How accurate is the UK coronavirus death toll? Our data expert answers your questions

Tom Dening, a professor of dementia research at the University of Nottingham, said: “It is worrying that there appears to be a sharp increase in deaths in the community that are not known to be due to Covid-19. So far, we don’t have good data on the possible reasons for this, but there are a number of possibilities.

“The first is simply that many of these are in fact caused by Covid-19 that wasn’t diagnosed. Testing remains extremely limited outside hospitals, so we probably won’t ever know how many people had the virus during this period.

“There are probably multiple reasons for other deaths. These include people not feeling able to attend their GP surgeries, call an ambulance or attend A&E as they may have done in the past. Therefore, some serious conditions may present too late for effective treatment.”

“Another possibility is that some people with serious conditions, like cancer or chronic kidney disease, are either unable or unwilling to attend hospital on the usual regular basis, so their treatment regimes may lapse.”

The new data follows the release of NHS figures last week which showed that, since the beginning of the outbreak, A&E attendance has dropped to its lowest level since modern records began.

That prompted health service leaders to implore members of the public with serious conditions to continue using the NHS.

Professor Spiegelhalter said coronavirus exaggerates existing medical risk, adding: “Mathematically, it’s roughly like the sort of risk we’d face over the whole year; it’s packing it into a few weeks while you’ve got the virus.

“It’s rapidly doubling your risk of dying this year if you get the virus. This increases massively with age.”

Below are questions on shopping from our readers that the Telegraph’s data journalists, Ashley Kirk and Dominic Gilbert, have answered. Send in your queries for upcoming Q&As to [email protected].

‘Do the statistics show that the lockdown is working?’

Terence Cooper asks: “Is the lockdown working, as evidenced by a reduction in hospitalised patients testing positive for Covid-19, or, the daily death rate in hospitals attributed to Covid-19?”

And we’ve got another great answer from Dominic:

Early indications are that lockdowns are beginning to work, and they appear to be following predictive models. When arguments were being made for the lockdown the consensus was that we would start to see a reduction in the number of new cases and deaths after around two weeks. 

In reality many countries have taken slightly longer than that, likely because the stringency of the conditions will be different, and depending on the time it took for each country to lock down. 

Italy’s cases and deaths peaked around two weeks after lockdown and are now in decline. In Spain, cases and deaths peaked around 20-21 days after lockdown, in part because they enforced quarantine later (43 days after their first case compared to 39 days in Italy). The consequence for Spain has been a more rapid increase in cases and deaths.

The UK appears to be tracking the same trajectory, but we locked down later – 53 days after our first case was detected. The government estimated we would hit our peak at Easter Bank Holiday weekend, and that would fit with the trajectories of other countries (Easter Monday was 21 days since lockdown was imposed). To date, we have had the highest number of new cases and deaths over the bank holiday weekend, and numbers have been falling since then. 

Those numbers do need to be treated with caution as there is an inherent lag in numbers from the last five days, and they will rise, but the early signs are that we are close to, or may have already passed, the peak.

‘How reliable are the statistics from China?’

We’ve got time for just two more questions. 

Our penultimate questions comes from Kamal Gilkes.

Kamal asks“How easy is it for us to verify the data and statistics coming out of China? If we were able to verify it to our satisfaction, their containment/control model might be even more attractive for consideration.”

Dominic says: The simple answer is we can’t verify the data coming out of China – each country has its own methodology and reporting systems, which are fed into their health ministries for their daily situation reports. There have been plenty of anecdotal reports from China with suspicions the death toll might be much higher than initially reported, but it is unlikely to be significantly different to the general trend we have seen in China. It would be counterproductive for them to begin lifting quarantine measures while the risks remain high, and according to the data we have the virus has been under control in China for more than a month.

‘What happens if you have pre-existing medical difficulties?’

A tricky question here from the comments section. 

Gabrielle Teare asks: “How do they record the deaths? So if you have a coexisting disease, say N stage cancer, and test for Covid-19 you are recorded as dying from coronavirus when you had cancer? Death solely caused by the virus is a massive difference in numbers?”

