Hacker News Reader

Comment by loveistheanswer | Hacker News Reader

Until we have challenge trials, how do we know whether or not other variables, such as natural herd immunity and seasonality are not also significant contributing factors to the 'substantial reduction'?

View on HN

Because its still winter and herd immunity from natural infections is a gradual process not a steep cliff? At the very least it seems like those factors could be accounted for.

Seasonal viruses often peak before the official end of winter. The key date may be the winter solstice in late December when sun exposure reaches its minimum.

There have been several new variants that have popped up lately, many able to spread rapidly, and some showing the same mutations but developing them independently.

With the rapid spread of some of the variants, it would be foolish to assume that what we are dealing with now is the same thing as what we are dealing with before, and we really don't know enough about the new variants. Maybe they are spreading, but not making people sick enough, increasing herd immunity.

It's difficult to account for these things because the same time these new variants started exploding is the exact same time we shifted resources from random testing to mass vaccination.

Why would seasonality be coming up now when we're a year into this thing?

Do you think we got a break at the beginning that we didn't identify in the numbers?

I think it has been shown that standard corona seasonality shows the same shape of all case loads we have seen?

That is, most of the recovery we saw last year going into summer could potentially be explained by regular cycles of similar viruses in the areas. The impression being that that could also explain or current recovery.

It is not an argument against vaccination. And I haven't seen people pushing we have heard immunity, yet. But the stark drops we are seeing do seen surprisingly sharp.

"the stark drops we are seeing do seen surprisingly sharp"

Surely that suggests an unnatural cause, such as millions of vaccine doses being rolled out, specifically targeting the most vulnerable (LTC residents) and those most likely to catch/spread the virus (medical profession, first responders, front line workers)?

The argument is, if that were the case, it would be sharper than the natural charts of corona viruses otherwise.

To be clear, I first saw this pushed by some epis on Twitter as caution to get to hopeful that we are seeing the vaccines as a resounding success so early in their rollout. It is expected that the vaccines are needed, but the dramatic drop was pushed as likely unrelated.

I will try and dig up the tweets. Could be they have changed their minds with more data.

Also, my first sentence was a question as I am not sure that is what the opening post meant.

Edit: https://twitter.com/jbarro/status/1363866144029949952?s=19 had a recent discussion where this came up. (Looking for epis in this one, but not finding them.:( )

Edit: https://twitter.com/EricTopol/status/1363551912021221377?s=1... is a look at a drop without vaccines

Could it be that there's a strain that results in largely asymptomatic cases and there is not enough data on those people? That could result in a decline now due to competition, or the slope of the line is historically wrong due to invisible statistics.

I know there has been some trouble identifying antibodies in people who were exposed months ago. So if you aren't really sick, you only get counted (maybe) if someone in your circle gets really sick.

For myself? I have no idea. I think this would fall under a general data quality concern.

I can say I have not seen this brought up too much on Twitter. With the caveat that I am not following everyone. :)

I do know that sometimes graphs with weird dog-legs are caused by either graphing the wrong derivative[1], or because there are more populations and someone is either being devious or is unaware.

[1] Developers are by and large flummoxed by S-curves for progress. An S curve for distance maps to a bell curve for velocity. If the sums don't make sense, look at the rates, or the rate of change. Don't keep staring at the S trying to fit trend lines.

Where are we at in mapping genotypes of the most affected and the superspreaders?

Perhaps exposure has reached saturation among certain populations.

Thanksgiving and Christmas holiday were both events where many people that were diligent about quarantine protocol for the past year decided to break protocol "just this once" because it involved the two primary holidays for gathering with loved ones.

You don't need a challenge trial if you have large enough control and experimental groups.

A challenge trial just reduces the time and number of people it takes--not everything needs a challenge trial.

The numbers from Israel seem to show the 60+ age group that has been vaccinated with two doses is seeing a sharp drop in hospital admissions compared to the under 60 group that is still mostly unvaccinated. This should rule out the herd immunity and seasonality arguments.

https://i.imgur.com/Tu3ckZN.jpg

https://i.imgur.com/1eP5sbi.jpg

I would have expected even lower hospital admission numbers for the 60+ population, I mean taking into consideration that almost 80% of them have already been vaccinated.

Older people with COVID bad enough that they end up in the hospital usually stay there for 3-6 weeks

Data from personal friends contradicts this. 2-8 days in my sample.

That's called an annectdote.

https://www.kpcnews.com/covid-19/article_8ab408ad-8fb0-5f74-...

Overall, the average hospital stay for COVID-19 for all ages is 22.4 days, just over three weeks. The length of stay is slightly longer, 23.5 days, for regular hospital admissions and shorter for ICU patients at 16 days, likely because ICU patients go on to die in the hospital.

That's based over thousands of patients.

> Patients in their 50s, who make up the third largest group of hospitalizations at 17.8% of all admissions, have, to date, had the longest average hospital stays at 27.5 days on average.

