Experts: Latest anti-lockdown paper falls short
Analysis: Public health experts say a new paper from Tailrisk Economics that casts doubt on Covid-19 modelling efforts isn't all it's cracked up to be, Marc Daalder reports
A new paper that has been hailed by anti-lockdown advocates in New Zealand as a thorough rebuke of existing Covid-19 modelling falls short of proving its hypothesis, public health experts have told Newsroom.
The analysis, authored by Tailrisk Economics' principal Ian Harrison - a former economist at the World Bank, IMF and RBNZ - claims that death tolls projected by models from the University of Otago Covid-19 Research Group (OCRG) and the University of Auckland's Te Pūnaha Matatini are unrealistic.
Instead of killing tens of thousands of New Zealanders in a worst case scenario, the virus would have only killed 157 people in a scenario where preventative measures were taken but the country didn't shut down, Harrison predicts.
However, public health experts and modellers affiliated with the OCRG and Te Pūnaha Matatini all dispute Harrison's findings, saying he doesn't understand New Zealand's health response or how the models work.
What Harrison says
Harrison's paper is nominally dedicated to determining the cost-benefit ratio for the shutdown of the construction industry. Using a new model that he has developed, Harrison predicts that opening up the construction sector while keeping the rest of the country under Alert Level 4 would double the number of Covid-19 cases in the country but result in just one additional death.
Harrison quantifies the cost of this decision, saying it would cost the country about $7.6 million, while the construction sector shutdown has a price tag of $3 billion.
"Of course, the benefit-cost ratio of .003 is from just model run. Different, and plausible, assumptions can readily generate benefits that are a order of magnitude, say, ten or twenty times, higher than the $7.6 million. But it is very difficult to see how they could be over 300 times higher," Harrison writes.
Nick Wilson, an associate professor in the University of Otago's Department of Public Health and a member of the OCRG modelling team, said that Harrison's cost-benefit analysis assumes the construction industry could operate in a business-as-usual scenario while the rest of the country was locked down.
"I think he doesn't actually understand his counterfactual. You make a comparison and there's no such thing as business-as-usual, in anywhere. I mean, Sweden [which has refused to lock down] has got a GDP hit," he said.
"This is because regardless of what New Zealand does, its economic wellbeing is reduced regardless from the declines in international travel and the global economy and because some workers may choose themselves to stay more at home to avoid infection."
Harrison also took issue with the fact that the projected death tolls didn't distinguish between high-risk groups - elderly people and those with preexisting conditions - and the general population. He proposed reporting "life years saved" instead of deaths, which places different emphasis on the death of an older person than that of a younger person.
The question of case isolation
The central thesis of Harrison's paper is that the modelling overestimated the death toll because it failed to take into account the tracing and isolation of cases by the Ministry of Health. If New Zealand had applied robust and rapid case isolation instead of going into lockdown, the impact of the virus could be minimised while the country continued to operate, Harrison hypothesises.
"We found that OSRG’s [sic] model run results grossly overstated the number of deaths, because they made an assumption about the critical tool in the Ministry’s arsenal. It was assumed that there would be no tracing and isolation of cases, or alternatively that the Ministry was so ineffective that their efforts could be disregarded," he writes.
"The same model, configured with effective tracing and isolation, and some other plausible assumptions, generated about 160 deaths."
This misses one important fact, University of Otago epidemiologist and OCRG researcher Michael Baker told Newsroom: New Zealand wasn't able to rapidly isolate cases at the time it went into lockdown.
The entire basis for going into lockdown, which Baker pushed for the better part of a week before it was announced, was that New Zealand wasn't yet in a place where it could test and contact trace to the degree it needed to.
"We weren't prepared. We learnt nothing from SARS. None of the Ministry's documents even mentioned SARS. We were so far behind. In a way, if the Government wanted to get a grip on things, it had no choice other than a lockdown, in my view," Wilson told Newsroom.
On March 23, the day that Jacinda Ardern said New Zealand would go into lockdown, just 1050 tests were completed across the country. The public health units in charge of contact tracing could trace just 70 contacts a day. But on March 26, just three days later, there were 78 new cases in the country, each of which would have numerous contacts to trace.
In other countries caught unawares by Covid-19, efforts to test and contact trace have fallen by the wayside and the focus is now on hospital care and triage.
"We've seen countries where their ability to case isolate has been overwhelmed, just because numbers grow," Te Pūnaha Matatini director Shaun Hendy told Newsroom.
"So, you shouldn't be assuming that your ability to do case isolation is independent of your lockdown. In fact, those two things should go hand-in-hand. If you've got low case numbers, you'd better do fast case isolation. If you've got very large case numbers, then your case isolation will be insignificant," he said.
