Official mortality data for England reveal systematic
undercounting of deaths occurring within first two weeks of
, Martin Neil
, Norman Fenton2, Scott McLachlan2, Joel Smalley1, Josh Guetzkow3,
Jonathan Engler1, Dan Russell1 and Jessica Rose4
3 March 2022
The accuracy of any data purporting to show covid 19 vaccine effectiveness or safety is
critically dependent on the accuracy of four measurements: (1) people classified as having
the disease; (2) vaccination status; (3) reported deaths; and (4) the population of
vaccinated and unvaccinated (the so called ‘denominators’). Errors in any of these could
undermine claims of vaccine effectiveness or safety. We have previously identified
anomalies in the UK Government’s ONS deaths by vaccination status data (ONS dataset) -
specifically that some deaths occurring shortly after vaccination are being wrongly
classified as unvaccinated deaths. In this paper we identify a further problem that appears
to explain anomalies in the ONS data: the total deaths reported by ONS are significantly
lower than we would expect compared to other government datasets, even allowing for
the fact that the ONS use only a subset of the population. For both non-covid and covid
deaths respectively the number of deaths reported for the within 21 days of first dose
vaccination category tally almost perfectly with the number of deaths that would be
expected should they have occurred in the third week alone. Thus, for both covid and non-
covid deaths, the two weeks of post first vaccination deaths appear to have been omitted
from the ONS dataset. This pattern is repeated in all age groups over 60. A variety of factors
could have led to deaths in the first 14 days being omitted in the ONS dataset, including
miscategorisation, reporting lags and data handling or transcription errors. The dataset is
therefore corrupted, making any inferences about vaccine efficacy or safety that are reliant
on the data, moot. Accordingly, the ONS should publicly withdraw their dataset and call for
the retraction of any claims made by others that are based upon it.
The UK Government has been significantly better than most countries in providing detailed data on
Covid cases and deaths indexed by vaccine status. However, despite these efforts we revealed a range
of fundamental inconsistencies and flaws in ONS mortality data . Specifically, two of the most recent
UK ONS (Office for National Statistics) vaccine mortality surveillance reports [2, 3] reveal a range of
fundamental inconsistencies and anomalies in the data. Analysis of these identified the most likely
explanations for the observed anomalies are a combination of four possibilities: (1) systemic
miscategorisation of deaths between the different categories of unvaccinated and vaccinated; (2)
Independent researcher, UK
School of Electronic and Electrical Engineering and Computer Science, Queen Mary, University of London, UK
3 Hebrew University Jerusalem, Israel
4 Institute of Pure and Applied Knowledge, Public Health Policy Initiative, USA
delayed or non-reporting of vaccinations; (3) systemic underestimation of the proportion of
unvaccinated; and/or (4) incorrect population selection for Covid deaths.
In this paper we focus on a newly discovered and alarming source of bias or potential corruption in
the ONS February 2022 report which provides the reported deaths after vaccination data for the whole
of 2021 . Specifically, this report reveals systematic undercounting of both covid and non-covid
totalled deaths occurring within the first two weeks of Covid-19 vaccination. This bias can be detected
by simply comparing the mortality rate we would expect historically, as published by the ONS, with
the mortality rates published in the ONS dataset for 2021, for non-covid deaths. For covid deaths the
bias is evident when we compare the published covid deaths for England as a whole against those in
the ONS dataset .
The scale of undercounting is such that, we estimate, it is equivalent to the number of deaths that
would have been expected to have occurred within the two-week period immediately after
vaccination. Only those deaths that occurred during the third week post vaccination match historical
expected non-covid death counts and concurrent covid death counts. This is true across the age groups
60-69, 70-79 and 80+. It was not possible to compare deaths in the period after second vaccination as
these have only been released monthly rather than by week, and the ONS have not released whole
population data for deaths by month with an age breakdown.
Additionally, we compared the population in the ONS dataset and the UKHSA vaccination dataset,
NIMS (National Immunisation Management System)  and found evidence that the population that
appears in the ONS dataset is missing millions of people categorised as within 21 days of first dose
vaccination, that are present in the NIMS dataset. The number missing exceeds what would be
expected based on the proportion of the whole population not included in the sample. These biases
appear to be systematic and cover covid and non-covid deaths.
