HIV is linked to higher mortality from COVID-19 compared to HIV negative: 60% of deaths were black ethnicity

Simon Collins, HIV i-Base

Complimentary to the BHIVA audit, Public Health England presented their data on HIV and COVID-19 mortality from the first months of the epidemic.

This was defined as any death in adults (>15 years old) within 60 days of a COVID-19 diagnosis, or the specific inclusion of COVID-19 on the death certificate or enhanced surveillance form used in the study.

The study linked records in the PHE HIV surveillance data to the national COVID-19 surveillance system. Crude mortality rates were calculated for both HIV positive and HIV negative groups, including by age, sex, region, ethnicity and level of regional deprivation.

The study identified 115 cases of HIV positive people having a COVID-19 related death, with 99 confirmed by their HIV doctor and included in this analysis.

Overall, mortality rates were 107 vs 109/100,000 of HIV positive vs negative groups respectively.

Clear and significant differences however became clear where separating results by age.  For people aged 15 to 59 years old, rates were 58 vs 10 per 100,000 and for those aged >60 they were 434 vs 355 per 100,000, with 5-fold and 1.2 fold higher rates for the HIV positive vs negative groups respectively. The also illustrated the statistical Simpson’s paradox suggesting strong confounding with age.

In multivariate analysis, sex, age, ethnicity and HIV status were significantly associated with increased risk of COVID-19 related death, all p=0.0001 (in both HIV positive and negative groups), with an overall adjusted risk ratio for HIV of 2.18 (95%CI: 1.76 to 2.70).

The characteristics of the HIV positive people who died included median age 60 years and that 68% of deaths were black, Asian or other ethnic minority (compared to only accounting for 35% of the HIV population. This included 61 black, 5 Asian and 1 other/mixed. Adjusted rate ratio by ethnicity were aRR (3.44; 95%CI: 3.06 to 3.87) for black, (2.24; 95%CI: 2.00 to 2.52) for Asian and (3.23; 95%CI: 2.86 to 3.65) for other/mixed, compared to white ethnicity.

The HIV clinical profile (based on 94 forms) included median 15 years since HIV diagnosis, 88% having attended outpatient care at least once since 2018, 94% on ART, 58% with latest CD4 count <350 cells/mm3 and 91% with viral load <200 copies/mL.

Overall, 90% had any comorbidity, with 87% having more than one and 68% having more than two. The most common included cardiovascular (69%), obesity (49%), diabetes (48%), chronic kidney disease (41%) and hypertension (39%). These percentages were all significantly higher than reported for cause of death in 2019 (pre COVID-19).


This is important data with this analysis linking mortality with a lower CD4 count. The lower age among the HIV positive deaths should be included in public information on transmission risk.

The significant impact of ethnicity is also needs to be included in public information, likely due to higher occupational and complex social risks. Similar results for the general population in the UK from the OpenSAFELY observational study were also just published in the Lancet. [2]

 This study could also have underestimated HIV-associated rates as people with undiagnosed HIV would be counted as HIV negative and that many HIV positive people might have minimised their exposure risk by more careful shielding and social distancing during the early epidemic.

An updated analysis of results for the rest of 2020 will hopefully also be available soon.

Simon Collins is a community representative on this study.


  1. Croxford S et al. COVID-19 mortality among people with HIV compared to the general population during the first wave of the epidemic in England. Joint BHIVA BASHH Spring Conference, 2021. Oral abstract O-009.
  2. Mathur R et al. Ethnic differences in SARS-CoV-2 infection and COVID-19-related hospitalisation, intensive care unit admission, and death in 17 million adults in England: an observational cohort study using the OpenSAFELY platform. The Lancet, doi: 10.1016/S0140-6736(21)00634-6. (30 April 2021).

(The BHIVA link should be open access from approximately 20 May 2021).

Links to other websites are current at date of posting but not maintained.