One in 5k: The Real Chances of a Breakthrough Infection | Hacker News Reader

One in 5k: The Real Chances of a Breakthrough Infection

nytimes.com

21 pointsrafaelc35 comments

The analysis is absolute garbage for forward risk.

1. 1/5000 daily is 7% on an annual basis. Who cares about a daily rate?

2. I downloaded covid data for jul 1-aug 26th and found daily average was 80,000. Us average 160,000 now. So right off the bat, 2x underestimate. Up to 14% annual risk.

3. Infections are estimated to be about 3x greater than reported cases. So now up to 42% annual risk.

4. Delta is associated with breakthroughs. It was 60% jul 1st. Now 100%. So this also should increase breakthrough risk.

Risk of infection at current case rates likes around 50% for a vaccinated person in the US, on average.

Not so comforting once you make the necessary adjustments above. I do not think they made a serious attempt at analysis.

I used Our World In Data for all stats.

Title buries the lede, One in 5k per day

> How small are the chances of the average vaccinated American contracting Covid? Probably about one in 5,000 per day, and even lower for people who take precautions or live in a highly vaccinated community.

About a 2% chance of catching it before the end of the year. Much higher than I expected. There could be another spike this Fall and Winter, which would increase the chance. But, hopefully, more people getting vaccinated and possibly booster-shot would drive the chance down.

No, 7%. And that’s at july-aug rates. Cases up 2x since the average in study period.

Also infections lower than reported cases by perhaps 3x. Also delta went from 60% at start of study period to 100% now.

Take 6x multiplier + delta prevalence and you get ~50% annual breakthrough odds at current average daily case rates.

What an absurd use of statistics.

Were they afraid to make it a year, and say 1 in 15?

For those that can’t access the article, here’s the key sentences:

“In recent weeks, however, more data has become available, and it suggests that the true picture is less alarming. Yes, Delta has increased the chances of getting Covid for almost everyone. But if you’re vaccinated, a Covid infection is still uncommon, and those high viral loads are not as worrisome as they initially sounded.

How small are the chances of the average vaccinated American contracting Covid? Probably about one in 5,000 per day, and even lower for people who take precautions or live in a highly vaccinated community.”

> These numbers don’t include undiagnosed cases, which are often so mild that people do not notice them and do not pass the virus to anyone else.

But some of those people still end up with long-covid, right?

For a long time now that's been my biggest worry; at this point it's pretty much my only (personal) worry

Wouldn't long-covid lead to enough symptoms to get a diagnosis?

There's very little official diagnosis of long-covid at this point. It's only beginning to be acknowledged, and it's one of those things that's so varied in symptoms and severity that it's hard to pin down, and easy to dismiss as psychosomatic.

Some of these people will end up bitten by a brown recluse, have a heart attack and then die in a car crash on the way to the hospital. For a really large population, a tiny but not zero number of people will undergo all sorts of crazy scenarios.

That's why we should be placing numbers in context. How many people get objective 'long covid', by which I mean objectivly measurable decreased lung capacity or blood oxygenation levels, compared to how many people get long flu, flu + pneumonia or all sorts of other shitty but rare afflictions?

I'm not dismissing the possibility that 'long covid' is a thing, I'm decrying the lack of actual data and the reliance on media propagated rumors. Just within last month Delta was supposed to be a calamity, 'the hospitals are overflowing (with Ivermectin overdoses no less)', 'children are dropping like flies', etc. I'm not hearing that lately, nor have I seen any numbers to support that level of damage.

That's just it, we don't know yet. The reality (in terms of frequency, duration, and average severity) could be an order of magnitude off in either direction relative to the anecdata.

Covid has been totally alien in multiple ways already. The long-term effects could also be totally outside the norms for other viruses.

That we don't know doesn't necessarily imply that we need to assume the worst case scenario.

That's up to each individual's risk profile. I'd rather play it safe for another several months than risk lifetime debilitation. You can pick your own risk level.

