Leaked Doherty Covid modelling shows just the most worrying scenario ? not the most likely | Melissa Davey

Much of the pandemic modelling to make news headlines since Covid-19 hit has depicted concerning scenarios involving high case numbers and hospitalisations.

On Thursday, modelling from the University of NSW and cited by NSW health minister Brad Hazzard found Covid-19 cases in the state could reach 25,000 a day by the end of January. The latest modelling to raise alarm bells came from the Doherty Institute and was leaked to the media on Tuesday after being sent to politicians ahead of a meeting of national cabinet on Wednesday. It said Australia could see 200,000 new Covid cases a day by late January or early February.

Early into the pandemic in 2020, other models predicted "hundreds of thousands" of Australian deaths, while health workers in NSW were told to prepare for 8,000 deaths in the first-wave.

But these are worst-case scenario models and are just one scenario out of dozens of different scenarios that are calculated by epidemiologists and biostatisticians.

What is often lost in the reporting of these worst-case models is that they rarely ever eventuate, but are calculated to allow governments to see what might happen if they did nothing at all to control an outbreak, and to plan accordingly.

The leaked Doherty Institute modelling projecting 200,000 cases came with an important caveat; this would only eventuate if nothing was done, including if people did not change their behaviour at all of their own accords. It also assumed no change to the pace of booster roll-outs, and that only very basic restrictions such as requiring masks in hospitals are maintained.

But already, people are changing their behaviour in states like NSW where mask mandates for retail don't exist, wearing masks anyway to shopping centres, or cancelling bookings for crowded venues. The Australian Technical Advisory Group on Immunisation [Atagi] is constantly reviewing data on boosters, and adjusting its recommendations accordingly.