By Megan Higgie and Andrew Kahn

Megan Higgie

·

Follow

8 min read

·

Mar 18, 2020

--

1

Listen

Share

Updated 19 March with predictions for Australia (which is now one day sooner) and adding a prediction for NSW.

Before reading further…

The model we present below is the worst-case scenario of exponential growth of COVID-19 cases continuing into April. The reason we wrote this article is because Australia has been tracking this worst-case scenario for the last two weeks, that is, experiencing exponential growth with the number of COVID-19 cases doubling every 3–4 days. We hope that the social distancing measures that have been introduced in Australia this week slow this increase, but only if people adhere to them and as well as the recommendations around hand and face hygiene. However, this is not yet visible in the data due to the lag time of 7–10 days between the actions we take to prevent the spread and the consequences of those actions appearing in the case numbers. As soon as the growth of cases is no longer exponential then this model below will not predict our future number of cases. We will be very clear to update this article as soon as this happens. Every day we hope to see the new data show a slowing of the spread and move away from exponential growth. Let’s all hope (and act to ensure) that happens very soon. However, until case accumulation stops being exponential, we are on track for the worse-case scenario outlined below.

1 | ICU beds are the difference between surviving and dying from COVID-19

Using data from > 70,000 cases in China, 5% of COVID-19 patients become ‘critical’ and require an Intensive Care Unit (ICU) bed with a ventilator in order to survive [note 1]. There are reports that patients require the ICU beds for approximately 4 weeks, while others out this at 7–10 days. The number of ICU beds is therefore critical in determining the number of people who unnecessarily die from COVID-19.

Below are the number of ICU beds, both in public and private hospitals, in Australia (by state) and NZ, from the last CORE Report of the Australian and New Zealand Intensive Care Society (ANZICS) [note 2]:

2 | Number of COVID-19 cases before we run out of ICU beds

It is simple maths to then figure out how many cases before we run out of those beds (and this is when people start dying due to a lack of suitable medical equipment):

3 | What date will we run out of ICU beds?

Using the 5% rate of COVID-19 patients requiring an ICU bed and the above number of ICU beds, we now have a very firm idea on the number of COVID-19 cases before we run out of ICU beds. For Australia, that is 44,580 cases (from Table 1 above).

Next we can use some mathematical modelling to estimate what date we will run out of ICU beds.

We can only use this modelling where there is enough cases to fit a model. Currently that is Australia as a whole and now NSW. As cases accumulate we will add this analysis for other Australian states and for NZ. See note [3] for details of our analyses.

3.1 | Australia

The exponential growth of COVID-19 cases in Australia as modelled below has a very high level of confidence (R² > 99.8%), and is shown here as the black line with a 95% prediction interval shaded in blue. The red line is taken from the Australia row of data in Table 1, that is the number of COVID-19 cases before ICU beds are full. From this we can then project the dates that Australia will run out of ICU beds:

With the current worst-case scenario of exponential growth of COVID-19 cases, Australia may run out of ICU beds between 7 and 9 April

3.2 | NSW

The exponential growth of COVID-19 cases in New South Wales (NSW) as modelled below has a very high level of confidence (R² > 99.5%), and is shown here as the black line with a 95% prediction interval shaded in blue. The red line is taken from the NSW row of data in Table 1, that is the number of COVID-19 cases before ICU beds are full. From this we can then project the dates that NSW will run out of ICU beds:

With the current worst-case scenario of exponential growth of COVID-19 cases, NSW may run out of ICU beds between 5 and 9 April

Assumptions

We are assuming a few things here, but none of them extend this date out much further; in fact, three of them bring the date sooner:

1. That all ICU beds are only used by COVID-19 patients and not for any other medical emergency (e.g., car accidents, cardiac arrests) — this assumption is plainly not correct, so our actual date will be sooner than predicted.
2. That Australia and New Zealand have not added any new ICU beds since the 2018 ANZICs report. This is not correct, but we could assume that at least those new ICU beds added in 2019 are full due to the non-COVID-19 critical cases.
3. That ‘only’ 5% of cases become critical. We have seen anecdotal reports of 6–10% of people requiring ICU beds. Obviously this would bring the actual date much sooner. We have also seen reports that it may be less than 5%. But while we still have exponential growth of cases then halving and quartering this 5% rate only extends the estimate on when we run out of ICU beds by days – not by months.
4. That the current strategies to control the spread of COVID-19 remain the same (e.g., schools, universities, daycares, and non-critical businesses and services remain open). When we first published this article this was still true. Since then some stronger measures of social distancing have been introduced. Whether the effects of these measures will be strong enough to halt the exponential growth of cases will only been seen in the data a week or so after people begin adhering to them.
5. That we are confirming all cases that are occurring. With the criteria for testing remaining being quite limited in Australia then we are missing cases. This brings the actual date sooner than what we are predicting here.
6. That there is no capacity to easily increase the number of ICU beds rapidly and massively. There are some reports that makeshift ICU beds can be made, but how many and how quickly? Both unknowns at this stage.

What else do we need to know?

There are two further extremely important points to consider:

1. That there is a large lag between becoming infected and being ill enough to require medical attention. This lag is on average 10–14 days [note 1]. Given that 7 April is 19 days away, we only have 5–9 days to take the serious action necessary to prevent this human disaster in Australia, and less time in NSW.
2. The majority of infections are passed on before someone even shows symptoms. Approximately 80% of infections occur when an infected person does not even know they are infected [note 4]. This is a much under-appreciated fact and explains why Australia is still experiencing exponential growth in spite of quarantining people once they are sick — they have already infected people before they went into quarantine!

What do we need to do to prevent unnecessary deaths from 7–9 April onwards (both individually and as leaders)?

The only effective measure is that we go into lock-down and practice extremely strong social distancing within the next 5–9 days.

If people cannot tell when they are sick and pass on the majority of infections before their first symptoms show, the only possible course of action is to avoid contact with other people.

There is strong evidence [note 5] to show that only the action of extremely strong social distancing (also known as ‘suppression’) has been the difference between the high death rate in Italy versus the much lower death rates in China (especially once they knew what they were dealing with), Taiwan, South Korea, and Japan.

Yes, this may mean schools and daycares closed (except for children of critical workers, such as medical people and absolute core infrastructure needs such as sewerage plants, etc.), it means most businesses closed. Yes, businesses and we as individuals will lose money, but we are going to lose money either way and much much more than money if we don’t act soon and effectively.

The benefits of this early and strong action will be multifold:

• we won’t lose a massive number of our family, friends, and colleagues (due to lack of ICU beds — see above analysis)
• the actual number of infections will be lower and so we will reach the peak number of cases per day sooner (because the daily number of infections starts to decline, and eventually we will have less cases one day than the day before — that is, we have passed the peak). This means the sooner we act, the sooner that we can relax the strong social distancing.
• it buys us time — time to find a vaccine, time to increase ICU bed capacity and train staff, time to make more protective equipment like masks. See an excellent article on why we need time here https://medium.com/@tomaspueyo/coronavirus-the-hammer-and-the-dance-be9337092b56.

Notes

1. Wu & McGoogan. 2020. Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in China: Summary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention. Journal of the American Medical Association
2. ANZICS Centre for Outcome and Resource Evaluation 2018 Report, page 7
3. Based on data from Juliette O’Brien at https://www.covid19data.com.au/, we fitted an exponential model of confirmed cases over time since 10 March for Australia (the date Australia recorded its 100th confirmed case) and 14 March for NSW (the data NSW recorded its 100th confirmed case).
4. Li et al. 2020. Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV2). Science
5. Ferguson et al. 2020. Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand