Oh wow, here is a game changer in estimates of infected cases

I've posted before my many reasons to suspect the CDC's estimate that we are missing 9 out of 10 cases due to asymptomatic and mild cases that never get tested. Sure, what this study shows doesn't totally invalidate the reasoning but I believe that it makes it much less likely:

https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.1003346

So, this study, a meta-analysis of 79 others, found that generally speaking, only 20% (or 1 in 5) people who contract the virus remain asymptomatic for the duration. Sure, there is some uncertainty in the data, and a sub-set of these studies (7 of them) suggest a rate of 31%.

Still, the bottom line is that, as the authors put it, most people infected with SARS-CoV-2 will develop symptoms of the disease COVID-19. These days, there isn't so much test shortage so chances are that the people who develop symptoms will get tested... So maybe we aren't missing as many cases as estimated. There are many other reasons I've detailed elsewhere for me to suspect that this 9 out of 10 claim is a gross exxageration.

If we aren't missing as many cases, what are the implications?

One, the idea that we'll reach herd immunity threshold by relying on natural infection alone becomes less viable, if not many more than our roughly 7 million cases exist.

Two, the idea that the infection-fatality rate (which takes into account every infection, not just the confirmed ones) is much lower than the case-fatality rate (which is how many cases have died among confirmed cases), gets a blow.

Three, if not many more than 7 million cases exist in the United States (say, 15 million), it means that we still have a HUGE susceptible population, which could have implications for second and third waves.

I find it, overall, concerning, because it was comforting to suppose that we're missing lots of cases and they are all out there developing immunity and not dying... If this is confirmed, it makes the aggressiveness of this pandemic, a bit worse.

@557 and @Bowerbird , what do you think of it?

 

Haven’t had time to review the study much, but the metric of “remaining asymptomatic” after diagnosis is very problematic in light of the nocebo effect. Especially since it’s not clear what the definition of “symptom” is.

 

Well, 557, it's a meta-analysis, actually 94 studies, not 79 as I mistakenly said, so, presumably each of the studies had a definition of symptoms. See Box 1, they said they used the authors' definitions of symptoms in the studies they looked at. One would have to go to the reference list and check one by one the 94 studies. Too time-consuming for us... I'm not doing it, but if you really want to know, you'd be able to do it. Conveniently if you click on References, their list has a hyperlink to each article. I clicked randomly on one of them and found that the authors considered as symptoms, "myalgia, fever, and a cough". Clicked on another random one, and found even more objective symptoms; they considered fever and pneumonia with radiological findings, hardly a nocebo effect. So, yeah, I think the symptoms are real. I think we can take the authors' word for it; after all, this is a peer-reviewed study so presumably the peer-reviewers did look into the integrity of the data. No guarantees, but if I were the one reviewing it, I'd have clicked on the references.

 

I’m not disputing the “symptoms” are “real” necessarily. We just don’t know if they are a result of disease or not. The nocebo effect can produce many of the symptoms you list. Especially inflammatory symptoms in the presence of stress. We certainly have all the conditions necessary for strong nocebo effect. The only way to rule out a nocebo effect would be to not inform positive test subjects of their status and that would be unethical I’m sure we both agree. In fact, just the act of being tested and then checked for symptoms can be enough to create a nocebo effect even if they were not informed of status.

I am not looking over that many studies either. LOL

I’ve been bouncing back and forth from the semi to the combine all day and haven’t had a chance to look into recent testing data, but I know locally availability of tests is a non issue. Symptomatic folks always got tested if they wanted to and the 30 or so people I know who know they have been exposed never sought testing. Even ones who had mild symptoms.

You bring up some good points but I’m going to have to ruminate and check out testing data which I haven’t kept up with to comment further.

One more anecdotal point I’ve mentioned before. If we aren’t missing a huge amount of cases this thing isn’t nearly as contagious as advertised. As I said, I know a bunch of exposed (some REALLY exposed) people who haven’t become ill or been tested. Either these people are infected and missed, or the virus isn’t particularly infectious. Can’t have it both ways.

