Changing measures

Why were “almost all reported deaths” before April 29th in hospital?

On the UK’s daily reported measure of COVID-19 deaths, I wrote:

Before 29th April, almost all reported deaths were in hospital.

In this article, I delve deeper into that statement.

When in doubt, go to the source.

Reporting differences within countries

International comparisons are challenging. The measure sounds simple: how many people die with COVID-19?

Countries count this measure in different ways. Differences exist within countries too.

The Department of Health and Social Care figure comprises the four nations. ‘Tests’ refer to tests for the SARS-CoV-2 virus.

Confirmed deaths must have a positive result from a Public Health or NHS laboratory.

Before 29th April, the definitions were:

  • England (NHS England): confirmed deaths in hospitals. NHS trusts notify NHS England. The England count is up to 5pm on the previous day.
  • Wales (Public Health Wales): suspected deaths from COVID-19 in hospitals and care homes. The deceased person must have tested positive for the virus. The Wales count is up to 5pm on the previous day.
  • Scotland (Public Health Scotland): confirmed deaths in all places. The positive test result must be at most 28 days before the death. The Scotland count is up to 9am on publication day.
  • Northern Ireland (Public Health Agency): confirmed deaths in all places. The person must have died within 28 days of their first positive test result. The NI count is up to 9:15am on publication day.

The Public Health Wales measure has a conceptual difference. PHW are counting suspected deaths from COVID-19.

The other three nations count deaths with the virus. In England, there is no time limit between the test result and death. Also, there are consistent disparities in cut-off times.

By definition, recorded deaths in England were only in hospitals. Limited testing identified small numbers of infected people out of hospital. By 28th April, NHS England reported 19,294 deaths. This was 89% of the UK reported total at that time.

From 29th April, Public Health England expanded the definition for England. The new count in England now represents deaths in all settings. This measure uses data from NHS England, Health Protection teams, and test surveillance.

It is important to recognise these counts are for the date of record. There are lags between a death occurring and notifying public health authorities.

Image for post
Image for post
This graph is for deaths lags in English hospitals. (Image: Kathryn Leeming/Significance)

Deaths outside of hospital

Reported deaths may not reflect reality. Not everyone gets tested: confirmed deaths are an underestimating measure.

The Office for National Statistics publishes death registrations which mention COVID-19. These mentions can involve positive test results or clinical suspicion.

For deaths registered up to 15th May 2020, 65% of deaths involving COVID-19 were in hospital.

Image for post
Image for post
This graph shows deaths by the date that the person died. (Image: ONS)

We can look at all-cause mortality above the average for recent years. Researchers and statisticians call this measure ‘excess deaths’. This is often difficult to interpret.

Image for post
Image for post
All figures for 2020 are provisional. (Image: ONS)

These excess deaths may not be a direct result of the virus. Pandemics strain healthcare systems, disrupting care.

There are conceptual issues too. The average of prior years is not the same as what this year would have been. Some excess deaths are ‘natural’: resulting from unchanged mortality rates and increased populations.

Definitions are important. What you measure counts.

This blog looks at the use of statistics in Britain and beyond. It is written by RSS Statistical Ambassador and Chartered Statistician @anthonybmasters.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store