Navigating a public spat between scientists
Was an epidemiologist excoriated for going off script or not being on top of the literature? David Williams reports
It wasn’t exactly flattening the curve. A scientific spat over Covid-19 reached peak contrarian yesterday thanks to the country’s contrarian-in-chief, Mike Hosking.
Following a Stuff opinion piece that said “We don’t want to squash a flea with a sledgehammer and bring the house down”, University of Auckland senior lecturer and epidemiologist Simon Thornley explained his views on Newstalk ZB.
He told broadcaster Hosking overall mortality figures in heavily affected countries haven’t gone up, concluding: “These deaths are occurring in people who are running out of time.”
“Exactly,” one-man-reckon-machine Hosking responded. “They were going to die anyway and something was going to get them. It just happens, now, to have been this. Or maybe it wasn’t. Or maybe this exacerbated it. Or maybe this complicated it.”
(Or maybe the fact these people were going to die “anyway”, of something, sometime, is a statistical irrelevance in a society that cares about preserving life and protecting the public from preventable causes of death.)
Thornley’s column angered arguably our country’s most prominent scientist, Siouxsie Wiles, who tweeted on Tuesday: “For anyone who comes across the opinion piece of an epidemiologist suggesting lockdown is like using a sledgehammer to hit a flea: he studies diet not infectious diseases. Don’t listen to his reckons.”
She later apologised for making it personal, albeit without naming Thornley. Not before Auckland University of Technology Professor of Public Health Grant Schofield jumped in to back Thornley. Schofield, too, made it personal.
Accusing someone of being out of their scientific “lane” without discussing data wasn’t acceptable, he said on Twitter, adding: “Some would criticise you a microbiologist in public health.”
(A microbiologist studies microbes – living organisms invisible to the naked eye, such as bacteria and fungi. In medical work, that includes identifying disease-causing bacteria. Epidemiologists study the patterns of disease outbreaks to find the causes and effects – information that can be used to prevent future outbreaks.)
Wiles tells Newsroom it’s fair to question her credentials. “But I am doing my best to stay on top of the literature, which it would appear others are not. And have also changed my position as the evidence has changed and explained why.”
She says her frustration was sparked by “a piece that used old data disingenuously to strongly push a message that has the potential to lead to people’s deaths by undermining the lockdown”.
Thornley, meanwhile, says science, at its heart, is about open and honest debate. “That is what I intended to bring to this discussion, which, I believe, has been very one-sided.”
Schofield maintains he’s all for the lockdown but he’s also for robust, and civil, scientific debate. He was disappointed that Wiles used her authority not for scientific argument but to dismiss science she didn’t agree with.
“There is considerable uncertainty,” Schofield says. “[Thornley] is the single smartest guy I know, and he does have some challenging and possibly inconvenient truths about the uncertainty.”
In a very sciencey virtual punch-up, Wiles and Schofield tweeted each other links to papers from medical journal The Lancet.
Weighing the evidence
How to navigate this as a non-scientist?
At a fundamental level, the Government’s response to Covid-19 must be challenged. The country is under unheard-of restrictions for people’s movements which will lead to an enormous economic crash – albeit to stop people getting sick and save lives. With so much at stake, all decisions, including those based on scientific modelling, should be heavily scrutinised.
Thornley says it’s an epidemiologist’s role to voice an opinion. He’s worked on a number of infectious disease epidemics, including Auckland’s recent measles outbreak, and has published work related to infectious diseases. “If I am not qualified to speak on this matter, I wonder who is?”
It seems prudent for those in public health to weigh up the benefits and harms of the response. As Schofield puts it, “we should do the most good and the least harm with the resources we have”. But like debates about smoking and climate change, the idea of “balance” has to be carefully weighted by evidence.
The Prime Minister’s chief science advisor Juliet Gerrard says it’s been striking how much agreement there’s been on all major issues in this country, and huge consensus of support behind the Government’s early action.
