The Australian #COVID19 modelling was published today. My thanks to James McCaw ( @j_mccaw) for checking this thread. I’ll do two threads - one explaining the results and how we might interpret them; and another to try to explain how these models work. https://twitter.com/peripatetical/status/1246330200054349824?s=20

Disclaimer: my contribution to the paper was to fact check some of the parameters and to help with the interpretation. This is the result of hard work by Jodie McVernon, @j_mccaw, @rob_mathbio their team with extensive experience with these complex methods.

What are the take away messages? Four scenarios were modelled: "unmitigated" (no interventions), "quarantine/isolation" only, quarantine/isolation + some social distancing, and quarantine/isolation + more social distancing.

First, the unmitigated scenario is apocalyptic. ICU is the pressure point, but all parts of the health system would be stretched. Up to 35,000 ICU admissions *per day* at the peak.

Even if we could expand ICU capacity by 5 times its usual size, there'd be months where this capacity would be exceeded many times over

Thankfully, this won't happen. All countries are taking strong action to flatten the curve. But it does justify the actions taken and pain endured to get to this point.

Second, isolation and quarantine by itself isn't enough. This model assumes that isolation reduces transmission by 80%, and quarantine reduces transmission by 50%, with 80% of people complying with quarantine.

If ICUs were 5 times bigger than they currently are, then isolation and quarantine in combination with some physical distancing (reducing transmissions by 25%) might just be enough to keep cases under the ICU capacity.

Isolation and quarantine in combination with more physical distancing (reducing transmissions by 33%) looks like it would probably keep things manageable. But even in this scenario where we don't run out of ICU beds and ventilators, many people would still get sick and die.

A rare bit of good news is that what the community has done in reducing mixing so far may have “crushed” rather than “flattened” the curve. It’s early days, but the fall in locally acquired case numbers suggests that R0 might be close to 1 at the moment
https://cmmid.github.io/topics/covid19/current-patterns-transmission/global-time-varying-transmission.html

Third, the time scale under all scenarios is many months. This isn’t a short term shutdown. Whatever we do we’ll need to do for a while. We need a “sustainable suppression” strategy.

These are possibilities, not predictions. No country will see a smooth epidemic curve as depicted in these models. If we started seeing a surge, we’d be locking down even more strongly, as the US and European countries have been forced to do.

No model will tell you exactly how effective the physical distancing has been, or will be, whether 25%, 33% or 80%. Why not? We can model airplanes flying, because the laws of physics are predictable.

Human behaviour is not predictable. There is little data on how people are behaving at this unusual time and how that impacts on viral transmission.

So how do we put this together with where we are now?

The decisions made by government to date suggest we’ve avoided being the next Italy or New York. We have bought ourselves time. As @mlipsitch and @yhgrad put it, we’re almost on the liferaft, but still need to plot out a course for dry land. https://www.statnews.com/2020/04/01/navigating-covid-19-pandemic/

In the short term, we should use this time to build up equipment and PPE to better cope with what may come if control measures slip.

We should develop better tools for finding cases and trace contacts so that we might be able to relax some of the social distancing measures. Find better treatments for severe COVID.

In the longer term, we need a plan, but this will take time to develop. A vaccine is our current best bet. But if what isn't possible? Intermittent lockdowns? Carefully calibrated social distancing? Could there be a safe level of transmission? Antivirals to stop infection?

None of these options are good. All may be long shots or dead ends. We need time to better understand this disease. The endgame strategies might be informed by modelling, but there will need to be much broader considerations about our course to dry land.

Australia is in a privileged position, even this has come at a terrible price. Rather than having to make these decisions from a state of crisis, we now have an opportunity to consider the options with COVID-19 relatively well contained.

Want to know how the models work? A second thread follows: https://twitter.com/peripatetical/status/1247403904666439682?s=20

And of course I forgot to link to the actual paper! (note that it's a pre-print) https://www.doherty.edu.au/news-events/news/covid-19-modelling-papers