Hacker News Reader

Comment by tunesmith | Hacker News Reader

I've seem people theorize about this before saying maybe it's good news because that means it's less fatal than we thought and therefore maybe we're closer to herd immunity. But I saw an epidemiologist talk about how an R0 of 2 meant you needed 50% of people to be immune for herd immunity, and an R0 of 3 meant you needed 66%. This would mean it'd need to be even higher - the article says 82%. We're nowhere close to that now - even northern Italy, they're estimating that maybe 10% have caught it.

View on HN

> ...R0 of 3 meant you needed 66%. This would mean it'd need to be even higher - the article says 82%

For anyone curious,

R0 = 3 means iff 2 in 3 people (66.66%) [0] are immune would R0 drop to 1 [1].

So, when R0 = 5.7 (the value proposed by tfa for SARS-CoV-2), 4.7 in 5.7 (82.45%) would need to be immune for it to drop to 1 [1].

Why should R0 drop to 1 [1]?

It means each case leads to only 1 successful transmission of the infection. This implies constant incidence over time. If a greater proportion are immune, then incidence will decline.

Why "herd immunity" isn't enough?

...problems arise because herd immunity is not the same as biologic (immunologic) immunity; individuals protected only by indirect herd effects remain fully susceptible to infection, should they ever be exposed. This has advantages, in protecting individuals with contraindications to vaccination or those who for other reasons miss vaccination, but it also has its disadvantages. Measles and mumps outbreaks among university students, and pertussis in adults, are among examples of the consequences of accumulation of susceptible individuals who have not been protected by vaccination, and escaped infection because of a herd immunity effect earlier in their lives... This means that there is a need for immunization programs to maintain high vaccine coverage, together with surveillance and outbreak response capabilities, as numbers of susceptible individuals accumulate in older age groups.

From: https://academic.oup.com/cid/article/52/7/911/299077

[0] This assumes a 100% effective vaccine (E = 1) in ((R0 − 1)/R0)/E. If, for a given vaccine, E < (R0 - 1)/R0, it'd be impossible to eliminate the infection through that vaccine. (R0 − 1)/R0 is known as the "herd immunity threshold."

[1] What I really mean is R0 x S = 1. See: https://en.wikipedia.org/wiki/Endemic_(epidemiology)

Nitpick: changes in the number of immune people change R, not R0

Yes, thanks. R or R-eff is more like it, but for the sake of simplicity I chose to instead leave a footnote about inaccurate usage of R0.

Your comment belies the inaccuracy of your user name :)

Would a greater percent of the population being immune lead to a greater chance for mutation since we're creating a higher barrier for survival of the virus?

Nope. Viruses can only evolve within a host. I.e. while they are infected. If you've go the right vaccine, there won't be enough time for a virus to get a strong enough foothold to make that evolutionary leap, since mutations are always a shoot of the dice.

Remember, there is no intelligence to evolution. Only happy coincidence. An immune population has the same probability of having a virus mutate in a particular infection, but has many, many fewer cranks on the mutation slot machine lever before the virus is eradicated.

Where you run into problems is with things like dengue fever, where multiple strains can all provoke a single immune reaction, but the other strains aren't crippled as much by the antibodies as the original, and can actually use the immune response to increase the severity of the infection.

I've read no papers that suggest that is a factor here.

Also, it seems that people with low severity symptoms don't build up a lot of antibodies, so they're not sufficiently immune. So out of that 10%, only some count towards the herd.

Citation?

From the last sentence in the conclusion: "Stronger antibody response was associated with delayed viral clearance and disease severity."

https://www.medrxiv.org/content/10.1101/2020.03.24.20042382v...

That's very interesting.

> First, we observed three types of antibody responses in COVID-19 patients, strong, weak and non-response. Second, we found that the earlier response, higher antibody titer and higher proportion of strong responders for IgM and IgG were significantly associated with disease severity. Third, the weak responders for IgG antibodies had a significantly higher viral clearance rate than that of strong responders. These data indicates strong antibody response is associated with disease severity, and weak antibody response is associated with viral clearance, which resembles SARS and MERS.

I have a wild-ass guess. I vaguely recall that antibody response and cellular immune response generally tend to be inversely proportional. And that IgM antibody response is an aspect of allergy. So maybe strong antibody responders tend to have weaker cellular responses, and more wet lung problems.

