PHE counting blunder means the gap between real and reported Covid deaths will only widen

Since the UK’s first confirmed Covid-19 fatality in early March the national daily death toll has been the terrifying barometer used to gauge the severity of the coronavirus crisis.

The figure has been central to the national narrative surrounding the virus, not to mention the public mood. It has been used to justify policy and has acted as a shorthand for the purpose of comparing the UK’s coronavirus “performance” with that of other countries.

Incredibly, the government has now admitted that the way that Public Health England – an executive arm of the Department of Health – has been counting Covid-19 deaths is flawed.

On July 17 the Health Secretary, Matt Hancock, ordered an “urgent review” into how the figure is produced after an article penned by two academics at Oxford University’s Centre for Evidence Based Medicine pointed out that the official death toll in England has been exaggerated. The government has paused UK-wide updates on coronavirus deaths until the issue is “resolved”.

The article’s authors, Professor Carl Heneghan from Oxford University and Professor Yoon Loke from the University of East Anglia, explain how the way in which Public Health England compiles “out of hospital” Covid-19 deaths is statistically unsound.

Currently, anyone with a lab-confirmed positive test for Covid-19 is added to a national database. Another database, the NHS central register of patients, holds a record of whether current and former patients are alive or dead. PHE arrives at its official death tally by cross-checking these databases. If someone who has tested positive for coronavirus has also died, then their death is recorded as a Covid-19 death.

The article’s title – “Why no-one can ever recover from Covid-19 in England – a statistical anomaly” – points to the absurdity of the premise underpinning PHE’s accounting. The authors claim:

“PHE’s definition of the daily death figures means that everyone who has ever had Covid at any time must die with Covid too. So, the Covid death toll in Britain up to July 2020 will eventually exceed 290k [total UK cases], if the follow-up of every test-positive patient is of long enough duration.”

If someone in England contracted and then recovered from coronavirus in March and was hit by a bus in July, they would be counted as a coronavirus death. The result is that, as far as the statistical record is concerned, nobody can recover from the disease.

The analysis helps to explain at least some of the discrepancy between England’s high mortality figures compared to the rest of the UK. In Scotland, Wales and Northern Ireland, there is a 28-day cut-off period after which, if someone who had tested positive for Covid-19 dies, they are not automatically classed as a Covid-19 death.

Loke told Reaction that the flawed accounting has meant that the official UK total of 45,318 coronavirus deaths is between 5% and 10% higher than it should be. While this discrepancy seems modest, the gap between the true number of deaths and the inflated figure will only widen.

“At the moment this isn’t too much of a problem”, Loke says. “But the percentage will continue to steadily increase across the coming months as more and more people who have suffered from Covid are released from hospital and enter the community where they may die from other causes.”

Loke also suggests that the flawed statistics are making it much more difficult to accurately gauge a more important figure: the daily change in coronavirus deaths.

“This is really important” he says. “The change in the daily figure tells you whether coronavirus cases are shooting up or coming down and whether we’re getting on top of it. This is crucial in assessing whether we can afford to relax the lockdown.”

“If you don’t get the number of daily deaths right, how do you know whether NHS treatments are having a benefit or not? Even if the NHS found a cure tomorrow, based on what PHE is using, you would never be able to measure it. You would still see Covid-19 deaths for years and years.”

In response to the latest revelations, Dr Susan Hopkins, Public Health England’s incident director, said:

“Although it may seem straightforward, there is no WHO agreed method of counting deaths from Covid-19. In England, we count all those that have died who had a positive Covid-19 test at any point, to ensure our data is as complete as possible.

“We must remember that this is a new and emerging infection and there is increasing evidence of long-term health problems for some of those affected. Whilst this knowledge is growing, now is the right time to review how deaths are calculated.”

Heneghan and Loke are not the first experts to point out fundamental flaws in the official methodology. As the retired pathologist, Dr John Lee, has been arguing since the beginning of lockdown, the death figure is being inflated in other ways.

From March 29, to count a death as a “Covid-19 death” it was enough for Covid-19 to be mentioned on the death certificate, whether the disease was directly responsible for the death or simply a background condition.

And while most care home providers are not medically trained, they are – astonishingly – the ones who have been asked to make judgement calls about whether a resident died from Covid-19 or not. The ONS has confirmed this statistical fudge, noting that these verdicts “may or may not correspond to a medical diagnosis or test result, or be reflected in the death certification.”

Yet from April 29, the official statistics have included these deaths in care homes which were only considered “likely” to be the result of Covid-19.

These accounting issues are not limited to the UK. Long before the PHE anomaly surfaced, experts had criticised cross-country comparisons of Covid-19 deaths because the way figures are compiled varies greatly from one nation to the next.

How, then, should we measure deaths in light of the shortcomings of the current model?

Although calculating coronavirus deaths is always going to require some guesswork, one metric often touted as the more sensible alternative is “excess deaths”, which measures the how far above the average weekly UK mortality rate is compared with a 5-year average. ONS data shows that the UK has recorded 65,000 excess deaths in the three months up to June 30 although the figure has now dipped below the 5-year average.

The difficulty with this metric is that it also includes deaths that attributable to any cause, including those that have been caused by delays to medical treatment due to lockdown, as well as undiagnosed coronavirus deaths.

To fix the flaw in PHE’s accounting, Heneghan and Loke have suggested “defining out of hospital Covid-related deaths as those that occurred within 21 days of a Covid-positive test result” bringing PHE’s accounting more in line with health services in the rest of the UK.

As Loke explains, “If you’re out in the community, chances are, once you catch Covid, you will either recover or die within about 21-28 days.”

However we choose to measure coronavirus deaths, the PHE blunder has highlighted the need to handle the data with extreme care. It might also prompt the government and public to consider whether relying on a single number to paint a complex picture is such a good idea after all.