In recent weeks, commentators and statisticians have started to focus on “excess mortality” or “excess deaths” to work out how many people have lost their lives as a result of the coronavirus.

This measure looks at how many more people are dying now – from all causes – compared to the same point in the previous five years on average.

The head of the ONS, Sir Ian Diamond, has described excess mortality as a “really important statistic”. It currently stands at 47,000 in the UK over the course of the pandemic.

At today’s Prime Minister’s Questions, new Labour leader Keir Starmer also raised the question of unexplained excess deaths in care homes.

We’re going to look at what “excess deaths” figures tell us about the coronavirus crisis – and their limits.

International comparisons

One thing that’s become clear during this crisis is that the way different countries count their coronavirus dead varies wildly.

As FactCheck reported last week, the Spanish government only includes people who tested positive for the virus in its official tally.

Here, the Office for National Statistics uses death certificates, which allows it to count people even if they weren’t tested for the virus – but its figure still depends on doctors establishing covid-19 as a cause of death.

Meanwhile in France, the national tallies do not account for people who died at home – which commentators have described as a “timebomb” (the government plans to add these to the figures in June).

Part of the reason the excess mortality metric is attractive is that it doesn’t rely on people having been tested for the virus, on doctors making their best assessment of the cause(s) of death, nor on the rules that different governments use to decide what counts as a coronavirus death and what doesn’t.

In theory, that makes it far better as a basis for international comparisons than official death tolls produced by governments.

Though we should remember that even excess mortality figures are not always entirely reliable.

For example, the French statistical agency, INSEE, produces estimates of excess mortality based on filings from the country’s many thousands of “communes” – districts that report local deaths to the central authorities each week.

Some communes still send their updates on paper – and others, including France’s second-largest city, Marseille, have suffered “technical problems” getting their data to INSEE at times during the pandemic. The agency describes its excess deaths figures as “very provisional”.

ONS estimates for the UK are also billed as “provisional”.

And even if reporting differences between countries were eliminated, the populations themselves are not always directly comparable: average age, population density, ethnic mix and movement across borders are all significant factors that affect overall death rates. We should keep these in mind when looking at other nations.

The true toll of the epidemic

The headline excess deaths figure does not differentiate between people who died because they caught the virus and “indirect deaths” from the epidemic.

In many ways, this gives it an advantage over those that focus solely on direct deaths.

Professor Sir David Spiegelhalter, Chair of the Winton Centre of Risk at the University of Cambridge, told FactCheck that excess mortality “is currently a reasonable measure of the total impact of the epidemic, both the virus and the measures taken against it”.

Professor Spiegelhalter told us: “I believe a ‘substantial proportion’ [of excess deaths] are indirect deaths caused by collateral damage of the massive disruption in the health system and people just not going to their doctor or hospitals.”

How many excess deaths are ‘indirect’?

Working out how many excess deaths are caused indirectly by the epidemic is important.

If direct excess deaths are increasing, it might suggest that tougher measures are needed to control the virus, or that existing policies should be maintained for longer.

Conversely, rising indirect deaths could indicate that strict measures need relaxing – or that authorities should take other steps to make sure people get medical treatment when they’re ill.

But it’s hard to put a reliable number on these indirect excess deaths.

In the latest week for which ONS data is available, 25 per cent of excess deaths in the UK did not mention coronavirus on the death certificate.

So is that the indirect deaths figure? Not necessarily.

As Professor Spiegelhalter tells FactCheck, we cannot be certain that all the deaths in that category definitely did not involve covid-19. It is possible that some are the result of doctors under-diagnosing coronavirus when writing death certificates.

But, as he points out, other experts say that under-diagnosis is increasingly unlikely to be the whole story.

Professor Carl Heneghan of the Centre for Evidence Based Medicine at the University of Oxford told journalists yesterday that the proportion of excess deaths that were not recorded as covid-related on a death certificate has remained roughly stable at around 25 per cent throughout the epidemic.

Professor Heneghan said that if there were significant under-diagnosis of coronavirus, we’d expect the proportion of deaths not attributed to covid-19 to decline over time.

In theory, doctors should become more confident in diagnosing the disease and putting it on the death certificate as the epidemic progresses and they learn more about the condition.

Yet that decline did not happen – the proportion of excess deaths not attributed to covid-19 has remained pretty much constant throughout.

This could indicate that doctors have been getting it about right all along and that the 25 per cent figure is a reasonable approximation of the indirect deaths caused by the epidemic. But we simply don’t know for certain either way.

The ONS says it is “continuing to investigate the number of non-covid-19 related deaths and will publish detailed analysis on this in the future.”

FactCheck verdict

“Excess mortality” tells us how many more people are dying than would be expected at this time of year, based on the last five years of mortality data.

It’s a helpful metric that allows us to make comparisons between countries more easily and may avoid some of the pitfalls of relying on official covid-19 tallies.

One feature of excess mortality is that it includes deaths from all causes – including people who didn’t die directly as a result of the virus itself, but who died of other things because they didn’t seek, or couldn’t get, medical treatment.

For this reason, excess mortality figures can give us a fuller picture of the true impact of the epidemic.

Nevertheless, it’s important to differentiate between the direct and indirect excess deaths from covid-19 – because those figures have very different implications for what policymakers do next.

It’s hard to say exactly how many excess deaths were caused indirectly from the virus in the UK, but we know that about 25 per cent of excess deaths registered here did not mention covid-19 on the death certificate. It’s possible that some of these people did die from the virus, but that doctors didn’t diagnose it properly – though experts consider that less and less likely as more data comes in.

The Office for National Statistics says it will publish further analysis on these non-coronavirus excess deaths in the coming weeks.