Ethnic differences in COVID-19 mortality in the second and third waves of the pandemic in England during the vaccine rollout: a retrospective, population-based cohort study

The disproportionate impact of the coronavirus pandemic on ethnic minority groups has been widely reported [1‐4]. In England, rates of hospitalisation for COVID-19, admission to intensive care, and death were higher among ethnic minority groups during the first and second waves of the pandemic [5]. However, compared with the first wave, excess COVID-19 mortality was reduced in the second wave among the Black African and Black Caribbean groups but increased among the Pakistani and Bangladeshi groups [6, 7]. Moreover, adjustments for geography, socioeconomic factors, and pre-existing health conditions accounted for a large proportion of the elevated COVID-19 mortality risk observed in the first and second waves. However, some residual risk remained unexplained, most notably for South Asian and Black African groups.

The UK began its coronavirus vaccination programme on 8 December 2020, starting with those most likely to experience severe outcomes (people aged 70 years and over and those with underlying health conditions) and those working in health and social care roles [8]. Survey data indicates that rates of vaccine hesitancy in the UK are highest among people from Black ethnic groups [9]. Consistent with this, differences in vaccination rates by ethnic group were evident early in the rollout of the vaccination programme and these differences widened over time [10, 11]; by 12 December 2021, 3.7% of White British adults aged 50 or over had not received any dose of a COVID-19 vaccine, compared with 26.2% from the Black Caribbean group and 17.4% of the Black African group [12].

This study used population-level linked administrative data for England to investigate whether inequalities in deaths involving COVID-19 by ethnic group have continued into the third wave. We also explored the extent to which elevated COVID-19 mortality in some ethnic groups can be explained by differences in age, residence type, geographical factors, sociodemographic characteristics, pre-pandemic health, and, for the first time, vaccination rates.

Our findings demonstrate that throughout the rollout of the vaccine programme, most ethnic minority groups have continued to experience greater rates of death involving COVID-19 compared with the White British group. Although the patterns of excess COVID-19 mortality risk by ethnic group have changed over the course of the pandemic, the Bangladeshi, Black African, Black Caribbean, and Pakistani groups remained the groups with highest rate of COVID-19 mortality in the third wave. We also found that adjusting for vaccination status explained some of the remaining increased risk of COVID-19 mortality that is not explained by other factors, particularly during the third wave for the Black African, Black Caribbean, and Pakistani groups. After adjusting for geographical factors, sociodemographic characteristics, pre-pandemic health, and vaccination status, the risk of COVID-19 mortality was similar to the White British group for all ethnic groups except the Bangladeshi group and men from the Pakistani group.

Previous analyses showed that differences in location, measures of disadvantage, occupation, living arrangements, and certain pre-existing health conditions explain a large proportion (but not all) of the excess COVID-19 mortality risk observed in some ethnic groups [6, 7, 13]. Based on these findings, we hypothesised that ethnic differences in COVID-19 mortality could be mediated by factors associated with the risk of exposure to SARS-CoV-2 (e.g. geographical factors, socioeconomic status, living arrangements) and the risk of severe outcomes once infected (e.g. pre-existing health conditions, vaccination status) (Additional file 1: Fig. S2). We found that adjustment for geographical factors was associated with the largest reduction in the HRs for most ethnic groups during the second wave. SARS-CoV-2 case rates in England were higher in more densely populated urban areas than in less densely populated rural areas during the second wave [27]. Therefore, the reduction in risk of COVID-19 mortality in ethnic minority groups following adjustment for geographical factors likely reflects increased risk of infection. Additional adjustment for socioeconomic and demographic characteristics further reduced the HRs for most ethnic groups, most notably for Bangladeshi and Pakistani groups. This suggests that socioeconomic inequalities also contribute to increased rates of COVID-19 mortality in ethnic minority groups, which is consistent with evidence from other countries [2, 4, 28]. Further adjusting for pre-existing health conditions had a modest effect on the HRs for most ethnic groups, with the largest reductions for Bangladeshi and Pakistani men during the second wave. The smaller reduction in HRs for other ethnic groups possibly indicates that differences in the prevalence of comorbidities associated with severe COVID-19 outcomes were already captured by adjusting for socioeconomic inequalities that are also associated with health status.

Our results are consistent with studies that investigated ethnic inequalities in SARS-CoV-2 infection and COVID-19 mortality in the first two waves of the pandemic. There is ample evidence that most ethnic minority groups were disproportionately affected in the first two waves of the pandemic [5‐7, 13, 29]. This study shows that the excess rate of death involving COVID-19 observed among ethnic minority groups early in the pandemic has continued throughout the rollout of the vaccine programme until late 2021. This is despite SARS-CoV-2 case rates being higher among the White British population from summer 2021 onwards [30, 31]. However, among people aged 65 years and over, cumulative case rates between March 2020 and October 2021 were higher for Black and South Asian groups than the White British group [32], suggesting that the age distribution of cases may partly explain the continued higher rates of COVID-19 mortality in these groups.

In addition, our findings suggest that lower vaccination rates (especially among Black Caribbean, Black African and Pakistani groups) contributes to explaining why some ethnic groups are more likely to experience more severe COVID-19 outcomes once infected. However, residual unexplained risk remained in the Bangladeshi group and men from the Pakistani group, even after full adjustment. People from Pakistani and Bangladeshi groups are more likely to reside in deprived areas, in large households and in multigenerational families [33]. Living in large, overcrowded, multi-generational households is associated with increased risk of SARS-CoV-2 infection [34, 35], and there is some evidence that living in a multi-generational household explains some of the differences in mortality [36]. Differences in occupational exposure may also account for some of the differences in mortality between groups, as a higher proportion of men from Pakistani and Bangladeshi groups work in key worker roles, such as healthcare workers, taxi drivers, shopkeepers, and proprietors than any other ethnic group [37], and these occupations have been found to be at elevated risk of COVID-19 mortality [38]. Whilst our study adjusted for a range of sociodemographic factors, including household composition and occupational exposure, these variables were retrieved from the 2011 Census, which may not reflect the situation in 2020 accurately. Consequently, adjustment for these factors might have been incomplete, possibly contributing towards the residual association.

