18 October 2021
Report delivered 9 September 2021, this version compiled 15 October 2021.
We use an individual-based, Aotearoa-specific Contagion Network Model (CNM) to simulate the spread of COVID-19 in the community for an outbreak comparable to that detected on 17 August 2021. The CNM explicitly simulates spreading processes and the effect of any interventions, in order to predict the spread of COVID-19. It is therefore able to avoid the need for input assumptions or external estimates of the effective reproduction number (Reff) during the outbreak.
We use this model to consider two scenarios:
- An intervention with parameters comparable to Alert Level 4 (AL4) is applied from 18 August 2021 onwards and remains in place for the duration of the simulations.
- An intervention with parameters comparable to AL4 is applied from 18 August to 15 September 2021, after which the intervention parameters are changed to match those expected for Alert Level 3 (AL3).
We complement the CNM simulations with results from a Branching Process Model (BPM). We use this to simulate scenarios where an Alert Level change on 16 September 2021 results in a multiplicative increase of Reff of factors of 1, 1.5, 2, or 3 times that estimated during AL4.
After 60 days at AL4, it is predicted that almost all trajectories will be either eliminated or contained (i.e. all cases are either in quarantine or isolation). In contrast, only a few trajectories are eliminated or contained for the optimistic and pessimistic AL3 scenarios. In the former, many trajectories exhibit suppression like behaviour, remaining at low numbers (under 20 new cases per day) or growing only slowly.
The pessimistic AL3 scenario shows a growing number of daily new cases for the majority of trajectories. The pessimistic AL3 scenario also suggests that it would initially be difficult to distinguish between uncontrolled growth and suppression or elimination like behaviour – it takes approximately another 10 days post-deescalation before the growing trajectories start to become distinguishable.