It’s all about the test: Counting COVID-19 deaths
by Allan LEONARD
14 April 2020
“The figures on deaths relate in almost all cases to patients who have died in hospital and who have tested positive for COVID-19. Slight differences in reporting in devolved administrations may mean that they include a small number of deaths outside hospital.” — Department of Health and Social Care (UK)
Who is and is not being counted in this daily reported figure? What are the regional variations and why? We explain below.
Two sets of COVID-19 data are regularly being produced:
-
- Daily reports on the deaths of people who had tested positive for the virus (mostly dying in hospital, though may include some people discharged to care homes); and
- Weekly reports based on COVID-19 being listed on death certificates no matter where the person died — but there’s a delay to account for the registration process. (The Northern Ireland Statistics & Research Agency intend to provide a further level of breakdown to include deaths in care homes.)
Every day at 2pm, the UK Department of Health and Social Care (DHSC) publishes the statistic for the number of people who have died of COVID-19 in the previous 24 hour measurement period (from 5pm to 5pm). This statistic is compiled from NHS England and Improvement, Health Protection Scotland, Public Health Wales, and the Public Health Agency (Northern Ireland).
Public Health England also updates a daily dashboard, which shows trends across the four UK countries on positive tests and deaths.
This daily figure is used as a world-wide comparative metric for those who have died and tested positive for COVID-19 in the last 28 days. The World Health Organisation (WHO) publishes a dashboard of these data, which represents “laboratory-confirmed cases” of COVID-19 (i.e. does not include those who were not tested for COVID-19).
Only COVID-19 positive test deaths counted daily
Professor Chris Whitty (Chief Medical Officer for England) explained during a daily press briefing that DHSC figures include all those people who are proven to have coronavirus on testing. He also said that DHSC figures are collected the same way the whole time and are comparable to international figures (such as those published by the WHO), which are collected in the same way — i.e. those who test positive for COVID-19.
People who die elsewhere, such as in care homes or in the community, may or may not have been tested for COVID-19. This could be due to a lack of tests, or the lack of a test result before death, but in any case, a doctor may certify the cause or contributory cause of death as being consistent with COVID-19 symptoms.
However, in theory, there may be cases whereby some people who tested positive for COVID-19 in hospital were discharged and subsequently died out of hospital. Because there is an associated positive test, their deaths are included in figures from devolved administrations if their deaths have been reported back to the health authorities.
For those who were never tested for COVID-19 (or tested negative) and subsequently die, regardless of place, if a doctor deems that coronavirus was a cause of death and records it as such on the death certificate, that death will not be included in the DHSC daily figure but will be included in country’s registration of deaths data.
Daily reporting of COVID-19 deaths in Northern Ireland
The Public Health Agency (PHA) is publishing daily Surveillance Reports, which include a figure of deaths of people who have died within a 24-hour period who had a positive COVID-19 test within the last 28 days (data correct up to 9.15am on the morning of the report being issued). This figure feeds through to the UK total and differs from the England & Wales figure in that, in theory, it could include deaths outside hospital, for example, following discharge after treatment if these deaths are reported to the PHA. The daily PHA figure does not include any deaths where there was no COVID-19 test (or tested negative).
For the calendar year to date (14/4/2020), PHA reported 132 deaths where there was a positive COVID-19 test result. The following graphic shows this figure, by date:
Figure: Northern Ireland deaths with positive COVID-19 test result
Source: Public Health Agency: COVID-19 Surveillance Reports.
Northern Ireland weekly official statistics
The Northern Ireland Statistics Agency (NISRA) publishes all registered deaths on the same basis as ONS and NRS. These figures are released every Friday and refer to the week ending the previous Friday, a lag time of 7 days. These are the definitive official statistics on the total number of deaths where COVID-19 is mentioned on the death certificate. The figures remain provisional until the quarterly National Statistics are published, as registrations can often lag by much more than a week.
The NISRA weekly death statistics include information about respiratory and COVID-19 deaths. Up to the calendar year ending 3 April 2020, there were 65 COVID-19 deaths registered in Northern Ireland. Meanwhile, for the same time period there were an estimated 79 COVID-19 deaths occurring. The difference between “registered” and “occurring” is explained by the time lag for an occurrence of death to be registered in Northern Ireland. In order to allow comparison with PHA figures, NISRA is indicating this best estimate of occurrences in the previous week, based on registrations received up to two days before publication. For comparison, for the same time period ending 3 April 2020, PHA reported (4 April 2020) 56 Northern Ireland deaths with a positive COVID-19 test result.
On 14 April 2020, NISRA stated (at 54:01) that from 17 April 2020, its weekly statistical bulletin on deaths registered in Northern Ireland will include information on the place of death — hospital and out of hospital settings — akin to that published by ONS (see below table).
Other weekly UK COVID-19 statistics
Every Tuesday, the Office of National Statistics (ONS) is now releasing a weekly set of figures that show the total number of deaths in England and Wales where the death certificate mentions COVID-19 and which were registered in the week up to the previous Friday (i.e. a lag of 11 days). The lag is a function of the death registration process, but when published, these official statistics are the definitive numbers of the total deaths attributed to COVID-19, regardless of where the death took place.
The following table shows deaths attributed to COVID-19, by place of death from 29 December 2019 to 3 April 2020:
Table: Deaths registered from 29/12/2019 to 3/4/2020
| Place of death | England COVID-19 deaths | Wales COVID-19 deaths |
| All Deaths | 3,950 | 157 |
| Home | 126 | 10 |
| Hospital (acute/community) | 3,567 | 136 |
| Hospice | 31 | 2 |
| Care Home | 207 | 9 |
| Other communal establishment | 3 | – |
| Elsewhere | 16 | – |
Source: ONS: Deaths registered weekly in England and Wales, provisional
The National Records of Scotland (NRS) also produces weekly registered deaths every Wednesday for the previous week ending Sunday, a lag time of three days. In these data for deaths in Scotland, there is as yet no published data on place of death, beyond the NHS Board of usual residence. For the calendar year to date (31/3/2020), NRS reported 354 deaths where coronavirus (COVID-19) was mentioned on the death certificate in Scotland.
Summary
It is important to consider that there are two sets of data in regards to counting the number of people whose death is attributed to COVID-19: (1) a daily figure of those who died who tested positive for COVID-19; and (2) a time-lagged figure of everyone whose death is attributed to COVID-19, regardless of any test. This second figure will always be larger than the first.
The two figures serve different purposes. As Prof. Whitty concluded: “The ONS data [registered deaths] [are] extremely useful for looking at the wider picture, but the [daily] NHS data [those who died who tested positive for COVID-19] is more useful for us to make decisions day to day and to make decisions, for example, about what we need to change, in terms of our current interventions.”