Dominic says: The recording method differs between organisations, which is one reason why there is an increase in the death toll when the ONS figures are released. 

When the Department for Health and Social Care release their daily death toll it is based on the number of people who have died in hospitals in the previous 24 hours who have had a positive test for Covid-19. The ONS data includes people who have died (inside and outside hospital) but have Covid-19 recorded on their death certificate by a doctor as a factor. Many of the deaths linked to Covid-19 have involved other underlying health conditions, so the two are not necessarily mutually exclusive. 

Often contracting coronavirus will simply exacerbate pre-existing conditions, so while a patient may have died from a chronic condition, they would not have died at that time had it not been for coronavirus.

‘What about recovery figures?’

We’ve had a good question come in via WhatsApp here. 

JC asks: ‘Why are we not reporting on ‘recovered’ numbers as other countries do?’

Ashley says: Part of the issue here is that many people who test positive are told to stay at home, which means that it’s impractical for the health authorities to confirm their recovery. Even though the UK is now only testing people in hospitals and NHS workers, not all of those who are tested positive need hospital treatment and recover at home. This means that the health authorities are unable to track and retest people to confirm a recovery, and so therefore the data is incomplete.

‘What is the margin for error?’

This one comes from the comments section at the bottom of this article (keep them coming in). 

Michael Jones says: “Another thing that needs to be factored into the figures is the margin for error. I heard on LBC this morning someone said the Covid-19 test was 70% effective. What does this efficacy mean and how has this changed over time? For example, are we missing 30 out of 100 cases with Covid-19 and how much worse was this at the start of the testing?”

Dominic says: Test efficacy is a difficult one to measure as there is no data on the number of people who were wrongly given a negative test, or vice versa. What we can say is that the detection rate in the UK is improving, with the caveat that the criteria is very limited – tests are generally exclusive to those who are at greatest risk or are already exhibiting symptoms. That said, the proportion of tests in the UK which come back positive is increasing, which may only be a symptom of the spread of the virus, rather than increased accuracy in testing. The percentage of people who are confirmed positive for coronavirus currently stands at around 30%. That’s up from between 11 and 15% in late March, and has steadily increased.

‘How are other countries recording deaths in care homes?’

An anonymous one here. A reader asks: “What are other countries doing in terms of recording Covid-19 deaths in Care Homes?”

Dominic: The UK is something of an outlier here. Many other countries are able to count the number of deaths in care homes linked to Covid-19 and are publishing them. If we look at the numbers coming out of other countries it can go some way to explaining the discrepancy between reports from UK care homes and the number showing up in our statistics. For example, France publishes statistics on the number of cases involving care home residents, how many of those have died in hospital, and how many have died at home. In their latest data from April 14, 5,600 care home residents have died at home and 1,499 at hospital. A total of 15,729 have died in total (inside and outside hospital). So care home residents account for around 45% of all deaths in France to date. In the UK, where we are not routinely testing in care homes, or accurately measuring the data, the ONS have estimated only around one in ten deaths occurred outside hospital, which is very likely an underestimate.

‘How can we put these figures into perspective?’

Telegraph reader Sacmar says: “Over 12,000 deaths is a serious number, but could we know what that figure is as a percentage of all UK deaths over the same period from other causes such as heart disease, cancer, flu etc.? Put it in some kind of perspective.”

And a very interesting question from Dominic here: 

The best measure we have for this so far is the number of excess deaths which have occurred in a given period, and which was most startling about the latest ONS release. Generally speaking, in week 14 of any given year there are an average of 10,305 deaths in England and Wales. In week 14 of 2020 that number was 16,387 – giving us an ‘excess deaths’ figure of more than 6,000 that week. 3,475 of those were linked to Covid-19 – so around one in five of all deaths. 

Weekly data isn’t published on deaths from heart disease or cancer, but the number of flu deaths for the same period this year was just over 2,300 – so only around two thirds of those involving coronavirus. 

