Riddle me this Vaxman

Tsk, tsk. From your link:

1. Confirmed case
   A confirmed case is “a person with laboratory confirmation of COVID-19 infection” as the World Health Organization (WHO) explains.

That is quite disingenuous to try to pass off lab confirmed cases.

 

Sorry about only using scientifically confimed data but my Ouija Board is on for repairs. How about you read the entrails of a cat or something and give me more accurate figures.

 

Do you understand what a lab confirmed case is?
Do you know that asymptomatic and symptomatic persons did not get tested?

Do you even understand you own link?

 

Do you understand that asymptomatic cases do get tested, are recorded in the data and that I provided examples in a previous post. Specifically: - hospitals and other large government and private organizations with mandatory testing regimes for staff & also countries with stringent testing and case tracking mandates. In these countries at least during the early stages of the pandemic if not now people would be electronically traced & notified of the fact they were a close contact of a positive case (at a particular time and location) and be directed to get tested and/or isolate. This allowed data on asymptomatic cases to be collected.

 

There is no need to know the exact number of people who were symptomatic and didn't get tested, or were asymptomatic. They are included in the total populations of vaccinated and vaccinated people used to calculate hospitalization and death rates. This is the primary objective, and therefore measure of efficacy, of the vaccine - to reduce the number of cases which do result in serious illness, hospitalization and death. The fact that some, or indeed many, are infected but are asymptomatic/untested is baked into (reflected in) the resulting rates.

 

And now it's become public knowledge that pharma grossly deceived in its record keeping during the 'testing' phase, and CDC assisted in keeping bad data hidden away.

It's not like the first time they've done it. Back when 60 Minutes practiced honest journalism, they showed the fraud and deception associated with the Swine Flu.

Pfizer has pleaded guilty twice to criminal fraud regarding their products.

In fact, the medical industrial complex is much more dangerous and harmful than the military industrial complex.

 

But even if you come up with the asymptomatic rate among the tested you can’t extrapolate directly to the untested. I’ve explained why. Because the untested have intentionally excluded themselves from the sample making the sample not representative of the general population. As I also said, the extrapolation may come close. The reason your meta analysis clearly stated this in the title is because the authors understand what I do—that you simply cannot predict anything without a representative sample snd that the studies they analyzed did not use samples representative of the entire population.

But let’s say we DO know the true asymptomatic rate of the general population in relation to the symptomatic. We still don’t know the true numbers of actual asymptomatic infections (which was the concern of the other posters, not the asymptomatic rate in relation to the symptomatic among confirmed cases.

The reason we can’t know the number in dispute is not a function of or a failure of statistics. It’s a function of us not knowing two things. We don’t have a representative sample to determine asymptomatic rates in the full population and we also don’t even know how many total (symptomatic and asymptomatic) infections have occurred. In other words we don’t know the actual total infection rate (in any population really).

Why don’t we know the total infections in a population and thus the infection rate? Two reasons. The first I’ve already gone over. Large portions of infected (asymptomatic and symptomatic) population intentionally exclude themselves from testing. Second, we do not have the tools to determine who is or isn’t actually infected even in the tested population.

Using PCR if an infected individual is tested prior to symptom onset (when is that in an asymptomatic individual) there is only about a 50% chance of returning a true positive. On day one to two of infection a false negative result is highly likely. On day of symptom onset the odds of a single PCR test returning a positive vs. a false negative is about 85%.

What this means is that in most random mass testing events, as many as 30-50% of actual infections are not “discovered” and the actual infection is recorded as an uninfected.

Antigen tests are even worse than PCR at returning true positives as opposed to false negatives.

Antibody tests have about a 20% false negative rate in individuals who have been recently infected. Depending on what test is used, fading antibodies after recovery can also result in a false negative result.

Because we have no reliable way to come up with a representative sample of asymptomatic rates or of overall infection rates we just don’t know how many asymptomatic (or symptomatic) individuals have been infected and recovered as was (as I understand it) the disputed claim here.

