Study of SARS-CoV-2 antibodies among blood donors in four provinces suggest actual infections may exceed officially diagnosed infections by a factor of ten

The South African National Blood Service (SANBS), in partnership with the Western Cape Blood Service (WCBS), conducted a survey (also known as a seroprevalence study) of healthy consenting blood donors in four provinces, to determine how many donors have SARS-CoV-2 antibodies, which would be an indication of previous infection with the virus that causes COVID-19 disease. This kind of study is one of the few relatively cost-effective ways of investigating the extent of SARS-CoV-2 infection at the population level. Donors can donate blood after they have recovered from the illness and have been free of symptoms for at least 14 days.


The testing platform used is more than 99.5% specific for antibodies against SARS-CoV2, identifying people who previously have been infected with the virus and who have since developed antibodies against it. Preliminary results of this study, reporting on seroprevalence in these four provinces (Eastern Cape, Northern Cape, Free State, and KwaZulu Natal) have just been released by deposit of a manuscript to the preprint archive (DOI: 10.21203/ Results are summarised in figure 2 and table 2. When weighted back to the general population an estimated seroprevalence of 63% in Eastern Cape, 32% in Northern Cape, 46% in Free State and 52% in KwaZulu Natal was found.

There was no significant difference in seroprevalence by sex or age groups (See figure 1). However, race was a highly statistically significant and epidemiologically important, factor – and this applied to each of the four provinces for which data are available. There is no biological reason for race itself being a factor in contracting SARS-CoV-2.  The reasons why we are seeing these differences among our donors is not exactly clear and needs further investigation. As these kinds of investigations are not core to the SANBS expertise, we certainly look forward to collaborating with others to unpack these findings further.


These results, and indeed results from other ongoing studies, are expected to have substantial impact on our understanding of important issues such as:

  • the ‘spectrum of disease’ (broken down into infections as asymptomatic, mild, and serious)
  • the age-dependence of infection rates and complication rates
  • key social factors impacting the spread
  • challenges faced by test and trace programmes


“Usually, in a nationally representative study, we would want to publish a complete analysis of all provinces in one step, but this is a unique situation, with such a high level of legitimate public interest in these results. As a result, we agreed to publish these numbers so that the discussions about lockdown levels, vaccine distribution, the circulation of novel variants, and so on, can take place with all available information in play,” said Principal Investigator Marion Vermeulen, acting Chief Operations Officer of SANBS. “Expansion of the testing to other provinces is being planned and will be concluded upon regulatory approval of the testing reagents available for the high throughput platform. The current analysis we have, paints a picture with some consistent features across four provinces, so we believe these results are suggestive of what we would find in the other five provinces,” Vermeulen added.

Wendy Sykes, the lead author on the scientific paper, who coordinated the study logistics and testing of specimens in her Durban laboratory, says: “We expected that the seroprevalence would be significantly higher than what is implied by the official laboratory diagnosed clinical case counts tracked by the NICD, but we were frankly surprised at how high it turned out to be.”

Prof. Alex Welte of the South African Centre for Epidemiological Modelling and Analysis (SACEMA), an analytical consultant to the project comments: “One can certainly debate how well this sample of blood donors represents the overall community of each province. For one thing, blood donors are, by design, HIV negative. Also, donations should be avoided for at least 14 days after COVID-like symptoms, and these samples were obtained when we were barely over the crest of the second wave. Nevertheless, from previous work, we know that blood donors are pretty representative of the general public, and there is no magic trick to produce a perfectly balanced sample of a whole population. We expect these results to withstand significant scrutiny, and I really hope resources are available for the blood services to continue this work, which is probably the most cost-effective way to monitor the overall scale of COVID infection in the country.”


Dr Karin van den Berg, lead consultant in Translational Research at SANBS commented: “One year into the pandemic, our understanding is still evolving almost every day. While this level of seroprevalence may mean that a significant amount of population-level immunity has been achieved, we know that other countries have seen large outbreaks even after such strikingly high seroprevalence was reported. The emergence of variant strains likely contributes to this phenomenon. To understand the situation better, we want to analyse the role of changes in major circulating variants, changes in the spectrum of disease (what proportion of infected people are asymptomatic versus those who become seriously ill) and the natural waning of individual immunity – which is gradual rather than abrupt. So while we can’t draw conclusions about how ambitious a vaccination campaign would now need to be to achieve something resembling elimination of COVID-19, we at least now have a relatively robust and cost-efficient process for tracking how the epidemic unfolds over time.“

Figure 1: Prevalence of SARS-CoV-2 antibodies among South African blood donors, compared by sex, for each combination of i) broad age group, ii) major race group, and ii) province. (EC = Eastern Cape, FS = Free State, NC = Northern Cape, ZN = KwaZulu Natal), expressed as a fraction between 0 (0%) and 1 (100%).

Figure 2:  Estimated prevalence of SARS-CoV-2 antibodies, expressed as a fraction between 0 (0%) and 1 (100%), among blood donors, in January 2021, for four South African provinces (EC = Eastern Cape, NC = Northern Cape, FS = Free State, ZN = KwaZulu Natal) by race (White, Asian, Coloured and Black), as well as a race-weighted (Tot)al provincial estimate and the official prevalence of having had clinical COVID-19 diagnosis (Dx) by PCR of nasal or oropharyngeal swab specimens.


