What raised the alarm?
Over the last few weeks there have been several statements made by the Government which raise concerns over the accuracy of Public Health England’s COVID19 novel coronavirus mortality figures, released daily and reported widely in the press.
To briefly summarise those issues:
The Chief Medical Officer and Chief Scientific Officer have given rise in comments to concerns that COVID19 is not necessarily being recorded as the cause of death when it should be.
There is a well established recording delay in the reporting of deaths in Britain, last reported on in depth by the ONS in 2018.
After a very low increase in recorded deaths on the evening of March 25 2020, a number of comments were made – including on BBC Newsnight – which introduced a number of ambiguities, for example “the figures can’t be counted without consent of the families.”
At present it is highly unlikely the COVID19 mortality figures in the UK are accurate as it is highly likely the figures are being under-recorded.
This is highly likely to give the public at large a false impression of the risk faced.
We believe the Public Health England COVID19 mortality figures would be more accurate if they were in the range 700 to 2600, as opposed to the currently reported (and by acknowledgment flawed) 463.
Are There Red Flags in Official Data?
The Office for National Statistics provides comprehensive weekly death data, going back to 2010. This is publicly available.
We downloaded the data for the years 2010 to 2020 and found that the complete data available for the year to date covers the first 10 weeks.
Even on a cursory analysis, immediate patterns which tend to show that 2020 mortality data is being under-recorded are clearly identifiable.
This supports the lack of credibility of the COVID19 mortality statistics being released by Public Health England.
We looked first at the total weekly deaths over the ten week January to March period of every year from 2010 to 2020.
2020’s data appeared in the middle of the range, indicating a roughly average death rate during the time period. This seems peculiar, given global circumstances and the level of national emergency measures.
We compared the annual 10 week period deaths and found that the 2020 figures were not aligned with what would be a fairly normal increase trend. Applying a 5% error margin marker gave some further indication that the 2020 figures are currently likely to be incorrect.
We compared the ten week period from 2020 to the 11 year and 5 year average weekly deaths in order to ascertain a likely under-recording range.
With this view, a clear pattern of deviation becomes visible, commencing in the last week of January 2020.
A very rough guideline figure for under-recording in 2020 would be around 500 deaths per week. Under-recording would, of course, include delays in reporting.
The respiratory death data follows a similar pattern when viewed over the 11 year period, with 2020 appearing below average – which is illogical given the national circumstances.
The deviation from the expected increase trend is larger within the respiratory death category.
Within the respiratory death category, it is clear that the under-recording began a week earlier (3rd week of January), and the 2020 figures are below both the 11 year and 5 year averages.
We can establish a potential under-recording range in this category of deaths between 90 and 330 cases per week.
On this basis, we believe the Public Health England COVID19 mortality figures would be more accurate if they were in the range 700 to 2600, as opposed to the currently reported (and by acknowledgment flawed) 463.
The ONS data contains a break down of weekly deaths by age.
In this data two age groups show signs of statistical deviance from expected norms. The 75-84 age bracket shows signs of potential under-recording towards the boundaries of the 5% error margin threshold.
The data for those aged 85 and over shows a clear deviation from expected norms, indicating under-recording.
What does this mean?
In simple terms, it means the current mortality rates for COVID19 are inaccurate. There are obvious signs in the ONS data that something is abnormal.
This should, in normal circumstances, result in a correction in the near future as reporting delay periods run their course.
However, because there are other issues at play, including government direction on when a COVID19 related death should be recorded, it is unlikely even a correction will be accurate.
In practical terms:
The general public should not focus on or be encouraged to view the COVID19 mortality rates as an indicator of reducing risk.
Journalists should focus on scrutiny of recording practices rather than reporting the figures, as this may misinform the public and create a false impression of reducing risk.
The government should cease engaging in, encouraging, or permitting any practice which distorts the mortality rate, or misinforms or is likely to misinform the public as to the risk presented by COVID19.