 

why do you ASSUME that those with symptoms will get tested? My friends simply stayed home and took tylenol for a day. They both lost their sense of taste and had mild fevers for a day. They felt better in 2 days but kept away for a week and are bopping along with life. We don't need to be told when we have a cold or flu, we know it. If we lose sense of taste and have a mild fever, we know that we have covid and covid IS NOT even close to being a death sentence. Why go get tested to learn what you know?

That my friend, despite your insistence to the contrary, if you moved out of liberal circles, you'd find that many folks just go about their lives, and don't seek medical care unless needed. That includes being tested for what you already know.

 

So I’ve read through the study more closely and see data was used from pre print journals as well as peer reviewed journals. Not a big deal but does negate the fact the meta analysis is peer reviewed to some extent. I seriously doubt the peer reviewers of of this meta analysis went back and reviewed all the sources from MedRxiv etc.

Anyway, the more I think about this, the more I’m convinced my initial reaction was correct. Nocebo effect is a huge factor in this kind of data collection. The authors of the meta analysis do say many of the data sets used were not designed for determining asymptomatic prevalence in populations. That’s as close to a mention of nocebo influences as they get. The fact nocebo effects aren’t realized or admitted to by the researchers and not a concern to reviewers either really surprises me actually.

So I’m in agreement @CenterField that if asymptomatic rates are really this low there are implications just as you say. However, I don’t see this analysis as being very good evidence we are catching most cases. I can’t overlook the foundational flaw in the methodology.

Kudos to the authors of the meta analysis admitting false negative PCR tests are a factor that must be considered. It’s good to see some researchers finally admitting that.

 

Just curious but what about those with symptoms not tested because of a lack of test availability and/or not meeting the criteria for test when only a limited number of tests were available?

Reason I ask is because I had 8 out of the 9 listed symptoms relatively mildly. I was not eligible for the test at that time because I never had the fever symptom. I doubt that I was the only one who had this experience. I agree that with testing now commonplace I would be tested if I presented with those same symptoms if for no other reason than to rule it out.

The experience in Manuas, Brazil indicates that it is possible to reach her immunity around 65% of the population and they had a relatively young population with only 6% over the age of 60. Their death toll was still a horrendous 2.9% of those infected and could have been higher with more seniors.

https://www.sciencenews.org/article/coronavirus-covid-19-brazil-city-manaus-herd-immunity

Assuming those figures are correct then 65% would be 214 million infected here in the USA and 6.2 million dead. However we have 20% of our population over 60 so that could well be a low end estimate that could be 3 times higher.

Herd immunity comes at a steep cost and I really do not want to go down that road and find out just how bad it could actually be. We need a vaccine ASAP.

 

ASSUME with all caps? What part of "chances are" you don't understand? What part of "if this is confirmed" you don't understand? My OP was very sober "doesn't totally invalidate the reasoning" "there is some uncertainty in the data" etc.

And when have I EVER suggested that Covid-19 is close to being a death sentence? Don't you think that I know that only about 1% of people die? Still, I've been saying that the death toll is not all, as we discover more longer term complications. Read this article about this paper, and the paper:

https://www.nbcnews.com/health/health-news/young-people-are-risk-severe-covid-19-illness-n1240761

https://jamanetwork.com/journals/jamainternalmedicine/fullarticle/2770542

And this is for young people... I know it's not a death sentence but it is a concerning illness.

 

I hear your concerns but still I think that papers that have pneumonia with radiological evidence as symptoms were really looking at SARS-CoV-2-caused lesions. Anyway, as in any meta-analysis the quality of the 94 is necessarily variable.

 

Exactly, that's why the herd immunity approach is a disaster. It's may be possible to control the virus this way, but the problem is, the cost to pay for it is likely to be intolerable, and it's quite unnecessary since the vaccines seem to be coming.

By the way, the CDC director is frustrated with that imbecile Scott Atlas too:

https://news.yahoo.com/director-cdc-overheard-saying-everything-145247059.html

"Everything he says is false." LOL. No wonder Trump likes Atlas so much. A kindred spirit.

 

I found the meta analysis definition of symptoms.

So it looks like the evidence you are depending on is not included in the analysis. Just data highly likely to be corrupted by nocebo effects.

 

True, good catch.