Table 1: The headline results of figure 4, in numbers: Provincial seroprevalence estimates, interpreted as attack rates, the total number of infections this implies, as well as the official number of persons diagnosed in each province, and the multiplicative factor (rounded to the nearest whole number) relating our estimate to the official case count. Note that the estimated number of infections is based on our estimated prevalence being applied only to the age group 15-69.


Province Estimated seroprevalence/ attack rate (%) Estimated total infections (Count, rounded) Officially diagnosed cases (Count) Diagnostic underestimate

(Fold, rounded)

Eastern Cape 62.5 (58.9–66.1) 2 729 448 186 771 15
Northern Cape 31.8 (25.3–38.3) 235 053 29 558 8
Free State 45.5 (39.9–51.1) 878 206 70 511 12
KwaZulu Natal 52.1 (49.1–55.2) 3 955 859 279 974 14



How many participants took part in the study? Sample size?

A total of 4 858 blood donors from four provinces (EC, NC, FS and KZN) of South Africa were tested for anti-SARS-CoV-2 antibodies. This included 2 107 (43%) from KZN, 1 457 (30%) from EC, 831 (17%) from FS and 463 (10%) from NC.


Was the sample size representative of the population size per province?

No, there is still a disproportionately higher number of White donors compared to the general population with Black and White donors making up approximately equal proportions of the donor pool.


Are specific race groups more susceptible to COVID-19? If not please unpack your findings and what your theory related to your findings on race and COVID-19.

There is no evidence, internationally or locally, to suggest that specific race groups are more susceptible to COVID-19. There is however strong evidence that the inability to socially distance and adhere to COVID preventative measures is associated with higher rates of COVID transmission. Exactly why we are seeing these differences among our donors is not exactly clear and needs further investigation. These kinds of investigations are not core to the SANBS expertise. We certainly look forward to collaborating with others to unpack these findings further.


When did you start this study and over what time period did you analyse the results?

This study was performed in January 2021.


Was this study rushed?

No, the study was not rushed; it has been in planning since May 2020. We proceeded once we received ethics clearance and regulatory approval for the reagents we used for the testing. In fact, we delayed sample collection to January as opposed to collecting samples during the December period.

The results of the study are extremely important to South Africa as this is one of the largest and most comprehensive seroprevalence studies performed to date. It is also of particular interest as it was performed towards the end of the second wave, which provides a very “current” snapshot of the status in these provinces. These results show that many more people were infected than what is estimated based on national testing statistics. This information is valuable to our government when they make COVID-19 related policy decisions. It is for this reason that we wanted to get the results out as quickly as possible so that decisions could be made in real-time.


There has been mention that blood donors are generally HIV negative, do you think there would be anything statistically significantly different if HIV positive participants were included in the study?

People living with HIV who are on effective antiretroviral treatment do not appear to be at greater risk of getting COVID-19. Those not on antiretroviral therapy may be at greater risk.

Regarding the risk of developing severe COVID-19, current evidence suggests that HIV is less of a risk factor for severe COVID-19 than other chronic health conditions such as diabetes and obesity.

We therefore do not believe that there would be anything significantly different if HIV positive participants were included.


What impact, if any, do these findings have in terms of the rollout of the vaccine?

This is a difficult question to answer, as there are many important factors to consider regarding vaccine rollout. There are already studies showing that infection with one variant may not necessarily protect you from a new variant. As a result, we need to be very cautious in over-interpreting the results of this study.


If a significant portion of the population already have antibodies, is there still a need for a massive vaccination programme, please explain your case for your answer?

Our study does not address this question specifically. We cannot, from our study, know exactly when our donors were exposed to the virus or to which virus they were exposed. As stated earlier, it is still unclear to what degree antibodies developed in response to one variant will protect one from infection with a second infection.


Are you able to identify which strain of the virus the antibodies are from?

No, antibody tests that detect binding antibodies (the type of test we used) are not able to identify which strain of virus the person was infected with.


What impact, if any, should these findings have on the lockdown levels?

As indicated above, the degree to which these donors have antibodies against the old versus the new variant is unclear. In addition, we also do not know how well antibodies from the old variant will protect you from infection with the new variant. Further work certainly needs to be done before one could try to estimate how these findings may inform national lockdown policies.


If I have donated blood in 2021, does this mean I have been part of this study, and can I access my antibody results?

We only tested you for antibodies if you agreed to participate in this study. On the day of your blood donation, we would have explained the study to you and you would have signed an additional consent form agreeing to participate. All blood donors who were tested will receive their results via SMS.

We did not advertise beforehand on which days the study would be performed as it was important that we did not influence who came to donate so that we did not have a biased sample.


What is being done to ensure that you comply with POPIA to ensure people’s health data is protected?

SANBS submitted a research protocol to our independent SANBS Human Research Ethics Committee to ensure that the study participants’ (donors) rights are protected. All participants signed an informed consent document in which the study was explained and assurance provided that their information will be kept confidential. In addition, when analysing the data, the donor information was anonymised with then only aggregated information being reported.


Will you communicate the findings of individuals who participated in the study if they request the information? If so how should they go about this?

Individuals who consented to participate will receive their own results by SMS.


Of the donor population what % took part in the study?

The 4 858 donors who participated in the study constitutes approximately 1% of the total SANBS donor population of about 450 000 donors who donated in the last year and 6% of the donor population (~150 000 donors) in the four provinces we included in the study.


Article available online: