'It's not a bad flu season' - Covid myths debunked with data
FT data journalist John Burn-Murdoch examines the coronavirus stats, and explains why the UK's Covid-19 death rate is likely to rise further
Written by John Burn-Murdoch, produced by Tom Hannen
Transcript
You can enable subtitles (captions) in the video player
JOHN BURN-MURDOCH: If anyone says to you that the conditions we're seeing in England's hospitals at the moment are no different to what we get in a bad flu season, this video is what you should show them. What I've done is I've taken data on the numbers of people being admitted to ICUs in England with flu over the last few years in a typical winter. First of all, we're going to look at the year 2013-'14.
So on the vertical axis, you've got the number of people admitted with flu per week per million people. To put that into more context, England has about 56 million people as a whole in the country. So when the number is 1 person per million per week, that's about 56 people per week.
So let's start in the winter of 2013-'14, where the flu season peaked around about 1.5 people per million per week. If we then move to the following winter, it was a bit worse, and things got up to about 2.5 per million per week. The subsequent winter was worse still. 2016-'17 was a bit more mild. But then 2017-'18 was a relative record, one of the worst flu seasons we've seen in about 40 years, '18-'19 also bad, '19-'20 not quite as bad.
But what we've seen this winter, in 2020-'21 with COVID is a completely different picture. It's an order of magnitude different. Whereas the worst flu seasons in the past were peaking at about 6 people per million per week admitted to ICUs, what we've seen this winte-- and, of course, the numbers are still developing, still, in fact, rising-- we've seen a peak so far of about 17 people per million per week. And that translates into around 980 people admitted to ICUs in England per week with COVID.
To put that 980 new admissions into context, the total ICU capacity across all of England-- that's the number of ICU beds typically available, even in winter-- is about 3,500. So that's more than a quarter of all available ICU beds taken up just by new admissions in one week. Now, of course, not all patients admitted into ICU stay there. Some are discharged. Some unfortunately lose their lives.
But it just shows you quite how full and, in fact, overwhelmed England's ICUs have been this winter relative even to bad flu seasons in the recent past. Now, some people will look at that last chart and say, well, we've not been testing everyone, including asymptomatic people, for flu in the past in the same way that we're doing for COVID this winter. Now, there are a couple of things to say about that.
First of all, the vast majority of tests being done are on people with symptoms. So to say that the lack of asymptomatic testing in the past is different is not really true. We test for flu in normal years when people are ill. We test for COVID this year when people are ill.
But the second and perhaps more definitive response to that kind of criticism is we can look at the total numbers of people in England's ICU for any reason. Now, if the spike in that last chart for COVID is simply due to the fact that we're doing more testing for respiratory illnesses this year than in other years, and what we're really seeing is people who got admitted to ICU for other reasons-- perhaps a broken leg, perhaps a heart attack-- and then tested positive for COVID, then the total numbers of people in England's ICUs shouldn't be any different. It's just that COVID is spreading among people who are already in hospital.
So let's see if that is, in fact, the case. The chart here looks again over the winter months from November, December through to February, March. And we're looking here at the total numbers of people in ICU beds in London for any reason at all, doesn't matter whether they're in for flu, for COVID, broken ankle, anything like that.
If we look at 2013-'14, the same winter we started with last time, the numbers start rising as we come into winter. Now, that's not unexpected. But they then level off at around 650 people in ICU beds in London. And that stays roughly the same as winter plays out.
Now let's pan through the years again. So here's the following winter, '14-'15. Around 700 people in ICU beds in London all the way through, similar the year after that, similar in 2016-'17. Even in '17-'18, that year that was especially bad for flu, the numbers never got above 750 people in ICU beds in London for the whole winter, 2018-'19, again around 700, '19-'20 the same.
But here's 2020-2021. And once again, just as in the last chart, we see that COVID is completely different to what we've seen before. Instead of the typical number of about 750 people in ICU beds, that number has shot up above 1,000. And again, the line has only been growing steeper. When the data is updated, it's almost certain that this will have risen even further.
The peak of that red line for this winter is at 1,074 patients in critical care, in other words, ICU beds in London. That's a 40% increase on that record-bad flu season in 2017-'18. So this is, again, a completely different scenario that we're dealing with, putting doctors and hospitals under immense pressure.
And when you look at this, the only reason that doctors and hospitals are able to accommodate 1,074 patients in ICU beds when they typically have about 700 to 750 is they're moving beds from elsewhere, and they're moving staff from elsewhere to care for these patients. So the care that patients receive in this case really is a zero-sum game.
For every additional patient being treated in ICU for COVID, a patient elsewhere is receiving either a lower quantity of care or a lower quality of care. We now know that in many ICUs in London, one doctor or one nurse is having to deal with four patients in ICU. Whereas, typically, even in a bad winter, that number would be one or two patients.
So people are more stretched. Resources are more thin. And it's likely the outcomes may become worse. Now, of course, the outcome after you see a huge surge in the number of patients in hospitals, and especially in ICUs, is that unfortunately we also see the numbers of deaths start to climb. But there have been a lot of confusing numbers bouncing around online over the last couple of weeks about whether deaths this year are indeed higher than they normally would be in a typical winter.