"It doesn't matter how good you are, you'll never really be able to do it. That's why we're seeing rebounds in some countries. Countries like the UK and the US have effectively given up trying to do that because they just can't do it anymore."
This was the reason models didn't factor case isolation into account - New Zealand couldn't do it to the degree needed and would soon be overwhelmed without some sort of intervention.
Harrison uses the public Covidsim model that the OCRG relied on to come to his own conclusion of just 157 deaths over a year, if New Zealand's case isolation efforts had reduced the spread of the virus by 60 percent. But at the time of the lockdown, the public health units could only trace a small fraction - far less than 60 percent - of each day's new cases' contacts.
Wilson said events overseas had vindicated the OCRG modelling.
"The fact that our modelling analyses - that typically showed thousands of deaths - were quite plausible has been borne out by some overseas data. For example, if New Zealand had around the death rate from Covid-19 that New York City has had on a per capita basis - 0.17 percent of the population dying - then we would have had more than 8000 deaths in New Zealand, and the epidemic there is far from over. While New York City is obviously more crowded – they possibly have higher ICU capacity, since New Zealand levels were typically fairly low compared to other countries."
Misunderstanding the model
Wilson and Hendy both said Harrison misunderstood fundamental aspects of the model.
For starters, Hendy said, the Covidsim model is best used to simulate long-term impacts, not what would happen in the first few weeks of an outbreak.
"They don't have a good understanding of the model that we've used. And I think they've misunderstood how you can use Covidsim. Covidsim's not really appropriate for analysing the current situation, which they've tried to do. It's really for doing long-term scenarios," Hendy said.
Wilson agreed, saying that while Covidsim is accurate over the long run, it can't account for freak events like the "super-spreader" at a Bluff wedding who led to the infection of 96 people, or the World Hereford Conference in Queenstown, which exported infections from a single case all over the country.
"It is not very accurate at the start, because it's not capturing the random variations of that. It plugs along with whatever you put in as the reproduction number," Wilson said.
"Say that number is two - every person infects two people. Covidsim just starts off doing that, whereas in the real world you could have a super-spreader on day one who infects everyone at a wedding, just like in Bluff."
Wilson also said Harrison had misused the case isolation input in his Covidsim projections. For Covidsim, case isolation indicates the quarantining of all Covid-19 patients - even if they have mild or no symptoms - away from their homes or families.
This is what has been done in China, Singapore and Taiwan, Wilson said. While some epidemiologists have said New Zealand could consider adopting this model, the country currently instructs non-severe cases to remain at home and isolate themselves from others in their household. This still leaves open the possibility of accidental infection within a bubble or deliberate non-compliance.
Because Harrison has misused the case isolation input, Wilson said, the model has also churned out a different result than had he used home isolation.
A trend of 'contrarianism'
Harrison's paper represents the latest entry in a series of attempts by academics and politicians to criticise New Zealand's strict lockdown.
Both National and New Zealand First politicians have signalled they want to see an end to Alert Level 4 as soon as possible.
National MP Nick Smith told his followers on Facebook that the country would be moving to Alert Level 3 come Thursday, even though the decision had not yet been made. A National Party spokesperson told TVNZ that Smith's "comments are consistent with the public's expectation".
Deputy Prime Minister Winston Peters told Newstalk ZB on Sunday that "it would be of no value if we save a lot of Covid people in our approach - and yet during it we've got all sorts of things like depression, suicide, breakdowns and a whole lot of other social things having far greater social damage."
"In the end the only thing that's going to save lives, isn't just medical expertise - but an economy that's capable of providing the utilities to face the crisis," he said.
Newsroom reported last week on an effort by six academics to present a "Plan B" to lockdown. Baker called the leader of the effort, University of Auckland epidemiologist and diet health expert Simon Thornley, a "contrarian".
Thornley, who was interviewed by almost every major media company in the country on the day his proposal was announced, receiving a front page treatment in the Dominion Post, said on Sunday evening he felt censored.
"I think there has been some censorship of alternate views throughout this period," he said.
Thornley made the comments during a primetime appearance on Three's nationally-broadcast Newshub Live at 6pm.
Help us create a sustainable future for independent local journalism
As New Zealand moves from crisis to recovery mode the need to support local industry has been brought into sharp relief.
As our journalists work to ask the hard questions about our recovery, we also look to you, our readers for support. Reader donations are critical to what we do. If you can help us, please click the button to ensure we can continue to provide quality independent journalism you can trust.