PCR Testing Is Not What You Think It Is

What a PCR test actually does

In 1985 the team around Kary Mullis published the method. They took a tiny amount of DNA, added short primers that matched one chosen stretch of the beta-globin gene, and ran an enzyme through heating and cooling cycles until that stretch had been copied about 220,000 times [1]PubMed. That is the entire mechanism. Each cycle roughly doubles the target. Thirty cycles is about a billion copies. Forty cycles is about a trillion. The machine stops counting when the signal crosses a line, and the number of cycles it took to get there is the cycle threshold, written Ct. Low Ct means there was a lot of the sequence to begin with. High Ct means there was almost none, and the machine had to copy it many more times to see it. Mullis said the same thing in his own words at a 1997 event in Santa Monica, recorded and archived: the test is a process used to make a lot of something out of something, and it does not tell you that you are sick. He was speaking about HIV testing at the time, and he died in 2019, before the pandemic, a point the fact-checkers make correctly. The sentence still describes the method, because the method has not changed.

One method, not one disease

The video that prompted this post opens on a list: PCR is used for genetics, infectious disease, parasites, crime scenes, paternity, mold, cancer and venom. That list is not an exaggeration. Aetna, one of the largest US insurers, keeps a policy bulletin on what PCR testing it will pay for. The bulletin runs to dozens of organisms and genes: Chlamydia and gonorrhea, tuberculosis, hepatitis B and C, HIV in infants, measles, mumps, malaria species, Candida, leukemia and lymphoma gene rearrangements, BRCA and Lynch genes, HLA typing for transplant, and more. It also lists dozens of uses the insurer calls experimental, from Lyme disease to urinary-tract panels. One method. Hundreds of targets. The method does not know which one you ordered. This matters because the public was taught to think of it as a test for a disease. It is a test for a sequence. Whether finding that sequence means you have a disease, are contagious, or are even carrying something alive is a separate question, answered by a different kind of evidence, and the test's own instructions say exactly that.

The test's own instructions: RNA is not the same as infectious virus

The CDC wrote the instructions for the first US SARS-CoV-2 PCR panel, and the FDA authorized them. Under Limitations, the document says, in its own words: detection of viral RNA may not indicate the presence of infectious virus or that the virus is the causative agent for clinical symptoms. Under the interpretation of results it adds that the agent detected may not be the definite cause of disease. That is the test's label, and it has said this since 2020. The first SARS-CoV-2 PCR workflow, published by Corman, Drosten and colleagues in January 2020, was designed before any virus isolate was available, built from synthetic nucleic acid and the genetic closeness to the 2003 SARS virus [2]DOI. The protocol later drew a formal retraction request from a group of outside scientists; the journal reviewed it with five laboratory experts, rejected the retraction, and updated the authors' conflict-of-interest declaration, which it acknowledged had been incomplete. Both halves of that are on the record, and both are linked below.

The cycle count is the whole story

Here is the part nobody put on your result sheet. The laboratory in Marseille that ran one of the largest culture series in the world took 3,790 PCR-positive samples and tried to grow the virus from every one. At a Ct of 25, about 70 percent grew. At a Ct of 30, 20 percent. At a Ct of 35, the value that laboratory used to call a result positive, under 3 percent of samples grew anything at all [3]PubMed. A Canadian group found no growth in any sample with a Ct above 24 or taken more than eight days after symptoms began, and wrote plainly that RT-PCR detects RNA, not infectious virus [4]PubMed. Public Health England found culture succeeded in only 8 percent of samples with a Ct above 35, and found no difference between people with and without symptoms [5]PubMed. The Marseille group concluded that patients above a Ct of 33 to 34 on their system were not contagious and could leave isolation [7]PubMed. Most US laboratories in 2020 called a result positive at up to 40 cycles. When the New York Times obtained cycle data from Massachusetts, New York and Nevada that August, it found that up to 90 percent of the people testing positive in those data sets carried barely any virus, and a University of California virologist said any threshold above 35 was too sensitive. The cycle count was known to the lab and almost never reported to the patient.

• Ct 25: about 7 in 10 samples grew live virus [3]PubMed.
• Ct 30: about 2 in 10 [3]PubMed.
• Ct 35: fewer than 3 in 100 [3]PubMed, and 8 in 100 in the English series [5]PubMed.
• Ct above 24, or more than 8 days after symptoms: 0 of 90 grew in the Canadian series [4]PubMed.

The WHO told laboratories to report the cycle number. In January 2021.

On 13 January 2021, the World Health Organization issued an information notice to every user of PCR-based SARS-CoV-2 tests. It says the cycle threshold needed to detect virus is inversely proportional to the patient's viral load, that careful interpretation of weak positive results is needed, that where a result does not match the clinical picture a new specimen should be taken, and that most PCR assays are indicated as an aid for diagnosis, not a diagnosis. Its fourth instruction to laboratories: provide the Ct value in the report to the requesting health care provider. It also states that the probability that a person with a positive result is truly infected decreases as prevalence decreases, irrespective of the claimed specificity. Read that again. A year into mass testing, the WHO had to tell laboratories to report the one number that decides what a positive means. Laboratory scientists were arguing the same point in their own journals: whether the cycle threshold belongs on the patient report at all was listed among the open controversies of 2020 [12]PubMed, and reviewers noted that kits with different sensitivities make a single cutoff hard to define, that few studies had related Ct to infectivity, and that the genome alone is not sufficient proof of infectivity [13]PubMed. Research PCR has had a minimum-reporting standard since 2009, requiring assay efficiency and analysis methods to be disclosed so a result can be judged [14]PubMed. Patient reports carried none of it.