The main strength of our study derives from using the ONS Public Health Data Asset, a nationwide large-scale population-wide data source combining the 2011 Census, mortality records, the General Practice Extraction Service (GPES) Data for Pandemic Planning and Research (GDPPR), Hospital Episode Statistics (HES), and vaccination data from the National Immunisation Management System (NIMS). Unlike studies based only on electronic health records, our study relies on self-identified ethnicity, limiting the potential for exposure misclassification bias. The PHDA also contains both detailed sociodemographic characteristics, such as household composition, housing quality, and occupational exposure, and extensive information on pre-pandemic health based on primary care and hospital records. To our knowledge, our study is the first to use nationally representative population-based linked data to examine the association between ethnicity and COVID-19 mortality in the third wave of the pandemic and explore the role of differences in vaccination uptake as a potential additional explanatory factor for the differences in COVID-19 mortality.

The main limitation is that most sociodemographic characteristics included in our models reflect the situations of individuals as they were in 2011, not necessarily those at the start of the COVID-19 pandemic. To mitigate this, we excluded people aged less than 30 years old, whose circumstances are the most likely to have changed since the Census. We also updated place of residence based on information from primary care records. As a result, information on area deprivation, rural/urban classification, region, and care home residence were up-to-date at the beginning of the pandemic. In addition, measurement error is likely to be smaller for the people at greater risk, since the sociodemographic factors are less likely to have changed for older people than younger people. However, some measurement error may reduce the explanatory power of the sociodemographic factors and pre-existing conditions included in the model, thus reducing their effect on the hazard ratios. For example, people who have retired since the 2011 Census will have incorrect data for current occupation. We have no reason to believe that there was substantial misdiagnosis of COVID-19 where it was mentioned on death certificates, especially in the latter part of the pandemic when testing was widely available.

Another limitation is that the study population is limited to people enumerated at the 2011 Census and therefore did not include people who were living in England but did not respond to the Census (estimated to be 5% of households) [39]. We also excluded people who did not link to the 2011 to 2013 National Health Service (NHS) Patient Registers. Whilst this does not affect the estimated association between ethnicity and COVID-19 mortality in our sample, it may affect the external validity of our study. Since the rate of linkage failure was higher for ethnic minority groups (with the highest rates of failure for the ‘other’, mixed and Chinese ethnic groups), the most likely result is an underestimation of the association between ethnicity and COVID-19 mortality in the population living in England at the beginning of the rollout of the COVID-19 vaccination programme, because a larger proportion of deaths occurring in ethnic minority groups will have been excluded. We sought to remove people who had emigrated since the 2011 Census from the population at-risk by restricting our study population to people registered with the NHS at the start of the pandemic. However, the study population also did not include people who immigrated or were born between 2011 and 2020. As a result, it did not fully represent the population at risk. Rates of immigration are higher for ethnic minority groups [40], which may mean that the denominator in our study population was less representative for some groups. In addition, deaths occurring among recent migrants would not have been included in our analyses. However, migrants overall tend to be young and the risk of COVID-19 mortality has been shown to be lower for young people [41]. Finally, single imputation using nearest-neighbour donor input for missing Census data may have inflated the statistical precision of Census variables.

Our study shows that adjusting for vaccination status eliminates the remaining elevated risk of COVID-19 mortality that is not explained by geographical factors, socioeconomic and demographic characteristics, and pre-existing health conditions in the third wave for Black African and Black Caribbean groups and reduces the remaining excess risk substantially for the Pakistani group (particularly for Pakistani women) compared to the White British group. The reduction in risk was more modest in the second wave, possibly reflecting widening ethnic inequalities in vaccination rates over time [10] and lower uptake of third vaccine doses during the third wave in some ethnic groups. As of 1 December 2021, the age-standardised rate of receiving three vaccine doses was 25.8% for the Black Caribbean group and 27.8% for the Black African group, compared to 46.1% for the White British group. Our results suggest that increasing vaccination uptake in ethnic minority groups could substantially reduce the inequalities in COVID-19 mortality. Strategies to increase vaccine uptake include improving trust in the efficacy and safety of vaccines by engaging with local community networks with lower vaccine uptake using appropriately tailored communication strategies. Identifying barriers to access, providing vaccinations in more convenient settings, and improving vaccine uptake in deprived areas could also increase vaccination rates in ethnic groups with low uptake [42, 43]. Future studies could also explore the potential impact that prioritisation for vaccination based on age, health conditions, and occupation may have had on long-term health outcomes (e.g. post-acute COVID-19 syndrome) in ethnic minority groups.

The elevated rate of death involving COVID-19 observed among ethnic minority groups in the first phase of the pandemic has continued throughout the rollout of the vaccine programme until late 2021. Whilst much of these differences can be explained by geography, socioeconomic factors, and pre-existing health conditions, differences in COVID-19 vaccine coverage are also a key driver of the elevated risk of COVID-19 death, particularly among the Black African, Black Caribbean, and Pakistani groups. Increasing vaccine uptake in ethnic minority groups would help reduce the inequalities in COVID-19 mortality.