Everyone has made educated guesses. The CDC at one point estimated the infection rate was 10 times the case rate. We know it’s lower than that. Probably 2-3 times the case rate. But it’s just a guess based on a bunch of assumptions—there is no way to generate hard data with representative samples.

This means all the statisticians in the world can’t solve the problem because statisticians can’t (or haven’t yet) make more reliable tests of force randomized testing. I guess they could do one or the other or both, but it’s statistically unlikely.

To be clear, my posts should not be taken as an argument for or against vaccination or a prediction of vaccine efficacy. It’s simply to point out it is the case we don’t have good evidence for the true numbers of asymptomatic (or symptomatic) cases. Too many infections can not be documented in any representative sample because of human nature and limitations in testing accuracy.

 

He saw the word 'predict' within the first explanation of regression analysis that he stumbled across, which totally obviously clearly unabashedly means that statistics "has the power of prediction"... It's quite laughable.

Regression analysis estimates relationships between variables (iow, it finds trends in a data set and one can use that information for making estimates); it does not have the power of prediction.

 

Regression analysis estimates relationships between variables (iow, it finds trends in a data set and one can use that information for making estimates). That is not predicting the future. There is no power of prediction inherent in statistics as there is in other types of mathematics.

For example, you can do a regression analysis on a person's weight over time and ESTIMATE that a person will gain 4 pounds in a month. After a month goes by and that person doesn't actually gain any weight at all, you will see why statistics ABSOLUTELY DOES NOT have the power of prediction as other types of mathematics have.

 

"You didn't build that."

 

No, statistics DOES NOT have the power of prediction. See my other response to you for an explanation as to why this is.

 

Regression analysis does not have the power of prediction.

 

That is the justification politicians dream of. Ignore results, throw accountability out the window; just accept. "we already screwed you, so take your inferior government product."

That is the worst reason to consume a pharma product ever!

 

That's analogy is incorrect. You don't use regression analysis (or any other form of statistical problem solving) to make estimates about one specific, individual data point. That one point has to be included in an entire set of data and the set has to be adjusted for all the relevant variables (e.g. gender, age, income. location, calorie intake etc) so that you compare apples with apples. The more people you have the set the more accurate your measurement is going to be better. Finally you should also have to repeat the measurement so you get a time series. Then you'd have a chance of estimating the amount of weight any one person was likely to gain. Bit you can't then pick one person out of that group and say 'look they didn't gain 4 pounds' because if the numbers tell you 4 pounds is the average weight gain, then one or more people in the group will weight more than 4 pounds. The estimate is for the group not the individual.

What your describing would be like trying to estimate anyone's chance of being struck by lightning by sitting around in front of someone waiting for it to happen!

 

There are multiple points in your post so I'll start with one and see how much time I have. For example in the first paragraph above you refer to fact that many (not all as I previously explained but many) of the untested population have intentionally excluded themselves from testing and so do as you claim form from a 'representative sample' of the general population.

This decision to exclude themselves from the data might I suppose matter when analyzing statistics in other cases e.g. say voting intentions. (I'm still not absolutely certain this is the case mind you because I suspect statisticians would have 'work-a - arounds' for such situations but lets assume its true.) The thing is COVID infections and the question of what % of cases are asymptomatic is NOT one of those situations where this would apply.

The point is that the population group your interested in is going to divide into two distinct groups automatically, those who have (or had) covid and those who don't. There is no 'choice' involved in this division its measurable physiological fact. The 'intentions' of all members of the data set are literally irrelevant to the question of whether or not they have COVID. They either do or do not. Then within the positive group (which is that part of the original sample group your really interested in any there are again only two district groups. Those who are asymptomatic and those who are not. Again intention has no place in the equation. You literally can't intend to be asymptomatic, it to is a measurable physiological state - end of story. Again you either are or are not asymptomatic

So all of that being the case as long as you have access to a large enough sample group of people who did test positive yet were asymptomatic a statistician will have all the the numbers to work with. And remember as I noted previously there were and are literally millions of people with positive test results who were either subject to compulsory testing campaigns at work or who were otherwise picked up by government run COVID tracking/testing campaigns. That's a lot of data on asymptomatic cases to work with. And remember as I said previously. Being unwilling to be tested (for virtually anything) has no medical bearing whatsoever on the question of whether or not you actually have whatever it is you've chosen not to be tested for. Or for that matter whether or not you actually have a bad case of 'it'.

