21 January 2021

Is the UK’s Covid-19 lockdown working?

Britain currently has the worst death rate in the world but the number of new cases each day appears to be falling.

One of the few positives of the year-long global fight against Covid-19 is that it is no longer an unknown enemy. Until the vaccines began to arrive, information was our main weapon against the virus. We all became familiar with R numbers and growth curves. Previous lockdowns and the introduction of tiers, with different levels of restrictions on our lives, gave us a chance to see in real time what was relatively effective in limiting numbers, and what was not.

One thing we thought we’d learned is that full national lockdown, while psychologically and economically damaging, was an effective way of curbing, then reducing, levels of infection. Then came this week’s grim headlines about Britain’s “record number of daily Covid-19 deaths”. Official data shows the number of Covid-19 patients in hospital and on ventilators remains far higher than during the April peak and remains on an upward trajectory.

Given that England is now more than two weeks into national lockdown, people could be forgiven for asking: has the virus spun out of control this time? Or is it just that it takes time for the effects of restrictions to filter through?

One way of approaching this is to compare the epidemiological data we are seeing now with those we saw after the first national lockdown in March. Then, hospital admissions for Covid-19 rose in England for around eight days after restrictions were introduced before falling at a steady rate. Coronavirus deaths rose for 17 days before doing the same. Given that Covid-19 can incubate for two weeks before becoming symptomatic, and that more serious cases can get progressively worse over a period of several days, those figures suggest the first lockdown was effective – and reasonably quickly.

How does the current situation compare? In terms of the deaths data, it is perhaps too early to say. When looking at the figures, epidemiologists (and data journalists) have tended to use a seven-day rolling average rather than the individual figures reported each day. This smooths out sudden dips and falls which can be caused by reporting delays and other statistical “noise”. Using this method we see that during the first lockdown deaths increased at a fairly even rate until 14 days into lockdown, when the curve flattened for three days, and then fell. This time, all we can say with confidence is that 12 days into lockdown the number of deaths is still rising. The next few days will be crucial.

The hospital data is roughly tracking the first lockdown, too – but there is a more immediate concern, which is that the absolute numbers are much higher this time. During the first lockdown, the number of new patients being admitted with Covid-19 peaked a full eight days after the restrictions were brought in. But the total number of people in hospital with the virus continued to rise for 11 days after that. Why? Because people with Covid who are sick enough to be in hospital tend to stay there for some time, so admissions, while falling, continued to outstrip discharges. This time around, new admissions with Covid-19 appeared to peak seven days after the January lockdown. Again, though, patient numbers have continued to rise – and from a higher base.

The one reason for apparent optimism has been that average daily case numbers, in terms of people testing positive for Covid-19, have started improving. They are down by 25 per cent compared to the day the lockdown came into force. But while in theory this should be followed by an improvement in deaths and hospitalisations, those haven’t yet arrived.

Case numbers are not a perfect proxy for the prevalence of the virus in the community as a whole, as they also reflect how many people (and which people) are being tested. If we stopped testing altogether, we’d record no cases. The virus would not have miraculously disappeared. Other methods can be used, such as polling-style sampling. The evidence here is not good: just today, an ongoing study by UCL and Ipsos Mori which sampled nearly 150,000 people between 6 January and 15 January found prevalence of Covid-19 was “very high”, with “no evidence of decline” after the lockdown.

Ultimately it is the numbers in hospital, and the numbers dying, that really matter. If it turns out the lockdown has been less effective this time, why might that be? The current restrictions are superficially similar to those introduced in March – for example they both include schools and universities, while the November lockdown did not. Yet if we look at mobility data tracking how much people move around, there are differences in behaviour. While overall movement has dropped, that drop is less dramatic than it was in March. That is particularly true of travel related to shopping and recreation. Work-based mobility is also higher than it was during the spring lockdown.

While the government has been quick to cite people allegedly flouting the rules, it is the vagueness of those rules which is more likely to blame. Telling people to work from home “unless they cannot” leaves substantial room for manoeuvre in terms of what employers expect, as well what employees choose to do. More shops are open this time, too. And while there is a lot of talk of “lockdown fatigue” and “lockdown scepticism”, a recent major survey by UCL found compliance rates with coronavirus rules had risen sharply in January.

The other major point of difference this time is that the new Covid-19 variant which ripped through the UK in December is estimated to be 40 to 70 per cent more transmissible than the original. So even a lockdown that was exactly as stringent, and well-observed, as the one in spring would likely be less effective in stemming the virus’ spread.

That brings us to another question – did the government leave it too late before the January lockdown? We know that the increase in cases in September was briefly curtailed in November, before rising again. We can see that the current surge started after the switch to the new, less restrictive tier system on 19 December. That was a deliberate choice by the government; and while it might have scaled down its plans to allow families to meet up for Christmas, non-essential shops were open and busy during that time.

How bad are things now, as a result? If we look again at the number of Covid-19 deaths per day, calculated as a rolling seven-day average to smooth out reporting delays, there have been more than 1,000 every day for the last six days. During the April outbreak, that number peaked at 942.

Some people have questioned whether these record death figures are really all they seem. It is important to understand what the daily “Covid-19 death figures” put out by the government represent: they count the number of people who died within 28 days of a positive Covid-19 test. If people haven’t had a test, but still died of the virus, they aren’t counted. If people tested positive but then were unfortunate to die from something else entirely within the next four weeks, they are.

What almost certainly follows from this is that we significantly undercounted Covid-related deaths during the first wave in spring. For a long time, Covid-19 testing was restricted to people seriously ill in hospital. Most of the people who died of the virus in care homes never had a test, and so didn’t count in these headline figures. This was proven by looking at total “excess deaths” during the peak Covid months: counting the number of people who died of all causes, and comparing that to the number of people who would normally die at that time of year. From this we know there were around 20,000 more deaths than expected in care homes in March and April.

Since then, Covid-19 testing has become far more widely available. So it is fair to question whether we can look at the headline figures (“deaths within 28 days of a positive test”) and conclude that more people are really dying of the virus than ever before. But it isn’t fair to put this forward as evidence that the current wave of Covid is somehow less serious. We underestimated the first wave; we aren’t overestimating this one. The fact the first wave was far worse than the official figures suggested doesn’t make many thousands of Covid-related deaths now any less terrible.

One could also wonder – given that testing is now so much more common – how many deaths within four weeks of a positive Covid test have nothing to do with the virus. Some elderly or vulnerable people might have asymptomatic Covid, get a positive test, die of something unrelated, and show up in the daily count. If there are enough of those, the official figures really might be inflating how bad the current wave really is. But that wouldn’t explain why our critical care beds are so full of patients.

We can also revert back to the excess death figures to take testing out of the equation. New data from the Office for National Statistics this week showed that in the first week of January there were 5,576 more deaths than would typically be expected, up from 3,566 in the last week of December and 1,463 the week before that. That is indeed lower than the worst week in April, when there were 11,854 more deaths than expected. But it is still a remarkable number which only covers the start of the current surge in deaths. Future weeks in January may well be worse.

Is the lockdown working? The number of new cases each day appears to be falling, and it is reasonable to look at the available data and think, rather than merely hope, that the daily death toll will soon follow.

But it is also reasonable to look at the global league table of Covid-19 deaths per million population over the last seven days, notice that the UK currently has the worst death rate in the world, and ask: did it have to be this way?