Study: One in five Covid-19 cases do not get immunity
thelancet.com
thelancet.com
The conclusion relies on the accuracy of the tests. But what if these tests had a 0.5% false-positive rate? This would explain the cases that seemingly do not develop immunity.
The authors' analysis mentions a 0.02% false-positive rate. However, there are those that estimate that the real-world number (for the UK, at least) is unknown and could be between 0.8-4.0% : https://www.thelancet.com/journals/lanres/article/PIIS2213-2...
Summarised, COVID has been defined as having a positive test. Therefore, by definition, the test has no false positives. We know this is unlikely to be true in reality because in prior PCR testing programmes where disease was defined correctly they did indeed have FPs. In one notorious case the FP rate was 14% and the error rate was 100% - literally every single reported positive was false. In other pre-COVID times PCR labs were challenged with random samples and some labs had 8% FP rates.
Last year I wrote about some of these problems on my blog:
https://blog.plan99.net/pseudo-epidemics-7603b2da839
This second article summarises the first in bullet-point form for those who prefer that to a narrative format:
https://blog.plan99.net/pseudo-epidemics-part-ii-61cb0566960...
None of the problems identified there have been fixed, as far as I know. Every FP rate you see quoted for PCR testing is based on some form of invalid reasoning like "one test might be false, but if we test the same person twice in a row then it's guaranteed to be correct and we can calculate the FP rate from that".
There are some subtleties to consider.
1. FP rates seem to vary a lot across labs and across time. There is no one single FP rate. Lab challenge programmes in the past returned everything from zero FPs to unusably high rates of FPs for what was theoretically the same test and the same challenge samples. This can occur if FPs are driven mostly by cross-contamination.
2. Public health officials seem to routinely mis-use logic in ways that converts false positives into false negatives. FNs are then used to justify ramping up the sensitivity of the test, which creates more FPs, which are interpreted as more FNs, in a loop. My essay provides some evidence for this.
3. PCR tests don't detect live virus. They start with a lysis stage that kills the viruses and breaks open the capsid, so the RNA is exposed and can be detected. By implication the test cannot tell the difference between someone who is currently infected and infectious, and someone who has debris from their own body destroying the virus in their bloodstream. Normally this doesn't matter because PCR tests aren't scheduled by doctors for people who are healthy, but for COVID clinical diagnosis has been cut out of the equation entirely, so anyone who tests positive is considered to be "sick".
It's fair to say that huge quantities of research papers that are predicated upon the correctness of PCR testing are actually in some limbo state of quasi pseudo-science. A test which detects a "disease" which is itself defined as testing positive is an exercise in circular reasoning and all conclusions based on it must be treated with extreme care.
These are different things. The covid case definition is an epidemiological case definition and wouldn’t be used for measuring test sensitivity.
So the sentences I quoted incorrectly conflate two different concepts: incidence/prevalence and clinical accuracy.
What’s screwy is that there needs to be some “gold standard” to measure “true prevalence” of disease so there’s some sort of estimates of this for however they calculate the sensitivity and specificity of these tests.
But it’s not as simple as a tautological “positive test==COVID; therefore impossible to have false positives”
https://www.cdc.gov/coronavirus/2019-ncov/symptoms-testing/s...
"Fever or chills, Cough, Shortness of breath or difficulty breathing, Fatigue, Muscle or body aches, Headache, New loss of taste or smell, Sore throat, Congestion or runny nose, Nausea or vomiting, Diarrhea. This list does not include all possible symptoms."
That's what you'd expect to see if a test with false positives was mandated to be treated as if it had none. Instead of the test being calibrated against a set of symptoms (the usual approach) the set of symptoms becomes calibrated against the test, and as the test has FPs, the set keeps expanding until it becomes a description of more or less any illness.
