Giving officials something practical to work with, was one of the hardest challenges facing modelers in the early stages of the COVID-19 pandemic in South Africa, says Prof Juliet Pulliam, director of the South African Centre for Epidemiological Modelling and Analysis (SACEMA), hosted at Stellenbosch University.

During an online Science Café talk at Stellenbosch University last week, she gave an overview of the thinking behind the development of the early models, starting on 28 February when the National Institute for Communicable Diseases (NCID) requested her to provide “a rough estimate" of the potential health impact of the disease.

Prof Pulliam has more than 15 years of experience in the modelling of infectious diseases, often in resource-poor countries. Currently she is also a member of a consortium of modelers advising the NICD and the government on the handling of the pandemic.

“Working with decision makers, my biggest challenge was the need to be practical. You have to find ways of capturing the uncertainty inherent in the modelling process, particularly in a situation like this where so little is known about the disease, while also giving them something useful they can work with."

“With very limited data, the early models had to be simple and flexible, and able to explore the many uncertainties we were faced with. The first model was therefore aimed at providing a rough understanding of what the situation could look like if nothing was done," she explains.

So instead of asking “how many people will be infected", they reframed the question to “if a specified proportion of the population became infected, then how many would seek care, and how many would die?"

Following this approach, in a scenario with an infection rate of 10%, the fatality estimate was between 20 300 to 87 900 people. In the scenario with an infection rate of 40%, the fatality rate could be between 81 300 and 351 000 people.

But this first model did not make provision for a time scale, and by 13 March they were confronted with the next question: How much time do we have? How fast is the epidemic going to take-off in South Africa? When can we expect the first 1 000 or 10 000 cases?

“For these short-term projections we developed a stochastic model, based on no interventions and the assumption that half of the existing cases had been missed. According to this model, the number of cases would reach about 1 000 between 28 March and 2 April.

“On 28 March we stood at 1 070 cases detected. According to our model, without the lockdown, we were likely to reach 10 000 cases by 11 April. Instead, we reached a figure of 10 015 cases a month later, on 11 May."

By 23 March she became part of the South Africa COVID Modelling Consortium, brought together by the NCID to formally assist the Department of Health in modelling the pandemic.

She says as the pandemic progresses, different models are needed to answer new questions. Currently she is working on the National COVID-19 Epi Model, one of two models developed by the Consortium that is used by the Department of Health to assist with the planning of resources.

“The model is being used to address the question of how long we have before current ICU capacity is exceeded. The approach was to compare detected cases on June 1 with predictions under an optimistic and pessimistic scenario to assess the current trajectory. Based on the current trajectory and long-term projections, the model can be used to evaluate the likely time when capacity will be breached by province."

Prof Pulliam says their predictions thus far are consistent with what is happening on the ground, especially in the public sector: “The ICU capacity in the public sector has already been exceeded in the Western Cape, and they've been working hard to expand that capacity and work out ways to transfer patients to the private sector."

The Eastern Cape will breach its ICU capacity in mid-June, while the rest of the country's provinces will reach that point in mid-July or later. The latest projections, taking into account the impact of the lockdown since 27 March 2020, were released this week. They are available online at https://www.nicd.ac.za/diseases-a-z-index/covid-19/surveillance-reports/

Science Café Stellenbosch is an initiative of SU's Faculty of Science to promote the discussion of scientific issues in a language everyone can understand. Like us on Facebook or join our mailing list - sciencecafe@sun.ac.za