2022: the year COVID-19 truly came to Australia.
Since January 1, more than 10 million COVID-19 cases have been diagnosed and reported through either a PCR or RAT test.
Depending on how you count them, we’ve had up to five waves of Omicron sweep across the country.
That’s already a staggering number of cases, when compared to Australia’s 2021 or 2020 totals, but it belies the true number of infections actually occurring.
We’ve always known the official case number was an underestimate of true transmission, but 2022 was the year that we lost grip of just how big an underestimate it is.
Epidemiologists call it the case ascertainment rate. Put simply, what percentage of infections that exist each day are we aware of?
As someone who’s been reporting on COVID-19 data since the beginning of the pandemic, this has been the question I’ve been asked most often this year.
As we enter 2023, the fourth year of this global pandemic, let’s attempt to answer it.
Beyond contact tracing
The short answer is we don’t know how many cases are going unrecorded. Are we catching half of all cases? Thirty per cent? One in 10? It’s hard to know for sure.
Professor James McCaw, head of the Infectious Disease Dynamics Unit at the University of Melbourne and one of the leaders of the Doherty Institute consortium, says his team knows most cases are being missed each day, but they only have a fairly vague idea of what the case ascertainment actually is.
“I would expect case ascertainment is somewhere around 10 to 25 per cent, with a lot of uncertainty," he says.
“It’s not up at the 50 to 80 per cent range … like it almost certainly was over the first two years of the pandemic.”
If we do have a case ascertainment range of 10-25 per cent, that meant that last week, when there were about 107,000 cases reported, there were actually between 428,000 and 1.1 million people infected.
The huge gap between those two numbers shows the high level of uncertainty we're facing.
But it's worth drilling into the long answer as well. It tells us a plenty about how we’re handling the pandemic going into 2023, what the novel coronavirus has taught us, and how we could be managing infectious diseases in future.
Freya Shearer, who works with University of Melbourne and Doherty Institute teams to produce modelling and ongoing assessments of the pandemic, says we had a “near perfect” case ascertainment rate in 2020, but it started to slip during the Delta outbreaks of 2021.
“We definitely lost a handle on what case ascertainment was in December 2021, when Omicron emerged,” Dr Shearer says.
That’s a direct consequence of the change in strategy adopted by governments around that time, which saw centralised contact tracing essentially cease.
“[In 2020] we were actively trying to find all cases and all chains of transmission, so we had intensive upstream and downstream contact tracing,” Dr Shearer says.
“The upstream contact tracing wasn't something that was done in all countries."
This meant starting at a positive case and working backwards to understand where they caught the virus, using genomic sequencing to search for unknown chains of transmission.
Does it matter?
Why do we care what Australia’s case ascertainment rate is? After all, the strategy that all Australian governments have signed onto is about managing COVID-19 consistent with other infectious diseases, and in particular, “focusing on the prevention and management of serious illness”.
“Australia has a complement of well-established surveillance systems that allow us to understand the aspects of disease transmission that have the greatest public health impact, including the prevalence of severe illness, health system capacity and emerging variants of concern,” a federal health department spokesperson told ABC News.
It's unlikely that knowing more about the cases we’re missing would change that national strategy.
The trouble is that uncertainty about numbers propagates through all modelling and forecasting of the pandemic in Australia.
Having a good idea of case ascertainment would improve the accuracy of forecasts, letting governments, hospitals and the public better understand when peaks are coming and how big they will be.
“I think it's a very useful piece of information because it would have enhanced the accuracy and the reliability of the epidemic intelligence and the projections that are provided to government,” Professor McCaw says .
“I don't think it fundamentally would have reshaped the approach that Australia has taken to managing the virus this year. It just would’ve helped manage acute elements of it.”
And, in a world where people are left to make their own decisions about COVID-19 precautions, it would’ve helped members of the public determine what to do based on their personal risk tolerance.
If you want to make an accurate forecast of how a case curve might grow, you also want to have a good understanding of how many people are susceptible to the virus. That’s difficult if you don’t know how many people have some immunity from recent infections.
