South Africa Expects 3.4 to 3.7 Million COVID-19 Cases by November.
South Africa will see between 3.4 and 3.7 million lab-confirmed cases of COVID-19 before November, with between 34,015 and 49,774 deaths. National and provincial departments of health will need to spend between 26 and 32 billion rand ($1.5 to $1.8 billion) to take on the coronavirus. And the number of hospital and ICU beds in the country could be exhausted as early as July.
That is according to a new report from the South African COVID Modelling Consortium, which is made up of researchers from South Africa and from the School of Public Health working to help the South African government make informed decisions about the pandemic.
“It’s important to note that these predictions are not set in stone—mitigation strategies may prove to be more effective than anticipated, and, by pushing out the peak of the curve, hospitals have time to prepare and increase capacity,” says Brooke Nichols, assistant professor of global health and a member of the South African COVID Modelling Consortium. She is also co-creating a similar consortium for Zambia, and advising on a World Health Organization (WHO) consortium that is developing COVID models for low- and middle-income countries around the globe.
South Africa completely shut down in mid-March, when the country had only 116 confirmed cases, “because everyone, including modeling groups, told the government, ‘Shut this down now if you want to have any impact, because if you wait until it looks bad, it’ll be too late,’” says Nichols, a health economist and infectious disease mathematical modeler who has worked in South Africa and Zambia for most of her career.
According to the consortium’s models, that “hard lockdown” reduced the transmissibility of the coronavirus in South Africa by 40 to 60 percent. The subsequent, slightly looser “level 4” lockdown reduced transmissibility by 25 to 35 percent, and later physical distancing measures reduced it by 10 to 20 percent.
“Our work doesn’t stop here,” Nichols says. “The models will continue to be further refined and updated in light of any new developments in the field.”
The South African COVID Modelling Consortium includes researchers from SPH who work with the Health Economics and Epidemiology Research Office (HE2RO) at the University of the Witwatersrand in Johannesburg, which is co-directed by Sydney Rosen, research professor of global health at SPH. Gesine Meyer-Rath, a research associate professor of global health at SPH based at HE2RO, is leading the budget modeling of the COVID response for South Africa with the consortium. Nichols, who was an SPH research scientist at HE2RO before transplanting to Boston to join SPH’s faculty last year, co-developed one of the consortium’s first models of COVID in South Africa in early March alongside HE2RO’s Lise Jamieson, and has been working on subsequent models since.
The other two modeling groups in the consortium are from the University of Cape Town and the South African Centre for Epidemiological Modelling and Analysis, and the consortium is led by South Africa’s National Institute for Communicable Diseases, which is coordinating the country’s COVID response. (Nichols notes that the consortium happens to be all women.)
The basic idea behind a modeling consortium, Nichols says, “is that one model is always wrong. If we have three different models—with some agreement on the range of assumptions, given the uncertainty and the underlying structure—we can say, ‘How are we in alignment? How do we diverge, and why?’”
More models also help to handle the gaps in South Africa’s COVID data. “You have to triangulate,” Nichols says. “We don’t actually know the number of people hospitalized with COVID in the public sector, but we do for the private sector, so we’re looking at some of the hospitalization data, we’re looking at the case data, we’re looking at the death data, and we’re trying to figure out which piece of the pie we’re missing for each.”
Using multiple data sources and multiple models, Nichols and her colleagues in the consortium have been able to figure out how to best model different elements of COVID in South Africa, ultimately generating data to help support the government’s decision-making.
The first round of the consortium’s work focused on more broad-stroke predictions about the scale of COVID in South Africa and the costs of working to contain it, which led to a national budget for COVID response. The government and HE2RO released these predictions on May 19.
Now, the three modeling groups are working together on one model—informed by their earlier work—to try to answer more nuanced questions about different mitigation strategies as different parts of South Africa move to more relaxed levels of lockdown.
But especially in a country still deeply divided decades after Apartheid, different strategies to slow the spread of COVID may be less effective—or even impossible—for different socio-economic groups, Nichols says. While many upper- and middle-class South Africans are able to isolate much like their counterparts in higher-income countries, she says, “you have very low-income areas where people live very close together and share communal resources like communal water taps and communal sanitation, so, even in a perfect lockdown scenario, people still have to interact,” she says.
Nichols notes that complete lockdown is also financially devastating for lower-income South Africans: “If you can’t work, you can’t eat.” So, the government has to weigh the risks and benefits as it maintains or eases lockdown in different areas.
These considerations are very different in neighboring Zambia, says Nichols, where she has worked for years with that country’s government and the United States Agency for International Development (USAID). Nichols is now helping create a similar COVID modeling consortium in Zambia, with modeling groups from Zambia’s Eden University and from Johns Hopkins University.
Zambia is less stratified and more communal overall, Nichols says, for better and for worse. “For example, individual isolation doesn’t make sense in communities that are very strong, and so an option that may be considered could be community isolation,” Nichols says. In other words, Zambia’s many—often geographically isolated—rural communities could continue sharing water and sanitation, and interacting normally, but barring visitors from other communities.
While working on understanding COVID in South Africa and Zambia, Nichols has also recently been appointed as an advisor to the COVID-19 Multi-Model Comparison Consortium (CCMC) technical group convened by the WHO. That consortium includes major modeling groups around the world, working to compare COVID models and model outcomes across low- and middle-income countries.
It is, in many ways, what Nichols and her colleagues in the South African and Zambian consortia are doing but on a much larger, global scale, she says.
Nichols is excited for this greater level of collaboration in the world of global health disease modeling, and says she hopes it can continue after the pandemic. “The whole idea behind this model comparison was to see which models are best suited to answer some of the questions that we’re facing in low- and middle-income countries,” she says. “It would be great to have more brains on this, because the questions are complicated, and the answers are unclear.”
Very interesting article. I didn’t really understand modeling and at least now I have a basic knowledge of it. Brooke Nichols PhD gives some great insight about the process that goes into using modeling as a tool to help poor & middle class people get better health care especially during the Covid-19 pandemic.
The concept of “community isolation” is very interesting. It makes great sense for communities that are strong and cohesive. I’m curious to hear more about this.
We need the best and the brightest working on the modeling, planning and distribution of resources, informing public of best practices to mitigate the impact and the treating of covid19 infected. Thank you Brooke for your work which will help to save lives.
And yet the SA Covid 19 Modelling Consortium has horrendously over-estimated the numbers of deaths and hospital and ICU beds needed. The PANDA model has been much more accurate, and yet is continually ignored by government. Why is this? Why is our government following models that are basically discredited?