Likewise, we compared the death counts registered for England in  with the ONS dataset and found
that 13,593 deaths were missing from the ONS dataset (taking account for the fact that the ONS use
only a subset of the population). The mortality rate in the vaccinated and unvaccinated population
omitted from the dataset is disproportionately high when compared to historical norms, whilst that
reported for the vaccinated are disproportionately low, as previously reported in .
In summary, three new key pieces of evidence suggest that the ONS failed to accurately report deaths
and omitted deaths that occurred within two weeks of vaccination:
• Deaths and population data omitted from ONS dataset
The ONS dataset only represents a part of the population, based on those included in the ONS
2011 census in England and in the GP register. Comparing total deaths in the ONS dataset with
those in the whole population shows the mortality rate for those outside of the ONS dataset
to be more than double that of those included in the dataset. Unless the population outside
the ONS dataset genuinely has a much higher mortality rate, the only explanation is that
deaths have been omitted from the ONS dataset, thus ‘pulling’ the mortality rate down.
• Non-covid deaths are implausibly low for the group within 21 days of first dose vaccination
The ONS dataset fails to include non-covid deaths that would be expected based on historical
mortality rates. Only those non-covid deaths expected, based on historical mortality rates, in
the third week after vaccination appear to be included. Deaths that would be expected to
occur in the first two weeks appear to have been omitted.
• Covid deaths are implausibly low for the group within 21 days of first dose vaccination
In the first two months of 2021, the covid mortality rate was higher for those in the more than
21 days after the first dose category than for those in the within 21 days of the first dose
category, which is the reverse of what might be expected. Using the covid mortality rate for
the whole population, the category within 21 days of first dose suffer far fewer covid deaths
than would be expected; the rates are approximately the same as would be expected in the
third week after vaccination alone.
Section 2 describes the background, analysis approach, and identifies sources of data. Section 3
examines total population mortality rates and death counts using data excluded from the ONS dataset.
Section 4 covers omitted non-covid deaths across all age categories and Section 5 performs the same
analysis for covid deaths. Section 6 examines possible explanations for these systematic omissions.
Section 7 discusses how vaccination and death data is collected in the myriad NHS systems and
identifies potential issues around the ‘chain of custody’ needed to guarantee accuracy of this data and
how errors, omissions and problems might occur. Finally, in Section 8 we draw some conclusions.
The ONS have been under pressure to release a dataset of deaths after vaccination, most likely with
the intent to reassure the public that vaccination had caused no harm. They first promised a release
of this data in March 2021  but they did not release any data until six months later , after which
there have been updates in November 2021 , December 2021  and February 2022 .
A thorough investigation of the rise in non-covid mortality of the unvaccinated which coincides with
peak vaccine rollout in each separate age category has been shown to be compatible with a data lag
or data miscategorisation . Some (including ONS themselves) claimed the explanation was a
“healthy vaccinee” effect. However, as shown in , this healthy vaccinee effect is not supported by
the data for two reasons. First, because the proportion of the unvaccinated population considered to
be in poor health fell during the vaccination rollout and remained low even after the unvaccinated
population fell to only a small number. Second, the same spike in mortality in the unvaccinated was
observed when looking only at deaths of those in very poor health.
An analysis comparing those in the ONS dataset to the population as a whole, has been undertaken.
By combining data from other data sources, including ONS data on total weekly registered death
counts , and from the UKHSA NIMS data on numbers vaccinated , the mortality pattern in the
whole population of England can be estimated and compared against the ONS dataset. First, the ONS
dataset can be compared to the UKHSA data on vaccinations and the differences uncovered can be
used to estimate the size of the vaccinated population that has been excluded from the ONS dataset.
