key: cord-0757779-qx3o4bsx
authors: Edge, Rhiannon; van der Plaat, Diana A; Parsons, Vaughan; Coggon, David; van Tongeren, Martie; Muiry, Rupert; Cullinan, Paul; Madan, Ira
title: Ethnic differences in risk of severe Covid-19: To what extent are they driven by exposure?
date: 2021-09-21
journal: J Public Health (Oxf)
DOI: 10.1093/pubmed/fdab347
sha: 0f745a97be9d67b5711d3cdd3a5b32f112f71254
doc_id: 757779
cord_uid: qx3o4bsx

BACKGROUND: This study quantifies the risk of Covid-19 among ethnic groups of healthcare staff during the first pandemic wave in England. METHODS: We analysed data on 959 356 employees employed by 191 National Health Service trusts during 1 January 2019 to 31 July 2020, comparing rates of Covid-19 sickness absence in different ethnic groups. RESULTS: In comparison with White ethnic groups, the risk of short-duration Covid-19 sickness absence was modestly elevated in South Asian but not Black groups. However, all Black and ethnic minority groups were at higher risk of prolonged Covid-19 sickness absence. Odds ratios (ORs) relative to White ethnicity were more than doubled in South Asian groups (Indian OR 2.49, 95% confidence interval (CI) 2.36–2.63; Pakistani OR 2.38, 2.15–2.64; Bangladeshi OR 2.38, 1.98–2.86), while that for Black African ethnicity was 1.82 (1.71–1.93). In nursing/midwifery staff, the association of ethnicity with prolonged Covid-19 sickness absence was strong; the odds of South Asian nurses/midwives having a prolonged episode of Covid-19 sickness absence were increased 3-fold (OR 3.05, 2.82–3.30). CONCLUSIONS: Residual differences in risk of short term Covid-19 sickness absences among ethnic groups may reflect differences in non-occupational exposure to SARS-CoV-2. Our results indicate ethnic differences in vulnerability to Covid-19, which may be only partly explained by medical comorbidities.

The disproportionate impact of the Covid-19 pandemic on minority ethnic groups in the UK is now well established 1 but not fully understood. During the first wave (24 January 2020 to 11 September 2020), people from all ethnic minority groups (except for women in the Chinese or 'White Other' ethnic groups) had higher rates of death involving SARS-CoV-2 than the White British population. The rate was highest for the Black African group (3.7 times greater than for the White British group for males, and 2.6 greater for females), followed by the Bangladeshi (3.0 for males, 1.9 for females), Black Caribbean (2.7 for males, 1.8 for females) and Pakistani (2.2 for males, 2.0 for females) ethnic groups.

These findings could arise from differences in exposure to infection and/or differences in vulnerability to more severe disease when infection occurs. Vulnerability to Covid-19 is related to age, sex and various comorbidities. One factor that contributes to exposure to SARS-CoV-2 infection is occupation. If minority ethnic groups were employed disproportionately in occupations entailing proximity to other people, particularly people who are more likely to be infected with SARS-CoV-2, then they would be at higher risk of infection. Exposure to infection will depend also on other factors such as household size and composition, housing density, and non-occupational activities and behaviours. 2 Large record linkage studies such as OpenSAFELY suggest that important differences in mortality by ethnicity persist even after allowance for region, social deprivation, sex, age and multiple comorbidities. 3 However, it remains possible that there are differences in exposures through work, and to date, few studies have been able to adjust well for occupational differences in exposure.

The aim of our study was to determine whether ethnic differences in risk of less serious Covid-19 (which is less likely to be influenced by differences in vulnerability) were apparent during the first wave of the pandemic among healthcare workers in England in specific job categories, after adjustment for potential exposure to infected patients and geographical variation in rates of infection.

As detailed in an earlier report, 4 we analysed pseudonymized data abstracted from the National Health Service (NHS) electronic staff record (ESR) for all personnel who had been continuously employed by NHS trusts in England during 01 January 2019 to 31 July 2020.