“So I'd advise the public to look at a range of views, and give weight to those commentators that have looked hard at the global situation and done deep analysis across many countries, and really interrogated all the data they cite.”
Best dive into the data, then.
In his Stuff piece, Thornley points to the Diamond Princess cruise ship as one of the few examples of a “closed population” who were all tested for the disease.
The so-called case fatality rate was 1 percent – seven deaths from 706 positive patients. Thornley links to a March 17 story on StatNews by Stanford University professor John Ioannidis, who says the pandemic may be a “once-in-a-century evidence fiasco”.
Ioannidis says the Diamond Princess mortality rate was largely elderly, and when projected onto the entire US population the rate would be 0.125 percent. But he takes the mid-range guess of his analysis, 0.3 percent for the general population.
If 1 percent, or 3.3 million, of the US population gets infected, that would result in about 10,000 deaths, Ioannidis says. That number is “buried within the noise” of deaths estimated from influenza-like illnesses, he argues. (As of last night, US deaths from Covid-19 stood at just over 4000, and even President Donald Trump admits 100,000 deaths would be a “very good job”.)
Wiles says using the case fatality rate from the Diamond Princess is unrealistically low. She prefers a “more realistic” 0.88 case fatality rate arrived at in a preprint article posted on the website medRxiv on March 13, based on roughly 70,000 cases and 3000 deaths.
Its estimate of China’s fatality rate was 1.38 percent – lower than estimates for other coronaviruses, like SARS and MERS, but well above H1N1, known as swine flu. The “age-stratified” infection fatality ratio for international cases (Wuhan residents returning from overseas) was 0.66 percent, the paper says. It notes those with underlying conditions, something likely to be correlated with age, are more likely to get severely ill. The fatality rate rises from about 1 percent in the 50-59 age group to 13 percent in those 80 and above.
University of Auckland's Thornley says another three passengers on the Diamond Princess have subsequently died, which means the fatality rate has lifted to 1.4 percent (10 from 712). Still, the overall point is it’s a very elderly group, he says, and when translated to a Western population the death rate will be higher than influenza, “but not by much”.
“I don't think the essence of the argument changes here. It gives the fatality rate a sense of scale.”
As mentioned, those with pre-existing conditions and risk factors – like cardiovascular disease, diabetes, and hypertension – are at higher risk of getting a severe case of coronavirus.
For this reason, many countries, including New Zealand, have put tighter clamps on the movements of those aged over 70. (Having simultaneous medical conditions is called comorbidity.)
Thornley’s piece says while Italy’s death rates and “crude case-fatality rate” of 9 percent are alarming, a “recent analysis” (published in Italian) shows only a “small fraction” were entirely due to Covid-19. “Many deaths were hastily labelled as Covid-19 related when they were not.”
Wiles says that report, in which only three deaths of 355 with no serious “comorbidities”, is two weeks old. “Italy now has over 11,591 deaths – this is some serious cherry picking.”
Fundamentally, she says: “Regardless of comorbidities these are people that would not have died.”
An early-release article from the US Centers for Disease Control and Prevention’s Covid-19 response team was published on Tuesday. It made preliminary estimates of the prevalence of selected underlying health conditions among coronavirus patients in the US between February 12 and March 28.
The report, based on 7162 reported cases with relevant data, said 38 percent (2692) had one or more underlying health conditions and other known risk factors, while 62 percent (4470) had none of those reported conditions.
Of those Covid-19 patients admitted to hospital, 71 percent (358 of 457) had at least one pre-existing condition or risk factor. That percentage lifted to 78 percent (358 of 457) for cases requiring intensive care. But the authors say those data are biased towards more severe cases.
Thornley says he wasn’t claiming it was an up-to-date number of the total number of cases. “It is the only window we have in to the sorts of people who are dying in that region, rather than overall number. This is important, as people in their 80s have a very high mortality rate, with or without COVID-19.”
