Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Death Is a Trailing Indicator
Every time I see one of these apocalyptic exponential projections based on a “doubling of the death rate every two days,” or whatever the current numbers suggest, I want to slap someone. At the moment, and for the next week or so at least, the death count is a trailing indicator of contagion.
It appears to take, on average, from about ten days to two weeks between infection with the Wuhan virus and subsequent death. That suggests that today’s death figures are a proxy for the rate of infection ten days to two weeks ago.
Ten days ago the schools were open, businesses were open, theaters were open, bars were open, colleges were open, and America was going about its business with little concern of infection. Unless one believes that our subsequent efforts have had little effect, one must assume that the rate of new infection has dropped off substantially. It’s hard to imagine a scenario in which it hasn’t, outside of one or two super-dense metropolitan areas.
Today’s death figures are an echo of conditions that no longer exist. This will be true for another few days, after which time the death figures will begin to reflect actions begun ten days ago.
I am pretty sure that the nation has never shifted its approach to interpersonal contact as profoundly and abruptly as it did last week. Today’s death figures can barely reflect that. Look at the numbers toward the end of this week. I’m expecting a good Friday report.
Published in General
There’s ample evidence that this is not a black swan, but like almost everything with this outbreak, it’s still too early to know anything for certain.
“Pandemic” conjures up notions of immediate homogeneous spread around the planet, but the way the virus is spreading is much more like a bomber dropping napalm bombs: discrete fires starting one after the other, some of which flare up strongly and others which don’t (or didn’t before the global lockdown was enforced). So if you take a snapshot at an early point in the spread, of course the first two or three hotspots will look like outliers – it’s because the bombs haven’t landed in other regions yet. And remember, when dealing with exponential growth, a few days’ difference in time can result in a huge difference in the statistics.
To put it another way, this is like Parisians interpreting the takeover of Sedan at the start of WWII by saying “Sedan must be an outlier”. On the day after the invasion of Sedan, it was indeed an outlier and the rest of France was still free – but not for long.
Also supporting the “non-black swan” theory is the fact that the trajectory in Spain seems very similar, only delayed by a few days. This supports the notion that the differences are primarily due to the timing of community transmission.
All that being said, I’ll repeat: there’s not enough data to know anything for certain. Perhaps there’s some good demographic (e.g. age)/lifestyle (e.g. smoking)/climate/infrastructure explanation as to why Lombardy and Madrid and Wuhan are different from the rest of the world. I certainly think there’s hope that the dynamics would be different in much of the US. We just don’t have any decent evidence for that yet, so the question is: how much of a risk should we take based on conjecture, gut feeling and hope? That’s not a snarky question, that’s genuinely the dilemma and I don’t have an answer.
The paper I linked to a few comments back shows the breakdown by week of flu cases. Based on that distribution, even the most intense week of flu season is still considerably lower than the current burden.
As long as I’m carpet bombing this thread, let me add my usual caveat:
I am not making any predictions or projections as to what I think would happen in the US if we immediately loosened the closures and restrictions. There are scenarios in which the problem goes away largely on its own within the next few weeks which I find highly plausible, and there are also scenarios in which the virus causes more devastation and economic harm than three weeks of global economic shutdown which I also find highly plausible.
All I’m trying to do here is point out that we seem to be getting quite strong evidence that this virus is a unique threat, and that simply trying to calculate case fatality rates and the like mischaracterize the nature of the threat. I’m not making any claims to the overall magnitude of the threat, because nobody knows that at the moment.
I’ve gone to very great lengths to explain my arguments, point out the sources of the evidence, and acknowledge the weaknesses in each argument.
I’ve also gone to great lengths to point out that I am not making any long-term predictions about this pandemic, and that I find it plausible that it may have a fairly simple resolution.
I don’t know how anybody could call that gaslighting, unless gaslighting in this case simply means “disagreeing with James Gawron”.
In any case, given the lack of adequate evidence at the moment, I think what we absolutely need the most is humility and civility with each other. In a situation in which nobody can predict the future with any confidence and with potentially disastrous outcomes in both directions, needlessly yelling at each other only makes an already-difficult situation worse. Therefore, I kindly ask you and everyone here (on all sides of this debate!) to treat each other with even more grace than usual.
Regards,
Mendel
Henry, is it ten days to two weeks between infection and when the virus manifests (fever, cough, etc.) rather than death (if you die)?
If so (that’s my understanding) then there’s a (n up to) two week lag between social interventions like social distancing and lockdowns and when any changes to rates of new infections due to these interventions can be measured and assessed.
I understand that the actual rate of infection is probably much higher than the numbers we’re getting (due to asymptomatic infections, we’re only testing and counting the symptomatic infections), but if you assume (just an assumption) that the ratio of symptiomatic and asymptiomatic is constant(ish, and that’s also an issue because if you start isolating aged care facilities then that’s going to skew the numbers) then it’s a good proxy for whether the infection rate is going up or down.
