Why Are Non-White Britons More Likely to Die of COVID-19?
Evidence suggests that geographic and socio-economic factors may not be sufficient to explain non-white Britons’ elevated risk of death from COVID-19
It has been widely reported that non-white Britons (as well as non-white Swedes and black Americans) are overrepresented among those who have died of COVID-19. In a recent paper, Robert Aldridge and colleagues computed standardised mortality ratios for all major ethnic groups in the UK. Compared to the general population, the age-adjusted risk of death was 4 times higher for Pakistanis, 4 times higher for Bangladeshis, 2.6 times higher for Indians, 5.7 times higher for black Africans and 3.7 times higher for black Caribbeans.
The causes of these disparities are not yet well-understood. As a consequence, the government has launched an official inquiry into non-white Britons’ elevated risk of death. In this blog post, I will consider some of the explanations that have been put forward in the literature so far. While it is too early to draw definitive conclusions, evidence suggests that geographic and socio-economic factors do not fully explain non-white Britons’ elevated risk of death.
Conceptually, there are two possible reasons for the overrepresentation of non-white Britons among those who have died from COVID-19: they could be more likely to become infected with coronavirus, and they could be more likely to die conditional on becoming infected (or some combination of the two). Why would non-white Britons be more likely to become infected?
The most obvious reason is geography. Non-white Britons are concentrated in densely populated urban areas, such as central London and Sheffield, where the number of coronavirus cases has been highest. At the last UK census in 2011, only 79% of whites lived in urban areas, compared to 97% of Asians and 98% of blacks.
Another possible reason is household structure. Non-white Britons are more likely to live in multi-generational households, where there is a greater chance that an asymptomatic grandchild could pass the virus to an elderly grandparent. At the last UK census, only 2% of whites gave their household type as “Other, with dependent children” (which includes multi-generational households), compared to 11% of Asians and 7% of blacks.
A third possible reason is occupation. Non-white Britons are more likely to work in professions that have been deemed essential by the government, including the NHS. According to an IFS analysis, the percentage of the working-age population employed as key workers is substantially higher for blacks than for whites. And the percentage employed in health or social care is higher for Indians, Pakistanis, black Africans and black Caribbeans.
A fourth possible reason is socio-economic deprivation. Non-white Britons are more likely to live in deprived neighbourhoods, where the level of overcrowding may be higher, and the availability of personal protective equipment may be lower. At the last UK census, 9% of whites lived in the most deprived 10% of neighbourhoods, compared to 17% of Asians and 20% of blacks.
Hence there are at least four non-medical reasons why non-white Britons would be more likely to become infected with coronavirus: geography; household structure; occupation; and socio-economic deprivation. However, evidence suggests that these factors may not be sufficient to explain non-white Britons’ elevated risk of death from COVID-19.
First, some of the comparisons between ethnic groups go in the opposite direction from what would be expected based on the reasoning above. For example, according to the IFS, the percentage of the working age population employed as key workers is slightly higher for whites than for Bangladeshis and Pakistanis. Likewise, the percentage of whites living in the most deprived 10% of neighbourhoods at the last census was slightly higher than the corresponding percentage of Indians: 8.7% versus 8.3%.
Second, two major multivariate analyses have found that ethnic disparities in risk of death from COVID-19 persist after controlling for some or all of the factors mentioned above.
ONS researchers linked data from death registrations to geographic and socio-economic data from the 2011 census. They estimated a model that controlled for (among other things): region, an indicator for urban versus rural, area deprivation, household structure, and individual-level socio-economic status. Even after adjusting for these factors, they found that blacks, Bangladeshis/Pakistanis and Indians had a higher risk of death from COVID-19 than whites:
In addition, the OpenSAFELY Collaborative analysed a sample of over 10 million anonymised NHS patient records. They estimated a model that controlled for (among other things) the level of socio-economic deprivation in the patient’s postcode, and the patient’s Sustainability and Transformation Partnership (a more fine-grained administrative region than the one used in the ONS analysis). After adjusting for these and other factors, they again found that blacks and Asians had a higher risk of death from COVID-19 than whites.
Third, analyses have found that non-white Britons are substantially overrepresented among NHS staff who have died from COVID-19. Such analyses are of particular interest because they focus on the section of the population that is most exposed to coronavirus. According to the ONS, blacks and Asians comprise 16.1% of the overall NHS workforce, and 34.3% of the NHS medical workforce. Non-whites in general (including people of mixed ethnicity, Chinese and “Other”) comprise 21% of the overall NHS workforce, and 44% of the NHS medical workforce.
However, Tim Cook and colleagues identified 106 NHS staff who had died up to 22 April, and found that 63% were non-white (specifically black or Asian). They also found that 94% of the 19 doctors/dentists who had died were non-white. Given that medicine and dentistry are high-status professions, this figure cannot easily be explained by socio-economic deprivation. In addition, Sky News reported that 72% of health and social care staff who had died of COVID-19 up to 22 April were non-white. (Note that they did not provide any information about their methodology.)