“We're constantly asked about peak timing and size, and this uncertainty is perhaps one of the most important unknowns in trying to predict peak timing and size,” Dr Shearer says.
Accurate case numbers would also help when responding to peaks, she says, like deciding whether to launch a vaccine booster campaign or ratchet up mask rules. Right now, the uncertainty around case ascertainment makes it hard to model the impact of any public health intervention.
“Suddenly intervention 'A' looks really similar to intervention 'B' because it’s all wrapped up in this uncertainty envelope,” she says.
While COVID forecasts have faded into the background this year, their importance could quickly rise if a new, more deadly variant, came into play.
“It doesn't matter as long as everything's going fine,” says Dr Michael Lydeamore from Monash University.
“But when it will matter is if we get some new class of variant or something like that, that very quickly changes the picture.
“When something goes wrong it could go very wrong."
With as much transmission as we have now, you’d never be able to achieve a 100 per cent case ascertainment rate every single day, but you’d get a near-perfect rate if the entire population did a daily PCR test.
That would be, of course, both massively unwieldy and absurdly expensive.
When it was considering the national plan framework and targets in 2021, the National Cabinet was told of the importance of case ascertainment. It made its decisions knowing that case ascertainment would drop.
The Doherty modelling that leaders extensively referred to canvassed the issue, even making some suggestions about measures they could take to improve it.
What could we do?
There are a few ways to go about measuring case ascertainment.
What’s the probability that, if someone is infected, they show symptoms? From there, if they’re showing symptoms, what’s the probability that they get a test? And if they get that test, and it shows up positive, what’s the chance (for rapid tests) they report that result to the government?
If you knew all those things, you’d be able to work backwards from the case numbers and come up with a decent estimate.
But we don’t have a great handle on each of those three ingredients, and it has exposed some gaps in how we conduct disease surveillance in this country.
Let’s step through them.
What are the chances of having symptoms if you’ve been infected? The expert modellers have been using some information learned in population surveys in the UK, but it's not perfect.
For one, the UK and Australia have different populations, with different levels of immunity from both vaccines and natural infection. The survey data has also got less reliable, since the BA.2 wave in the first half of the year.
New variants can change things, so there’s quite a lot of uncertainty about how likely you are to have symptoms if you’re infected.
What’s the probability you get tested if you have symptoms? Even anecdotally, it’s clear that this has gone down significantly this year, and it likely varies quite a lot between different demographics and communities.
The federal health department is conducting a weekly behavioural survey, aiming to understand how people are acting in the real world regarding public health advice and rules. The results of that survey are available to governments and the researchers working with governments, but not to the public.
“It’s fair to say that case ascertainment has dropped dramatically, including since around the BA.4/BA.5 wave, because we can see people’s behaviour has changed,” Professor McCaw says.
“The probability of seeking a test, given [a person has] symptoms, has gone down notably over the last few months.”
And importantly, the number of people getting tested is not a fixed number; people are probably more likely to get tested if they’re aware that a COVID wave is underway.
Finally, the chance a positive test gets reported to government is a newly important ingredient, now that you’re not required to report a positive test. We don’t yet have a good idea of how many people are actually reporting their RAT results.
The forecasters put all those ingredients together, making the best estimates they can for each of those probabilities. That's given them the very rough estimate of 10-25 per cent.
Having a low case ascertainment rate isn’t necessarily a problem. For example, we know that it’s very, very, low for influenza.
“What you do need to know is what that case ascertainment rate is,” Professor McCaw says.
What else could we use?
Short of estimating case ascertainment directly, there are other surveillance methods that give us some information.
One of them is the quarterly seroprevalence surveys led by the National Centre for Immunisation Research and Surveillance. This survey tests blood samples for COVID-19 antibodies, allowing researchers to estimate the number of people who have been recently infected.
The federal government points to this survey as an important part of its surveillance.“This helps us understand how many people have been infected, as well as the proportion with immunological markers, that indicate protection against future infection,” a spokesperson said.