Secondly, taking the difference between the deaths in the ONS dataset and the ONS publication for
covid and non-covid deaths in England and Wales , prorated to the population of England only, gives
an estimate of the total number of deaths excluded from the ONS dataset. Studying the death and
population estimates excluded from the ONS dataset reveals a radically different mortality pattern
than that provided by the ONS dataset in isolation. Details of the methods used are listed in the
3. Mortality rate in excluded population disproportionately high
The ONS dataset is a carefully selected large sample of the whole population; it is restricted to people
in England who were both included in the 2011 census and registered with a GP in 2019. However, it
can be compared with other mortality datasets published by the ONS for the whole population. It is
therefore possible to compare the mortality rate for people included in the dataset with the rate for
those not included in the dataset. For the 80+ age group included in the ONS dataset up to 26th March
2021, 1% of the population had a covid-attributed death. Using other official data sources, total deaths
 as reported by the ONS and the ONS’s own total population estimate , 2% of the English 80+
population, not included in the ONS dataset, had a covid-attributed death. So, the reported mortality
rate for those not included by the ONS is twice as high as for those included. The only possible
explanations for this are either that: (i) deaths have been omitted from the dataset; (ii) the population
is twice as large as the ONS estimated; or (iii) the mortality rate is genuinely twice as high for the
residual population not included in the ONS dataset. There is no reason to believe the latter two
explanations are plausible.
The total deaths included each week in the ONS dataset can be compared to the total registered
deaths the ONS have reported for the whole population. As shown in Figure 1, most of the deaths
omitted from the ONS dataset occurred in the early weeks of the year, peaking at the beginning of
February 2021. A further comparison can be made between the whole population reported by UKHSA
as belonging to the within 21 days of first dose vaccination category in NIMS and the numbers reported
to be within 21 days of first dose in the ONS dataset. The population not included in the ONS dataset,
belonging to the within 21 days of first dose category, correlates very closely with the deaths not
included in the ONS dataset, as shown in Figure 1. The rise and fall in this population deficit reflect the
period during which people fell into the category of within 21 days of first dose.
Figure 1: Estimated deaths for England that were not included in the ONS dataset plotted against population
in NIMS belonging to the within 21 days of first dose’ category (using a three-week moving average)
4. Implausibly low non-covid deaths for the ‘within 21 days of first
dose’ vaccination category
Using historical weekly mortality rates for 2015-2019 we can estimate how many non-covid deaths to
expect as a proportion of the population by age. The number of people who were within 21 days of
first dose vaccination category changed over time but from the population data, included in the ONS
dataset each week, an estimate of expected non-covid deaths can be calculated.
Omitted deaths 60-69
Omitted deaths 70-79
Omitted deaths 80+
Population within 21 days of first vaccination included in NIMS but omitted from ONS dataset
Surprisingly, the non-covid deaths reported by the ONS for the within 21 days of first dose vaccination
was between one third to a half of the number that would be expected using historical mortality rates.
This difference is consistent across each age group. This is shown in Table 1.
Table 1: Total expected and reported non-covid deaths up to 26th March
2021 for the ‘within 21 days of first dose’ vaccination category included
in the ONS dataset (pro-rated to England population)
Figures 2 to 4 show the weekly non-covid deaths reported in the ONS dataset for those in the within
21 days of first dose vaccination category; the non-covid deaths expected for this group as calculated
from historical data, and the expected non-covid deaths occurring in the third week alone since first
dose vaccination, as calculated from historical data. The proximity of the non-covid deaths reported
in the ONS dataset to those expected to occur in the third week alone since vaccination is remarkable.
The same pattern occurs across each age category and is highly suggestive that the ONS dataset has
not included the non-covid deaths that occurred during the first two weeks post first dose vaccination
administered to each age group. This finding is consistent with the hypothesis in  that the anomalies
found in the ONS data are most likely caused by undercounting deaths occurring shortly after
vaccination and that this undercounting was due primarily to misclassification.
Note that, because this is the population within 21 days of first dose the peaks are synchronised with
the vaccine roll out for each age group and hence are not natural or due to random error.
Figure 2: Expected non-covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and non-covid deaths included in the ONS dataset for the 80+ age group (to May
Non-covid deaths 80+ expected in first three weeks
Non-covid deaths 80+ expected in third week only
Non-covid deaths 80+ in ONS dataset
Figure 3: Expected non-covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and non-covid deaths included in the ONS dataset for the 70-79 age group (to May
Figure 4: Expected non-covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and non-covid deaths included in the ONS dataset for the 60-69 age group (to May
The healthy vaccinee hypothesis, that those close to death will postpone or decline vaccination might
hypothetically account for a lower rate of death in the first two weeks. But as an explanation it is only
plausible if every possible death that might occur in the first two weeks, after the offer of vaccination,
was foreknown whilst those deaths in the third week were not, and hence those dying in the third
week did not postpone or decline vaccination. See Appendix in  for further discussion of the
implausibility of the assumptions required for the healthy vaccinee hypothesis.
