Featured researcher: Professor Catherine Bennett

Delivering the facts without the spin

Professor Catherine Bennett, Deakin’s Chair in Epidemiology within Deakin’s Institute for Health Transformation, is a leading researcher and teacher in public health, with a specific interest in infectious disease epidemiology and community transmission. During the COVID-19 pandemic, she has become a trusted and reassuring voice in the media, clearly presenting the facts around the daily case numbers and reducing anxiety and uncertainty by stripping away the misinformation and speculation surrounding the virus and its impact on our lives.

Her engaging commentary and expertise ranges from analysing and interpreting the numbers, to discussing the reasons why people can’t or won’t comply with restrictions, and the importance of facts over opinion.

Discover more about Catherine’s research and career

For the latest evidence-based comments and analysis, follow Catherine on LinkedIn

About the daily numbers

  • 27 November 2020: "Twenty-eight days without new cases for Victoria, and hopefully NSW will announce today that they have reached 20. Adelaide looks like they have this cluster under control (contacts of contacts pays off yet again) with only one new case reported. Victoria has reported some positive sewerage tests and so we do need to keep our testing rates up, and we need to practice those precautions that mean even if the virus is out there, it won’t have the opportunity to take off quickly. We have this Australia, but we all have to keep actively working on it individually and collectively to hold on to this... the positive wastewater results in our south west are a reminder."