What’s most concerning about the excess death figures are those which are not attributed to coronavirus (around 2,600). That’s a quarter of the number of deaths you would expect to see that week with no attribution yet.

That could be because those people did die from coronavirus but haven’t been tested, or it could be because people are dying from other factors at a greater rate because of the pressures caused by the pandemic. Either way it’s very worrying and the number we should all be watching to try and assess the real death toll going forward.

‘Why are figures not recorded by date of death?’

Another great question from WhatsApp here. Alpin M asks“Why are death figures not reported by date of death, which over time would give a much more useful trend?”

Dominic: The ONS release does attribute deaths according to date of death, which is the reason for the time lag. They work backwards and calculate all deaths which occurred by a certain date (in the latest release April 3), but were registered at a later date. This is why the more accurate numbers we get from the ONS are always retrospective.

‘How long is the delay from a death occurring to it appearing on ONS records?’

A very interesting question via WhatsApp here. Saiam asks“What is the lag time from a death occurring to when it appears on ONS records?”

Dominic says: It depends on the time the death was registered in most cases, but generally speaking there is a lag of around 11 days before the ONS is confident it has attributed deaths to the date of death. In the latest release around a third of deaths took more than a week to be registered. 

For deaths occurring before April 3 in England and Wales the ONS had recorded 4,122 by April 3. By April 11 that number had risen to 6,235. 

‘Why do we count people who have died with and from Covid-19?’

Our next question comes from Chris Hall via emailChris asks: 

“Why are the deaths of people with Covid-19 being included with those people who died from Covid-19? What effect does this have on the numbers?”

Here’s what Ashley has to say:

When the Department for Health and Social Care release their daily death toll it is based on the number of people who have died in hospitals in the previous 24 hours who have had a positive test for Covid-19. The ONS data includes people who have died but have Covid-19 recorded on their death certificate by a doctor as a factor. Of course, the hospital numbers could indeed inflate the numbers slightly if someone with multiple conditions as well as the coronavirus dies.

A similar counting issue has also been seen in Italy. Prof Walter Ricciardi, scientific adviser to Italy’s minister of health, told The Telegraph: “The way in which we code deaths in our country is very generous in the sense that all the people who die in hospitals with the coronavirus are deemed to be dying of the coronavirus. On re-evaluation by the National Institute of Health, only 12 per cent of death certificates have shown a direct causality from coronavirus, while 88 per cent of patients who have died have at least one pre-morbidity – many had two or three”.

‘Why does the UK not report how many people have recovered from Covid-19?’

Our next question comes from Damian Sutcliffe in the comments section. Damian asks: 

“Slightly off topic, but does anyone know why the UK is not reporting the number of people who have recovered from Covid-19 ?

“All other countries seem to be doing this except the UK and I think it’s an important number so we can understand how many active cases there still are.” 

Here’s Ashley’s answer:

Part of the issue here is that many people who test positive are told to stay at home, which means that it’s impractical for the health authorities to confirm their recovery. Even though the UK is now only testing people in hospitals and NHS workers, not all of those who are tested positive need hospital treatment and recover at home. This means that the health authorities are unable to track and retest people to confirm a recovery, and so therefore the data is incomplete.

‘How can you compare death rates from different countries?’

Carpe Jugulum has a question in the comments sectionThey ask:

“How can you compare death rates when there are no agreed criteria for determining whether a death is as a result of Covid-19?”

Here’s what Ashley has to say…

This is a very good question – and it is very hard to compare different countries’ death rates, for several reasons.

Firstly, different countries are testing their populations at different rates. Germany has twice as many coronavirus cases as the UK right now, and yet has a third of the deaths. This means that Germany has a lower case fatality rate than the UK – which, on the face of it, could lead you to say that they’re handling the crisis better than us. But experts say that, because Germany stepped up testing faster than the UK, they’ve recorded more of their overall cases and therefore their lower fatality rate is simply a product of their testing catching more cases.