In the analysis for this paper, we focused on two main outcomes-(i) Covid-19 sickness absence beginning between 09 March 2020 and 16 July 2020, at least one episode of which was prolonged (i.e. with duration >14 days); and (ii) Covid-19 sickness absence during the same period that was only ever of shorter duration. Covid-19 sickness absence was defined as sickness absence ascribed to any of five diagnostic categories (cough/flu, chest/respiratory, infectious diseases, other and unknown) with Covid-19 recorded as a related reason.

The main explanatory variables of interest were ethnicity and staff group. Ethnicity was classified initially to the 12 categories listed in Table 1 , but in some analyses, we aggregated all South Asian ethnic groups and all Black ethnic groups to ensure statistically meaningful numbers. Staff group was classed to nine categories (Table 1) , following a scheme that was employed in the ESR, but with students aggregated into a category labelled as 'Other or unknown', which also included some individuals who held multiple jobs simultaneously. As in our earlier report, 4 where individuals had changed staff group over the study period, we aimed to classify them according to the job held on 09 March 2020.

In addition, we considered five other explanatory variables-trust (191 categories) sex, age group (8 categories), number of episodes of sickness absence in 2019 (4 categories) and exposure category. The last was assigned by application of a job-exposure matrix to the occupation (659 possible categories) that the individual held on 09 March 2020. It was assigned to two levels according to whether or not the occupation was judged to involve face-to-face or hands-on care of patients who were more likely to have Covid-19 than the general population. In earlier analyses, such exposure was associated with clearly elevated risk of Covid-19 sickness absence. 4 The other variables were classified as in our previous report. 4 Statistical analysis was carried out using R statistical software. We used logistic regression to estimate odds ratios (ORs) with 95% confidence intervals (CIs) for the two outcomes in relation to combinations of ethnicity and staff group with adjustment for other explanatory variables.

Ethical approval to conduct the study was obtained from the NHS Health Research Authority (reference 20/SC/0282).

After exclusion of 3811 employees who were absent from work continuously between 09 March 2020 and 31 July 2020 (mainly because of maternity or study leave), analysis was based on 959 356 individuals (77% female) from 191 trusts. Most (89%) were aged between 25 and 60 years. Detailed information on the numbers of individuals by age band and by frequency of sickness absence during 2019 has been reported elsewhere. 5 From application of the job-exposure matrix, 383 097 (39.9%) employees held jobs at 09 March 2020, which were classed as providing hands-on or faceto face care for patients who could be expected to have a higher prevalence of Covid-19 than the general population. Table 1 shows the distribution of the study sample according to staff group at 9 March 2020 and ethnic group. Among staff of Asian ethnicity, the proportion employed as doctors or dentists was some five times higher than in White workers. Relatively high proportions of the Black ethnic groups, and especially Black African, were registered nurses or midwives.

In total, 20 988 individuals (2.2%) had at least one episode of Covid-19 sickness absence that started between 09 March 2020 and 16 July 2020 and continued for >14 days (prolonged Covid-19 sickness absence). In addition, a further 70 863 (7.4%) had episodes of Covid-19 sickness absence during that period, all of which were of shorter duration. Table 2 shows associations of Covid-19 sickness absence with ethnicity and staff group, according to whether absence was only ever of short duration (≤14 days), or at least one episode was prolonged. In comparison with White ethnicity, the risk of short-duration Covid-19 sickness absence was modestly elevated in Indian (OR 1. Table 3 presents risk estimates by ethnic group for Covid-19 sickness absence that was only ever of short duration, when analyses were restricted to specific staff groups. To ensure adequate numbers, for this analysis we aggregated all South Asian ethnic groups and all Black ethnic groups. The reference was no Covid-19 sickness absence at any time during the study period. The higher risks of short-duration Covid-19 sickness absence in Asian and/or South Asian ethnic groups were apparent in most staff groups but were not observed among doctors and dentists (OR 0.99, 95% CI 0.92-1.07). Table 4 gives findings from analyses analogous to those for Table 3 , but with at least one prolonged episode of Covid-19 sickness absence as the outcome. Within each staff group, risk was highest in the South Asian and/or the other/unspecified Asian ethnic groups, with ORs (relative to White) substantially higher than for short-duration Covid-19 sickness absence. In contrast to the findings for shorter duration Covid-19 sickness absence, Black people were at an increased risk (relative to White) of prolonged Covid-19 sickness absence in several staff groups.

