The publication of the national epidemiological model recently spread fear and dismay as it forecasts 40,000 deaths by November from Covid-19, with 1-million people falling ill. This sort of projection risks creating the perception that we cannot limit the contagion — that flattening the curve just means spreading out the timeline of people contracting Covid-19. It has become a reference point for the argument that we should just get on with our lives and drop the heavy burden of seeking to prevent infections.
In the real world, countries as diverse as Vietnam, New Zealand and Germany have demonstrated that it is possible to reduce the number of cases, not just to delay them. The key is to change social behaviour, reorganise work, public transport and retail, and implement strong public health measures to identify and isolate cases. The costs of these measures are high, but the savings in terms of lives, health-care spending and ultimately sustainable social and economic development, are immeasurable.
Epidemiological models bear an uncanny resemblance to economic models. Both aim to simulate critical relationships in the real world to explore the effects of changes in parameters and policies. In both disciplines the findings depend on what questions the models are designed to answer, and on how well their simulations reflect real-world relationships.
The national epidemiological model originated not to analyse how best to prevent infections but to anticipate the need for health-care resources. Its main output is the number of ICU beds required for Covid-19 patients. Changes in government policies or citizens’ behaviour are inputs to the model, not outcomes.
Overestimating the number of infections isn’t a problem for the health-care response. Most of us would prefer an oversupply of health inputs to a shortfall, even if the resulting excess might pain the National Treasury. Based on similar models, New York acquired thousands more ICU beds than it ultimately used — and most residents were grateful.
But this sort of model is less useful in identifying policies or behaviours to prevent infections. Indeed, the apparent inevitability of the death spike leads some politicians and lobbyists to argue that trying to change behaviour and implementing public health responses are pointless as well as costly.
A second challenge is that any model is only as good as the underlying evidence. With Covid-19, we still don’t know, among others, how many cases are asymptomatic, how many need intensive care, whether to use ventilators, the riskiness of specific behaviours ranging from singing to shopping to eating out, or when more effective treatments will emerge. Yet these relationships are crucial for simulating contagion.
The national epidemiological model, for instance, assumes about 75% of cases are asymptomatic, according to its May 19 update. In contrast, most recent research puts the figure at anywhere from 10% to 45%. Overestimating asymptomatic cases can support the belief that a population can reach herd immunity without excessive deaths or a vaccine. In practice, most authorities no longer believe that is possible.
Finally, the inherent rhetoric of models often makes their outputs seem more reliable than they are. Almost by definition, modelled projections are quantified, even if the numbers change regularly as research into the underlying variables evolves. The modellers usually know the models really just generate an order of magnitude, not detailed forecasts. But the media and the public often assume that the specific numbers must mean they are accurate predictions.
SA cannot afford to let modelling exercises limit its ambitions. We need to try to copy the successful countries, which have contained Covid-19 by focusing on rigorous public health interventions combined with physical distancing in social interactions, at work and in public transport and shopping. If we get it right, we can collectively change the modelled projections, rather than being constrained by them.
• Makgetla is a senior researcher with Trade & Industrial Policy Strategies.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.