This is why the few papers that ask questions about test FPs invariably seem to use an invalid definition, usually one based on double testing, i.e. no followed by yes is a false negative.
there needs to be some “gold standard” to measure “true prevalence” of disease so there’s some sort of estimates of this for however they calculate the sensitivity and specificity of these tests
Indeed, but unfortunately no such gold standard exists, as the papers I quoted in the essay explain. Or rather, the test is considered to be the gold standard. I read somewhere that the WHO may finally have changed the official definition of COVID to actually incorporate symptoms ... on the day of Joe Biden's inaugeration. That won't help with people claiming they're politically biased. However at this point such a move seems unlikely to fix anything because it's too late: the official set of symptoms was already derived from test results.
Here’s a systematic review of many tests for accuracy [0] and they describe how different tests determine sensitivity and they aren’t based on the epi definition. They do stuff like chest X-rays and none of them, that I read, rely on symptoms or any circular case definition.
Here’s the surveillance case definition that CDC puts out from back on April 5 [1] and it’s not meant to be “real covid” for healthcare purposes but used to try to estimate incidence. So I think of it as the most likely to be useful.
The lab test accuracy is closest to “real covid” in that they try to measure whether the disease actually exists and corresponds to the test result. For novel diseases there’s no gold standard so different tests attempt different techniques. I don’t think all these different tests are all politically biased especially since many report pretty bad sensitivity (50%).
[0] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7457918/ [1] https://wwwn.cdc.gov/nndss/conditions/coronavirus-disease-20...
It's a good study. They reviewed a lot of papers: 130! Despite that, the paper abstract clearly states it cannot be used to conclude anything because despite reviewing a lot of papers, ALL of them were missing critical study details. Literally despite selecting 816 articles post-dedupe and then whittling it down to 130 articles, and then whittling it down again to 49 for the final paper, none of them were usable because all of them had methodology problems! That's prima facie evidence that the medical establishment has absolutely no idea how to answer a most basic question: what actually is COVID and how accurate are the tests?
We can see the problems I'm talking about by reading further. You mention chest CTs and imply that calibrating the test against them is a common technique, vs relying on circular definitions or symptoms. That doesn't seem to be true. In fact only a single study looked at chest CTs and it had the following problem:
"Ai et al. compared the positive rate of rRT-PCR (59%) with the positive rate of Chest CT in order to draw conclusions about the accuracy of Chest CT for the diagnosis of COVID-19,7 ... In terms of quality assessment, the studies lack specific details as to how patients were classified as having suspected COVID-19 infection"
In other words, the study is useless. They calculate a false negative rate (not an FP rate) yet don't say what definition of COVID they used as the gold standard. This sort of thing seems to be totally rampant in medicine, it's ridiculous.
You said, "none of them, that I read, rely on symptoms or any other circular case definition". I don't understand that statement because the paper clearly says that virtually all of the studies did rely on circular definitions. For example:
"Eight studies ... attempted to determine the accuracy of rRT-PCR by comparing the initial rRT-PCR result to the result after multiple repeated samples from the patient submitted for rRT-PCR testing, which was called the reference standard ... Suo et al. considered a positive result of either repeated measurements of rRT-PCR or serology to indicate a positive test according to the reference standard"
That's a circular definition because the test is being calibrated against itself. The paper concludes:
"Clinical diagnosis by the case definition for COVID-19 used in the early period of the pandemic does not correlate well with positive rates of COVID-19 rRT-PCR"
In other words, COVID as initially detected (unusual pneumonia) is in fact a totally different disease to current COVID, which speaking very pedantically and technically does not actually exist at all: a disease defined as testing positive for the disease cannot be said to really exist in a meaningful, logically coherent sense.
Now, you can get upset at this explanation, but I challenge you to find a better explanation for starting with the exact, but arbitrarily picked false positive rate that leads to a politically desirable outcome.
I’d like to think that antibody test would be the one to use as it can determine whether or not a person has ever contracted COVID. Plus antibody has lower false positive and lower false negative than PCR.
Just a bit unclear as to why one would make a statistical case using short-ranged (short-insight) PCR that could have missed the ones who have already had been infected but recovered and then go on to get reinfected (whitepaper’s missed data points).
Oh wait, a case of FREE (short-insight) PCR tests provided by Dutch government. Never mind.