Background: Multiple reports suggested disproportionate impact of Covid-19 on ethnic minorities. Whether ethnicity is an independent risk factor for severe Covid-19 illness is unclear.
Purpose: Review the association between ethnicity and poor Covid-19 outcomes including all-cause mortality, hospitalisation, critical care admission, respiratory and kidney failure.
Data Sources: MEDLINE, EMBASE, Cochrane COVID-19 Study Register, WHO COVID-19 Global Research Database up to 15/06/2020, and preprint servers. No language restriction.
Study Selection: All studies providing ethnicity-aggregated data on the pre-specified outcomes, except case reports or interventional trials
Data Extraction: Pairs of investigators independently extracted data, assessed risk of bias using Newcastle-Ottawa scale (NOS), and rated certainty of evidence following GRADE framework.
Data Synthesis: Seventy-two articles (59 cohort studies with 17,950,989 participants; 13 ecological studies; 54 US-based and 15 UK-based; 41 peer-reviewed) were included for systematic review and 45 for meta-analyses. Risk of bias was low, with median NOS of 7 (interquartile range 6-8). In the unadjusted analyses, compared to white ethnicity, all-cause mortality risk was similar in Black (RR:0.96 [95%CI: 0.83-1.08]), Asian (RR:0.99 [0.85-1.16]) but reduced in Hispanic ethnicity (RR:0.69 [0.57-0.84]). Age and sex-adjusted-risks were significantly elevated for Black (HR:1.38 [1.09-1.75]) and Asian (HR:1.42 [1.15-1.75]), but not for Hispanic (RR:1.14 [0.93-1.40]). Further adjusting for comorbidities attenuated these association to non-significance; Black (HR:0.95 [0.72-1.25]); Asian (HR:1.17 [0.84-1.63]); Hispanic (HR:0.94 [0.63-1.44]). Similar results were observed for other outcomes. In subgroup analysis, there is a trend towards greater disparity in outcomes for UK ethnic minorities, especially hospitalisation risks.
Limitations: Paucity of evidence on native ethnic groups, and studies outside US and UK.
Conclusions: Currently available evidence cannot confirm ethnicity as an independent risk factor for severe Covid-19 illness, but indicates that disparity may be partially attributed to greater burden of comorbidities.
Registration: PROSPERO, CRD42020188421
Funding source: none