Hacker News Reader

Nobody knows the asymptomatic infection rate. Without that, differences in mortality numbers between countries are meaningless, because testing rates are all over the place. Explaining the differences by overrun hospitals is jumping the gun I think.

Exactly. Fatality rates are mostly just a function of a country's testing (how much testing they're doing, and how biased the sample is).

If you're only testing a small # of people with severe symptoms, obviously the fatality rate will look really high. If you're testing a much wider section of the population, and testing people who aren't severely sick, the number will look lower.

Some examples: - Iceland randomly tested people in the general population (no symptoms), they found ~1% of people tested positive, and half of those had no symptoms (and so probably wouldn't be tested). [1] - Italy tested an entire village (the village had a cluster of cases), 3% tested positive, of which half had no symptoms. - One professor used the Diamond Princess cruise ship as a baseline, and estimated anywhere from 0.025%-0.625% [2] - China had a fatality rate of 17% initially, which went down to 0.7% if you only consider symptom onset _after_ Feb 1st. [3]

Of course, there are a lot of other factors, and it's possible that the rates end up in fact being really high, but there's an argument that we just don't know how many people are infected and fatality ends up being a lot lower then just dividing # of deaths by # of confirmed cases would have you believe.

Just to be clear: even the lower case scenarios are really serious and worrying, and I'm definitely not in the "this is just another flu" camp. But there's a huge range of uncertainty here.

[1] https://www.government.is/news/article/?newsid=f96a270c-66e8... [2] https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-a... [3] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

I can't stress enough how important this factor is... without widespread randomized testing of the population, we have no idea what the true infection rate is or how widespread the virus really is. Our current stats have a huge selection bias problem.

This is why it is so important that we take social distancing, self hygiene, and shelter-in-place measures really seriously in order to protect those among us who may be more susceptible to severe infection.

The silver lining: the more widespread the virus, and hence the number of asymptomatic cases... lower the actual the severity statistics are.

i.e. we know "number of deaths", "number of hospitalizations", and "number of positive cases". Right now, when we calculate our various severity rates, we are putting "number of positive cases" in the denominator because we don't know the number of actual infections. There is an unknown multiplier (>= 1) relating positive cases to infections. All our severity statistics need to be divided by this unknown multiplier. But right now, since we don't know what it is... the responsible thing to do for reporting and policy decisions is assume it is 1 (worst case). It is not 1... it is greater than 1... we all know that. But how much is a tough question to answer properly without proper statistical sampling of the population.

There are modelling studies and estimates that suggest value ranges for the multiplier, but I don't want to spread potentially false information on a public forum.

It's not a silver lining: the number of serious cases remains the same and our ability to contain the epidemic is less effective.

It is, because assuming that having the virus gives some immunity, then if we find out that 90% of people are asymptomatic, that implies that the number of cases we have right now won't go up much more (since many people have already had and recovered from Coronavirus).

Nobody knows what is coming, but I'm in the 1M+ camp as well. US is so ill-equipped plus administration is very aware, how devastating the economic consequences are even now (2T stimulus and infinite FED liquidity are already in effect today). Let this spread for two more weeks and the administration might loose their countenance and will declare: "we cannot save both the infected and our country, so we will save our country".

Agreed, I so wish one of the networks or YouTube would give someone like Fauci an hour or two to walk through their methodology in detail instead of just answering topical questions on projections from pundits. We’re not getting a whole lot of context right now that would be useful and engaging.

There's a 15 min interview on YouTube with Fauci being interviewed by Trevor Noah today. Strongly recommended: https://youtu.be/8A3jiM2FNR8

frustratingly not available in my region

That's what our media has to offer? An interview by a comedian? I agree we need Fauci to walk us through it. I'd like a panel interview conducted by a reputable interviewer.

Your worst case is correct but % population infected drops dramatically with lock down. The countries with better infrastructure (richer, Western) will likely see ~1-2% of their population infected and a death rate under 1%

The question to me, aside from assumptions about infection rate, is: are the people that are getting hospitalized and/or dying today just unlucky, or are they more susceptible to the virus or its effects?

If they are just unlucky, the death rate may hold. If, however, its just the highest risk population having the worst outcomes, the death rate will plummet as they die off. In order for the US to hit 1M+ deaths, you have to assume that people getting infected in 3 or 6 or 12 months from now are just as likely to die as those dying today.

Can you share your math for the various situations for those of us not as clear on the differing rates in those countries? Has anyone built a calculator that shows how it could play out in different countries based on the experience we have seen in countries so far?

No. It all depends on mass testing.

If mass testing is adopted, it would surface the asymptomatic carriers.

In addition, it would allow the creation of "clean" zones (.e.g on flights/buses) where you would need to prove that you are clean.

So, once the virus cannot "hide", and R0 is bought down to below 1, you would actually see the rate of infection crash to zero (I.e. the exponential rate works both ways).

To sum up, as long as there is no vaccine, it all depend on mass testing.

A best case scenario might be to look at the number of confirmed CV-19 deaths in China as a percent of total China population. This of course assumes competent public health policy, but we’re assuming what the best case could be.

There's no reason why 40-70% of people will inevitably get infected. California daily new cases already plateued

I think that number might be coming from the idea that at that level of infection you'd get herd immunity - i.e. this thing is going to keep going around until we hit herd immunity.

Reaching herd immunity level is a possibility but not inevitable

Agree it's not inevitable, but seems plausibly to be the highest probability path at this point.

With the current tools we have, it seems inevitable to me. Look at china or hong kong... lockdown led to getting local infections under control, which leads to an easing of restrictions, which lead to new infections from outside of the region. So by that logic, life doesn't get to go back to "normal" until either herd immunity is reached, or new tools come online.

That assumes that this virus doesn't mutate quickly, mitigating the effectiveness of herd immunity and vaccination.

Totally true. Obviously nobody knows right now if that will happen or not, but that question was posed on a recent episode of This Week in Virology (great podcast, btw) and they seemed to be saying that the mutated seasonal variant is a fairly unique characteristic of the flu virus, and it would not be _likely_ to happen with COVID-19.

True, but that seem to be a reasonable assumption according to at least some virologists looking at it.

There are so many unknowns right now that any modeling is going to have issues, best you can do is state your assumptions clearly, and explore the ones that seem most likely in some depth. Otherwise you just end up with an ensemble of models that could result if nearly anything.

California cases are plateaued due to a lack of testing capacity, I would be hesitant to suggest that the growth has stopped.

The thinking behind that is that short of having a vaccine (18 mo off at best, probably) we are simply not capable of containing it, just slowing it down. Once you you somewhere in that 40-70% range, it self contains due to lack of available hosts.

Talking to people who study this stuff for a living, it's not an unreasonable assumption at this point.

Plateauing during stringent isolation (if that is the case) isn't' evidence against this, I think.

The big problem is that there still are not anything like good population estimates in the US, and unlikely to be since serious testing isn't done. Death and hospitalization statistics are better (but can get inaccurate if systems are overwhelmed) so the best the modelers can do is try and project from countries with better data ... inherently this is not as good as having good local data.

Reported new cases will be flat at the capacity they can perform tests. You need to look at death rate (change) to get better idea.