Between 8 December 2020 (the start of the vaccination programme) and 12 June 2021 (the approximate end of the second wave of the coronavirus (COVID-19) pandemic), people from all ethnic minority groups (except the Chinese group and women in the White other ethnic group) had higher rates of death involving COVID-19 compared with the White British population.
During this period, the rate of death involving COVID-19 was highest for the Bangladeshi ethnic group (5.0 times greater than the White British group for males, and 4.5 times greater for females), followed by the Pakistani (3.1 for males, 2.6 for females) and Black African (2.4 for males, 1.7 for females) ethnic groups.
Since the start of the third wave of the coronavirus pandemic (from 13 June 2021 onwards) the rate of death involving COVID-19 was higher for all ethnic minority groups (except Chinese people, men from the Mixed ethnic group and women from the White other ethnic group); the risk remained highest for the Bangladeshi ethnic group (4.4 and 5.2 times greater than for the White British ethnic group for males and females, respectively).
Location, measures of disadvantage, occupation, living arrangements, pre-existing health conditions and vaccination status accounted for a large proportion of the excess rate of death involving COVID-19 in most ethnic minority groups; however, the Bangladeshi ethnic group and men from the Pakistani ethnic group remained at higher risk than White British people in the third wave, even after adjusting for vaccination status.
Males and females of Black Caribbean and Black African background were at elevated risk in the third wave after adjusting for location, measures of disadvantage, occupation, living arrangements, pre-existing health conditions; after also adjusting for vaccination status, there was no evidence that the risk was greater than for the White British ethnic group.
"Today's analysis shows that since the vaccination programme began, the risk of death from COVID-19 has continued to be higher in most ethnic minority groups than in the White British ethnic group.
As already highlighted in our analyses of earlier periods, these differences in mortality are largely explained by socio-demographic and economic factors and health. For the first time, we show that the lower vaccination coverage in some ethnic groups also contributes to the elevated risk of COVID-19 death, particularly in the Black African and Black Caribbean groups."
Vahé Nafilyan, Senior statistician, Health and Life Events Division, Office for National Statistics.Back to table of contents
The Office for National Statistics (ONS) previously published analysis of deaths involving the coronavirus (COVID-19) by ethnic group for England between 24 January 2020 and 31 March 2021. In this article, we update the previous analyses to present provisional analysis of deaths involving coronavirus (COVID-19) by ethnic group for deaths occurring in England between 8 December 2020 (the start of the vaccination programme in England) and 1 December 2021, that were registered by 15 December 2021.
In previous analyses we adjusted for differences in location, measures of disadvantage, occupation, living arrangements (retrieved from the 2011 Census), hospital admissions (from Hospital Episode Statistics) and certain pre-existing health conditions, which are known to increase the risk of dying from COVID-19 (from the General Practice Extraction Service Data for Pandemic Planning and Research). In this report, we extend previous analyses by also adjusting for vaccination status, by linking to data from the National Immunisation Management System (NIMS). This enables us to also assess the extent to which the increased risk of COVID-19 mortality in some ethnic groups is explained by differences in vaccination status.
To explore whether differences in the risk of COVID-19 mortality between different ethnic groups have changed over the course of the coronavirus pandemic and vaccination programme, analyses are presented for two time periods. We classify deaths occurring between 8 December 2020 (the date of the first vaccination in England) and 12 June 2021 as having occurred in the latter part of the second wave, and deaths from 13 June onwards as having occurred in the third wave. Reflecting the progress of the vaccination roll out, our analysis of the second wave adjusts for first and second vaccination doses, and for the third wave adjusts for first, second and third doses.
To understand the drivers of differences in the rate of death involving COVID-19 between ethnic groups, we used Cox proportional hazards models to estimate how differences in the rates changed when adjusting for a range of factors affecting both the risk of infection and the risk of death if infected. In addition to the models presented in previous analyses, in this latest analysis we also adjust for vaccination status.