Secondly, perhaps surprisingly, definitions of a ‘coronavirus death’ can differ across countries. For example, Italy classes every patient who dies while carrying the coronavirus as being a death caused by the virus. This could potentially inflate the death toll. Other countries count differently. Until you understand these potential inconsistencies, country comparisons shouldn’t be done.

But, at the end of the day, the data that we have is the best we’ve got. We’d be fools not to analyse it and compare different countries. So yes, we can compare them – but we have to be aware of the caveats. Whether that’s testing regimes or the way in which countries count deaths, there are plenty of things to consider before we can make claims based on the data.

How do coronavirus deaths compare to flu deaths?

Our next question comes from Jeremy via WhatsApp. Jeremy asks: 

“How do the current coronavirus deaths compare to the best and worst winter flu deaths?”

Here’s what Dominic has to say…

In terms of flu deaths we have now seen the number of deaths from coronavirus overtake those from flu or pneumonia for the first time in England and Wales – but we are also seeing above average deaths from flu for the time of year.

At this time of year generally around 2,000 people die each week from flu or pneumonia. In the week to April 3 that was over 2,300 – more consistent with February figures. That could be down to a number of reasons, including the pressure on the health service, coronavirus being present but not detected, or a reluctance from people to attend hospital. Last week we saw the lowest ever attendances to emergency departments, which is likely to be a trend across the health service over corona fears.

The last release from the ONS put deaths linked to coronavirus at 3,475. That’s closer to the average death toll from flu in early January, when it is at its peak. But we should expect those numbers to rise even further in the coming weeks.

‘Has there really been a 10 fold increase in care home deaths?’

Our next question comes from Phil via email:

“I have to wonder if there really has been a 10 fold increase in care home deaths in a week.

“We know that reporting of deaths has been delayed. Could this increase be just the catch up of previously unreported deaths?”

Here’s Dominic’s answer…

I think the better way of putting it is that there has been a tenfold increase in care home deaths that we know about. As mentioned elsewhere there is a lag of at least a week in registering many deaths linked to Covid-19, so in a sense we will always be playing catch up with the data. The hope is that as the government begin to measure care home deaths more accurately we will eventually get a better picture of what is happening.

‘What about the data for Week 14?’

Our next question comes from Sarah Anderssen in the comments. Sarah asks:

“Could you please comment on the data for Week 14 which shows that Covid-19 is mentioned on 3,475 death certificates, however the number of deaths which are attributed to the underlying cause of respiratory disease is 2,106?”

Dominic has the following to say…

There is an element of double counting here, because there is a substantial overlap between the two. The 3,475 figure is those deaths which had a mention of Covid-19 on the death certificate, but may not necessarily have been attributed as the main cause of death. 

So if a patient had died from pneumonia but had also tested positive for coronavirus, they would show up in both sets of statistics. Equally, if a patient dies from coronavirus but had no underlying respiratory condition, or had died from another cause altogether, they would show up in the Covid-19 figures only, because the virus would have been a contributory factor in their death.

‘What is the explanation behind Germany’s low death rate?’

Good afternoon! Our first question of the Q&A comes from Peter Whitehead in the comments section. Peter asks: 

“Does Germany test those who have died with symptoms but were not hospitalised or tested previously?

“I would rather see information about how Germany succeeds in having such low numbers dying. Covid-19 not an easily preventable disease yet strangely, even though many have tested positive in Germany, apparently not that many have died.”

Here’s what Dominic has to say…

Germany’s low death rate is very simply linked to the number of tests it carries out, and how early they introduced mass testing. 

They have been conducting around 350,000 tests a week since early March, and according to the latest numbers we have – they have tested nearly a million people (more than one person in every hundred). The UK has tested less than half that (0.58% of the population). Germany has a death rate of 42 per million, while the UK has a death rate of 182 per million.

By expanding the scope of the testing regime to the general population (Germany uses antigen tests with a very low criteria – anyone with even the mildest symptoms can be tested with a doctor’s referral) they are able to catch mild cases earlier and isolate those cases as appropriate.

Source Article