Preliminary results from a Dutch study I was reading about a few hours ago, in Dutch, that I came across through news.google.nl - although because I'm in a different timezone now, google news has been refreshed, and I can't find this specific one any more. So take that as you will...

That's not what i've seen. They're looking at it for blood transfers and finding asymptomatic people have sufficient loads.

edit: i think this is logical (for my rudimentary knowledge) given the cause of death is not directly related to the virus. The symptoms of dry cough and fever are all the bodies response. Even the shortness of breath and pneumonia might be attributed to inflammation. Perhaps the asymptotic people just don't have such a strong response for whatever reason.

Do you have a source for this? Haven’t read that yet.

I've seen it mostly in the Dutch news until now. The source is the Dutch CDC (RIVM) via Marion Koopmans (Professor of virology and head of the virology department at the Erasmus Medical Center).

Article in Dutch (translated title: Mild corona virus symptoms seem to lead to lower creation of antibodies)

https://nos.nl/artikel/2329846-milde-coronaklachten-lijken-m...

This article says that people with mild symptoms have fewer antibodies and says it’s unclear if they can be reinfected. Further it doesn’t mention asymptomatic cases.

You don't even need to be infected to develop the right antibodies:

https://www.randombio.com/coronavirus2.html

You may not have a lot of them, but you have already evolved targeted killers.

> even northern Italy, they're estimating that maybe 10% have caught it.

How do we know unless we do randomized antibody testing in the population?

It's just a lower accuracy level of estimate, not an invalid method. For example you can know that an infected person that had no known contact with another infected person means that there is in fact at least one other infected person out there. Yes, not as accurate as random population testing, but it's a difference of degree, not kind.

One way to extrapolate: excess deaths. Look at the overall death rate for an area and compare it to the normal value when there isn't an epidemic going on. If there is a difference that is larger than your number of known Covid deaths (which is true for Italy by a factor of somewhere between 2x and 5x), you can assume you're undercounting your Covid deaths by that factor, and your overall Covid infections by at least that factor.

This still involves guesswork wrt the ratio of deaths to infections (IFR). Proper statistical sampling is the only way to get an actual answer.

The usefulness of that approach decreases as people's behavior otherwise changes. For example, accidents account for ~10% of deaths. It's quite feasible (and I think we already have data to support this) that that has dropped a lot due to people's behavior changing in response to covid-19.

Forget accidents, there’s a dropoff in heart attack, stroke, and cancer deaths as every death now is attributed to Coronavirus.

It is unknown what percentage of people who are infected get symptoms and it is largely unknown what percentage of people who get symptoms ultimately die. So there is no way to calculate anything.

https://www.imperial.ac.uk/media/imperial-college/medicine/s...

With R_0 of 2, you need to have 50% immune to have herd immunity, but in practice, you’ll have more people than that that will end up infected before the epidemics dies out. Think about it: when over 50% of people get and survive the disease, rendering them immune, the epidemics doesn’t just immediately stop. What happens instead is that since average sick person transmits it to 2 people, but one of them is immune, less than one person gets the disease, and the epidemics starts the exponential die-out. However, as it dies, people still get infected, albeit slower and slower until it stops altogether.

The epidemics final size equation is F = 1 - exp(R_0 F). For R_0 = 2, F is around 80%, despite the herd immunity kicking in already at 50%.

This is one reason that "herd immunity through everyone getting exposed and lots dying" is not the same thing as "herd immunity through vaccination".

If you vaccinate a big chunk of the population when only 1% are currently infected, you hit R(effective) = 1 much sooner. While becoming immune through infection essentially guaranties there are a tremendous number of infectious persons at the point R(eff) drops to 1.

> because that means it's less fatal than we thought

Don't we already have solid numbers on the fatality rate because of the Diamond Princess? A change in estimated R0 wouldn't affect that.

Diamond princess is a non-stochastic sample, with all its attendant biases.

What does that mean?

People on cruise ships are usually not a good sample of the general population, for example, the mean age is quite high.

But it is also a sample with few in the sickest bracket, who are a disproportionate fraction of deaths.

And the number of deaths is sufficiently small that it is consistent with a wide range of death rates.

Another way to put it is that the study is enumerative and not analytic. Unfortunately, just about every study being done on COVID is enumerative.

https://en.m.wikipedia.org/wiki/Analytic_and_enumerative_sta...

"Diamond princess is a non-stochastic sample, with all its attendant biases."

