Originally posted by Cockney Sparrow
View Post
As before here are some notes in case there are people who aren't conversant with the differences:
ONS
The ONS figures are by date of death. They go up to the date of death 17th July and include deaths registered by 25th July. There can sometimes be delays of many weeks in the deaths being registered and there will be some future adjustments for deaths registered after 25th July where the date of death is up to the 17th July.
They include all deaths where Covid-19 is mentioned on the death certificate, whether or not there has been a Covid-19 test, and are regarded as the most accurate source, but with a time lag
NHS England
The NHS England figures are by date of death for those dying in hospital of CV19, whether confirmed by test or not (approx 10% of Covid-19 deaths in hospitals since late April have not had a positive test result). From what I gather the figures are considered reliable.
They reflect the prevalence of the virus in the general community.
The difference between the ONS and NHS England figures is predominantly care home deaths.
PHE/DHSC
These are the figures that have in the past been announced by the government on a daily basis and published on the gov.uk website.
The deaths are only available by date reported and are intended to include deaths in all settings where the deceased has been tested positive for CV19. They do not include instances where Covid-19 is mentioned on the death certificate but where there has been no positive test.
Publication of the figures has been suspended pending the outcome of a review into their accuracy and suitability for purpose, however they are still available in one of the downloadable data files.
It is puzzling why they are significantly higher than the ONS figures.
I am no expert at all, so I might be completely wrong but some possible reasons might be:
- They will include some deaths where there has been a positive CV19 test but where the death was caused by unrelated events (falling under a bus). However if you do the maths these will only be a small contribution to the difference [Edit] unless the percentage of those previously tested who die from other causes is higher than average by an order of magnitude, which might be the case.
- Timing differences: the deaths will appear in the PHW/DHSC figures significantly earlier than they will in the ONS figures (i.e. it can take some time for a death to be reflected in the ONS figures). Once again this doesn't account for the differences.
- Problems in methodology. The drive through centres (and probably the posted kits) were set up in England with the patient only having to give their full name, date of birth and post code. The idea was that the data would be integrated into the NHS records somehow, some time later. This has reportedly been difficult to do and has caused problems. The Wales and Scotland both refused to accept the system proposed for England and insisted on one which made it simpler to integrate the data into the health records.
One element of the PHE system to collate the number of deaths is to take the details of all the recorded deaths and run them against a database of people who had previously tested positive for Covid-19. (See above for possible difficulties.) There is then an additional stage to eliminate duplicate deaths (e.g. where a death had also been recorded in the NHS England Hospital data). All this seems, with the best will in the world, susceptible to things going wrong.
Nevertheless, even though the absolute level of deaths reported in the PHE/DHSC figures is puzzling some scientists believe that the data is still useful in the early identification of trends.
Comment