'Flatten the curve': why predicting coronavirus infections and deaths is so tricky


If there is a message that has united Australians since the Covid-19 epidemic began, it is that we must “flatten the curve”. It’s a mantra being shared by politicians, doctors, nurses and members of the public.

What is more contentious, though, is how to flatten the curve. There has been a proliferation of graphs on social media and in news reports containing all kinds of data, predicting deaths, projecting infection rates and intensive care cases, and estimating the impact of school closures and other containment measures on slowing spread. Some of these models come from peer-reviewed papers, others from individuals – including respected doctors, journalists and health workers – doing back-of-the envelope calculations.

Dr Kathryn Snow is an infectious disease epidemiologist at the University of Melbourne, working in health services research for vulnerable populations, and says it’s important to remember a lot of models, for example those that predict deaths, make assumptions about the health system or the behaviour of a population. But even small changes to these elements could drastically change the projections. So governments are more likely to have dozens of different graphs and models to predict infection, bed capacity and deaths, rather than relying on one.

For example, what if at the start of an epidemic you have a model that predicts deaths over the next six months, but then one month in, the government closes its borders? What if a graph suggests that closing schools slows infection spread, but it turns out that is in cases of diseases children are especially vulnerable to, such as measles? What if a model on the impact of closing schools assumes children will stay at home for six months, but people get complacent after four weeks and children begin to mingle in shopping centres and parks, or with vulnerable people such as grandparents? Uncertainties, errors and presumptions in models are rarely communicated when images of graphs are shared.

Snow says: “On social media, a lot of anxiety is being driven by homemade models and graphs, which have been produced by well-meaning people who aren’t actually epidemic modellers.

READ  People with weaker muscles are 50 percent more likely to die early, study suggests

“Epidemic modelling is extremely complex, and small changes in the way it’s done can have huge impacts on the results,” she says. “We shouldn’t trust homemade epidemic models any more than we would trust homemade climate models or economic models.”

Prof Hassan Vally, an infectious disease epidemiologist, says even looking at good quality models from researchers overseas such as in China, where experts have had more time and data to examine the virus, can only tell us so much. It is hard to separate out every intervention to see how much curve-flattening impact any one has, he says.

“It is worth noting that there have been countries where schools haven’t been shut down and they have been able to gain control of the outbreak,” he says.

“However, right now what is being balanced by the government is looking at where we are at in the epidemic and judging the most appropriate response that is both proportionate and sustainable. There is no doubt there would have been long discussions about this and this would have included weighing up the other costs of this, which are significant, including possible drains on the health workforce, many of whom may be parents and may need to stay at home to look after children if they are not going to school.”

As the epidemic progresses and all factors are taken into account, things may change. There may be a time where closing all restaurants or schools will have a significant impact.

Prof Jodie McVernon is director of Doherty Epidemiology, with expertise in mathematical modelling of infectious diseases and clinical vaccine trials. She has also been advising the government almost every day on Covid-19. She is well aware of all of the evidence about flattening the curve, including widely shared modelling from Imperial College London on interventions to reduce Covid-19 deaths. The modelling assumes closing schools in the case of influenza reduces further infections. It also shows that for Covid-19, closing schools needs to be combined with other measures to be effective.

READ  Risks of 'domino effect' of tipping points greater than thought, study says

McVernon says in Hong Kong school closures are commonly employed for influenza control and can reduce the spread of infection by as much as 15%. “Children are known to be highly susceptible to and infectious with influenza, so this is the best school closures might do,” she says. “Children appear to play a much lesser role in transmission of Covid-19, emerging evidence indicates that they are less susceptible and infectious than adults and very rarely have severe infections.

But surely even a moderate effect on infections means closing schools and businesses down is better than not doing so at all? The answer to this can also depend on your value systems. For some, economic considerations are important because of the devastation that occurs when people are in long-term unemployment and lose their homes and livelihood. Others are willing to brace for widespread economic harm and close every business and school right away for months on end, if it means even a small amount of infection and death is prevented. Others will say economic and health impacts are inextricably linked, and it’s too hard to separate the two out so simplistically.

There are unique considerations in overseas examples of school and business closures too that may not apply to Australia, such as population density and unique population demographics.

McVernon says consistent evidence shows ongoing efforts to find and isolate cases, and trace and quarantine close contacts of those cases, is critical in flattening the curve, having significantly greater impact than school closures. And Australia is undertaking both.

“We know that we won’t be able to find everyone as numbers increase, and we know that some people may spread infection before or without symptoms,” she says.

“To deal with that, whole of society behaviour change is needed to reduce the spread of infection. That’s what the social distancing recommendations are about. We need to be realistic that whatever changes are in place are likely to be needed for at least six months, so they have to be sustainable.” She says hygiene, too, is essential.

READ  In-flight turbulence is getting worse thanks to climate change heating up the jet stream 

The Australian Health Protection Principal Committee, comprised of all state and territory chief health officers and chaired by the commonwealth chief medical officer, has published the rationale behind its latest recommendations, including that schools remain open for now. It makes clear this advice may change – school closures are not permanently off the table.

However, some teachers have said they have not received hand sanitiser or tissues for their classrooms, and parents are already pulling students out of school. Providing support for teachers in high-risk groups to stay home and supporting parents to take immunocompromised children out of school are measures being taken by some public schools, but not all teachers have reported receiving those supports. Even if keeping schools open makes sense in terms of flattening the curve, it may be that it is too late to convince the public when there is so much distrust and when teachers feel undervalued and expendable.

McVernon says it is important not to downplay the reality that Covid-19 can be very severe, and more people will die. But she says she is confident the strategy rolled out by the government so far is about saving lives, even if that message is not getting through to everyone.

“We focus obsessively on every bit of new information we can glean about this virus, and the more we understand about outcomes in health systems like our own the better,” McVernon says. “Using new data we can refine our scenarios to help the planning further. As the numbers of cases in Australia grow, we will continue to review our local evidence to make better predictions about our own epidemic.”



READ SOURCE

LEAVE A REPLY

Please enter your comment!
Please enter your name here