Covid 19 coronavirus: Could the Auckland cluster flare up again?

What difference would an extended Auckland lockdown have made to its current outbreak? Or what about if the city had stayed at level 1 instead? And why is this situation a trickier one to predict than the one New Zealand faced before its first lockdown in autumn? Science reporter Jamie Morton asked Te Punaha Matatini investigators, Professor Michael Plank and Dr Alex James of the University of Canterbury, Professor Shaun Hendy and Nic Steyn at the University of Auckland, and Dr Rachelle Binny and Dr Audrey Lustig at Manaaki Whenua-Landcare Research.

What does the modelling tell us about the risk environment we're in now? You work off a reproduction number - or the average number of people that one infected person can pass the virus on to. Do we have any idea what that number is under these settings? And what can we say about the likelihood of cases beginning to climb again?

The data shows that the reproduction number was higher at the early stages of this latest outbreak than it was towards the end of March.

This means case numbers could have grown very rapidly if the Government had not reacted as quickly as it did.

Under alert level 3, we estimate that the reproduction number fell to around 0.5 to 0.7.

This is not far off what we achieved under alert level 4 in April, reflecting the role of improvements in contact tracing and managed quarantine.

At alert level 2.5, we hope that the reproduction number will be less than one, but it could be just below or just above one.

We won't know for a week or so, until we start finding cases that were exposed under this new alert level.

If the reproduction number is greater than one, then we would expect to see a slow increase in new case numbers over the next few weeks.

It is also possible we could see a large spreading event, if businesses, schools, and the general public aren't able to follow the rules at alert level 2.5.

If this happens, we might have to go back to level 3, so it is important that people avoid gatherings of more than 10.

If Auckland and the rest of New Zealand was operating in a level 1 setting right now, how might that change the picture?

We still have active cases in the community and we know the reproduction number at level 1 is much greater than 1, even with excellent contact tracing.

This means if we were to go back to level 1 now, then we would probably see rapid growth of the cluster in Auckland and it would spread quickly to other parts of the country.

What might have happened had Auckland remained at level 3 for longer, as you recommended?

The reproduction number at alert level 3 was less than one.

This means that if Auckland had stayed at level 3 for longer, the number of cases would have declined over time, although there probably would have been some ups and downs along the way.

Commuters step off the Fullers Ferry from Waiheke Island in an Auckland operating under alert level 2. Photo / Dean Purcell.

Level 3 also makes our contact tracers' jobs easier because there are far fewer difficult-to-trace casual contacts, like people who were on the same bus or in the same supermarket.

This means that extra time at level 3 gives us a better chance of extinguishing active chains of transmission.

However, extra time at level 3 also has economic and social wellbeing consequences that are not felt as much in level 2.

In the first lockdown, you were able to correctly predict the day that our case numbers would drop to zero. Why is it trickier to say exactly when this outbreak might tail off to nothing? What are the big variables we're dealing with here?

It's very hard to predict what a single cluster will do, because random occurrences, such as who happened to take the bus on a particular day, play a big role.

In the last outbreak, we were dealing with seven or eight large clusters.

That meant we could average out some of the random variation and make reasonable predictions.

But even then, it was difficult to predict accurately when cases get down to small numbers. Instead, we try to estimate the probability of elimination.

This is a way of measuring confidence that community transmission of the virus has ended.

It's been pointed out that South Auckland, particularly, is one of the toughest places in the country to control an outbreak. From a modelling standpoint, how do you factor in this area's unique demographics, as opposed to modelling spread of the virus from a national perspective?

The main factor from a modelling standpoint is that we know big parts of this community are at high risk of developing severe illness or dying from Covid-19.

This makes it all the more important to make sure the virus doesn't get out of control.

The higher density housing and larger household sizes also mean that transmission rates can be higher.

This was seen very strongly in the first wave of cases in New York where the reproduction number was almost double what had been seen elsewhere.

Is it possible to say what difference the new measure of compulsory masks on public transport - and advocated use elsewhere - might make?

There is a lot of international evidence that masks are highly effective and can reduce risk of transmission by up to 85 per cent.

It's difficult to translate this into the overall effect of mask use on transmission rates because it depends when and where people are wearing masks, and the types of contact this affects.

The more people have masks and the more places they wear them, the more effective they will be, and the less likely we will be to need costly interventions like lockdowns.

This time around, have you learned any new lessons that might improve modelling going forward? Specifically, has it told us about any differences between level 3 and level 4 as effective controls?

A big difference this time around is that our contact tracing system is functioning at a higher level.

As a result, alert level 3 this time has been almost as effective as alert level 4 was last time in reducing spread.

This has been a big help in getting the Auckland cluster under control.

We also have much better data this time around, including information about who caught the virus from whom.

This is useful for modelling as it helps us build a picture of the epidemic tree and make sure it is consistent with the model.