Sweden as a test case
Thornley tells Newsroom this country’s response to Covid-19 is a matter of judgment, “as the Australian and Swedish examples shows”.
After making his “squash a flea with a sledgehammer” comment, he wrote in his Stuff story that he believed other countries, such as Sweden, are steering a more “sensible course”. He linked to a Guardian article, which mentions schools, kindergartens, bars, restaurants, ski resorts, sports clubs, and hairdressers remain open, unlike in neighbouring Denmark and Norway.
The Swedish Prime Minister, Stefan Löfven, has said coping with Covid-19 is about commonsense behaviour. “We all, as individuals, have to take responsibility. We can’t legislate and ban everything,” That country’s Public Health Agency’s position has been criticised in a joint letter from 2000 Swedish university researchers.
On Monday, in another Guardian article, Professor Cecilia Söderberg-Nauclér, a virus immunology researcher at Sweden’s Karolinska Institute, said: “We’re not testing enough, we’re not tracking, we’re not isolating enough – we have let the virus loose.” She concluded: “They are leading us to catastrophe.”
Comparing Sweden, Denmark, and Norway on coronavirus counter worldometers.com shows a concerning trend. Norway (4651) has more confirmed cases than Sweden (4435), but fewer deaths – 39 versus 180. Seventy of those Swedish deaths were reported on March 30 and 31. Denmark, meanwhile, has 90 deaths but far fewer cases than Sweden, at 2860.
Per capita, the figures are far different. Norway has 90.3 confirmed cases per 100,000 people, Denmark 53.6 cases per 100,000 people, and Sweden 49 cases per 100,000.
On Twitter, Wiles says: “As for holding Sweden up as an example of a country to follow, let’s revisit that in a week or so shall we?”
“I certainly wouldn't be advocating for the Swedish policy settings.” – Juliet Gerrard
Gerrard, the Prime Minister’s chief science advisor, says Thornley has raised Sweden as a different path and Wiles challenged his assumptions. “I think that is healthy and how science work.”
She adds: “Having independently looked at the Swedish data in comparison to their neighbouring countries (in particular the number of deaths), I certainly wouldn't be advocating for the Swedish policy settings.”
Adopting Wiles’ language, and the “harder, earlier” mantra of the Government, Gerrard adds: “In a couple of weeks we will see how well they worked in comparison to countries who went harder, earlier, and can all learn from that.”
There’s huge uncertainty about Covid-19, Gerrard says. Numbers are often a proxy for what’s really happening and countries with a lot of contact tracing and testing are a better proxy than others.
People get “distracted” by the mortality rate as a percentage of cases, she says. The more cases that are missed by contact tracing, the lower the real mortality rate. It also means “R”, the average number of cases that each person infects, must be higher.
“But the much more important numbers are harder to argue with – the actual number of deaths, and the number of ICU admissions that precede them.”
Restoring the balance
Thornley says he received some hate mail after his Stuff piece, but about 80 percent of emails “were from ordinary New Zealanders who agreed with my view”. “I also had several academics and doctors congratulate me for restoring some balance to the public conversation on the topic.”
Schofield, the AUT professor, argues anyone going "off script” is perceived as letting people die, being callous, and irresponsible. “If we go that way, then the contrary opinions which might end up mattering a great deal will never get discussed.”
Gerrard confirms the vast majority of scientists and public health experts support the Government’s action – action based on a scenario warning that this country’s health system, like heavily affected countries, wouldn’t cope with a sudden surge in cases.
Experience from overseas shows Covid-19 results in sudden peaks in intensive care admissions – “and therefore unnecessary deaths”, Gerrard says – as the health system overloads. “What people care about is these real numbers of real people whose lives were lost early.”
In the end, Wiles says, it boils down to what people value. “Far too many people seem to value better outcomes for themselves but horrendous outcomes for others.”
* This story has been updated to include extra comments from Simon Thornley and the per capita comparisons of coronavirus cases between Sweden, Denmark, and Norway
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.