Back to top

Stopping the spread

Restrictions and lockdowns

Masks and face coverings

Testing

Community transmission and contact tracing

Back to top

Points of view

    • 20/11/20: A leading epidemiologist has argued workers in hotel quarantine need to be paid more so that they don't work other jobs and risk coronavirus infection. Could higher-paid hotel quarantine workers better protect Australia from COVID-19?
    • 18/11/2020:  “South Australian six-day circuit breaker lockdown: let’s hope it’s enough to ensure they have found the edge of this cluster. The next few days will tell. Five generations of spread in less than two weeks makes it a challenging fight.
      They were not aware of the leak till an elderly second generation case attended an emergency department and the doctor decided to run a COVID test even though the respiratory symptoms were mild. This could have made all the difference.
      Testing the staff at med-hotels would have been better again, picking this up a generation earlier .and there might not be 4000 in isolation or an entire state locked down for six days.
      Let’s hope this works for South Australia’s sake, and because it matters a lot to the rest of Australia. It will also tell us if this short burst of extreme lockdown is a useful strategy to consider when dealing with a point source cluster that is not discovered until there have been a few generations of spread.” South Australia's 6-day lockdown shows we need to take hotel quarantine more seriously
    • 12/11/2020: “Is Victoria a model covid-19 response? Our story is one very different from what is happening in Europe now. We had the first wave contained through early introduction of six weeks of stage 3 lockdown, but we missed the start of the second wave when the virus was reintroduced from hotel quarantine breaches. We then had a large and long second wave largely driven by healthcare outbreaks, especially aged care, whilst Victoria was held in lockdown until these were eventually closed off ... and we had a public health response that could cope with opening up. Add to that a government so concerned about a third wave that it was believed we had get numbers extremely low “to keep them low”. That all ends up looking like 4 months of lockdown... and at what cost? The Victorian response must be judged on the extent and duration of lockdown, and the reasons behind the need for such a long lockdown, .. just the case number outcome.” What is behind Victoria's suppression success, and will it last?
    • 7/11/2020: “Coronavirus could "die out" in Australia entirely if New South Wales and Victoria stay on their current trajectory of low infections, according to two leading epidemiologists.” Coronavirus could disappear in Australia if NSW and Victoria maintain control over next few weeks, experts say
    • 3/11/2020: “Europe is beginning what Melbourne has only just finished: a second lockdown.” Podcast: Has Europe left it too late?
    • 25/10/2020: As coronavirus cases plummet, it's time to ask: Is Australia ready for the third wave?
    • "An opinion piece with colleagues .. we did not come up with the title and of course the comments relate to the title rather than the piece.. but I hope you appreciate the read!" COVID success will only come when Premier trusts the public
  • “Throwing some numbers around... if we have to count anything, then we may as well count in a way that captures where the real risk is (unlinked cases) and the linked cases where there is still some work to do… especially if these are casual contacts who may not yet be in quarantine.” A 14-day rolling average of 5 new daily cases is the wrong trigger for easing Melbourne lockdown. Let’s look at ‘under investigation’ cases instead
  • Doctors stoush over Victoria's extended lockdown
  • With cases stubbornly staying in the double digits, is there any light at the end of the tunnel? What was that about being OK for Xmas? Lockdown life in Melbourne shows no let up
  • “The CHO said today that once a close contact develops symptoms, DHHS will now consider them to be presumptive cases before they receive their tests results, and follow up their contacts immediately. So contacts of contacts will then be notified and asked to isolate as a precaution in case they are infected and heading into their pre-symptomatic infectious period. That’s exactly what’s needed when we have rapid spread with nearly 30 cases in 10 days in first and second level contacts. It sounds like DHHS might do this more extensively in outbreaks now - The real advantage of having lower numbers. The complexity of this outbreak shows what was happening many times over when there wasn’t the clarity or resources to see and investigate outbreaks with cases embedded in amongst 100s of new cases a day. We are in a whole new world now.. still extremely hard work for the department, but much more possible for the health teams to contain these outbreaks.” Victoria enters uncharted waters, urged to trace contacts of contacts
  • "Time to gather the information needed to fuel the difficult but necessary debates about how we live with the virus - we had limited choices when COVID-19 was such an unknown, but all action, and inaction, comes at a cost (lives, health, economic and the myriad of societal impacts), and now we must comprehensively evaluate our strategies." Lockdown is working, but is it excessive?
  • "We know hotel quarantine is not the perfect option, and not just in Vic. WA is a test case for us to determine how safe home isolation is for returned travellers, especially coming from low risk countries. If home isolation was successfully for many of our 20,000 known positive cases in wave 2, and for the quarantine of their close contacts who had a much higher risk of being positive than travellers returning from low risk countries, then this is an option to consider. Let’s hope evaluation of compliance in WA shows this to be a workable option, i’m sure Aussies wanting time come home (even from interstate) will be very relieved if this is deemed safe!" Coronavirus experts cautiously support Scott Morrison's plan for 'safe' international arrivals
  • Health experts question 'incredibly conservative' road map
  • “More modelling? The Burnet Institute team’s nuanced agent-based model calibrated to our second wave in Vic has more to offer. They found a 41% risk of another wave if we had jumped to the final roadmap step yesterday with our 14-day average at 22 (the previous conservative model said 62% for 14-day av. of 25).
    Many are now asking whether we still need the 5km rule. Or when will it be safe to move to step 3, and should the 14-day case count thresholds for this step be updated. These are the kind of scenarios that could be explored using this new model.
  • The cases seeded into the model should reflect the different mix of case risk we have now (92% are linked and most therefore likely to be in isolation/quarantine when tested), but could be really helpful in guiding decisions on thresholds and timing of steps, as well as whether the 5km rule could be lifted. This is stopping some families from meeting outdoors or permitted businesses opening if clients aren’t local.
    It could also help people adapt to the roadmap changes (inc lifting of curfew etc), and help allay fears they may have about opening up. This will be very safe, our detection and response systems too, so we need to bring the people along with us who no longer believe that!” Melbourne's five-kilometre rule: is it really worth it?
  • “Two months ago there was consensus among epidemiologists that to open up we would need new case numbers in low double digits and mystery cases to single digits. Since then it has all gone a bit crazy, but even after the road map was announced, there has been agreement that 5 cases was too ambitious and 10 cases a day is manageable. Over 70% if Vic’s cases are known to be linked when first reported, and therefore in people already in quarantine, and 85% of our cases under investigation are also linked eventually. We by now must have less than one mystery case a day on average (judging from changes in 14 day total). So why is a small number of contained but persistent outbreaks and 5 community cases a week not manageable? Sure we don’t want months of chasing cases like NSW has had to do, but continuing strict lockdown when we have already had 15 or more incubation cycles under it is clearly not the answer. We need to start easing restrictions, keep testing up, have fast and comprehensive case and contact follow-up, and remember the basics - distance, hygiene and masks. Then we can focus on prevention, early warning systems and monitoring to stay safe." What Australia needs to do to avoid a third Covid wave
  • Chatting again about all things Covid-19 with Tass and Fiona on Joy radio’s Saturday Magazine Catherine Bennett, Epidemiologist
  • “Divisive speak is dangerous - true always but particularly so in a pandemic. Opinions are being polarized; open-up completely or strict lock-down, you’re onboard or “you think it’s over because you want it to be over”, this virus is wildly infectious and cant be controlled without complete lockdown or you don’t believe it exists. Anyone with an opinion that doesn’t fit with either extreme is placed in to one to suit the commentator’s argument. These are worrying times. Most recently the narrative at the daily press conference is about regional and peninsula folk being terrified of Melburnians because of their “vastly higher amounts of virus”. Regional Vic’s 12-day av is a very low 1.9 cases/million, and the metro rate is also very low at 7.9 cases/million (half already in isolation). The Health Dept say they can finally see community transmission chains that were there all the time but hidden behind high numbers (data analyses could have shown sooner?). Stage 4 is about suppressing risk of transmission beyond these pockets where infection persists, creating the optimum backdrop for the Dept to do its work. 9 weeks in lock down has not been enough time for this (5-12 incubation periods), let’s hope another couple of weeks is.” - Catherine Bennett's LinkedIn
  • RMIT ABC Fact Check: Were NSW's coronavirus numbers higher than Victoria's thresholds for lifting curfew, as Scott Morrison said?
  • Has Australia really had 60,000 undiagnosed COVID-19 cases?
  • Another week, and we will know more about our October in Melbourne. The Age has taken the lighter moments from today’s press conference to build hope. Also a useful discussion on whether those identified in breach of CHO rules in contact tracing interviews should be fined. Our enforcement strategies have placed us in a difficult situation - respect privacy, open up conversations in order to close outbreaks down, or issue fines. This is part of a bigger public health communication conversation about the way we engage community. Public health communication strategies needs to provide transparency, clarity and certainty, so that people come forward for testing quickly, then have the conversations to allow us to chase any transmission - without rapid testing and information handover, it is that much harder to contain outbreaks. Outbreak control always works better if enforcement is a last resort, and that is something we should try to find ways to avoid going ahead. Premier raises hopes of easing restrictions sooner than planned
  • What is ‘herd immunity’ and what role does it play in a pandemic? Explainer: Reaching herd immunity in a viral pandemic
  • “Think before you voice an opinion — consistent messaging is critical in a public health crisis. Expert debate can be valuable, but publicly opposing current interventions on the basis of opinion rather than fact can reduce efficacy by undermining public confidence, and therefore compliance... both of which are already harder to garner in a second wave.” Latest restrictions became a self-fulfilling prophecy