Positive for weeks after you could infect anyone

Genetic fragments linger. A systematic review of 79 studies found viral RNA detectable for about 17 days on average in the upper airway and for as long as 83 days at the extreme, while viral load peaked in the first week and no study isolated live virus beyond day 9 of illness [8]PubMed. The first detailed virology of hospitalized patients found throat shedding peaking on day 4 and infectious virus in throat and lung, but never in stool, blood or urine, even when the RNA signal there was high [6]PubMed. A Lancet Respiratory Medicine comment put it in one sentence: no data suggest that detection of low levels of viral RNA equates with infectivity unless infectious virus has been confirmed by culture [9]PubMed. So a person can be ill for a week, recover, and test positive for a month. The test is not lying. It is counting copies of a fragment that is still there. The lie, if there is one, is in the sentence someone else attached to the number.

Negative does not mean clear, either

The same method fails in the other direction, and on a schedule. Pooling seven studies and 1,330 samples, researchers at Johns Hopkins estimated that a PCR test on an infected person comes back falsely negative 100 percent of the time on day 1 after exposure, 67 percent on day 4, 38 percent on the day symptoms begin, and at best 20 percent on day 8, rising again after that [10]PubMed. One in five infected people, tested on the single best day, were told they were clear. This is why a modelling group concluded that for screening, the frequency of testing and the speed of the result matter far more than how sensitive the assay is [11]PubMed. The sensitivity that was advertised as the test's strength was never the thing that decided whether it caught anyone.

The low-prevalence trap, in plain arithmetic

No test is perfect, and the imperfection cuts hardest when the thing being looked for is rare. The UK estimate for operational false positives on swab tests was between 0.8 and 4.0 percent, and the authors warned that at low prevalence those false positives erode the positive predictive value of the test [9]PubMed. The WHO notice says the same thing in its own words. The arithmetic is simple and worth doing once. Take 10,000 people, a true prevalence of 1 percent, a test that catches 80 percent of real infections [10]PubMed, and a false-positive rate of 1 percent, the low end of the UK estimate [9]PubMed. The test finds 80 of the 100 infected people and labels 99 of the 9,900 uninfected people positive. Of 179 positives, 99 are wrong. More than half. That is not a flaw in the arithmetic. It is what any test does when prevalence is low, and it is why every medical textbook says to interpret a result against the pretest probability, which the Lancet authors spelled out as symptoms, history, exposure and the likelihood of another explanation [9]PubMed.

What the record says about the video's hardest claims

The video calls the test the ultimate lie. Here is where the documents land on each piece of that, so you can decide for yourself.

• The video argues the test was never designed to detect disease. The record: PCR was designed to copy a chosen sequence, and its first published use was a prenatal genetic test [1]PubMed. Detecting disease is an inference laid on top of the copy count, and the test's own instructions say the RNA it finds may not indicate infectious virus or the cause of symptoms.
• The video argues the cycle counts were set too high. The record: at the 35 to 40 cycle thresholds in common use, fewer than 3 percent to 8 percent of positives grew live virus [3]PubMed[5]PubMed, and the WHO had to instruct laboratories to report the cycle number a year in.
• The video argues positives were counted that meant nothing. The record: RNA persists for weeks after the window in which any study could isolate live virus [8]PubMed[6]PubMed, and the authors of the false-positive comment say low-level RNA does not equate with infectivity without culture [9]PubMed.
• The video argues the CDC pulled its own test because it was fraudulent. The record: on 21 July 2021 the CDC announced it would withdraw the emergency authorization request for its original panel after 31 December 2021 and encouraged laboratories to adopt a multiplexed method that detects and differentiates SARS-CoV-2 and influenza. The CDC said the test had not demonstrated performance issues and that demand had declined. The alert and the fact-check that quotes it are both linked below. Draw your own conclusion about why a test that could not tell flu from covid was retired as flu season returned.
• The video leans on Kary Mullis. The record: he said the method makes a lot of something out of something and does not tell you that you are sick, in 1997, about HIV testing. He never spoke about covid. Both facts are true, and the first one is about the method, not the year.

What to ask before you accept any test result

This is not an argument against testing. It is an argument for reading a result the way a scientist does. Informed consent is a process, and a process has questions. Before a PCR result changes what you do, ask the laboratory or the clinician:

• What sequence was the test looking for, and what else carries it?
• What was my cycle threshold? If they will not tell you, the WHO says they should.
• At that Ct, what fraction of samples grow live virus in culture [3]PubMed[5]PubMed?
• How many days from exposure or symptoms was the swab taken, and what was the false-negative rate that day [10]PubMed?
• What was the pretest probability: symptoms, exposure, history, another explanation [9]PubMed?
• Has anyone confirmed infectious virus by culture, or is this a copy count alone [4]PubMed[9]PubMed?

Why this belongs on a site about cells

Your body is not a yes-or-no machine, and neither is a laboratory. A number on a report is a measurement with conditions, whether it is a cycle threshold, a ferritin or a thyroid panel. Health Sovereignty is the habit of asking what a number measures, what it cannot measure, and who decided where the line was drawn. The same habit that reads a PCR result honestly is the one that reads your own labs honestly: the mechanism first, the verdict last, and the studies in between. We covered how a story becomes true without anyone adding a fact in the plague post. This is the same lesson, one layer down, at the level of the test itself.