Answers from an Ex COVID Tester: Test Sensitivity

Are PCR tests picking up too many positive results?

There’s been a lot of questions about the sensitivity of the PCR tests. Unsurprisingly the answer is a nearly resounding yes:

• When we were training with COVID positive samples we tried some troubleshooting samples by removing critical reagents such as digestive enzymes and RNA magnetic beads. The samples still turned up positive, which was extremely surprising at the time.

• We kept old COVID positive samples that had Ct (Cycle Threshold) values in the 10s/20s and used them as positive controls to make sure our extraction worked. We did a 1/30 dilution and kept them in the fridge. The samples stayed positive for weeks.

• A coworker did a serial dilution of a COVID+ sample. Even at a 1/50,000 dilution the sample still turned out positive. (Note: The correct value is 1/10,000. See the Correction below.)

• The Emergency Use Authorization required us to use the specified amount of viral sample (~200 uL). We did not need that much volume for a sample to turn up positive.

So here’s just a handful of things I experienced that indicated that these tests have been highly sensitive. So why could these samples still test positive with such varied testing procedures? The answer may be in the cycles and in exponential growth.

The RT-PCR process sets the cycle threshold during the exponential growth (early) portion of the amplification curve. Because the number of copies is experiencing exponential growth, it means that its growth is based on powers and not factors.

If we start with the assumption that each PCR cycle leads to a doubling of amplicon (amplified DNA- the COVID RNA is converted to DNA during PCR, most likely due to the possibility of RNA degradation during the PCR process) we can see what effects that difference in Ct counts can have on the concentration of copies.

Let’s take 2 samples; one has a Ct Count of 25 while one has a Ct count of 35. Both would be considered positive, and has a Ct count difference of 10. But because we said that each cycle should double the amount of amplicon, this isn’t a difference by a factor of 10, but is a difference by a factor of 2^10 = 1,024! Hypothetically, this would mean that the sample with a Ct count of 25 should contain ~1,024 times more viral particles than the sample that has a Ct count of 35 (remember that the relationship is inversed; higher viral load means lower Ct count).

That is a huge difference, and we can see why people may be concerned about using such a high cutoff. If we compare the 1/50,000 dilution sample with its original concentration we can see that it will take the 1/50,000 diluted sample between 15-16 cycles (2^15=32,768, 2^16=65,536 → this is a value for the 1/50,000 dilution and not 1/10,000. See the Correction below) to reach its original concentration. That would mean that an undiluted sample with a Ct Count around 20 would still test positive (possibly a Ct count around 36).

This makes a huge difference when we look at COVID test results. If we are clumping anyone below a Ct count of 40 or 37 as positive, it creates a wide margin for acceptable viral load, and it becomes even more concerning that we may not have a definitive connection between sample viral load and the viral load needed to infect someone.

At the extreme ends, one person with a Ct Count of 10 vs one with a Ct Count of 35 could be a viral load difference by a factor of 2^25= 33,554,432!

There are many considerations that we have taken here. Every PCR cycle is not likely to lead to a full doubling of the number of amplicon, and human error plays a large role in what happens during the extraction and amplification procedures. We also have to take into account limiting reagents and the maximum amount of viable amplicon.

Even so, this indicates a severe flaw in solely relying on PCR tests. The use of PCR as a diagnostic tool has caused it to come under heavy scrutiny; many of these tests are poorly standardized and may draw the wrong conclusions.

Remember that changing the Ct threshold would mean that you can artificially determine what is considered a positive/negative sample. With a continued push to have wide testing available we may have pushed ourselves into a position where many people who should not be considered positive are, and several people have already argued this position with no real change happening to testing procedures.

PCR tests were never previously used as a diagnostic test; people who present with symptoms of an infection may go see a doctor, be examined, and may get a diagnosis. A doctor may then ask for a PCR test as a confirmatory test, and here’s the issue; we usually obtained a diagnostic test that is confirmed by a PCR test. By solely relying on PCR tests irrespective of symptoms we may be greatly overestimating the value of asymptomatic patients who never show symptoms.

This is part of a larger issue of what I would call translational science/medicine; how well does the data and test translate to real world scenarios. Is there evidence that the Ct Count cutoffs should be put to 40? This would require there to have been research relating Ct count and viral load that would indicate a 40 cycle cutoff as a good measure. We also would need to have evidence that relates viral load to infectious load. Putting it together that means that in order to consider these tests highly valid, we should have data that indicates the minimum amount of viral load needed to be infectious and examine samples for that viral load amount.

We need to take PCR tests with a grain a salt; the precautionary principle may need to be used for those who may be ill as well as when we approach testing procedures.

As an aside, I have on good authority that even in extremely vaccinated populations (>95%) many vaccinated people were testing with a Ct Count in the low teens/high singles, so make with that what you will.

Please remember that many of the ideas here are anecdotal, and please let me know if you have any questions about COVID Testing in the comment section below and I’ll try to write answer them, possibly in separate newsletters.

Correction: I made an error with the 1/50,000 dilution. The actual dilution is 1/10,000. This would change the Ct Cycle difference to fall between 13 and 14 (2^13 = 8,192 and 2^14 = 16,384) but would lean closer to 13 cycles. Note that this correction will only change the Ct difference by about 2 fewer cycles.

Thank you for reading my newsletter. If you enjoy my articles please consider becoming a free subscriber in order to receive notifications.

And share with others who may find these newsletters interesting.

Also, please consider becoming a paid member. The research and work put into these articles takes many hours and being a paid subscriber allows me to continue to do this full time.

Additional Information

1. Qasem et. al. 2021. Coronavirus Disease 2019 (COVID-19) Diagnostic Tools: A Focus on Detection Technologies and Limitations. Taken from https://www.mdpi.com/1467-3045/43/2/53

2. Tali et. al. 2021. Tools and Techniques for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2)/COVID-19 Detection Taken from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8142517/

3. Younes et. al. 2020. Challenges in Laboratory Diagnosis of the Novel Coronavirus SARS-CoV-2. Taken from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7354519/

4. The Basics: RT-PCR. Taken from https://www.thermofisher.com/us/en/home/references/ambion-tech-support/rtpcr-analysis/general-articles/rt--pcr-the-basics.html#3

5. Real-Time PCR: Understanding Ct. Taken from https://www.thermofisher.com/us/en/home/life-science/pcr/real-time-pcr/real-time-pcr-learning-center/real-time-pcr-basics/real-time-pcr-understanding-ct.html]