DocJay! The dude who didn’t come back….but he did?
Thanks Mendel. This was very helpful and informative. I humbly withdraw my “Probably a black swan…” comment.
And Lombardy is located where? And they have how many Chinese working there? And what is the average age of the population? And what’s their health care systen like?
I think the comparison is apples and oranges . . .
I think it more accurate to say the comparison is between apples and a Schröedinger box containing an apple/orange probabilistic wave function. That wave function has two extremes, best represented here on Ricochet by @iwe and @unsk. (I’d guess you, @stad, are a bit closer to iWe, and Mendel is a bit closer to unsk…)
The box will be opened in a week or two, I think.
Zafar, good to hear from you, and I hope you and yours are well.
I don’t know how accurate our understanding of the Wuhan progression is, but what I’ve read suggests that the average patient whose case is serious enough to send him to the hospital presents himself there about a week into his infection, and that he develops serious pulmonary failure within a day or so of that. I’m not clear just how soon after death occurs, typically, for those unfortunate enough to go that route, so I figure a range of ten days to two weeks from initial infection probably cover most of them. (But that’s my guess.)
As for new cases being a proxy for actual infections….
At the moment, I think there is no good proxy, at least in the United States. I think that will change very quickly, within the next week or less. Right now, our testing frequency continues to climb, and it’s reasonable to assume that our methodology will shift with that, subjecting new populations to testing that might previously have been missed. That should settle down as the process becomes more regularized.
Soon, the death count should become a reasonable proxy for infection post-lockdown, since it will begin to reflect cases originating after the population started taking this very seriously. I don’t think we’re there yet, but, as I said, I’d expect it late this week or sometime next week.
A complication is that treatment techniques are also changing and, to the extent that those reduce the fatality rate (and of course we hope they do) they also make deaths a weaker proxy for new infections. We should probably assume that some improvements in treatment are occurring now, and so be careful of interpreting a small slowing of the death rate as significant.
That’s an optimistic take on the situation and quite reasonable. I hope you’re right.
Italy put social distancing into effect very late. They have long run out of ICU beds and are now refusing to put anyone over 60 on a ventilator. As a result the case fatality rate there is over 7%, and the number of infected people is still climbing.
And they are running out of ICU beds in New York City.
Ready Kilowatt’s nose should be under the mask.
Thank you very much for that line, it gave me a well-needed chuckle.
In fact, I think I would say I’m on Team Schrödinger on this one, even if it’s a team of just one. (I’ve disagreed with Unsk on many items related to coronavirus)
My point has never been that I think this virus is going to ruin society if left unchecked. I think there’s a very likely chance it reveals itself (without our help) to be much less harmful than even some of the less-dire projections are predicting.
Rather, my point is that the existing data and the underlying science are consistent with multiple hypotheses simultaneously – hence Team Schrödinger.
And one of those hypotheses is a state of financial and social chaos that is worse than 3 weeks of completely shutting down the economy. Basically, my rattling on and on is not because I believe such a ruinous state of affairs would come to pass if we rapidly drop the drastic measures; but rather, that the data does not let us rule it out with the degree of confidence we should require when it comes to a threat of that magnitude.
Regarding New York: if, as I increasingly suspect, this disease is one thing in high-density urban areas and another, much less frightening, thing in non-urban areas, then New York will be a terrible but nearly unique outlier in America. We do not, as a nation, have the ability to offer extraordinary relief to every single ICU in the nation. We do, however, have the ability to offer extraordinary relief to every ICU in New York City. It will be expensive, and there will be several days required to ramp that process up. I’m sure it’s already underway.
We aren’t Italy.
I highly doubt this is the case.
We now have lots of consistent data that the real case fatality rate is well below 5% (likely below 1%).
The reason the case fatality rate is so high in Italy is almost certainly because they simply can’t test anywhere near enough people to get reliable figures.
Of course, the fact that people are dying due to the lack of equipment/staff will technically raise the case fatality rate, but nothing of that magnitude.
It’s 7.2% according to this article from JAMA. The authors scratch their heads over the fact that people over 70 are dying in Northern Italy at such a high rate when compared to elsewhere. Then we find from an Israeli doctor working in Italy that they are not treated old people with ventilators. It’s pretty obvious that this is the reason.
New York City is an interesting test case. It has a high concentration of cases, the rate of testing is the highest in the United States — and we have a case fatality rate of 0.8% as of this morning.
Now that a new treatment is being tried, there’s a possibility the case fatality rate will drop because of it and not because of a change in the number of patients presenting at the hospital. That would be good news, though it would reduce the value of deaths as a proxy for disease spread — and we’d be back to having no truly independent numbers we can look at to know where things stand.