HIV positive people in South Africa at increased risk of dying from COVID-19: first data from country with high prevalence of HIV and TB

Polly Clayden, HIV i-Base

Preliminary data from the Western Cape, South Africa show people with HIV and TB have a 2- to 3-fold increased risk of death from COVID-19.

But the increases were much higher for other known risk factors. Older age increased the risk by 10 to 20 times (if over 50 or 70, respectively) and diabetes increased this by 5 to 13 times (depending on whether controlled or uncontrolled).

These findings were presented by Mary-Ann Davies from the Western Cape Department of Health, on a webinar organised by the Bhekisisa Centre for Health Journalism in collaboration with the Aurum Institute, South Africa.

For this analysis, which Professor Davies explained was not a formal study, the department reviewed 12,987 COVID-19 cases seen in the public sector – this included 435 deaths. The analysis revealed that slightly above half of the deaths were associated with diabetes, about 12% HIV and 2% active TB.

These findings are important as this is the first data looking at risk factors for with COVID-19 from a country with two key high burden comorbidities: HIV and TB. There has been limited data so far on whether or not these comorbidities will increase the risk of poor outcomes from COVID-19.

To date, known risk factors from other settings include: older age, male sex, diabetes, cardiac disease, respiratory disease, kidney disease, liver disease, overweight/obesity, organ transplant and recently diagnosed cancer. Some risk factors may be linked, such as diabetes and overweight/obesity, but this data does not include information on BMI or smoking.

The analysis looked at factors associated with COVID-19 death in all adult public sector patients 20 years of age and above.

Western Cape public sector data is brought together in the Public Health Data Centre (PHDC) using a unique identifier across all systems: primary care, hospitals, emergency, disease specific, laboratory, dispensing, community, births and deaths.

Several comorbidities can be inferred from lab tests and medication dispensed: diabetes, hypertension, chronic kidney disease, chronic respiratory disease/asthma, TB and HIV. But the data does not capture other risk factors such as overweight/obesity, smoking and socio-economic status.

Table 1 shows the adjusted hazard ratios for dying from COVID-19 for different risk factors.

Table 1: Chances of dying from COVID-19 for different risk factors  

Patient characteristics

Adjusted hazard ratio

95% confidence interval


Female 1
Male 1.4

1.16 to 1.7

Age (years)
Less than 40 1
40–49 3.12 1.88 to 5.17
50–59 9.92 6.34 to 15.54
60–69 13.55 8.55 to 21.48
70 and above 19.53 12.20 to 31.26
Non-communicable disease
None 1
Diabetes well-controlled 4.65 3.19 to 6.79
Diabetes poorly-controlled 8.99 6.65 to 15.54
Diabetes uncontrolled 13.02 8.55 to 21.48
Diabetes – no measure of control 3.34 12.20 to 31.26
Hypertension 1.46 1.18 to 1.81
Chronic kidney disease 2.02 1.55 to 2.62
Chronic pulmonary disease 0.98 0.75 to 1.30
Never TB 1
Previous TB 1.41 1.05 to 1.90
Current TB 2.58 1.53 to 4.37
Negative 1
Positive 2.75 2.09 to 3.61

Older age of 70 years and above was the highest risk factor giving an approximately 20-fold risk of death compared to that in people aged 40 and below. Diabetes was associated with an approximately 13-fold risk if uncontrolled and just below 5-fold risk if well-controlled.

Both HIV and active TB were associated with a 2- to 3-fold risk of dying from COVID-19. Notably there was no difference by viral suppression among people with HIV dying from COVID-19.

Men were more at risk than women but this difference was modest.

Professor Davies explained that for every 100 people in the public sector who have died from COVID-19, 52 can be attributed to diabetes, 12 to HIV, 2 to current TB and 4 to previous TB.

As these data are limited to the public setting the group calculated the Standardised Mortality Ratios (SMR) for the increase in COVID-19 death in people with vs without HIV in Western Cape: 2.33 (95% CI 1.83 to 2.91). So across the public and private sector, about 8% of COVID-19 deaths were due to HIV.

She added that the risk might be over-estimated if the analysis was not able to disentangle all comorbidities and risks eg overweight and socio-economic status. And that those with HIV and TB tend to be younger where overall risk of COVID-19 death is low.


These data are extremely important as the first from a setting with large numbers of people with HIV and TB. Professor Davies said that although the numbers of people dying of COVID-19 with these comorbidities might have been expected to be much higher, HIV and TB need to be included in the risk groups.

Although the results are adjusted analyses, the numbers of people in each category were not given. Overall, however, 86 HIV positive people have died from COVID-19. Also, importantly, viral load suppression has a very different definition in South Africa compared to the UK. 

The analysis used a very long window for assessing last viral load. People were classified as suppressed if either:

  1. The most recent viral load test in the last 3 years was <1000 and ART was dispensed in the last year, or
  2. The most recent test in the last 18 months was <1000 and ART was not dispensed in the last year.

Viral load measurements are not done frequently, and even when they are done, they do not always get recorded. This means that many of the complications associated with HIV might have been complicated by levels of immune activation. However, it is unlikely that the recent effects of service interruptions would be reflected in the viral load measurements.

Francois Venter from Ezintsha at Wits University’s faculty of health sciences, and discussant on the webinar, stressed  the importance of these data: “First proper African data to compare ourselves to the rest of the world…(with) huge implications for how we manage our health programmes.”

He noted that the allocation of resources to COVID-19 as well as the draconian measures originally taken in lockdown has had an impact on HIV and TB services in South Africa. Health seeking behaviour for HIV has been affected including people being afraid to go to health facilities and pick up their ART. They are scared of catching COVID-19 (including from visiting their clinic), scared of being arrested going the the clinic (for breaking strict lockdown regulations) and scared of being tested and forcibly quarantined. 

This is likely to have led to treatment interruptions of both HIV and TB treatment and delayed diagnosis of new TB infections. Nevertheless, approximately two-thirds of COVID-19 deaths in HIV positive people were in people with undetectable viral load.

Venter expects huge challenges over the six months and suggests that policy decisions should have instead learned lessons from HIV experiences with community engagement for successful treatment and prevention programmes rather than using police and military.

But he added that a few things have been fast-tracked, since COVID-19 that were already being discussed. This includes multi-month dispensing and triaging people with respiratory illnesses.

This article was first published online on 17 June 2020.


Davies MA et al. Western Cape: COVID-19 and HIV/Tuberculosis. Webinar: when epidemics collide. Bhekisisa Centre for Health Journalism. 9 June 2020. (webinar) (Mary-Ann Davies’ presentation pdf)

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