Covid modelling revamp needed after ‘wrong’ NSW predictions

A leading Covid expert is urging governments to look overseas for virus data after modelling predicted a very different October for NSW.

A mounting death toll, escalating cases and inundated ICUs – October was foreshadowed as NSW’s deadliest month of the pandemic.

But doomsayers have been proved wrong as case numbers consistently lingered in the 200s in the past month while restrictions were eased.

Premier Dominic Perrottet on Monday hinted at changes to the NSW road map to freedom and a potential delay to the December 1 reopening date for unvaccinated residents with the aim of pushing the vaccination rate to 95 per cent.

The soaring number of immunisations helped keep new Covid cases to 177 on Sunday, the lowest since July 30, and dropping even further on Monday to 135.

The promising downward trajectory came despite then-Premier Gladys Berejiklian warning constituents in September to expect dramatic scenes that had never been seen before.

“October is likely to be the worst month,” she told reporters on September 3.

“When I say images of things we’ve not seen before, it means many additional spaces within hospitals dedicated to Covid patients in intensive care (and) patients perhaps having to be taken to a hospital that might be quite a distance away from where they live,” she said on September 20.

Her warnings were backed by modelling from the Burnett Institute which predicted that cases would reach a peak in mid-September, with a seven-day average between 1219 and 2046 for Greater Sydney.

Those peaks were never reached, and Covid-19 expert Peter Collignon said authorities must take a cautious approach to modelling because it “plans for the worst”. Prof Collignon said the best approach was to look at real time data from international jurisdictions.

Infectious Diseases physician and microbiologist Dr Peter Collignon. Picture: AAP Image

“You ought to put real world data from countries that have gone through this. I would base what happens in our winter on Canada, US, UK and their winter,” he said.

“This isn’t going away, and we need to get better with our predictions because they have been so wrong.”

Health Minister Brad Hazzard said he was told by modellers in March and April 2020 that 25,000 people would be dead by that Christmas.

“Sadly, some people did die but nowhere near that number. This year’s modelling for the Delta variant certainly had me and Dr Chant on high alert but again, the weapons were always in the hands of the community,” he told The Daily Telegraph.

“Neither Dr Chant nor I dared to dream that the community would get on the journey to such a great degree.”

The minister shot down Prof Collignon’s advice that modelling should be swapped for real-time international data. “Very often modellers need to use data from so-called similar overseas communities but in reality, they are not the same. They are very different,” he said.