Unfortunately, the same analysis cannot be applied to the group within 21 days of second vaccination
dose because ONS have only released this data as a monthly value and have not published overall
deaths by month with an age breakdown.
Non-covid deaths 70-79 expected in first three weeks
Non-covid deaths 70-79 expected in third week only
Non-covid deaths 70-79 in ONS dataset
Non-covid deaths 60-69 expected in first three weeks
Non-covid deaths 60-69 expected in third week only
Non-covid deaths 60-69 in ONS dataset
5. Implausibly low covid deaths for the within 21 days of first dose
The vaccine ought to reduce the number of covid deaths but is not thought to be fully effective until
two weeks have passed since vaccination. It has been suggested from several studies (cited in ) that
the vaccinated are vulnerable to covid infection during this period; therefore, the covid mortality rate
should – if anything - be higher in the within 21 days of first dose vaccination category than in the
more than 21 days after first dose vaccination category. This is due to increased susceptibility in the
first 21 days and possibly correspondingly lower later susceptibility (from immunity acquired after
infection) . Yet, conversely, in the first two months of 2021, the covid mortality rate was higher in
the more than 21 days after first dose vaccination category than in the within 21 days of first dose
The covid mortality rate in the within 21 days of first dose category can be estimated using a similar
approach to that already used to estimate non-covid mortality, using the whole population covid
mortality rate, available from other official data. The covid deaths included in the ONS dataset for
those within 21 days of first dose group is considerably lower than expected, as shown in Table 2.
Table 2: Total expected and reported covid deaths up to 26th March 2021
for the ‘‘within 21 days of first dose’’ vaccination category included in the
ONS dataset (pro-rated to England population)
Figures 5 to 7 show the weekly covid deaths reported in the ONS dataset for those in the within 21
days of first dose vaccination category; the covid deaths expected for this group as calculated from
concurrent data from other official sources, and the expected deaths occurring in the third week alone
since first dose vaccination. The proximity of the covid deaths reported in the ONS dataset to those
expected to occur in the third week alone since vaccination is – again – remarkable. The same pattern
occurs across each age category and further suggests the ONS dataset has not included the covid
deaths that occurred during the first two weeks post first dose vaccination.
The fact that covid deaths occurring during the first two weeks have been omitted from the ONS
dataset, in the same manner as non-covid deaths have been omitted is a significant and rather
troubling anomaly warranting an explanation.
Figure 5: Expected covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and covid deaths included in the ONS dataset for the 80+ age group (to May 2021)
Figure 6: Expected covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and covid deaths included in the ONS dataset for the 70-79 age group (to May 2021)
Figure 7: Expected covid deaths 21 days after vaccination versus expected non-covid deaths in the third
week since vaccination and covid deaths included in the ONS dataset for the 60-69 age group (to May 2021)
Covid deaths 80+ expected in first three weeks Covid deaths 80+ expected in third week only
Covid deaths 80+ in ONS dataset
Covid deaths 70-79 expected in first three weeks Covid deaths 70-79 expected in third week only
Covid deaths 70-79 in ONS dataset
Covid deaths 60-69 expected in first three weeks Covid deaths 60-69 expected in third week only
Covid deaths 60-69 in ONS dataset
It could be argued that the 18-day average delay between diagnosis and death would mean that the
lack of deaths within 21 days of first dose might be explained by covid contracted prior to vaccination
(it is recommended that vaccination should be postponed if covid positive). However, studies [10, 11]
have reported significant numbers of covid infections occurring immediately after vaccination, with
numerous reports of around 40% higher incidence in the first two weeks after vaccination than in the
unvaccinated . Given the period between infection and death is heavily age dependent, with a
much shorter period in the over 60s , this explanation is not credible.
6. Alternative explanations for anomalies
The mortality rate is defined as the number of deaths divided by the size of the population, and
therefore any inaccuracies in either the number of deaths or the size of the population will produce
an inaccurate and misleading result.
