You asked
Please may you provide me with:
In April 2020 it was stated that your mortality figures now included estimated numbers of deaths, can you confirm, under this freedom of information request, what the total number of estimated deaths added was, up to the end of December 2020 and what was the total number of deaths removed from the figures after the actual death numbers were added?
We said
Thank you for your enquiry.
Our weekly mortality bulletin is published on a Tuesday and provides provisional counts of deaths registered in England and Wales in the week ending on the Friday, ten days before publication date. This number is provisional as it is derived from registration data processed by ONS by the following Wednesday, six days before publication. Registration data is not estimated and no records are added or removed through a statistical process.
There is a delay between a death occurring and being registered and so the weeks registration data will typically contain to 45% of records of deaths that occurred in the same week as registration, 39% that occurred in the week before and an ever decreasing proportion of records relating to deaths occurring in the weeks before. Though over a long period the average number of deaths registered per week will equal that average number of deaths occurring per week, in times of rapid change in occurrence rates the number of deaths registered per week will be different from the number of deaths occurring per week. This is a consequence of registration delay.
On 29 May 2020, we published a method to estimate the number of death occurrences that occurred in a given week based on the registration data available: https://www.ons.gov.uk/releases/rapidestimationofdeathoccurrences
This method has been used to provide estimates of weekly death occurrences that have been published in our weekly mortality bulletin since 3 November 2020, which provided provisional figures on the death registrations in England and Wales for 2020 week 43, which ended on 23 October 2020. This bulletin can be seen here:
In the main points of this bulletin (for 2020 week 43, which ended on 23 October 2020) there is the following paragraph "Based on a statistical model that allows for the time taken for deaths to be registered, we estimate that the number of deaths actually occurring (rather than registered) in Week 43 in England and Wales was between 9,750 and 12,097". The number of registrations that week was published in the bulletin at 10,739. However, just 4,836 of these registrations occurred in week 43. Using the method published on the 29 May 2020 above, we estimated that the number of deaths that occurred in week 43 was 10,809 (9,750-12,097). This estimate, and all subsequent, have been revised for the 26 following weeks and published with the weekly mortality bulletin as a dataset, the most recent of which is here:
The downloadable spreadsheet has a tab called "Estimated total deaths 2021" and it can be seen that the most recent revision of death occurrences for 2020 week 43 is 11,031 (10,959 to 11,120). This value is based on 10,654 registrations for deaths occurring 2020 week 43 that were registered by the end of 2021 week 9 (05 March 2021; 19 weeks and 19 revisions, all published, later).
In addition to this, we did publish comparisons of weekly death occurrences in England and Wales weekly between 14 April and 21 July 2020. This can be seen here:
This was subsumed into the weekly mortality bulletin as section 7. This provides counts of death occurrences by week that are registered by week; they are not estimates of total death occurrences and as such are not subject to statistical revision.