> Older patients have slightly lower average stays than middle-aged Hoosiers — again, likely because they are more prone to die in care than younger patients

> The average stays for patients in their 30s is 16.4 days

> That's called an annectdote.

Oh look: this data matches my annecdote(sic):

https://bmcmedicine.biomedcentral.com/articles/10.1186/s1291...

That study specifically says

"There were too few studies to conduct any comparison by age or disease severity. "

It also dates back from a year ago, under very different circumstances to today

This appears to be similar to what’s being seen in Scotland: deaths in the over 85 age group dropping sharply compared to younger age groups.

BBC Scotland article has some more data on the progress of the vaccine rollout and the impact it’s having on outcomes: https://www.bbc.co.uk/news/uk-scotland-56097899

Because you look at similar cohorts at the same time. So if group A has received the vaccine, but group B has not, and they're being observed during the same time/geographical area, and the groups are sufficiently randomized otherwise, then you would expect to capture the effect of any other confounding variables.

Given the large number of people vaccinated so far, and the magnitude of the effect it's pretty safe to say that the vaccine is causing a significant reduction in hospitalizations independent from the broader background trend towards lower prevalence of the disease overall.

Sounds like cherry picking to me. Doctors know if you have been vaccinated when making this decision. I.e. perhaps 90% of people with 1 dose who are admitted are serious enough to need intensive care vs a small percentage of non-vaccinated patients admitted more often as a precaution.

I guess everyone here thinks double blind trials are for fools?

I think the clinical guidance is still the same, that is you get admitted when you are showing certain symptoms regardless of vaccination status.

DanBC seems to have posted the study itself:

https://www.gov.uk/government/publications/phe-monitoring-of...

cases:mortality would be a weebit less influenced by placebo and it would be hard to reach numbers like 80%..

IMO, there's no way a significant number of discussions between doctor and patient are not going to reach different conclusions about whether to go to check in to a hospital or wait a few more days based on a significant fact like a jab 3 weeks ago.

So I'd have to say this is cherry picking..

Isn't there also a confounding factor that the vaccinations were largely introduced following the thanksgiving and christmas infection events. Even without a vaccine, I would have expected hospitalizations to fall approximately 1 to 1.5 months after the Christmas holiday.

Basically, we had many clusters of fresh unburnt tinder (the household covid pod) and the Thanksgiving and Christmas holiday was a perfect event for many people to "just this one time" break quarantine protocol, leading to the many infections we saw. That's a 2-3 week increase in direct hospitalizations from those events, and then you have another 2-3 weeks of indirect hospitalizations impacting the remaining members of each covid pod. Anecdotally, I've personally witnessed this happen as I know fare more people that acquired immunity from becoming infected during the holidays than for most of last year.

Depends on where you live, in the Midwest the peak was more around Halloween in October. You can hardly find Thanksgiving in any data (and then you probably have to squint and ignore proper statistics) By Chirstmas/new year things were clearly in decline. Other areas of course have different results.

In those areas, has the rate of decline just been consistently downward? Has vaccine introduction had any measurable impact?

Yes. I follow the daily numbers for my county.

The vaccines seem to have an impact, but it is hard to be sure. There are a lot less old (>65) in the new infected list, which used to dominate the list. The younger groups seems to be be about even by age group. However there are many potential confounding factors, and I haven't done a proper statistical analysis so I don't want to claim something.

We don't, but a challenge trial wouldn't change that.

Best you could do with a challenge trial is to get faster to the results we already have. These vaccines work. We know that. This is just observational data supporting that what works in trials also works in the real world (which is not particularly surprising).

https://www.gov.uk/government/publications/phe-monitoring-of...

Because we're comparing vaccinated people right now to unvaccinated people right now. Herd immunity and seasonality would affect both groups equally.

At the very least it would be easy to measure the vaccines effect on severity by comparing the rate of hospitalisations to the rate of infection

You have a control group of unvaccinated people.

These people are starting to get vaccinated now though[1], which actually causes issues with the study.

[1]:https://www.npr.org/sections/health-shots/2021/02/19/9691430...

That's the clinical trials, but it should be reasonably straightforward to do observational studies on COVID hospitalizations cross-referenced with vaccine status.

There are plenty of millions of people who haven't yet (or won't ever) get vaccinated to serve as a control.

This WSJ article mention herd immunity. Covid spread really fast, so herd immunity through a combination of the large amount of people that already had covid and vaccination of the vulnerable should make a big difference.

https://www.wsj.com/articles/well-have-herd-immunity-by-apri...

Or maybe they just lied about the numbers and the whole story is pure fabrication from the start.

I wouldn't believe any of those guys.

The government will tell us of course! Are you one of those lolbertarians who thinks covid is a myth and doesn't trust government?

Give up trying to have any reason or rationality about this.

The world wants this mass hysteria. I'm starting to think it's not even about Corona virus anymore. Everyone has an agenda. From the remote workers to the politicians to the news media to the people getting unemployment.

You just have to let the madness pass as it looks like it's making progress towards being behind us.

Reply on news.ycombinator.com