If geographic and socio-economic factors do not fully explain non-white Britons’ elevated risk of death from COVID-19, what might explain the remaining gap? One possibility is pre-existing conditions. People with pre-existing conditions are more likely to die from COVID-19 if they become infected, and there is evidence that some such conditions are more common in non-white Britons. For example, South Asians are more likely to develop coronary heart disease, while Afro-Caribbeans are more likely to develop high blood pressure, and both groups are more likely to develop type 2 diabetes.
However, the two multivariate analyses mentioned above both included controls for pre-existing conditions, and still found that non-whites had an elevated risk of death. The ONS researchers controlled for self-reported health (from “very poor” to “very good”), as well as an indicator for whether the individual had any activity-limiting health problem or disability. Given that their data came from patient records, the OpenSAFELY Collaborative were able to control for a large number of specific health conditions: obesity, smoking status, type 2 diabetes, cancer, chronic respiratory disease, chronic cardiac disease, and others. (They were not able to control for household structure or individual-level socio-economic status.)
The ONS researchers noted the following:
This means that a substantial part of the difference in COVID-19 mortality between ethnic groups is explained by the different circumstances in which members of those groups are known to live, such as areas with socio-economic deprivation. Geographic and socio-economic factors were accounting for over half of the difference in risk between males and females of Black and White ethnicity. However, these factors do not explain all of the difference, suggesting that other causes are still to be identified.
The OpenSAFELY Collaborative noted the following:
Non-white ethnicity has previously been found to be associated with increased COVID-19 infection and poor outcomes. Commentators and researchers have reasonably speculated that this might be due to higher prevalence of medical problems such as cardiovascular disease or diabetes among BME people, or higher deprivation. Our findings, based on more detailed data, show that this is only a small part of the excess risk. Other possible explanations for increased risk among BME groups relate to higher infection risk, including over-representation in ‘front-line” professions with higher exposure to infection, or higher household density.
In addition, Tim Cook and colleagues (who analysed NHS staff deaths) noted the following:
In the USA, the preponderance of deaths among BAME groups has also been noted and is of rising political importance. The causes of this excess mortality are not clear and could be biological (genetic susceptibility), medical (due to association with diseases such as hypertension and diabetes which are risk factors for poor outcome from covid-19), or sociological (due to employment and working patterns that increase risk of exposure and transmission).
It appears, then, that taking into account geography, socio-economic factors and pre-existing conditions may not fully explain non-white Britons’ elevated risk of death from COVID-19.
One possible explanation for the remaining gap — which has been suggested by a number of commentators — is higher prevalence of vitamin D deficiency in non-whites (specifically blacks and South Asians). Exposure of the skin to UVB rays promotes vitamin D synthesis, but pigmentation is believed to interfere with this process. UVB intensity is lower at at northern latitudes, meaning that people with dark skin obtain less vitamin D via photoproduction. (Whatever the precise mechanism, it is a fact that black and South Asian Britons are much more likely to be vitamin D deficient.)
In a recent editorial, The BMJ asked, “Is ethnicity linked to incidence or outcomes of covid-19?” This editorial prompted a number of responses, two of which argue that vitamin D deficiency may be part of the answer. As Shamil Haroon and colleagues note:
The effect of vitamin D on immune function and respiratory health has been known for some time. Vitamin D reduces pulmonary inflammation and enhances innate and adaptive immunity to respiratory pathogens […] Vitamin D also increases the phagocytic potential of macrophages and inhibits the maturation of dendritic cells […] Vitamin D inhibits a potentially harmful chemokine and cytokine response while maintaining antiviral activity, which may prove particularly important in COVID-19, in which critically ill patients experience a cytokine storm.
Indeed, it has been known for some time that vitamin D deficiency can increase both the risk and severity of respiratory tract infections. Hence even in the absence of any specific evidence, there would be reason to believe that vitamin D deficiency increases the risk of death from COVID-19—which is primarily a disease of the respiratory tract. Nonetheless, there have already been a number of studies showing that vitamin D deficiency is related to the risk and severity of COVID-19 itself. Some of these are reviewed in a recent article by Maryam Ebadi and Aldo Montano-Loza. And many additional studies are available as preprints.
Of particular interest is a recent preprint by Adam Li and colleagues, who carried out a multivariate analysis of the number of cases and deaths from COVID-19 across US counties. They controlled for a large number of variables including population density, GDP per capita, overcrowded housing, geographical mobility, violent crime, poverty, air quality, obesity, and many others. They found that, when adjusting for all these factors, counties with a higher percentage of blacks had more cases and deaths from COVID-19. As the authors note:
U.S. counties with a higher proportion of Black residents are associated with increased COVID-19 cases and deaths. However, the various suggested mechanisms, such as socioeconomic and healthcare predispositions, did not appear to drive the effect of race in our model. […] Several theories are posited to explain these findings, including prevalence of vitamin D deficiency.
In conclusion, evidence suggests that geographic and socio-economic factors are not sufficient to explain non-white Britons’ elevated risk of death from COVID-19, and that taking into account pre-existing conditions may not fully explain the gap. The most plausible explanation for the remaining gap is vitamin D deficiency. While this explanation is still tentative, there is a reasonable amount of circumstantial evidence supporting it. Given the low costs associated with vitamin D supplementation, recommending this for at-risk groups such as non-white people and the elderly would seem to be a top priority for Western governments.