The ONS dataset shows a spike in all-cause mortality in the unvaccinated that coincides with the first
dose vaccine rollout for each age group . This higher mortality rate seen in the unvaccinated during
the vaccination programme has been explained as being caused by these possible phenomena:
1. There is a reporting lag such that deaths are reported a week late. The mortality rate would
then be distorted by the shrinking population denominator in the unvaccinated group and the
growing population denominator for the vaccinated.
2. Post vaccination deaths have been miscategorised as unvaccinated.
3. There is a “healthy vaccinee” selection bias such that people who are mortally ill are not
allowed or decline the offer of vaccination.
The December 2021 publication by the ONS  supports the latter hypothesis and stated that:
“The all-cause ASMRs for the year-to-date were lower in the first three weeks after a vaccine
dose than in subsequent weeks after that dose. This could be because of a “healthy vaccinee
effect” where people who are ill (either due to COVID-19 or another relevant illness) are likely
to delay vaccination. Therefore, the people who have been recently vaccinated are, in the short
term, in better health than the general population. The same is true for deaths involving
COVID-19 after the second dose. This is likely because the healthy vaccinee effect where people
who know or suspect they have COVID-19 delay vaccination until recovered, has a bigger effect
here than the difference in protection offered by the vaccine within and following the first three
weeks after vaccination.”
However, the same report also states:
“Changes in non-COVID-19 mortality by vaccination status are largely driven by the changing
composition of the vaccination status groups because of the prioritisation of clinically
extremely vulnerable and people with underlying health conditions, and differences in timing
of vaccination among people who were eligible.”
These points appear contradictory. If the extremely vulnerable were prioritised for vaccination, how
can the clinically vulnerable disproportionately be overrepresented in the unvaccinated group?
The ONS hypothesis relies on two extreme and implausible assumptions:
1. That a terminally ill person, when offered a ‘safe and effective’ vaccine to reduce the risk of a
distressing death from respiratory failure, would decline it.
2. The dying, or their carers, can accurately assess their remaining life span and those with only
two weeks left to live decline the vaccine. However, those with three weeks to live were
vaccinated anyway and then died three weeks later.
The ONS hypothesis also does not fit with real-world experience. There were numerous outbreaks of
covid in nursing homes shortly after vaccination leading to many deaths. For example, in Northeast
Scotland, Basingstoke, Cornwall, Corby, Renfrewshire [13, 14, 15, 16, 17].
7. Data chain of custody
To track vaccinations three separate reporting processes were set up by the NHS: (1) Using the General
Practice Extraction System (GPES) to extract vaccination status from the EMIS system used by General
Practitioners (GPs); (2) The Pinnacle system for use in community vaccination sites and those GP clinics
not using EMIS and therefore connected into GPES; (3) Immunisation Management System (IMS) and
the Commissioning Data Sets (CDS) in large hospitals that are able to transport vaccination status
directly into the National Immunisation Management System (NIMS); and (4) the National
Immunisation and Vaccination System (NIVS) for use in other hospitals and care homes.
In many community and pop-up vaccination sites, and where access to recording an electronic record
was a problem or in the absence of an NHS number , details were recorded manually on paper
. Many vaccinators have prioritised administration of vaccinations over these administrative
requirements and information may often have been electronically uploaded in retrospect. It is not
clear whether records were (ever) uploaded when patients died between vaccination and the transfer
of these paper record to an electronic equivalent, and even less likely where the patient was
vaccinated at one of the community or pop-up centres that were not required to collect identification
details where the recipient did not wish to provide them. These issues, along with delays in compiling
and transporting data between various database systems as shown in Figure 8, have created the
potential for under-reporting of vaccination figures.
Figure 8: Vaccination record data flow highlighting potential sources of error
Similarly, and again shown in Figure 8, there are situations that have potential for over-reporting. One
example has been where: (a) the person was vaccinated at a community centre or pharmacy, where
they were not required to collect personal identification data, but in circumstances where the person
gave a name only and this was manually entered into the Pinnacle system. In these cases, NIMS creates
an NHS number for the ‘unknown person’ when an NHS record could not be easily ‘matched’ to them.
