New Zealand's Covid-19 tracing system could be much more effective if officials were able to isolate "close contacts" within at least four days of the person who infected them being quarantined.
That's the upshot of just-released modelling, which also estimated that our small army of contact tracers squashed the virus' ability to spread by more than a third during the height of our Covid-19 crisis.
It's prompted new calls for more support for quarantining close contacts with limited sick leave, as is happening with cash payments in Australia.
The Ministry of Health considers a "close contact" as anyone with who may have been exposed to someone either suspected, probable or confirmed to have Covid-19 while they were infectious.
Normally, that included anyone who happened to be sitting within 2m of the infected person, or who had face-to-face contact with them, for 15 minutes or longer.
Our contact tracing system – today capable of tracing thousands of people in a day – remained one of New Zealand's biggest safeguards to crush a new outbreak.
But the researchers behind the new modelling, from centre of research excellence Te Punaha Matatini, say the ability to trace is just part of the picture, given that those close contacts also needed to be isolated as soon as possible.
The centre's director, Professor Shaun Hendy, said there were many questions about the service at the time New Zealand was plunged into lockdown in late April and May.
"Would it be good enough to catch the last few cases or any new cases that make it past quarantine at the border?" he said.
"Indeed, if a case does make it through the border into our community, then contact tracing will be the main way we try to contain the outbreak before it spreads out of control. If contact tracing fails, we will likely have to go into lockdown again.
"For this reason we decided it was important to incorporate a good model of contact tracing into our Covid-19 simulations."
The researchers set out to answer what metrics could best be used to tell if the system was going to be effective, but also to reveal the difference it had made in eliminating the virus here.
The model they used simulated the disease spreading at random in New Zealand, and assumed each case passed the virus on to a certain proportion of their close contacts.
They then looked at what happened if some of those contacts were successfully traced and quarantined.
"When someone is traced, it reduces the number of people they infect from that time onwards," explained co-author Professor Michael Plank, a mathematician at the University of Canterbury.
"We could set the speed at which tracing happened, as well as the likelihood of all contacts being traced. This allowed us to model a contact tracing system."
That meant they could compare this to how the real system in New Zealand performed, as well as asking how well a simulated system was at stopping the spread.
"We also modelled how effective quarantine and isolation was in stopping the spread," Plank said.
"There are reports from Melbourne about people who had been traced not staying at home. In New Zealand, public health workers follow up daily with people to make sure they are staying at home.
"We wanted to see how important this was, compared to the speed of tracing for instance."
They ultimately found that the ability of people to be able to go into quarantine - even when they had not yet had a positive test – could make a big difference to how effective the contact tracing system is.
"It is not much use tracing people if they can't afford to stay at home for two weeks," said co-author Associate Professor Alex James, also of the University of Canterbury.
"This suggests that we need to make sure people have access to things like paid leave when they are asked to quarantine."
They found New Zealand's system slashed what's called the effective reproduction number of the virus – or the average number of people that are directly infected by a single infectious person - by 30 to 45 per cent during the April and May outbreak.
Had no control measures been in place, the reproduction number could have been between two and four.
"Contact tracing played a big role, but it is important to remember that it would have been more effective when we were in level 3 and 4, because most of us had few contacts," James said.
The team also found the system could be improved by slashing the time taken to trace contacts.
While earlier advice to the ministry suggested criteria of isolating 80 per cent of contacts within four days of the infected person becoming symptomatic, the researchers suggested the target be within four days of the infected case actually being quarantined themselves.
And if that window could be narrowed even further to an average two days, then the virus' reproduction number could be reduced from 2.5 to 1.5.
However, James acknowledged that could prove difficult if New Zealand was faced with a large outbreak.
"Speed is of the essence when it comes to contact tracing because the virus spreads so quickly and people can be infectious several days before they get symptoms," she said.
"If you can find people within the four-day window, you cut out much of the period where they might be infectious.
"If you leave it to individuals to get a test after they develop symptoms, then they will often have infected others. And finally, supporting people so they can isolate or quarantine themselves is also very important."
While there was only five days' statutory sick leave in New Zealand, traced contacts might have to quarantine for up to two weeks.
"If you are asking people to take time away from work, obviously they need paid sick leave – and if you've only got five days, you are going to chew through that pretty quickly," said Plank, who added contacts in Victoria were given A$1500 payments to support them.
The paper comes as the Government has redoubled efforts to get more Kiwis using the NZ COVID Tracer app, amid fresh warnings another outbreak is inevitable.
As at today, the app had more than 646,400 registered users. There have also been 86,927 posters created, and 16,151 manual entries recorded in the app.
The researchers have also stressed that testing – rates have dropped to around 2000 people per day, from 7,500 in late June - would be vital in backing up the tracing system.