This is fun BTW. Good arguments.

 

My basic analogy is fine, and your expanding upon it is fine, but the end result is the same.

Statistics does not have the power of prediction. This is due to its use of random numbers. Statistical analysis depends on probability theory, which in turn depends on the generation of random numbers, and it is the use of those random numbers that destroys the power of prediction that is normally inherent in mathematics. Thus, statistical mathematics can only summarize past or present events. It cannot predict the future.

This is where you admit to me and this forum that you are wrong about this topic.

 

The argument isn’t that you can choose to be symptomatic or not. The argument is you can choose to be tested or not. You are trying to make the tested population representative of the untested population and this can’t be done. It’s why the meta analysis you posted and others I’ve seen are all clear they do not predict asymptomatic rates in the general population, only the tested population.

Let’s try this. Here is the CDC and Minnesota Department of Health advice on when to get tested.

CDC:

MDH:

We see testing is recommended if you have symptoms or if you have had known contact. So an individual that wore their cloth mask to Walmart and got infected but remains asymptomatic never tests. That individual is more likely to test if the infection is symptomatic. Essentially you not only have asymptomatics choosing on their own to not test, you have “official” recommendations on testing that exclude at times the majority of asymptomatic cases from being tested, diagnosed, and recorded.

We still have your examples of mandatory testing in healthcare. While there are numerous confounders that make it difficult to extrapolate their rates to the general public (including ag—the median age of nurses is about 52 while the median age of the US population is 38, stress levels, general health —nurses for example have a much lower incidence of type 2 diabetes than the general population and comorbidity decreases the asymptomatic rate).

The tested population can not be representative of the untested population. Asymptomatics are excluded by choice (personal and by the recommendations of public health entities) from the tested population.

It is !

I’m going to elaborate a bit here on the OP that started all this. Here is the OP.

As I pointed out already, the question was never what the asymptomatic rate was/is in relation to the symptomatic rate. The question of the OP was more of actual numbers of asymptomatic and symptomatics that recover without being vaccinated. Both the OP and you (to some extent) have really invoked the debate of what role absolute risk reduction (and numbers needed to treat NNT and numbers needed to vaccinate NNV) should play in the vaccination debate. The OP’s questions are really a down to earth description of absolute risk reduction and NNT/NNV vaccinate statistics. Unfortunately these statistics have been suppressed with C19 vaccines.

I remember a couple years ago when pharmaceutical companies were big evil corporations selling overpriced drugs people didn’t need, NNT statistics were gaining traction in the medical community. Now if someone brings up absolute risk or NNV statistics you are a science denier. LOL

Kudos to the OP for figuring out these statistical tools have value in evaluating C19 vaccines and how they benefit different demographics in varying quantifiable degrees.

We can get into NNT and absolute risk reduction if you want. It would be valuable information for everyone here to see and contemplate.

When the OP questions the total number (not asymptomatic rate) of recovered infected individuals they are essentially looking for information on the background risk from C19. This is necessary to know to determine the absolute risk reduction imparted by C19 vaccines.

In short, somewhere between 1% and 6% of infected individuals require hospitalization when unvaccinated. This is the results of many studies before vaccines were available etc. So if your background risk of hospitalization is 3%, a relative risk reduction from vaccination of 80% (vaccine efficacy is reported as relative risk reduction—RRR) actually decreases risk of hospitalization from 3% to 0.6%. Of course this is great, but not as emotionally stimulating as the 80% efficacy statistic.

