Although not cause for celebration, reaching the plateau is perhaps cause for a sombre degree of relief.
First published in April 2020.
While most British people support the lockdown, they will still be keen to know when the epidemic has reached its peak. Well, they don’t need to wait any longer – the answer is in. Data suggests that the UK is most likely over the peak.
Data released by NHS England, in which deaths are aggregated by the date of death rather than the date of reporting, shows a clear decline in recent days. While the figures are subject to constant revision, the numbers are starting to give us a coherent picture of the shape of the epidemic. Knowing that we have passed the peak is important because it shows that we can, with great effort and sacrifice, bring this disease under control.
Number of deaths reported in England, aggregated by date of death. / NHS England
The data (in the above figure) seems to suggest that the peak in the number of deaths may have been reached in the second week of April. But this reality has not always been clear in the numbers. It’s not a simple task to read and interpret the figures about cases of and deaths from COVID-19.
For a start, there are large numbers of different sources for the figures in the UK – the different branches of the NHS, government websites and the Office for National Statistics – all of whose figures differ slightly. Consequently, it’s not always clear which figures are the most reliable.
Perhaps the most obfuscating factor is the lag between people dying and their deaths being reported. In rare instances, this can be as long as a month, although the vast majority of deaths make their way into the government’s daily totals within a week (as the figure below shows).
How long does it take for deaths to be reported? / Kathryn Leeming, University of Warwick
Even when these daily numbers are reported by date of death (as in the NHS numbers in the top figure) there are reasons to doubt that they are a true reflection of the number of deaths. Official numbers for deaths caused by COVID-19 may be underestimates. Although most deaths from COVID-19 are in hospitals, the latest figures (from April 10) suggest that there have been more than 1,000 deaths in care homes and 500 more in the wider community.
There are also instances of over-reporting: people who died with coronavirus, not because of coronavirus. On March 22, it was reported that the UK’s youngest victim of coronavirus was an 18-year-old man. But it transpired that the young man had tested positive for coronavirus just the day before he died of unrelated underlying health conditions.
Given that the daily number of deaths peaked around April 8 (at around 800) there is good reason to believe that the number of cases peaked significantly before that – perhaps earlier in April or even in March. Yet this fact is not evident from simply looking at the number of reported cases.
In the UK, we have seen roughly constant new case numbers since cases topped out at 8,719 on April 12. But it’s important to note that the UK increased its testing capacity during this period. Constant levels of reported cases despite increased testing is a likely sign that the true number of cases in the community is coming down.
It would be useful to work out when the number of daily cases started to drop, so we can pinpoint which intervention measures had the biggest effect. Tim Harford on the BBC’s More or Less programme suggested that infections may have peaked three weeks (the average time to death) before the date of peak deaths, on March 18 – before the lockdown was introduced. If this were true, it would have serious ramifications for how we move forward in controlling the spread of the disease.
Unfortunately, back-calculating the date of peak infections from peak deaths isn’t as simple as subtracting the average time to death from the peak death date. The calculation requires more sophisticated mathematical modelling.
The language we use to describe the epidemic curve may also paint a misleading picture. For many, the word “peak” will bring to mind a sharp mountain crag, steep on the incline and steep on the decline. In reality, many countries have seen cases plateau for some time before decreasing. We should perhaps have Table Mountain in our mind’s eye rather than the Matterhorn.
We should perhaps have Table Mountain (L) in our mind’s eye rather than the Matterhorn (R). / Needpix
There’s also no reason to believe that the curve will be symmetrical. If Italy’s experience is anything to go by, it could take us far longer to bring cases down to a manageable level than it did for cases to increase in the first place. Every week that was spent allowing the disease to progress unchecked could cost us several weeks in lockdown in order to reduce the number of cases to a manageable level.
Still, being over the peak is indisputably positive news. It seems strange that studying a graph of death statistics can provide us with comfort, but such is the bizarre situation in which we find ourselves. It’s important to remember that each one of the deaths recorded on the chart above was a person with a family and friends who will be struggling to mourn them in these incredibly difficult circumstances. Although not cause for celebration, reaching the plateau is perhaps cause for a sombre degree of relief.🔷
Dr Christian Yates is the author of The Maths of Life and Death – which has a whole chapter about epidemics.