But isn't it a non-stochastic sample in the right direction ? Which is to say, the demographics of budget cruise line passengers are almost a worst-case. They represent the most susceptible cohort.

What we can infer from Diamond Princess is that the general population will, all else being equal, fare better statistically.

Yes, but I was responding to a claim that we already had "solid numbers on the fatality rate". They're far from solid. We have a small sample that gives us an idea of the CFR, for that demographic, but this tells us very little about global IFR.

That's a crowd that skewed older in age, amongst other issues.

You can adjust for age just fine as every country has different demographics. The real issue is the crew and passengers where significantly healthier than the demographics suggest.

Aka people undergoing chemo are less likely to go an a cruse.

I've heard tale that people have opted to be permanent cruise ship residents over living in assisted living facilities.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC526150/ https://www.cnbc.com/2016/07/26/ahoy-matey-more-folks-retiri...

Permanent is rather misleading in that context. These these people don’t have major mental or physical impairments yet. Their old enough to retire, but heathy enough to enjoy it.

That's not obvious to me. You can imagine people might take a cruise over a walking tour of Tuscany because the cruise is much more accommodating of physical disability, you can get around with a wheelchair, there's lots of staff to help, etc.

You can, but the cruise will not let you on once you reach a certain level of disability. Wheelchair is no problem, but if you can't get yourself out of bed into the wheelchair they will kick you off. Similar for other disabilities, minor cases are fine, but if you can't mostly take care of yourself you are off.

A cruise is one of the cheaper assisted living arrangements if you qualify, (but governments will not subsidize your cruise while if you need assisted living they will). Part of it is discounts for inside cabins, part of it is they know you are there and will give various jobs when they need help which helps pay for your cruise.

The town of Vò, 3300 tested repeatedly, 89 tested positive, 1 death.

If fatality rate is 0.1%, there was only a 10% chance of 1 death i.e. it is very likely death rate is higher than 0.1%. Also the Diamond Princess numbers are relevant, after adjusting for age spread in a normal population, it is a lot more than 0.1%.

How do you get 10%? On avg you expect 3.3 people to die if IFR is 0.1%, so it shouldn’t be as low as 10% likely to get 1.

IFR relates death to total infected, not total population.

Iceland has the most comprehensive testing, and their fatality rate is around 0.3%.

There are a couple of reasons for that, most likely. First, only 4.5% of the known cases in Iceland are over 70 years old. In Italy, around 16% of the population is 70+, so given that this virus has much higher death rates for the elderly, we would expect a much lower death rate than Italy. Second, there are still 11 people in intensive care (compared to six deaths). I'm not sure what the mortality rates are for patients that require intensive care, but I assume they would be much worse than average. I think I remember seeing about 20-30% mortality for ICU patients from a study in Italy. Finally, the majority of cases in Iceland are still ongoing, whereas many villages in Northern Italy have gotten through the worst of it, with fairly few cases unresolved. I bet that the age adjusted fatality rate in Iceland will end up being fairly similar to Italy.

I agree. So, the scalar fatality rate should be adjusted to the normal population distribution.

Diamond princess was tested early and often and therefore able to get immediate quality care before medical systems were overwhelmed. This gives something like the ideal minimum fatality rate if, for the most part, everything goes right with detection & treatment.

Out in the wild, this is very much not the case. Testing it still thin on the ground in many locales and the pressure exerted on the medical infrastructure is not evenly distributed, making levels of care vastly different.

Think of it like a highly treatable form of cancer: If everything goes right and it's caught in early stage 1, fatalities rates are much lower than when it's only detected after reaching stage 4 and fatality rates are much higher.

Diamond Princess, for distinct demographic group only, represents detection at Stage 1.

Wait for the serosurveys before assuming how far it has already spread - would it be surprising if yet another estimate about this thing turned out to be way off base?

Which also means that a larger percentage of the population will end up being infected.

yes, this is why measles vaccination rates must be nearly total. measles has an R_0 somewhere beteween 14 and 18 (!!!)

A second is that the measles vaccine fails more often than many others. The combination drives the necessary vaccination rate quite high, and the risk isn't just the unvaccinated kids, it's a % of the ones who thought they were fully vaccinated.

Is it not the case though that R0 is dependent on all sorts of factors, including social distancing?

So if we keep some social distancing, presumably we can reach the equivalent “herd immunity” with far far fewer people infected, such that society can operate in a reduced capacity until a vaccine arrives.

Reply on news.ycombinator.com