AbstractBackgroundPreliminary studies suggest that people from Black, Asian and Minority Ethnic (BAME) backgrounds experience higher mortality from COVID-19 but the underlying reasons remain unclear.MethodsProspective analysis of registry data describing patients admitted to five acute NHS Hospitals in east London, UK for COVID-19. Emergency hospital admissions with confirmed SARS-CoV-2 aged 16 years or over were included. Data, including ethnicity, social deprivation, frailty, patient care and detailed risk factors for mortality, were extracted from hospital electronic records. Multivariable survival analysis was used to assess associations between ethnic group and mortality accounting for the effects of age, sex and various other risk factors. Results are presented as hazard ratios (HR) or odds ratios (OR) with 95% confidence intervals.Findings1996 adult patients were admitted between 1st March and 13th May 2020. After excluding 259 patients with missing ethnicity data, 1737 were included in our analysis of whom 511 had died by day 30 (29%). 538 (31%) were from Asian, 340 (20%) Black and 707 (40%) white backgrounds. Compared to White patients, those from BAME backgrounds were younger, with differing co-morbidity profiles and less frailty. Asian and Black patients were more likely to be admitted to intensive care and to receive invasive ventilation (OR 1·54, [1·06-2·23]; p=0·023 and 1·80 [1·20-2·71]; p=0·005, respectively). After adjustment for age and sex, patients from Asian (HR 1·49 [1·19-1·86]; p<0·001) and Black (HR 1·30 [1·02-1·65]; p=0·036) backgrounds were more likely to die. These findings persisted across a range of risk-factor adjusted analyses.InterpretationPatients from Asian and Black backgrounds are more likely to die from COVID-19 infection despite controlling for all previously identified confounders. Higher rates of invasive ventilation in intensive care indicate greater acute disease severity. Our analyses suggest that patients of Asian and Black backgrounds suffered disproportionate rates of premature death from COVID-19.FundingNoneResearch in contextEvidence before this studyWe searched PubMed, Google Scholar, Medrxiv, Trip Medical Database and internet search engines from inception to May 10th 2020, using the terms “(COVID-19 or 2019-nCoV or SARS-CoV-2) AND (ethnicity)”, with no language restrictions, for research articles, editorials and commentaries. We identified 25 articles. Ten were international opinion pieces, fifteen were research articles reporting analyses of national and cohort datasets, predominantly in the United Kingdom (UK) and United States (US). Each of these studies indicated an increased risk of adverse outcomes in people from BAME backgrounds; either in terms of COVID-19 acquisition, disease severity or mortality. However, the underlying causes were unclear. Aggregated US data determined the relative risk of death for those of Black ethnicity compared to White ethnic groups to be 3.57. Three UK biobank cohort studies, limited by low BAME representation, described ethnicity as an independent risk factor of COVID-19 infection, partially attenuated by socio-economic status (SES). Analysis of a London hospital cohort of 520; experiencing 144 deaths, revealed an age and co-morbidity adjusted mortality odds ratio of 1.72 in Black populations of borderline significance. Age and geographical region-adjusted standardised mortality ratios, derived from UK composite hospital data, emphasised ethnic differences; being 2.41 for Bangladeshis and 3.24 for Black Africans. The impact of gender and deprivation was not explored. Another study of 5683 in-hospital deaths (England alone; 629 (11%) BAME) confirmed increased mortality risks in people from Black and Asian groups only partially attributable to social deprivation and co-morbidity but did not adjust for other vulnerability factors. There remained a need for a more detailed analysis of outcomes across different ethnic groups in a large, high acuity dataset, adjusting for broader clinical and laboratory prognostic factors, alongside SES, smoking status, age, body mass index (BMI) and sex.Added value of this studyWe conducted a large observational cohort study of COVID-19 hospital admissions within an area which experienced the highest rates of COVID-19 infection and mortality in the UK. It offers detailed insight into a majority (60%) ethnically diverse cohort and adds substantial evidence that ethnicity is a predictor of poor outcomes for COVID-19 patients at, and beyond, 30 days. Using robust multivariable survival analyses we have quantified and described the impact on this association of a number of additional prognostic factors such as frailty score and markers of inflammation alongside age, sex, deprivation, co-morbidity, BMI and smoking status. Those of Asian and Black ethnicities were consistently found to have an increased risk of 30 and 90 day mortality and an increased risk of requiring mechanical ventilation as compared to those of White ethnicity. The peak CRP and D-dimer levels in those of Black ethnicity were significantly higher than those of other ethnicities suggesting that these biological differences may accompany greater disease severity and increased risk of adverse outcomes.Implications of all the available evidenceIt is clear that ethnicity is a predictor of a positive SARS-CoV2 result, disease severity and mortality, regardless of age, sex, geographical location, deprivation, smoking status, BMI, co-morbidities and frailty. The association appears to be underpinned by a combination of factors including SES, pre-existing health conditions, biological risk factors such as D-dimers, environmental and structural determinants of health; but their relative contribution is unclear. Understanding these drivers is critical to designing interventions and refining clinical and Public Health policies. The evidence also emphasises the need for robust surveillance of ethnicity in health care research.
Biometric covariates and outcome in COVID-19 patients: are we looking close enough?