In sensitivity analyses, we repeated the calculations for Tables 2-4, after exclusion of 6854 individuals for whom one or more of age, sex or ethnicity was imputed because of inconsistencies in the raw data. The results, which are presented in Supplementary Tables S1-S3, were virtually unchanged.

Our analysis confirms that during the first wave of Covid-19 in England there were differences between ethnic groups in risk of short and longer duration Covid-19 sickness absence among NHS staff. Once staff group, age, sex, prior sickness absence, trust and occupational exposure category were accounted for, the risk of short duration Covid-19 was similar for Black people compared with White and only marginally elevated for people of South Asian origin. In contrast, staff from Black and other ethnic minority groups were at a higher risk of prolonged Covid-19 sickness absence compared to White NHS employees, suggesting important ethnic differences in vulnerability, whether because of comorbidities or for other reasons.

Multiple population-based studies have suggested that people from both Black and South Asian ethnic groups face an increased risk of SARS-CoV-2 infection compared to White people. 6, 7 However, this increase in risk can be at least partially explained by differences in socio-economic circumstances such as household size, number of dependent children and living in a deprived area. 6 A cohort study found that critical care admissions in the UK were more common in South Asian (OR 1.28, 95% CI 1.09-1.52), Black (OR 1.36, 95% CI 1.14-1.62) and other minority ethnic groups (OR 1.29, 95% CI 1.13-1.47) than White people. 8 A study of UK Biobank participants found that Black and Asian participants were at an increased risk of Covid-19 hospitalization compared to White participants; adjusting for socioeconomic factors and cardiorespiratory comorbidities led to some attenuation, but not complete elimination, of the increased risk in Black (OR 2.38 95% CI 1.52-3.74) and Asian participants (OR 1.75 95% CI 1.08-2.85). 9 However, unlike the work presented here, these studies did not adjust for occupational exposure.

This large study is the first to examine the associations of ethnicity with Covid-19 sickness absence in UK healthcare workers while accounting for occupational group and potential for exposure to infected patients. The sample size of almost a million individuals gave the investigation high statistical power and allowed us to investigate ethnic groups in detail (e.g. separating workers of Indian and Pakistani origin). Occupational groups were analysed separately, and an attempt was made to adjust for occupational exposure by using a bespoke jobexposure matrix. The effect of geographical differences in exposure to infection was accounted for by adjustment for hospital trust.

We explored the risk of short-duration sickness absence attributed to Covid-19 among NHS staff as a proxy for less serious Covid-19, which is less likely to be influenced by differences in vulnerability. By adjusting for the potential occupational exposure to infected patients (assessed by the job-exposure matrix), as well as trust (a specific geographical marker), sex and age, we have shown that any differences in risk of mild Covid-19 by ethnicity were small. The residual variation may reflect differences in exposure that were not adequately captured by staff group and exposure category.