One piece of information that some of those who don't think this winter is any different to normal have been using is data from the European Mortality Monitoring website, EuroMOMO. Now, what they'll do is they'll point to a chart that looks a bit like the one you see in front of you here, which is tracking whether or not deaths occurring in different countries in Europe are roughly normal for the time of year or above what we'd normally expect. They'll focus on this period right at the end of the chart, where it looks like the line is going downwards, suggesting that deaths are falling back towards roughly normal levels for this time of year.
The problem is the way this data is processed means that the lines always trend downwards at the end, because there is essentially a delay between when deaths occur and when they're registered. So if we look at the data for England, for example, if we visited this web page about a week ago, the data point for week 52, the last week of 2020, gave a value of 0.15. That means essentially deaths occurring in the last week of 2020 were no different to a normal year.
However, we come back to this page today a week later, and that value for England, for the same week, is now 5.98, almost 6. In other words, it's considerably higher than normal and in the range that EuroMOMO says means significant excess deaths. That's the same week. It's just that a week later, more of that data has been processed and the line has gone back up.
Once again, the final point of the line looks like it's tipping down. But we know from this experience, from past experience, that is a data processing issue. And once all that data is fed through, that point, like the one before it, will rise up significantly. That's just one of the ways that we can look at excess deaths and try to add some clarity to the data here.
But another way is to compare what we're seeing this winter, as we've been doing in those previous charts, with what we've seen in other bad flu seasons. So here's a chart showing the total numbers of people whose deaths are registered in England and Wales over the calendar year. And we can see this figure for the years through from 2012, '13, 2014, all the way through to the last couple of years, '17, '18, '19, and then here we have the data for 2020.
Now, that huge spike that you see in March, April and into May in England and Wales is the spike that we're all grimly familiar with. This was the huge excess that we saw in the spring, back when the virus was really out of control in England and Wales. And this peak was so high, in part, because until late March, we'd all be mixing freely with no concerns for how this might be affecting viral transmission. That led to thousands and thousands, in fact, tens of thousands of excess deaths.
What has been more contentious is whether or not the numbers we've been seeing in the last few weeks, in September into October, November and latterly December, have been especially abnormal for a winter. But the simple fact is they have. Just as we've seen ICU admissions far higher this winter than they would be even in a bad flu season, the same has been true of deaths.
So the winter of 2017-'18, as we mentioned earlier, was a record-bad flu season. And one of the statistics that is used to measure exactly how bad a flu season is this concept of excess winter deaths. Now, that terminology adds a lot of confusion when people are comparing the death tolls of different years because excess winter deaths and the general concept of excess deaths that we've been talking about this year are actually two completely different things.
Excess winter deaths are defined as the number of deaths in the four months from December, January, February and March compared to deaths in the four months immediately before that, so the autumn, and deaths in the four months after that, the next spring. So it's about saying how many more deaths did we see over one year in winter than in the autumn before and the spring after that period? It's a way of measuring the impact of seasonal winter viruses, like flu, and of freezing temperatures, which can themselves cause increased deaths.
But when we're measuring excess deaths during the COVID-19 outbreak, what we're doing is we're seeing how many more deaths there have been this winter compared to previous winters. So that measure is always going to give a smaller number because it's comparing against a winter baseline, where deaths are known to be high, rather than against a spring and autumn baseline, where deaths are known to be low, as is the case with excess winter deaths. So when people point to 2017-'18 and say, well, there were about 50,000 excess winter deaths in that year, but we've not seen 50,000 excess death this winter, they're actually comparing two numbers that cannot possibly be compared, two completely different methodologies.
And the closest we can actually get to a like-for-like comparison of excess mortality this winter versus 2017-'18 is to compare deaths from September to December versus the same months in 2017. And the result? The result is that autumn in winter 2020 has seen 15,000 more deaths than the same period in 2017-'18.
The final couple of problems we have when we try and deal with excess mortality this year relate to the fact that there are very, very long delays between the occurrence of a death and that death appearing in death registration data. So if we look at the very latest data that we have from the Office for National Statistics, that was published on January the 6th, and it refers to people whose death was registered in the week ending on Christmas day, the 25th of December. That's a lag of 12 days between when the death was registered and when it is published in the official data.
There's then another lag between when someone dies and when their death is registered. That's typically around five days in England and Wales. So the very latest data that we have on excess deaths in England and Wales refers to people who lost their lives around the 20th of December.
Now, we know that it was in the immediate lead up to Christmas in the UK that the virus was spreading at its fastest. It was from the 20th of December onwards that we saw case rates, positivity rates surging up higher and higher towards those April levels. And it's only been in the last two weeks that we've seen hospital and ICU admissions reaching and then, in fact, breaching those spring records.
So based on that, we would expect most of the rise in deaths from this winter wave of the virus to take place long after the 20th of December. So the fact that we've seen considerable excess mortality over the last few months is one thing. But we actually fully expect, and the data, I'm sure, will bear out the fact that we will now see an acceleration, a steepening, of that excess deaths curve, as the data from people who lost their lives after the 20th of December philtres through into the published data that we can actually use.
We've been tracking these statistics and many other metrics for COVID not just in the UK, but through dozens and dozens of countries around the world. And you can find all of our data at ft.com/coronavirus-latest.