Modelling versus the real world. The case for the latter over the former

After months of waiting, the federal government finally released targets (but not the underlying modelling) prepared by the Doherty Institute to guide Australia’s eventual reopening to the world. The stated target for “Phase C” of 80% of the eligible population is extraordinarily high compared to the rest of the world.  

However, it is far more achievable than the recommendation released a day earlier by the Grattan Institute. 

Headlined “Race to 80”, the Grattan report urged the federal government to maintain a “COVID Zero” policy until 80% of all Australians had been vaccinated, including 95% of those aged over 70.

The media lapped up the 54-page report, almost certainly neither fully reading nor understanding it. There are three key aspects of the report that benefit from further analysis and critique. 

The bar is too high 

While the Phase C target set by Doherty is very challenging (and no one bothered to set a full-reopening “Phase D” target at all), the Grattan target of 80% of the total population is extraordinarily difficult (Grattan acknowledges that it is ambitious) to reach without significant sanctions on unvaccinated people.

Let’s consider the global benchmarks.

Israel, the early vaccine bolter, appears to have hit the wall at 62% vaccination levels. While it has experienced a recent rise in infections courtesy of the Delta variant, deaths remain well below flu levels (around one per day).

Only the UAE has reached 70% vaccination levels. The UK, which leads the pack of Western nations, is around 56% fully vaccinated (importantly, though, the at-risk older age cohorts are above 90%) and is expected to peak at under 75%. Spain, Germany and France have vaccinated around 50% while the US is at 49% (and looks unlikely to get much above 60%). 

Few countries (other than the strictly governed Singapore and UAE) are currently within striking distance of Grattan’s target. 

The main problem with an all population target is that it requires a huge proportion of adults or a significant number of children to be vaccinated. Even if 85% of adults are vaccinated (which has rarely happened because vaccination rates drop off significantly for younger adults), we’d also need 60% children to be vaccinated. That is highly unlikely without serious sanctions for non-vaccinated people. 

But there is a difference between polls and actual action. Australia’s peak flu vaccination rate has been 76% — this includes a significant number of children who many believe are more at risk from the flu than COVID.

Can we count on the counting? 

Grattan’s report created a number of different scenarios based on various vaccination levels (50, 70 and 80% of total people) and various starting effective reproduction (Reff) rates. 

A quick refresher: the r0 is the basic reproduction number of a virus (for example, an r0 of 6 means each person will infect six others). The Reff is the “real life” r0 — that is, the r0 adjusted for things like natural immunity, vaccinations or interventions like lockdowns. While Australia has minimal natural immunity, we have already made good progress with vaccination — more than 40% of adults receiving their first dose and rising fast. 

Grattan used high starting Reffs of four, five and six (with a zero vaccination rate) and then discounted the Reff to between 1.5 and three to allow for vaccine usage. However, the Reffs used appear higher than we’ve seen in other countries (for example, the report notes the UK Reff during Delta briefly hit 1.5).  

For example, with 50% of the population vaccinated, this is what Grattan modelled would happen if Australia ended its COVID Zero approach:

Starting Reff	Infections	Deaths
4	~8,000,000	~18,000
5	~14,000,000	~32,000
6	~16,000,000	~38,000

Using the mid-range estimate, Grattan concluded a starting Reff of 5 would result in 14 million COVID infections. Note, under this scenario 12.5 million Australians have already been vaccinated. Grattan is therefore concluding that every non-vaccinated Australian will be infected with COVID — plus a few million vaccinated ones would as well.

Modelling meets real-world experience

The very high rate of infections modelled by Grattan don’t appear to reflect what has happened globally. 

In Canada 3.7% of the population have been infected, in Sweden and the US around 10% of the population tested positive. But these infection rates occurred mostly before vaccinations, so it’s reasonable to estimate with 50% starting vaccination coverage, 2-5% of the population would likely contract COVID. In a scenario of 50% vaccinations, we’d see potentially one million infections. (Remember, people also take their own precautions when the virus is spreading, like working from home and minimising contact, which reduces the observed Reff.)

In the UK, a 56% vaccination rate caused the case fatality rate to fall to only 0.20%. The fatality rate during NSW’s current wave is 0.27%. So even if we assume a (relatively high level) of 5% of Australians contracting COVID, based on the current NSW fatality rate, that’s just over 3000 fatalities — a far lower number of deaths than what was modelled by Grattan. 

COVID has made plenty of public figures like Neil Ferguson, Raina MacIntyre and Norman Swan look foolish with modelling that ultimately was incorrect, but in their defence, those forecasts were made very early in the pandemic before we had real data to guide our views. 

Grattan claims its assumptions took into account behaviours that reduce the spread of the virus and vaccine coverage to determine their effective reproduction rate and categorically denied that it overstated the number of infections. But based on actual data out of places like the UK, their conclusions appear to paint far too dire a picture — and one that doesn’t account for, among other things, the cost of Australia maintaining a COVID Zero policy. 

The costs of COVID Zero 

While Grattan’s report acknowledged the “significant costs” of COVID Zero, being a billion dollars a week on top of terrible education impacts, mental health and domestic abuse, for some reason it didn’t consider these costs in its consideration of what vaccination rate is optimal.

Grattan modelled infections and fatalities at different vaccination rates, but it didn’t account for the marginal costs of maintaining COVID Zero until we reached those high levels.

For example, at a 70% vaccination rate (starting Reff = 5), Grattan forecast under 20 deaths. At 80% vaccinated, it forecast just two deaths. But if it takes three months to go from 70 to 80%, we would be spending billions to save around a dozen lives. 

It doesn’t make sense to maintain a COVID Zero policy which prevents 15 COVID deaths but costs us money which could have been used to build multiple specialist cancer hospitals, which would have saved thousands of adults and children every year.

It’s possible that a vaccination rate as low as 20 or 40% could be the appropriate level. That is the level where the costs of COVID Zero start exceeding the benefits — we don’t know, though, because Grattan didn’t bother to model the marginal costs of COVID Zero — it only modelled the marginal benefits.

Every country in the world is grappling with the same challenge — that is, how to balance the costs of restricting COVID and the inevitable deaths and illnesses of a highly contagious and potentially fatal virus. However, Australia has the distinct advantage of global data. It’s imperative we rely on observed facts rather than “finger in the air” modelling, because the cost of COVID Zero for many is mentally, physically and financially catastrophic.

Adam Schwab is a Crikey and SmartCompany columnist, author of Pigs at the Trough: Lessons from Australia’s Decade of Corporate Greed, and the founder of LuxuryEscapes.com. He is a director of Private Media, the publisher of Crikey.