Victoria’s modelling
Mathematical models are theoretical scenarios which are used to help plan our responses.
These models have been a cooperative effort between the Victorian Department of health and Human Services, Monash University and modellers based at The University of Melbourne led by the Peter Doherty Institute for Infection and Immunity.
Models do not predict the future. They are used to answer ‘what if’ questions, and to assist in planning when data may not be available.
Potential scenarios in Victoria
The scenario modelling considered in this work is based on the same transmission model used by the Commonwealth government and released by the Peter Doherty Institute for Infection and Immunity [Moss, R., Wood, J., Brown, D., Shearer, F., Black, A.J., Cheng, A.C., McCaw, J.M., McVernon, J. Modelling the impact of COVID-19 in Australia to inform transmission reducing measures and health system preparedness. Preprint published online – 7 Apr 2020. The model includes isolation of confirmed cases, and quarantine of close contacts. Key parameters are shown below.
|
Parameter |
Estimate |
|
Victorian population |
6,359,000 |
|
Number of infected individuals arriving |
5 |
|
Weekly new infected arrivals |
2 |
|
R0 in absence of public health controls |
2.53 |
|
Incubation period |
3.2 days |
|
Duration of infectiousness |
9.68 days |
|
Time from disease onset to detection |
1 day |
|
Reduction in infectiousness once detected |
80% |
|
Reduction in infectiousness due to quarantine |
50% |
|
Percentage of cases requiring hospitalisation |
6% |
|
Percentage of admissions requiring ICU |
30% |
|
Percentage of ICU admissions requiring ventilation |
70% |
|
Mean stay in ICU |
10 days |
|
Mean stay in ward |
8 days |
Physical distancing policies have been implemented as a reduction in R0 applied over the entire epidemic.
The likelihood of developing symptoms necessary for hospitalisation is heavily age dependent.
|
Age |
Likelihood of severe disease |
|
0-9 |
0.03% (0.02-0.03) |
|
10-19 |
0.03% (0.02-0.03) |
|
20-29 |
0.4% (0.3-0.5) |
|
30-39 |
1.4% (1.20-1.7) |
|
40-49 |
2.6% (2.2-3.0) |
|
50-59 |
4.9% (4.2-5.9) |
|
60-69 |
7.7% (6.5-9.2) |
|
70-79 |
17.9% (15.1-21.2) |
|
80+ |
33.0% (27.8-39.2) |
These scenarios have focused on calculating raw bed demand only, and so have assumed effectively infinite capacity for health services to triage and treat cases and are independent of workforce and equipment constraints.
Estimation of the effective reproduction number
The effective reproduction number, Reff, is representative of how many people are infected by a single person on average. It is calculated based on the ‘serial interval’, which is the time between clinical onsets, using a method developed by colleagues at the London School of Hygiene and Tropical Medicine [Abbott, S., Hellewell, J., et al Estimating the time-varying reproduction number]. This method accounts for reporting delays, to account for whether an observed drop in the number of cases is reflective of an actual drop in the number of cases.
As an example, if everyone was infectious for 7 days, and there was 10 cases a week ago, and 20 cases today, then we know that every person passed the infection to two other people (on average), and so Reff=2.
The effective reproduction number is smoothed over a 7 day window to reduce the impact of localized events that may cause large fluctuations.
Conversion from the effective reproduction number to doubling time
To calculate the doubling time from the effective reproduction number, the same dynamic transmission model as used in the scenarios (in the absence of public health measures) is run for a given Reff until the total number of cases doubled.
Theoretical modelling to inform Victoria’s response to coronavirus (COVID-19) (Word)