Back to top

Estimated Reproduction Number for Victoria

(updated 30 September 2020)

The effective R number estimate and case count graphs: Reff can bounce a lot when case numbers are this low - if we have a low case count for a couple of days then it can appear to jump up after an incubation period (5-6 days) even if numbers are not that different. But remarkably we have only had two occasions when Reff peaked just above one, and the 10-day average is sitting around 0.6-0.7 and in fact has dropped this last week. This probably reflects that case and close contact follow-up is so much more effective when working with such manageable numbers, and working in partnership with the community. So easy for epidemic curve tails to be stubborn or unpredictable, so it’s great to see this steady downward trend continue right through to this end of the curve!

Notes:

  • Many interventions were rolled in; we focus on the date where the main intervention became mandatory
  • Not all cases are tested – only those with symptoms can be tested in the general community, and not all those will come forward
  • When comparing the counts across the 5-6 day incubation period, we use raw new case numbers in the numerator, and the average of the case counts taken from 5 to 7 days previously in the denominator.
  • The graph includes a 10-day moving average for the estimated Reff to highlight the trends

What is R0 and how do we calculate our estimates?

The Reproduction number (R0 or “R naught”) has been talked about a lot during this COVID-19 pandemic. It is an important statistical term that reflects how infectious the virus is in different settings. Each time we change levels of restriction, or mandate masks, we are changing the number of close contacts people have and the risk that they will infect another person, and so R0 changes. We call the specific R0 in any setting Reff; the effective Reproduction number.

Reff summarises the average number of people who will be infected by each current positive case and tells us how our outbreak control measures are working. Those secondary infections will be counted as ‘new cases’ after they develop symptoms, seek testing, and when a positive test result is recorded. If the average number of secondary cases linked to particular cases is greater than 1, then Reff is greater than 1. The number itself tells us how many people are infected by each case. So an Reff of 3 indicates that each case infects another 3 new people on average, and that the outbreak is growing, with case numbers tripling every incubation period. For COVID-19, the median incubation period is 5 to 6 days.

On the other hand, if existing cases don’t all infect at least one other person, then the number drops below 1 and we know the outbreak is under control and case numbers are shrinking. If Reff is, say 0.5, then that tells us that we can expect the number of new cases to drop to 50% of today’s new case numbers after one incubation period.

If we make certain assumptions, we can use the publicly available data to come up with a very rough estimate of the reproduction number for Victoria. If we assume that the time from infection to testing is reasonably consistent over time, then we can look at how new case numbers compare across incubation periods to estimate this Reff. It is likely, however, that the time to testing for people with symptoms may have changed as the outbreak became more serious, and with hardship payments making it a bit easier for people to forgo work and get tested and isolate until they have their result.

Another assumption we’re making is that the proportion of true cases out there that are identified through testing remains reasonably constant as well. If you had only 50 % of cases being tested in one week, and then 100% in the next, it might look like Reff had doubled, even if the true number of cases per day had not changed. This is true for all modelling where there is likely to be incomplete case capture, but as long as the proportion of cases that are counted stays relatively constant, then comparing even partial counts across incubation periods is still useful.

