Advertisement

Coronavirus UK death toll: why what we think we know is wrong

New figures reveal that what we think we know about the Covid-19 death toll in the UK is wrong. Here’s why.

Every day we get one big figure for deaths occurring in the UK. Everyone jumps on this number, taking it to be the latest toll. However NHS England figures – which currently make up the bulk of UK deaths – in fact reflect the day on which the death was reported, not the actual date of death, which is usually days, sometimes weeks, before it appears in the figures.

The truth is we don’t know how many deaths have taken place the previous day. In fact the headline figure is likely to under-report the number of deaths that actually happened the previous day.

The number we hear about usually counts deaths which took place at an earlier date. The difference matters because by undercounting the number of deaths we are skewing the curve.

Interactive

Prof Sheila Bird, formerly of the Medical Research Council’s biostatistics unit at Cambridge University, explains: “We’re on a rising epidemic trend, and so the death counts are currently increasing, and we’re trying to track how steeply they are increasing. If today I’m getting to know about a series of deaths that occurred in the past 10 days, then what I’m getting is not a reflection of the steepness of the curve at this moment.”

On 30 March, NHS England reported 159 deaths in the 24 hours to 5pm on Sunday 29 March. However, the actual number of people who died in that 24-hour period was revised up to 401 in Thursday’s report and again to 463 on Friday as more deaths which occurred on that date were reported. And this figure could be revised up again as more deaths come to light.

“When you’re on a rising trajectory, the reporting delay is likely to mean that you underestimate the steepness [of the curve] and so we may think that we’re doing better than we are. And when we come to the downturn in the epidemic, the slowing, and there’s a decrease in deaths, we’ll be too slow to recognise the change. Hence, we risk getting it wrong in both senses,” Bird adds.

Another complicating factor is that the Department of Health and Social Care’s daily count covers deaths in hospitals, omitting those in the community. Although the ONS this week started releasing the number of deaths including community deaths in England and Wales, there is also a time lag in this data being reported.

There are other datasets we can look at. The number of confirmed cases of the virus is a useful indicator but it relies on testing, which has not been rolled out to cover a broad enough swathe of the general population to give us a sense of how many people are possibly infected.

Interactive

The number of triage calls and online assessments through the NHS are also useful to give us a sense of potential infection levels – 1.9m at the time of writing in England. But these are people with Covid-19 symptoms, not those with confirmed cases of the virus.

The most solid data we have showing the trajectory of the impact of this virus are deaths. That is why it is imperative that we have timely and reliable data – and why the seriousness of the problem is growing along with the death toll.

“It’s not uncommon that this happens in a new epidemic,” Bird says. “Reporting delays are something to be managed, not to be ashamed of. You manage them down but you don’t want to do that by making people think they will be blamed for reporting late and therefore run the risk of failing to report. That’s the worst possibility.”