True. Everybody has a right to decide their own risk profile according to their own life circumstances and risk tolerance. Not to scold, but please notice that some people have quite a bit larger choice space than others. For example, white collar WFH vs. blue collar in-person service.

>That's just it, we don't know yet.

So why is it your biggest worry?

Because of how wide the margin of error still is, and the fact that lifelong debilitation is on the table.

Long covid right now is in the stage of analysis that regular covid was over a year ago. People had no idea back then how bad the mortality rates actually were, what to do about them, etc. We have a pretty good handle on that side now, which is why I'm no longer really worried about being killed or even hospitalized, as a healthy young vaccinated person.

But we're just beginning to study the long-term effects. Their severity in the grand scheme of things could be tiny, or it could be massive, because - again - we don't know yet. I don't want to roll those dice.

But. We. Just. Don't. Know.

Literally anything could happen. After years of studying it, we just don't understand it. Five years later, boom, you wake up with nasal demons.

If only there were existing respiratory coronaviruses we could compare it to, instead of this completely alien pathogen.

While COVID is closely related to other coronaviruses, it doesn’t act the same way. So, studying them doesn’t help us that much with COVID. And we’ve been studying some of them for over a hundred years.

There is new and novel activity going on here, and we don’t really understand how it works.

Not really.

SARS is closely related. (The virus is literally SARS-CoV-2.)

COVID-19 might be unique taking every possible characteristic together, but nothing it's done so far is unique among coronaviruses.

This was one day ago:

> An exhausted Iowa ER doctor pleaded with people to get vaccinated against COVID-19 in a Facebook Live video: 'We are drowning in people who are dying with this illness' https://www.businessinsider.com/iowa-er-doctor-pleads-get-va...

And:

> Austin region has zero available staffed adult ICU beds, state data shows https://spectrumlocalnews.com/tx/austin/news/2021/09/06/aust...

I don’t think the problem has gone away. Maybe you’ve just stopped listening?

Perhaps I am indeed exhausted by the continuous barrage of rumors of doom. The last CDC snapshot (as old as 07/13!) estimates that nationally 8% of the hospital beds were occupied with covid patients.

Perhaps I'm unfazed by the ICU beds stats because I have been born and raised in a country where there were no ICU beds. I file them in the 'amazing bonus' rubric, not feeling entitled to have an ICU waiting for me worst case scenario.

In my neck of the woods, the local pedriatic hospitals are at 70-80% utilization and have admitted a grand total of 8 covid cases in the past week. Some of the local adult hospitals are indeed going over 90% capacity, but others are cruising at 50%. The sum of covid cases admitted in the past week is in the low hundreds. This for an urban metro with 4M people.

On the bright side, Florida, which has been the focus of national media covid reporting appears to have turned the corner on covid hospitalizations, currently running at 87% capacity, of which about 25% covid, and trending down.

Iowa runs at 62% hospital capacity, with only 10% covid. Texas is worse with 80%/30%, but still functioning.

While the situation is serious and tragic, there are no signs of the Apocalipse. By Apocalipse I mean healtcare system overwhelmed 10-100x, bodies piling in the streets, etc. The kind of catastrophe the likes of Thomas Puyol were predicting in early 2020, and at which the media has been alluding ever since, continuously rotating the focus on the worst affected areas.

https://www.cdc.gov/nhsn/covid19/report-patient-impact.html

https://coronavirus.jhu.edu/region/us/florida

https://coronavirus.jhu.edu/region/us/iowa

https://coronavirus.jhu.edu/region/us/texas

>How many people get objective 'long covid', by which I mean objectivly measurable decreased lung capacity or blood oxygenation levels

One of the scariest aspects of long covid is the neurological effects. It's hard to "objectively" measure things like "food tastes weird now" and "I'm exhausted all the time" or even "I can't focus like I used to" in a clinical setting without having done baseline tests pre-infection - yet these things will be obvious to the sufferer. Sometimes, in the absence of better data, you have to take people's word for it.