So, it is possible for a person to have one record with a generated NHS number created for this
vaccination at a community centre or pharmacist, and an additional separate record for their
vaccination with their GP. They will thus appear to NIMS as separate two people who both received
their first vaccinations. We have observed instances where this has occurred, and the person has
ended up either being vaccinated an additional time by a GP because the record reported that they
hadn’t received the ‘missing’ community injection. There is also the possibility that the GP might enter
‘missing’ vaccination details from the vaccination card, issued at the community centre or pharmacy,
thus creating a third vaccination record in NIMS. Such cases do occur and mean that the NIMS dataset
is likely to be very messy. There may also have been issues with patients who had been hospitalised
when receiving their first vaccination where there was no record of that first dose on the system.
Knowing the proportion of people who experienced issues with the recording of their first dose would
be useful to estimate how much of a problem that may have been for records associated with those
The ONS are reliant on NIMS data from UKHSA to know who had been vaccinated. The numbers
reported as vaccinated each week in NIMS reduces between weekly reports as people are removed
from the dataset, having died. The changes to allow for birthdays are less frequent such that the
decrease due to deaths is noticeable week to week. Aside from these removals, the week 28 2021
NIMS report from UKHSA shows 12,864 more first doses were given in the first week of the
programme than the week 25 2020 report. Since then, rather than this total reducing with deaths, a
further 1,000 vaccination records were added more than six months after the vaccination took place.
It is not clear whether records for the dead were also added in tandem and whether records were
removed after death. Were these all replaced before sharing with ONS? It is critical that all UKHSA
shared data included all the deaths that were subsequently removed. Moreover, the NIMS dataset
was only made public for the first time in April 2021. It is likely there was immense pressure to release
a publication to enable data sharing on the speed of the vaccine rollout. However, for months prior to
the first data release, work would have had to be carried out to ensure the database was accurate by
removing duplicates and collating data from all the different vaccination sites. It is entirely possible
that including those who had died between vaccination and the first release in April 2021 was a low
priority and hence the data for those who had died was not accurate. However, this does not explain
why the omitted deaths appear to be systematically concentrated in the first two weeks post first
The accuracy of any data purporting to show vaccine effectiveness or safety against a disease is
critically dependent on the accuracy of four measurements: people classified as having the disease;
vaccination status; death reporting; and the population of vaccinated and unvaccinated (the so called
‘denominators’). If there are errors in any of these, claims of effectiveness or safety are unreliable.
The deaths reported in the ONS dataset are significantly lower than expected. There are lower
numbers of both non-covid and covid deaths in the first two weeks after vaccination. Those non-covid
deaths for the within 21 days of first dose vaccination category included in the ONS dataset tally
perfectly with the non-covid deaths that would be expected should they have occurred in the third
week alone. Thus, the two weeks of post first vaccination non-covid deaths appear to have been
omitted from the ONS dataset. This pattern is repeated for the covid deaths occurring in the same
within 21 days of first dose vaccination category and again across all age groups.
Total deaths not included in the ONS dataset are disproportionately higher than that expected for the
population excluded from the ONS dataset. This suggests deaths have been omitted from the ONS
dataset. The fact that total deaths excluded from the ONS dataset correlate with the population of
those in the within 21 days of first dose vaccination category supports this assertion.
A variety of factors could have led to deaths in the first 14 days being omitted in the ONS dataset,
including miscategorisation, reporting lags and data handling or transcription errors. The dataset is
therefore corrupted, making any inferences about vaccine efficacy or safety, reliant on the data, moot.
The ONS should therefore publicly withdraw their dataset and call for the retraction of any claims
made by others that are based upon it.
The paper has benefited from the input of senior clinicians and other researchers who remain
anonymous to protect their careers.
 Neil M., Fenton N., et al. Official mortality data for England suggest systematic miscategorisation
of vaccine status and uncertain effectiveness of Covid-19 vaccination. January 2022. DOI:
 Bermingham C., Morgan J. and Nafilyan V. ONS. “Deaths involving COVID-19 by vaccination status,
England: deaths occurring between 2 January and 24 September 2021”. 1 November 2021.
 Bermingham C., Nafilyan V., Morgan J., and Ward I. ONS: Deaths involving COVID-19 by vaccination
status, England: deaths occurring between 1 January and 31 October 2021: Age-standardised and age-
specific mortality rates for deaths involving COVID-19 by vaccination status; deaths occurring between
1 January and 31 October 2021 in England. December 2021.
 Munro M., Bermingham C., Nafilyan V et al. Deaths by vaccination status, England, 1 January to 31
 Caul S. Tweet from ONS representative about release of data.
