PCR Test FAQs
- What is a PCR test? What is a qualitative test v. a quantitative test?
- What can PCR do and what can it not do? Or rather, what does a positive or negative actually mean?
- How does PCR testing work (in as simple terms as possible)?
- What is meant by ‘viral burden’?
- How foolproof is the home self-swabbing process? What are some common errors and what effect would they have on the accuracy of the test?
- Are there controls to keep a sample from being contaminated or any other external events that could affect the results?
- Why does cycle threshold matter in the context of infectiousness? Can you explain the paragraph below?
- What does ‘PCR test performance’ mean?
- What are sensitivity and specificity? Why do they matter, in practical terms?
- What do you know about a person when they get a positive or negative result? Besides just testing positive or negative, what does a positive result indicate? That they are sick? Infectious? At risk for severe outcomes?
- What is the problem with having different positivity thresholds when diagnosing C19? For instance, what happens if one person gets a positive test at a CT of 45 and one at 30? Do these results mean different things, even if they are both positive?
- If we wanted to standardize Ct to only detect people who are infectious, what Ct value would we set the positivity threshold at?
- How would standardisation of the Ct threshold impact the case numbers?
- What do false positive and false negative mean?
- How do these impact at scale?
- How does the prevalence of disease within a population affect false positives & negatives?
1. What is a PCR test? What is a qualitative test v. a quantitative test?
A PCR (polymerase chain reaction) test seeks the genetic material of a virus in a clinical sample, typically a nose or throat swab.
A qualitative test gives a positive/negative result; a quantitative test gives a numeric result. Often a quantitative result is translated into a qualitative one — blood pressure is read as two pairs of values, then translated into ‘low’, ‘normal’ or ‘high’. In the case of PCR, a quantitative test may give a numerical estimate of viral load, whereas a qualitative test will simply give a positive/negative result.
2. What can PCR do, what can it not do? Or rather, what does a positive or negative actually mean?
PCR is incredibly sensitive for seeking what it’s looking for – in this case parts of viral genes.
It can’t, however, tell if the virus is dead or alive. Nor does it tell us how sick the patient currently is. And it can only seek the genes it has been designed to seek.
3. How does PCR testing work (in as simple terms as possible)?
A sample is taken – typically a nose or throat swab. If the patient is infected this will contain the SARS-CoV-2 virus, which has RNA (not DNA) as its genetic material.
At the laboratory, this RNA is converted to DNA, which PCR requires, using an enzyme called reverse transcriptase.
The DNA is then ‘amplified’ through a series of reaction cycles. Each copy becomes 2, then 4, 8, 16, 32 copies, etc. Once enough copies are made, a detectable color signal is triggered.
If there was a lot of virus in the original sample, meaning that the patient was heavily infected, then fewer doubling cycles are needed to reach detectability. If there is little virus, more cycles are needed.
This gives us the concept of a ‘Cycle threshold to positivity’, called a ‘Ct’.
A low Ct – indicating lots of virus present – is typical at the peak of infection, but a high Ct occurs both pre-symptoms in a newly infected patient as well as in a recovered patient, who may carry fragments of the virus’s genetic information for days or weeks after they cease to be infectious.
It is very rare to be able to recover infectious virus if the Ct is >30, but some labs run tests up to 45 cycles. This can increase the hazard of finding traces of lingering virus or simply of ‘false positives’ through test failure.
4. What is meant by ‘viral burden’?
‘Viral burden’ (or ‘viral load’) is the amount of virus in a patient’s airways or respiratory secretions. It can also be used, more rarely, in relation to populations and countries, rather than individuals.
5. How foolproof is the home self-swabbing process? What are some common errors and what effect would they have on the accuracy of the test?
Taking a nose or throat swab is unpleasant, as most of us can attest. So, in self testing, it’s easy to take a ‘bad’ specimen, by not probing deeply or thoroughly enough. A poorly-taken specimen increases the odds that the virus isn’t found even when it is present.
6. Are there controls to keep a sample from being contaminated or any other external events that could affect the results?
A well-run laboratory should
- be accredited
- run positive and negative controls with each batch of samples
- take great care to avoid sample cross contamination
- participate in quality assurance scheme
- should run internal quality assurance, for example by re-testing some of yesterday’s samples today and making sure the results remain the same
7. Why does cycle threshold matter in the context of infectiousness? Can you explain the paragraph below?
“Contagiousness is generally thought to be broadly dependent on the degree of the viral burden carried. Only people who shed complete viruses are infectious. The viral burden can be estimated by the clinical course, medical and drug histories, and in the case of PCR by reporting the CT. The lower the CT, the higher the likely burden. Without such information, no one can be reasonably labeled as infectious.
In addition, comparison across “positive” laboratory results can only be made if some standardisation has been carried out across the different PCR tests available. Standardisation requires knowledge, coordination and follow-up. Without it, adding figures of “cases” makes as much sense as adding apples and pears.”
The higher the Ct – the number of PCR cycles that must be run before the test trips positive – the less virus there was in the sample, and therefore the less likely the patient is to be infectious.
As a rule of thumb, it’s very hard to recover infectious virus if the Ct is >30.
A wrinkle is that a high Ct occurs both early in the infection, as a viral burden begins to rise, and late, when the patient is recovering, but viral fragments persist.