Case reports tend to come in batches from the outbreak investigations and the labs. For example, we see less reports on Mondays and Tuesdays as these are from tests completed on the weekend. Therefore, it’s useful to take an average of the number of existing cases on any one day. We use a 3-day average to smooth out those reporting fluctuations in the denominator of our calculation when we look at the ratio of new cases to those reported 5-6 days ago.

Finally, it’s important to note that the SARS-CoV-2 virus responsible for COVID-19 does not move through the population evenly. The number of people any one person infects is dependent on their number of close contacts and the environments they live and work in. It may also vary person to person based on their viral load, their symptoms, or other factors that may alter their potential to infect someone else. There have been many super-spreader events, whilst the majority of cases may not infect anyone else at all.

According to Victoria’s Chief Health Officer, Professor Brett Sutton, around one third of our new cases each day are from outbreaks, mostly in aged care and health facilities now. These have a big influence on our daily case numbers until these outbreaks are closed down and we stop seeing new clusters form.

All you need is a large super-spreader event, or a large new outbreak starting when someone takes the virus into their work place, and the reproduction number will change quickly. So, as always, this is not about numbers alone: the epidemic path is determined by where the virus is and who is exposed. Therefore, whilst useful to see how we are tracking, it remains very difficult to predict exactly where Reff will be in another five days’ time.

Want to know more? What Is R0? Gauging Contagious Infections

Case counts and roadmap targets

(updated 5 October 2020)

Mystery case update (revised):

The 14-day weekly average case number and total mystery case number both continue to head down, though the curve is now slowing as the change becomes incremental. Even so, the average case number the week before last was 15, this week it’s 10, and the Reff remains around 0.7. The average will continue to drop steadily as those higher numbers drop off the other end of the 14-day window. Aged Care and Health Care related cases are dropping now, so when the Chadstone-related cluster is closed off we should have very few cases.

Brett Sutton said last week that maybe they might look at the number of cases that are still under investigation for the roadmap targets (those already linked in morning they are reported are likely to be in quarantine awaiting results). These includes some cases that will yet be linked to clusters, and the mystery cases. I too have been tracking this number for some time, and the good news is our 14-day average for all those under investigation on the first day of report is only 5. Yay!

The effective R number estimate and case count graphs: Reff can bounce a lot when case numbers are this low - if we have a low case count for a couple of days then it can appear to jump up after an incubation period (5-6 days) even if numbers are not that different. But remarkably we have only had two occasions when Reff peaked just above one, and the 10-day average is sitting around 0.6-0.7 and in fact has dropped this last week. This probably reflects that case and close contact follow-up is so much more effective when working with such manageable numbers, and working in partnership with the community. So easy for epidemic curve tails to be stubborn or unpredictable, so it’s great to see this steady downward trend continue right through to this end of the curve!

Notes:

This graph tracks both daily cases and reported deaths over the second wave. The axis for the cases is on the left, and for the deaths it is on the right and, of course, on a much smaller scale. The reason we have them both in the same graph, even though the numbers are so different, is to highlight the difference in timing between new cases and deaths.

There is on average a week’s lag from someone developing symptoms and being tested and counted as a case, and then developing serious enough illness that might require hospitalisation. It might be a further week before they require intensive care (ICU). Some people can remain in intensive care for a very long time; some might succumb to the virus even without being in ICU. But, on average, there will be a two to three-week lag between peaks in cases and the peak in the deaths that are the terribly sad consequence of those large numbers of new infections.

Cases started to plateau at the end of July, and we saw sustained high numbers for a couple of weeks before the case numbers started to fall in the second week of August. The numbers of reports of deaths finally also started to plateau in mid-August and, whilst fluctuating, has been reasonably stable since; averaging about 15 deaths a day. As we are now at the end of the month, it is expected that we will soon see daily reports decline, and the 7-day average with them. Because the day to day reports can fluctuate a lot, we also show the 7-day average so that you can see the trends over time.

‘Disease detectives’: The role of epidemiologists during a pandemic

Epidemiology is the study of health and illness in human populations, from the occurrence and distribution of disease and the factors behind to the dynamics of infectious disease outbreaks. Epidemiologists use the information they gather to design and test interventions for prevention and control, promote public health education and inform government policy making.

Epidemiologists are invaluable during a pandemic because of their skills in gathering information about disease in a population and using it to anticipate what’s needed, identify risks and prioritise health and other resources.

Listen to Catherine talk about life as an epidemiologist: So you want to be an epidemiologist?

Become an epidemiologist