 Bermingham C., Nafilyan V., Morgan J. Deaths involving COVID-19 by vaccination status, England:
deaths occurring between 2 January and 2 July 2021. 13 September 2021.
 Caul S. Deaths registered weekly in England and Wales, provisional. 15 February 2022.
 UKHSA. Official Statistics – National flu and COVID-19 surveillance reports: 2021 to 2022 season.
17 February 2022.
 Park N. Estimates of the population for the UK, England and Wales, Scotland and Northern Ireland.
ONS, 25 June 2021.
 Egan C., Knight S. et al. Hospitalised vaccinated patients during the second wave, update April ’21.
 Bernal et al. Effectiveness of the Pfizer-BioNTech and Oxford-AstraZeneca vaccines on covid-19
related symptoms, hospital admissions, and mortality in older adults in England: test negative case-
control study. BMJ 2021; 373 doi: https://doi.org/10.1136/bmj.n1088 (Published 13 May 2021)
 Harrison E. Docherty A and Semple C. COVID-19: time from symptom onset until death in UK
hospitalised patients. 7 October 2020.
 BBC News. Covid: Eleven deaths at Aberdeen care home investigated. 11 January 2021.
 Basingstoke Gazette. Coronavirus outbreak: 22 deaths at Pemberley House Care Home. 5 January
 Cornwall Live. Eleven residents die in three weeks after care home Covid outbreak. 15 February
 Northamptonshire Telegraph. Several deaths at Corby care home after mass Covid outbreak. 28
 Daily Record. Scots care home residents test positive for Covid-19 days after receiving vaccine. 6
 Harrison E et al. COVID-19: time from symptom onset until death in UK hospitalised patients.
Mortality rates in the vaccinated and unvaccinated population were calculated for each week using
the reported deaths divided by the size of the population for each age group and vaccination status
within the ONS dataset.
The baseline mortality rates for 2015-2019 were calculated by taking the registered death figures by
age, averaged, for each week of the year prorated to England only and dividing by the ONS estimate
of the average population over that time in England.
Deaths not included in the ONS dataset
ONS weekly registered deaths for each age group were taken for England and Wales. In order to
estimate the number for England alone, the percentage of total deaths each week that occurred in
Wales was used to prorate the deaths for each age group. The same method was used to calculate
total deaths and covid deaths separately. Non-covid deaths were calculated by taking the difference.
Size of whole population within 21 days of vaccination
The cumulative people vaccinated with a first dose was subtracted from the value for the previous
week to give the number of first doses given in each week and the number of people within one week
of a first dose. This number was lagged by a week to give the number of people within two weeks and
by two weeks for the number within three weeks. The total within one, two or three weeks was
summed to give the whole population within 21 days of a first dose.
Expected non-covid deaths
The weekly 2015-2019 mortality rate for England was multiplied by the size of the population of the
‘‘within 21 days of first dose vaccination’’ category, for a given age group, to give the expected number
of non-covid deaths within 3 weeks for that age group. The same calculation was carried out for the
population three weeks after being vaccinated with the first dose to give expected non-covid deaths
in the third week. These expected non-covid deaths were compared to the non-covid deaths reported
in the ONS dataset for the group within 21 days of a first dose. This was then prorated to the whole
population of England. This was done by dividing by the population included in the ONS dataset and
then multiplying by the figure for the whole population that week.
Expected covid deaths
The weekly covid mortality rate for the whole population of England was calculated by dividing the
ONS registered covid deaths, prorated to England, by the ONS mid 2020 population estimate for each
age group. This mortality rate was then multiplied by the size of the population within 21 days of first
vaccination dose category to give the expected number of covid deaths within 3 weeks for that age
group. The same calculation was carried out for the population three weeks after being vaccinated
with the first dose to give expected covid deaths in the third week. These figures were compared to
the covid deaths reported within the ONS dataset for the group within 21 days of a first dose. This was
then prorated to the whole population of England. This was done by dividing by the population
included in the ONS dataset and then multiplying by the figure for the whole population that week.