Overview of COVASIM modelling released by NSW Government

Since June 2021 NSW has experienced a resurgence in Delta variant COVID-19 cases. The majority of infections were originally concentrated in only a few Local Government Areas (LGAs), but then expanded throughout the state with 12 LGAs now classified as “of concern”. Cases have continued to grow in the rest of greater Sydney with cluster outbreaks appearing in regional NSW.

Burnet’s modelling team were asked to provide projections to the NSW Government. The COVASIM model was used to simulate the greater Sydney population as two different units: the 12 LGAs of concern, and the rest of greater Sydney. The model has been calibrated on the NSW case numbers to date and vaccine roll-out, including an additional 530,000 doses of Pfizer vaccine delivered over three-week intervals for people aged 16-39 years in the 12 LGAs of concern. Data was also drawn from a variety of other sources including work-type, people movement, vaccine roll-out and uptake. It also took account of restrictions that were incrementally imposed in NSW, including ongoing lockdown and closure of non-essential services throughout Sydney, restrictions on mobility between the LGAs of concern and the rest of Sydney, and shorter lockdowns of regional areas in response to outbreaks. Vaccination rates have also been increased.

This work is ongoing with the model being regularly updated and recalibrated based on new data as it becomes available.

Download Modelling Explainer

Key assumptions for the current projections:

• Projections were based on data as at 27 August 2021
• The released graphs were of the 12 LGAs of concern, not all of NSW
• Case numbers in projections are 7-day average, not single day numbers
• Hospitalisation numbers were based on an estimated average 18-day stay in hospital
• In the reported scenario, public health restrictions are maintained
• 80 per cent+ first dose coverage is achieved among 16-39 year-olds in the 12 LGAs of concern by early September 2021.

Key findings:

Under current restrictions and projected vaccination numbers, diagnoses and hospitalisations in Sydney will continue increasing for several weeks before reaching a peak and declining once sufficient vaccine-acquired immunity is achieved.

1. Daily diagnoses:

- In the LGAs of concern are estimated to peak at a 7-day average of 1,129-1,967 between 13-20 September 2021.

- Across all of greater Sydney, the 7-day average is estimated to peak at 1,219-2,046 between 13-20 September.

However, it is recognised that the modelled estimates of daily diagnosis for the rest of greater Sydney an underestimate of actual current daily diagnosis as they are part of an initial model calibration.

2. There are corresponding estimated peaks in:

• Hospital demand of 2,222-3,938 in the 12 LGAs of concern.
• Hospital demand of 2,286-4,016 across all of greater Sydney.

Please note, Burnet did not provide state-wide estimates of the number of infections. However we are aware that NSW Health has access to a number of models and the peak ICU demand of 947 beds is for the entirety of NSW.

3. Responses implemented so far are estimated to have averted a substantive number of infections and deaths.

• In the 12 LGAs of concern, without any restrictions being introduced or additional Pfizer doses there would have been an estimated 589,817 and 5,808 cumulative diagnoses and deaths respectively over Jun-Dec 2021.
• The restrictions combined with the original vaccine roll-out have averted an estimated 488,020 infections and 4,830 deaths.
• The additional 530,000 Pfizer doses are estimated to have averted an additional 24,267 diagnoses and 254 deaths.

Model assumptions:

Model inputs included data on demographics, contact networks, workforce composition, contact tracing systems and age-specific vaccination rates. Model parameters for testing and the impact of packages of restrictions were calibrated to achieve the observed testing, diagnosis and hospitalisation numbers.

As well as a baseline projection, scenarios were run to assess:

• The impact of past restrictions and the estimated impact of the additional 530,000 Pfizer vaccine doses
• The additional cases / hospitalisations if high vaccination cannot be achieved uniformly across all sub-populations.

Vaccine properties:

In the model, vaccination acts to reduce the probability of acquiring an infection when a contact occurs with an infectious case, as well as the probability of developing symptoms (both mild and severe) for people who are vaccinated and become infected. The assumed efficacy values are based on estimates from Imperial College London, London School of Hygiene and Tropical Medicine and Warwick University (the same source used in the Doherty modelling report but parametrised differently for the COVASIM model).

Find out more about COVASIM at burnet.edu.au/Know-C19