Excess Deaths and the Daily Mail
On 20th November, the Daily Mail asserted: “fatalities aren’t any higher”.
This is false. The Daily Mail showed figures “adjusted for population growth” from a Twitter account. The change is in error — appearing to assume an implausible population growth in a year. Journalists should use official statistics for weekly deaths in their articles.
“Fatalities aren’t any higher”
As part of their article, Ross Clark (Daily Mail) wrote:
Despite what the fear-mongers would have you think, deaths are not far above average for this time of year as the graph above shows.
The accompanying graph is:
The caption is important:
Sources: ONS Weekly Deaths Report November 10. Upper/Lower Range ONS. Adjusted for Population Growth. Produced by: Statistics Guy
The amended series are not official statistics. The Office for National Statistics publishes weekly death registration statistics. These statistics — published each Tuesday — cover England and Wales. Instead, the article used figures “adjusted for population growth” from a Twitter account.
The graph is incomplete. The graph ends on week 44. That week ends on Friday 30th October. When the article went to its pixels, the ONS had already published the week 45 report. That report was on 17th November.
The adjustment is implausible. As Tom Phillips (Full Fact) highlights, the method for amending the death counts is unclear. We can look at the past five years to see the range of registrations in week 44:
- Year 2015: 9,618 registrations;
- Year 2016: 10,152;
- Year 2017: 9,984;
- Year 2018: 9,529;
- Year 2019: 10,164 registrations.
The implied changes range from 7% (from 2019) to 14% (from 2018). The population has not grown that much. In the last decade, annual UK population growth has been about 0.6% — 0.8% each year.
Making that change does not affect the story:
Such an amendment is also simplistic. The composition of the population matters when considering mortality.
Registrations and standardisation
The Office for National Statistics publishes weekly death registration statistics. Those figures are England and Wales. In Scotland, National Records Scotland has this duty. In Northern Ireland, that falls to the Northern Ireland Statistical Research Agency.
It is important to understand the number of death registrations in each week. In response to COVID-19, the ONS also conducts monthly mortality analysis.
- Calculating the age-specific mortality rates: deaths per 100,000 people in each age group.
- Weighting those age-specific mortality rates to the 2013 European standard population.
The mortality statistic accounts for both population size and composition across age groups.
It would be difficult to produce such figures on a weekly basis. The figures for this year depend on provisional mortality data and projected populations.
In the last four reporting weeks, deaths from all causes were above the greatest of the past five years. This does not accord with the paper’s continued assertion that deaths are “barely any higher”. Journalists and the public should trust official statistics for weekly death numbers.