With a PCR alone, and without a clinical history, these two circumstances can’t be distinguished… But, in the first case the patient is becoming infectious whereas, in the second, they’ve ceased to be infectious.
8. What does ‘PCR test performance’ mean?
It’s a rather loose term. What matters are ‘sensitivity’, ‘specificity’, and the prevalence of the pathogen in the population being tested. Taken together these determine ‘performance’.
9. What are sensitivity and specificity? Why do they matter, in practical terms?
Sensitivity is the proportion, among all ‘true’ positives, that the test correctly scores as positive.
Specificity is the proportion, among all ‘true’ negatives, that the test correctly scores as negative.
These are the two standard measures of any test’s accuracy. But a test’s ‘performance’ also depends on how prevalent the pathogen is in the population being tested.
10. What do you know about a person when they get a positive or negative result? Besides just testing positive or negative, what does a positive result indicate? That they are sick? Infectious? At risk for severe outcomes?
By itself the test tells us that the virus, or fragments of the virus, are present in the patient’s sample. It also can provide, from the Ct, a measure of the quantity of the virus present, which relates to infectiousness.
It doesn’t tell us about the patient’s prognosis nor – if we see a high Ct – whether they are in the early stages of infection, or are recovering, though this is critically important concerning infectivity.
Ideally PCR tests should be used together with a clinical examination of a patient and clinical history – standard medical procedure. Tests used in isolation, with no proper examination of the patient, provide an incomplete picture.
11. What is the problem with having different positivity thresholds when diagnosing C19? For instance, what happens if one person gets a positive test at a CT of 45 and one at 30? Do these results mean different things, even if they are both positive?
If one lab counts a test positive only if the Ct is <30 and another counts them positive up to a Ct of 35, then there will be cases that the first lab calls negative and the second calls positive.
12. If we wanted to standardize Ct to only detect people who are infectious, what Ct value would we set the positivity threshold at?
The optimal value is around 30. It is very rare to be able to recover viable virus if the Ct is higher. But, there are potholes: some tests are ‘black box’ systems and the user only sees a +/-, not a Ct.
Some tests, very reasonably, seek more than one viral gene, reducing the risk of false positives but leaving an interpretive problem when one gene has a Ct of 27 whilst Cts for the others are 33 and 34. Is that positive or negative? Difficult to say.
13. How would standardisation of the Ct threshold impact the case numbers?
That depends on what Ct is adopted relative to the diversity presently used. Standardisation is further complicated by different labs using different tests, some perhaps more efficient than others.
14. What do false positive and false negative mean?
A ‘false negative’ is a patient who is infected, but for whom the test delivers a negative result.
- Perhaps they had a badly taken sample
- Perhaps the virus was hiding in some atypical body site
- During the first wave, intensive care units saw some patients, with clinical COVID, from whom they repeatedly failed to obtain positive results
A false positive is a patient who is not infected but for whom the test delivers a positive result.
- Perhaps the lab ran the test up to an higher than normal number of PCR cycles
- Perhaps the patient had a virus a few weeks ago, recovered, but still has non-infectious viral fragments lingering
- Perhaps the lab cross contaminated specimens that they were testing
15. How do these impact at scale?
PCR tests were designed to be used together with a proper examination and clinical history.
But this is not how they are being used.
This creates a particular problem when they are used, at scale, in a population where only a small minority are infected.
16. How does the prevalence of disease within a population affect false positives & negatives?
Remember, first, those two classic measures of accuracy:
- Sensitivity is the proportion, among all ‘true’ positives, that the test correctly scores as positive.
- Specificity is the proportion, among all ‘true’ negatives, that the test correctly scores as negative.
Suppose we have a test with 99% sensitivity and 99% specificity. That sounds pretty good.
And it is pretty good if we apply it to a population with lots of infection – say 1000 people who have respiratory symptoms, and 200 of whom really do have SARS-CoV-2.
In that case, the sensitivity and specificity predict:
- Of the 200 true positives our test will find 198 (99% x 200)
- We miss 2 true positives (200 minus 198, or 1% x 200)
- Among the 800 true negatives we will correctly identify 792 (99% x 800)
- We misidentify 8 negatives as infected (800 minus 792, or 1% x 800)
- Though there are only 200 true positives, 206 positives are identified
- Though there are 800 true negatives, only 794 negatives are identified
So far so good. But it’s quite different when we test 1000 uninfected folks – prospective travelers for example, or the worried well, or those in recent contact with symptomatic people. Here let’s say only 1% (10 of the 1000) are infected.
- Our test should find all 10 infected cases (99% x 10), though it has a 10% risk of missing one
- We don’t miss any true positives (10 minus 10, or 1% x 10)
- Among the 990 true negatives we correctly identify 980 (99% x 990)
- We misidentify 10 negatives as infected (990 minus 980 or 1% x 800)
- Among the total 980 negatives identified, all are true negatives
- Among the total of 20 positives identified, only 10 (50%) are true positives
In short, our test was good when we applied it to a symptomatic population likely to be infected but, once we used it for population screening, it has a 50% false positive rate. In the jargon, ‘the positive predictive value is only 50%’. In practice this can mean lives unnecessarily disrupted when false positives are found. The answer, when disease incidence is low, is to re-test positives to confirm; however, this is not routinely done.