A few weeks ago, Stanford professor John Ioannidis published an article questioning the dramatic response to the current crisis. His argument was basically that we are engaging in an unprecedented response with very little evidence. You can read this article here.
Now, I know there are a lot of experts in adjacent fields sounding off (I'm looking at you Michael Leavitt), but when I read the original article by Ioannidis I paid attention. He's someone I actually greatly admired before all of this started. About 15 years back he published an academic article titled "Why Most Published Research Findings Are False." The article showed that many experiments in academic research can't be replicated. Subsequent reactions to the paper showed that entire lines of research were deeply flawed. It was an extraordinarily influential paper that I read years back and have really taken to heart.
About a week ago, Ioannidis gave a long interview following up on his coronavirus opinion piece. Now, before I share it, I should say that the main point of Ioannidis, as I understand him is that we need to justify our actions with evidence, not that our current actions might never be justified. I'm also not sure he accounts for the fact that even at a lower fatality rate our hospital system could still be overrun due to the sheer number of people becoming infected simultaneously. However, I think that his points are extremely important, and we're not getting nearly enough constructive debate around what's actually happening according to the data.
So I'm including the video below, for those of you who have a spare hour. For the majority of you who don't have time to watch, I took a few notes. Keep in mind the interview was done over a week ago, so the facts may have changed from what is below. If I've misrepresented anything, please let me know:
We are missing the key data needed to make decisions
The actual fatality rate is probably far lower than the 0.9% used in the Imperial College paper
Early estimates of fatality rates in pandemics almost always overshoot
That doesn't mean that is happening here but it could be
That doesn't mean another future pathogen couldn't exceed initial estimated fatality rates
We need to be more prepared to get the key information early on, for example:
We need random samples of the number infected in the general population
The cruise ship probably approximated a random population sample (in his opinion piece he estimates a fatality rate far lower than 0.9% based on that data)
Iceland tested 1/3000 of its citizens
In Italy there is evidence that many more could be infected than reported (indeed I read a paper by Italian epidemiologists that put estimated up to 45 times as many were infected, and the Imperial College team released a paper recently that had a similarly large range of potential underestimation in Italy)
Why is Italy so bad?
Italy is likely an outlier
Demographics - the average age of death is 81
Italian system was already close to full capacity when the disease hit
The system probably overadmitted initially in trying to do the best possible job with a new disease
One of the hardest hit areas (Bergamo) held a soccer match right before everything got shut down. People were dancing in the street, hugging, kissing, etc.
He states that the overrun problem was a small minority of hospitals (I have no way to verify this. It's what he said. I did see an Italian doctor in Northern Italy who conceded that his particular hospital was doing fine, although there was certainly severe strain in other areas)
On social distancing
Measures we're taking now should be taken but we need a decision in 2-3 weeks (now 1-2 weeks) on what to do next
We can't indefinitely shut people in their houses
There are a lot of health impacts associated with an economic depression, possible more than what we face with coronavirus
Some models suggest that to be effective a lockdown requires 18 months which clearly isn't feasible
Again, even in Italy, not all the hospitals were overwhelmed. There is a very good chance we'll be able to weather the storm
On coronaviruses:
There are 4 in common circulation. We don't talk about them because we really don't track them closely
Some of what are considered influenza deaths in America are likely from existing coronaviruses
He expressed concern about anecdotes and looking at every single death
We can't throw out all social distancing but need evidence based approaches
Should we let healthy go about their daily lives and shelter elderly? Maybe, but we don't have the data
Risk of a domino effect with state leaders as one tries to outdo the other
9 billion search results on Google for coronavirus, 100 million for influenza
It's too early to accuse anyone of doing too little or too much, but too early to know what's actually happening and there's no value in someone being wrong
Have to be careful to not lose public respect for science (personally my greatest fear is that the backlash to totally missing on these forecasts could cause extraordinary damage to future scientific endeavors)
When should we emerge from total lockdown?
He personally is abiding by shelter-in-place orders, but we need to decide our long-term plan in 2-3 weeks (now 1-2 weeks)
We don't currently have the data to know next steps.