These don’t let you directly calculate case ascertainment, but they put an upper bound on what it could be, and they give researchers information about how many people are susceptible to COVID-19 at the time of the survey.
Professor McCaw says these surveys have been “incredibly valuable”.
“They are the direct empirical evidence — still hard to interpret — that showed we were missing people,” he says.
The other surveillance tool used in many parts of the country is wastewater testing. Samples of sewage are regularly taken and tested in laboratories that detect fragments of the virus.
That testing was useful when there was very little virus in Australia, because it was often an early sign that a new outbreak had started.
Now, it’s helping uncover broad trends and answering important questions about the different COVID-19 variants circulating in Australia. But it has limited use in helping to determine the case ascertainment rate.
One of the big reasons is that the amount of COVID-19 in a sample is determined not just by how much of the virus is in the community, but also by how much sewage is flowing through pipes and environmental factors like rainfall, which can dilute the sample.
Professor McCaw says wastewater sampling could be more useful in future if researchers across disciplines work together to develop the method.
“I think there are huge prospects in the long term — the next five to 10 years — of wastewater surveillance being a very exciting area that could give us amazing quantitative analytic tools on all sorts of pathogens," he says.
“But we’re not there yet.”
Population prevalence surveys
So that’s a heap of indirect information, with lots of uncertainty. The best thing to do would be to attempt to measure it directly, with something like a population-wide prevalence survey.
The concept is relatively simple. You take a random sample of Australians, test them for COVID-19, do a bit of statistical analysis, and voila, you have an estimate of how many infections are actually occurring.
This was the theory behind surveys done in the UK since the beginning of the pandemic.
One prevalence survey from the UK’s equivalent of the ABS has been running weekly since May 2020, although with a smaller sample now than when it was launched. A similar survey called REACT has been run by Imperial College London.
In Australia, there have been a couple of one-off prevalence surveys on the Gold Coast and in Perth — but nothing national.
The latest COVID-19 national surveillance plan discusses a proposal for a widespread population infection survey.
The plan, published in June, says “the utility and feasibility of a nationally coordinated survey is currently being evaluated”. It suggests such a study would help track infections over time, monitor the rate of spread to plan public health measures, and estimate vaccine effectiveness.
Professor McCaw says such a survey would help greatly with the uncertainty, but that there "are no silver bullets".
A federal health department spokesperson says that evaluation is still underway, but in the meantime:
“Australia has existing systems in place to support our primary surveillance objectives, i.e. monitor disease trends with the aim of reducing the burden of severe illness on individuals and the broader population”.
Health Minister Mark Butler was unavailable for an interview, and the health department responded on his behalf.
While most Australians wouldn’t look at the United Kingdom’s COVID-19 response as one to copy, several modellers, including Dr Michael Lydeamore, see the prevalence survey as a missed opportunity.
“They have a pretty strong idea [in the UK] of where COVID is going to circulate again … even if their ascertainment isn’t up to scratch,” Dr Lydeamore says.
“I think Australia really missed the boat on that kind of survey.”
The next pandemic
The Albanese Government has committed to creating a centre for disease control, noting that Australia is the only OECD country without one.
It would be a national coordinating body for disease management, and the government is currently consulting with experts about how it should work.
“With all our successes, [the pandemic has] also revealed some of the challenges around the lack of a full coordinating system,” Professor McCaw says.
“A CDC is a wonderful development for Australia.”
While we don’t have the full details yet, the government has already said one of its first focuses will be on surveillance.
The pandemic has transformed a lot of how public health works in Australia. Public health officials and political leaders have never had the kind of information and data available to them that they’re now used to receiving.
That remains the case even through 2022, as parts of Australia’s emergency response were dismantled bit by bit.
“The absolute emergency phase of the pandemic is fading,” Professor McCaw says.
“What do we do for COVID surveillance over the next indefinite future?
“What do we do for influenza surveillance? We used to just say ‘oh let’s just monitor the cases and a few sentinel GPs and the hospitalisations’.
“But that’s not good enough anymore, because we know we can do better.”