Here is a good article on the subject, mostly dealing with NNV to prevent one infection, not hospitalization.

https://www.thelancet.com/journals/lanmic/article/PIIS2666-5247(21)00069-0/fulltext

Many scientists believe both ARR and RRR should be part of the discussion and guide public policy. Unfortunately, the ARR metric has been suppressed. I think that has been a mistake because as we can see, people are beginning to question why even when they don’t know the terms absolute risk reduction or relative risk reduction.

The OP figured out the background risk is very influential to the absolute risk reduction from vaccination. That’s why the question is asked how many undocumented infections there are and how many total infections there have been. The answer is we don’t know exactly but it’s likely somewhere around double the reported case count. This means absolute risk reduction calculated on the official case count will be double that of the absolute risk reduction calculated on the actual number of infections. If we use the infection rate not the case rate in our calculations it makes vaccination look half as effective as if we use case count. That’s why this discussion is avoided—it doesn’t promote vaccination like big relative risk reduction numbers. But absolute risk tells the individual more about their real risk and real potential benefit from vaccination.

I’m a bit stressed for time at the moment as well but will continue…

 

You provided data of “lab tested” cases.
Again, not everyone who has had Covid has been tested. Not everyone works for a company/government where testing was required.

So, this is all conjecture.

 

Yes, there is a need to know the exact number if you are trying to use statistical models to make the predictions. Otherwise, this is all a guesstimate.

 

IIRC, the mandates forced anyone who wanted to continue to be employed with the government or travel abroad to get the vaccine. The irony in that, is while it helps to protect the vaccinated person from the infection setting in it does not prevent them from being a carrier!!! None of this is new! The efficacy and limitations of all properly designed vaccines are well understood so the idea that the government pushed that the vaccine reduces how transmittable it is, is at best a 1/2 (or less) truth from the info I have read so far.

 

Its fact. For the umpteenth time you don't need to interview (test) every person in the country to get an estimate. If I want to know what % of the population (wherever you live eats) cornflakes for breakfast I don't have to ring every single family there and ask. I can use a representative sample. If I want to know what % of COVID cases were asymptomatic I don't have to ask every single COVID sufferer, I can use a representative sample. Choosing not to get tested makes SFA difference to the outcome - as long as there is still a large % of the population who do get tested and questioned. (Which in fact there are, at least in most western countries.) Exactly the same way as choosing not to answer a survey about Cornflakes makes no difference to the result as long a enough people do choose to answer. Repeat the process at intervals often enough and I can get a time series and make predictions. How do you think market researchers analysis make a living? Reading tea leaves?

 

Negative. Zero fact. Just conjecture. How very scientific, eh?
No matter how much you stomp your foot in here and deflect to snarkiness, you cannot overcome the fallacy of using statistics when you do not have the figure of total persons infected.

See post #116 for further details. I noticed you tucked tail and ran from that one.

 

blah, blah, blah

https://www.frontiersin.org/articles/10.3389/fpubh.2021.645405/full

 

did you bother to even read that?

What do you think that has to do with the stated database issues?

 

My whole buy-in to the this debate form the beginning was the claim that 'Statistics cant be used to make predictions'. Nothing more. Nothing less. There is no question that problems, with the accuracy or reliability of the data used to make predictions are an issue that needs to be addressed/taken into consideration regardless of whether statistic your measuring relates to COVID or cornflakes etc.

Furthermore the problems that beset the accuracy of data accumulated in the US do not by default effect the same data collected in other countries. Given the effectiveness (or rather ineffectiveness) of the US response to COVID inaccuracies in data collection is hardly surprising. Again other counties do not face the same specific issues - necessarily.

So to reiterate, statistics can be sued to make predictions for COVID, provided you have accurate data. The article linked confirmed there was problems with the data but also clearly identified the techniques that could be used to use good data to provide accurate predictions.