In our baseline model, we adjusted for age only. We then adjusted for:
- residence type (private household, care home, other communal establishments)
- geography (population density, region, rural urban classification)
- socio-economic factors (deprivation, household composition and occupational exposure)
- certain pre-existing health conditions
- vaccination status
For full details of the variables included in the models, see the Definitions tab in the dataset.Back to table of contents
Updating ethnic contrasts in deaths involving the coronavirus (COVID-19), England
Dataset | Released 26 January 2022
Age-standardised mortality rates (ASMRs) (deaths occurring between 24 January 2020 and 1 December 2021) and hazard ratios (deaths occurring between 8 December 2020 and 1 December 2021) for deaths involving COVID-19 by ethnic group, England.
Age-standardised mortality rates
Age-standardised mortality rates (ASMRs) are used to allow comparisons between populations that may contain different proportions of different ages. The 2013 European Standard Population is used to standardise rates.
Cox proportional hazards regression model
The Cox proportional hazards regression model is a multiple regression procedure that measures the association between a time-to-event outcome and a characteristic of interest (such as ethnic group), while adjusting for other characteristics expected to also be associated with the outcome.
A hazard ratio is a measure of the relative differences in the instantaneous rate of mortality between groups. A hazard ratio greater than 1 indicates the rate of mortality is higher, and likewise, less than 1 lower in the population group under study compared with a reference group.
Deaths involving coronavirus (COVID-19)
Deaths involving coronavirus (COVID-19) include those with an underlying cause, or any mention, of ICD-10 codes U07.1 (COVID-19, virus identified), U07.2 (COVID-19, virus not identified) or U09.9 (post-COVID condition). A doctor can certify the involvement of COVID-19 based on symptoms and clinical findings — a positive test result is not required.Back to table of contents
These analyses use data from the Office for National Statistics' (ONS) Public Health Data Asset (PHDA), and build on the methods used in previous publications. The PHDA combines Census 2011 records, death registrations, Hospital Episode Statistics (HES) and primary care records retrieved from the General Practice Extraction Service (GPES) Data for Pandemic Planning and Research (GDPPR), with England coverage only. Information about these data sources, how they have been linked, and the methods used for previous publications can be found in the Technical Appendix.
The analyses in this report build on the methods described in the Technical Appendix to also incorporate the following updates.
Firstly, in addition to the data sources described in the Technical Appendix, these analyses also include vaccination data from the National Immunisation Management System (NIMS) to adjust for vaccination status. We used NIMS data for the period 8 December 2020 (the day of the first vaccination in England) to 1 December 2021. This data was linked to 2011 Census based on NHS number. Our analysis of the second wave includes first and second vaccination doses, and for the third wave includes first, second and third doses. The analysis does not differentiate between booster doses and third doses provided for other reasons. Vaccination status was included in the model as a time-varying covariate, and we considered a person vaccinated once 14 days had passed since the dose was administered. Of people aged 30 years and over who received at least one dose of a vaccine, 79% were linked to the ONS PHDA.
Secondly, we also updated the list of pre-existing health conditions we adjusted for to align with the update of the COVID-19 risk prediction model known as QCovid2 used by the NHS.
The study population comprised 28.8 million people (aged 30 to 100 years) enumerated at the 2011 Census and living in either private households or communal establishments in England at the start of the vaccination roll-out. We modelled the hazard of death involving COVID-19 during the outcome period using a sample of the study population. The sample included all those who died from any cause during the outcome period and a weighted random sample of those who did not (1% of people identifying as White British and 10% of each of the other ethnic groups).Back to table of contents
The primary strength of the study is using nationwide linked population-level data that combines a rich set of demographic and socio-economic factors from the 2011 Census with death registrations data, pre-existing health conditions from electronic health records, and vaccination status from the National Immunisation Management System. Unlike studies based only on electronic health records, our study relies on self-identified ethnicity, with very few missing data, limiting the potential for exposure misclassification bias.
The Public Health Data Asset only contains information on people who were enumerated at the 2011 Census. It therefore excludes people living in England in 2011 but not taking part in the 2011 Census (estimated to be about 5%); respondents who could not be linked to the 2011 to 2013 NHS Patient Registers (5.4% of census respondents); people who have immigrated since 2011; and people not registered with a general practitioner at the start of the pandemic.
Many of the socio-demographic variables were derived from the 2011 Census. Some of these characteristics might have changed since the 2011 Census and may not accurately reflect individuals' circumstances during the pandemic. To mitigate measurement error, we restricted our analysis to people aged 30 years and over.Back to table of contents
Contact details for this Article
Telephone: +44 1633 455 046