You are right that we need more data on long covid - it's inherently hard to study, and early data is frightening. Until we know more, caution is advisable.

I assume anyone commenting in here has an online subscription to the New York Times, and thus they can read the article.

I can't. (read the paywalled article.) What does it say?

For NYTimes articles you can't read because paywall, they're usually available at archive.ph. Just paste the original link for the article in the lower search bar.

Thank you, Hacker News friend! Worth the downvote, to get this information.

And that got you one upvote from me at least. Welcome.

This is, of course, bullshit. Multiple tricks are used to further the narrative in this.

First: the CDC decided NOT to track breakthrough cases. For obvious reasons. That leaves us with less than ideal data. In this case, the data comes from three places: Utah; Virginia; and King County.

Second: These are the DAILY odds. Even assuming everything else is correct and stays the same, this is still a 1 in 7 chance over the course of 2 years.

Third: half of breakthrough cases are probably missed. Maybe more.

Third: the vaccine is near its highest efficacy now, and has only started to drop. It's all down hill from here.

Fourth: vaccine testing bias is not discussed.

I'll do my own math for the vaccinated:

Only half the cases are caught with a PCR test. = 1 in 7 chance over 1 years.

Let's do two years. 1 in 3.5 chance.

Now throw in some vaccine durability issues over 2 years. 1 in 2.5 chance???

Then a new variant.

Ah fuck it, everyone gets Covid.

Will everyone get COVID? Likely, yes.

Will the vaccine limit the damage to the population as a WHOLE? Evidence says yes.

Your math is rather suspect; 1 in 5000 daily chance (giving you the benefit of the doubt since I can't read the paywalled article) means that if someone is at risk for a breakthrough infection for 10 days (that's a WAG, but seems plausible given how long it takes to clear the virus) that's a 1 in 500 chance, not your 1 in 7.

Does that chance increase as your immune system "forgets" the virus? Yup, that's how our immune system works.

So aside from your very... loose math, I'm not sure what your point is? Take that away and there's nothing to your comment.

I know anecdotally a few of my friends have been vaccinated, got sick with symptoms, and wouldn’t get a COVID test because it couldn’t be COVID, so there is going to be some self selection bias happening in the number of cases that test positive.

There are other biases at play as well, for example many businesses are forcing unvaccinated employees to take tests, but the vaccinated are exempt from the tests. So someone could be an asymptomatic carrier and never reflect in the published statistics that NYT is drawing conclusions on. And even then, the 1:5000 odds don’t sound particularly reassuring.

> Your math is rather suspect; 1 in 5000 daily chance [. . .] for 10 days [. . .] that's a 1 in 500

The poster was correct. A 1/5000 daily chance is about 1/7 chance for a two year period.

Your math, on the other hand, is wrong.

If it didn’t seem wrong to you on an intuitive level (that is, if you didn’t know a binomial distribution would be necessary) you are at great risk for being easily misled.

Apologies; I wasn't able to read the article and it seems I didn't interpret his statement correctly. My comment is left for posterity, and because I don't like editing after other's have replied.

> Your math is rather suspect; 1 in 5000 daily chance means that if someone is at risk for a breakthrough infection for 10 days that's a 1 in 500 chance, not your 1 in 7.

I couldn't understand your comment or your math. The point OP is trying to convey is that 1/5000 is not a reassuring number even though the title makes it sound like one. That is, the title is deliberately misleading. (And you are a prime example of someone being mislead) That's because a 1/5000 daily chance implies a 1/7 chance over the course of 2 years which is actually pretty effing huge.

The other thing is that 1/5000 chance isn't fixed. It depends on the amount of virus circulating in the reservoir of unvaccinated dummies. The prudent thing to do is just wait for the virus to infect those guys and burn itself out.

It was because I interpreted his numbers one way, and it was incorrect, due to not being able to access the article.

Shouldn't be doing that. My comment is left because other's already replied, but I was incorrect.

Reply on news.ycombinator.com