How to interpret the modelling on Victoria’s projected coronavirus case numbers

if Victoria eased coronavirus restrictions too early, there would be a more-than 60 per cent chance of a lockdown having to be reimposed before Christmas, modelling carried out by the University of Melbourne for the state's Health Department has found.

It's a startling statistic, and the supercomputer-generated analysis is being used to pave the way on the state’s road from lockdown.

On Sunday, Premier Daniel Andrews called the work one of the most "comprehensive modelling exercises that we have even seen in the state" and the sort of analysis "you can't argue with".

The graph below helps to illustrate the analysis. If it looks a bit hard to read at first glance, don't worry, the rest of this article gives an overview of how to interpret it and goes over the information the projections took into account.

The first thing to note on the graph above is the 14-day average, which shows up as a blue line. It peaked at 505 on August 8 (that is to say, there were 505 cases a day on average in the two weeks to August 8) and has been steadily declining ever since as the lockdown measures have driven down the spread of the virus.

At the moment, the 14-day average is 101, and will probably drop into double-digit figures on Monday.

That red dot is the modelling's estimate of where case numbers will likely be on the Friday after next. Their analysis suggests that the state should hit a 14-day average of around 63 cases on September 18.

Then there is the dotted line on the chart, which indicates a 14-day overage of 25 cases per day. As you can see from the shape of the data, it appears that it will take some time before the blue line reaches this threshold.

(Remember as well, that we talking about the 14-day average, not the first day there are 25 new cases. For the blue line on the graph to dip under 25 there will need to have been about 25 cases per day recorded over two whole weeks.)

The University of Melbourne’s modelling suggests that if the state government was to ease restrictions once the 14-day average reached 25, there would be a 64 per cent chance of the infection rate increasing to the extent that restrictions would have to be re-imposed by Christmas.

But at a press conference on Sunday, Premier Daniel Andrews said if restrictions were eased when the 14-day average dropped to five, there was a 3 per cent chance of case numbers increasing to the point that the state would have to go back into lockdown.

He called the analysis “one of the most comprehensive modelling exercises that we have ever seen in the state, arguably nationally, and a piece of work that has international significance as well”.

"You can't argue with this sort of data, you can't argue with science, you can't do anything but follow the best health advice, otherwise… we will just be beginning to lose control again of this virus and the consequences will be even greater than the difficulties -that I know and appreciate – many Victorians are experiencing today," he said.

On Sunday, the Premier announced stage four restrictions would be extended for two weeks beyond the original end date of September 13. If case numbers remain around 30-50 per day on average by September 28, gatherings of up to five people from two households will be allowed, some workplaces will return to work and some children will start to return to school.

From October 26, if the 14-day average has dropped below five, the curfew will end, there will be no restrictions on leaving home, public gatherings outside will increase to 10, outdoor dining will reopen at cafes and restaurants.

Victoria's Chief Health Officer Brett Sutton said the target of a 14-day average of five cases per day by that date, and none of those cases being from unknown transmission sources, was achievable.

The model weighs up a lot of different factors such as the infectiousness of the virus, the percentage of cases that are asymptomatic, the number of people each new case infects on average, whether compliance to social distancing and mask-wearing directives drops over time and whether people with the virus fail to self-isolate.

But as its authors point out, there are some variables that the model can’t account for, such as the level of risk in particular areas, in particular industries and even the impact of weather conditions.

All these variables were fed into a supercomputer simulation, which played out the infection curve 1000 times. In 640 of those scenarios, it found cases rose to the point that restrictions had to be re-imposed before Christmas.

Deputy Chief Health Officer Allen Cheng said the modelling was not about predicting the future, but provided the most likely outcome based on current information and the impact of lockdown measures.

University of Melbourne professor of epidemiology Tony Blakely, one of the authors of the modelling, said there were two ways the model could be "beaten" – by improving contact tracing, and infectious diseases control protocols in hospitals and aged care settings.

"If we do our contact tracing better than we did three months ago, the contact tracers may be able to hold the case count without it going up again as badly as our model suggests," he said.

Professor Blakely said the use of 14-day average as a metric of case numbers was a choice of the Health Department.

But Deakin University epidemiology chair Catherine Bennett said the modelling was of limited use because the research was based on international parameters and general principles and was not built around Victoria's infection data.

"We saw absolutely no evidence today there has been any analysis of Victoria's data – the modellers don't have access to it," she said.

She said the modelling was most likely carried out scrupulously but "it is not the analysis we need".

“We’ve had 17,000 cases, that should have allowed the state government to make more evidence-based decisions about what restrictions should look like,” she said.

She said the tail of the epidemic – when numbers have started to drop – was uncertain and unpredictable, which meant it was difficult to predict whether the 14-day average would reach the targets set for the state to reopen at each step of the way.

You can read the full summary of the University of Melbourne analysis here:

With Paul Sakkal

Most Viewed in National

Source: Read Full Article