In contrast, the difference in risk of prolonged Covid-19 among Black and ethnic minority groups compared to White was more exaggerated than for short-duration Covid-19 sickness absence. Within each staff group, the risk of prolonged Risk estimates are relative to no Covid-19 sickness absence during study period and were derived from two logistic regression models (one per outcome), each of which also included trust (200 categories), sex, age group (8 categories), number of episodes of sickness absence in 2019 (4 categories) and exposure category at 9 March 2020 (two categories)-for further detail, see text.

ref. is in italics in the table, this denotes the reference group for the odds ratios, but again this does not need a footnote. ref. is in italics in the table, this denotes the reference group for the odds ratios, but again this does not need a footnote. Covid-19 sickness absence was highest in the South Asian and/or the other/unspecified Asian ethnic groups, and often the odds were twice those of White people. Our findings that ethnic minority groups are at higher risk of severe Covid-19 are supported by several other studies. In our study, ethnic disparities in short-duration Covid-19 sickness absence were not observed among those employed as healthcare scientists or doctors and dentists, in contrast to those employed in other roles within the NHS. It may be that non-occupational risk factors for infection differ less by ethnicity within these groups than in other job groups. Within healthcare scientists, doctors and dentists, ethnic differences were apparent, however, for longer duration Covid-19 sickness absence, again suggesting differences in vulnerability to severe illness when infection occurs.

Ethnicity was coded in the ESR with varying degrees of specificity and not always consistently. Exposure category was defined based on employment at 9 March 2020 and did not capture redeployment to different clinical settings during the pandemic. We were not able to account for use of personal protective equipment which may have biased our analysis if it differed by ethnicity within job groups. A British Medical Association snapshot survey taken early in the first wave of the pandemic suggested that a higher proportion (68%) of doctors from minority ethnic groups felt pressured to work with inadequate personal protective equipment where aerosol-generating procedures were being carried out, than those who identified as White (33%). 10 A further limitation is that sickness absence is an imperfect marker for the occurrence of Covid-19, and it is possible both that true cases were missed (due to asymptomatic illness) and that other respiratory illnesses were sometimes incorrectly attributed to coronavirus. However, our previous analysis showed that Covid-19 sickness absence correlated with seropositivity for SARS-Cov-2. 4

Supplementary data are available at the Journal of Public Health online.

Is ethnicity linked to incidence or outcomes of covid-19?

Disparities in the Risk and Outcomes of COVID-19

Factors associated with COVID-19-related death using OpenSAFELY

Occupational risks of COVID-19 in NHS workers in England

Changing patterns of sickness absence among healthcare workers in England during the COVID-19 pandemic

SARS-CoV-2 antibody prevalence in England following the first peak of the pandemic

Ethnic and socioeconomic differences in SARS-CoV-2 infection: prospective cohort study using UK Biobank

Ethnicity and Outcomes from COVID-19: The ISARIC CCP-UK Prospective Observational Cohort Study of Hospitalised Patients

Social Science Research Network

Race, socioeconomic deprivation, and hospitalization for COVID-19 in English participants of a national biobank

BAME Doctors Hit Worse by Lack of PPE

We are very grateful to the following, without whom the study would not have been possible: Sam Wright, Workforce Information Advisor, NHS ESR, and Mike Vickerman, Workforce Information and Analysis, DHSC. Dr Gavin Debrera (Public Health England) and Dr Kit Harling gave invaluable help in planning the study.

All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf and declare: no support from any organization for the submitted work; no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years; no other relationships or activities that could appear to have influenced the submitted work.

This work was supported by the Colt Foundation UK.

With permission, source data are available upon request from the NHS Electronic Staff Record (ESR) Warehouse (NHS England).

All authors contributed to the planning, conduct, analyses and reporting of this manuscript as outlined below.R.E. was responsible for advising on study design, analysis and interpretation of results. D.A.v.d.P. was responsible for the statistical aspects of analysis and interpretation of the quantitative aspects of the study. V.P. was responsible for overseeing the set-up and delivery of the study and facilitated data collection. D.C. was responsible for advising on methodological design, analysis and interpretation of results. M.v.T. was responsible for advising on study design, analysis and interpretation of results. R.M. was responsible for scoping out and reviewing the emerging literature. I.M. was co-chief investigator with responsibility for advising on study design, analysis and interpretation of results. P.C. was chief investigator with responsibility for advising on study design, analysis and interpretation of results. He had overall responsibility for the management and delivery of the study.