Minister offers sight of Covid-19 models
• While health minister has lifted lid on projections, numbers cannot be properly evaluated without peer reviews or release of codes
In the face of growing pressure from business and citizens, health minister Zweli Mkhize this week lifted the lid on the modelling work guiding the government’s response to the Covid-19 crisis.
In the face of growing pressure from business and citizens, health minister Zweli Mkhize this week lifted the lid on the modelling work guiding the government’s response to the Covid-19 crisis.
For the first time on Tuesday evening the public was given sight of the projections from a consortium of experts convened by the National Institute for Communicable Diseases (NICD), which indicate there could be as many as 3.7-million detected cases and between 40,000 and 48,000 deaths by November.
Millions more cases will go undetected, and 475,000680,000 patients are likely to require hospitalisation, said the consortium, which includes experts from the SA Centre for Epidemiological Modelling and Analysis at Stellenbosch University, the Modelling and Simulation Hub, Africa at the University of Cape Town (UCT), and the Health Economics and Epidemiology Research Office at Wits. The consortium’s models capture the public health impact of Covid-19, but do not consider the social and economic burden of the disease or the lockdown.
The scale of the projected numbers explains why the government has sought to delay the inevitable avalanche of infections with restrictions on trade, travel and social interaction, and buy precious time to ramp up the health system’s capacity.
While the anticipated number of deaths is dwarfed by those associated with HIV, TB and other infectious diseases, the extent of Covid-19 infection and the sheer volume of patients requiring admission to hospital is set to stretch the health system to breaking point.
In parts of the country, that day may arrive soon: the Western Cape, which has more than 60% of SA’s cases, is seeing a rapid uptick in hospital admissions, and already the 23 ICU beds in Tygerberg Hospital are reportedly all occupied. At the last count, there were over 19,137 cases of Covid-19 in SA, more than 12,000 of them in the Western Cape. The province has recorded a similar proportion of the 369 recorded deaths.
To put this in context, in 2017 there were more than 205,000 deaths from infectious diseases in SA, and 115,000 from non-communicable diseases, according to the Actuarial Society
of SA. And in 2018 there were 61,000 deaths from tuberculosis alone, according to the World Health Organisation.
However, most Covid-19 deaths are expected to occur in no more than two to three months, rather than the course of a year, posing an extreme challenge to the health system.
The minister has also sought to show that his officials are paying attention to other groups working on the potential trajectory and effect of the disease, and on Thursday showcased a raft of models produced by Deloitte, the Actuarial Society of SA (Assa), various independent academic groups at Wits and UCT, and the Pandemic Data and Analytics (Panda) consortium.
Deloitte’s models have informed the work of Business for SA (B4SA), which is supporting several government departments and helping with the procurement of essential medical supplies. It estimates there will be about 2.9-million cases and 40,000 deaths, and offered a glimpse of its projections for the demand for personal protective equipment, ventilators and oxygen, which have not previously been published.
Most of the other work highlighted by the minister is already in the public domain. Assa, which has elicited a deluge of comments from the profession in response to its preliminary model, has projected cumulative deaths for various scenarios, which hover around the 48,000 mark.
Wits academic Alex van den Heever has been vocal in his criticism of the economic and social costs of the lockdown, as has the Panda consortium’s Nick Hudson. Van den Heever’s work has focused on the effect of various interventions and concluded it would be far more cost effective to radically scale up testing and contact tracing than continue with the lockdown.
Panda’s analysis concluded that the lockdown would trigger such economic damage that it could cause 29 times more deaths in the long run than Covid-19.
But presenting results is not enough. It is impossible for experts to evaluate the majority of these planning tools: few groups have subjected their work to peer review or released their model code so other scientists can replicate their work.
The stakes are so high, in terms of public health and the economic and humanitarian consequences of the lockdown, that the government, B4SA and independent groups alike need to provide far greater disclosure of their work.