at worst less than three days), it feels to me that it wouldn’t take too much of a change to bring it under control.
But, given that you can’t legislate for a change in people’s attitudes, what that change is, I am not sure.
Angharad Shaw is a lecturer in Computer Science at Aberystwyth University and has a PhD in bioinformatics.