In this post I will review a few pieces of research on testing, a couple of which I briefly covered before, and explain why this is such an important issue again.
First up is an article on the interpretation of CV-19 tests. (Medrxiv Paper) The author covers both infection and antibody tests. He expresses concern about the unintended consequences of unlimited testing. No test is perfect–always accurate. In regard to a test’s ability to detect a true positive or a true negative, there is limited information. The circumstances under which a test’s accuracy is verified are not like the real world, where human error and contamination can occur. In regard to PCR tests, the author suggests that sensitivity–the identification of positives accurately, may be only 50% to 70%. The value of a test and its likely accuracy are directly tied to overall prevalence and to the potential risk of the person being tested. People without symptoms in a low prevalence world are highly likely to have a positive result be wrong.
The next paper deals with using viral cultures to assess the accuracy of diagnostic tests. (Medrxiv Paper) This was a review of various studies which attempted to link actual viability of virus with the threshholds set for a “positive” results by various infection tests, including PCR. The point of any diagnostic test for an infectious disease should be to both assess the likelihood that a patient has the disease and also if the patient is infectious and therefore should quarantine or take other actions. PCR is a poor test from this perspective, as it rarely includes information about infectiousness and it can pick up long inactivated virus or viral fragments. The gold standard for determining infectiousness is the ability to culture replicating virus. The authors suggest that so far the studies don’t give good guidance on the relationship between PCR cycle numbers and infectiousness, but that studies could develop an algorithm for that purpose. The authors strongly recommend that WHO or some other group establish standards for PCR testing and for the interpretation of results. They further suggest that the available research indicates that people are not infectious more than 8 days from symptom onset, because they don’t have viable virus after that. They also suggest that the PCR cycle number threshhold be set no higher than 24.
This paper also looked at the utility of PCR tests. (Medrxiv Paper) The author is an economist, but economists are doing far better work than epidemiologists. He notes the same issue about the critical nature of determining infectious status, which isn’t the same as being infected. He points out that if the purpose of a test is to generate usable clinical information, then infectious status needs to be determined. Otherwise we are making decisions on incomplete and misleading data.
And this piece of research demonstrates how such useless testing information can give a completely misleading picture of the epidemic. (Medrxiv Paper) The author uses modeling to demonstrate how PCR tests that are picking up viral fragments for days or weeks after the person has cleared an infection are misleading, as are antibody tests that happen to be conducted before a person has developed the full adaptive immune response. According to the author’s modeling work, it is likely that these problems place the peak of an epidemic 10 days or more after it actually occurred.
I must confess that it is only recently that I have understood that clinicians were not routinely getting information on the likelihood of infectiousness, nor were public health officials getting such data or they were ignoring it. Other than in regard to the individual’s personal health, in an epidemic all that matters is whether and when a person is infectious and able to transmit the pathogen to others. It is astounding to realize that we are collecting all this testing data that basically has no practical utility. But even worse, most public health officials are treating every case as though the person is infectious, when the majority probably aren’t. The complete incompetence with which this epidemic has been handled is horrifying.
“The complete incompetence with which this epidemic has been handled is horrifying.” — Yes it is.
My son is living in off-campus housing at a university in the Northeast, and attending only Zoom classes, yet they are testing him twice-a-week with the RT-PCR method. Preposterous. The kids are living in total fear of testing positive & having their semester derailed. Colleges are reporting lots of “cases” but not yet one hospitalization nationwide. I’m looking to find out what cycle threshold their testing lab is using, but am not sure what answer I will get. I’ve seen elsewhere that it’s been 40 for most of the US, and that Europe uses 35, which could explain why MLB is having constant issues and European Soccer just completed a season without much interruption. These administrators and local public health people are using the abundance of caution principle while assaulting the freedoms of these young people just starting out their adult lives. Time for some balance.
Can you perhaps post something explaining in cycles in detail (or point me to an earlier post on that subject)? I don’t really understand what a cycle is or how it’s measured.
PCR relies on thermal cycling since there are several different reactions which must happen in a specific order but all require different temperatures. One cycle occurs after all the reactions are complete. At that point the DNA has been replicated once. Over each thermocycle the amount of replicated DNA doubles until it reaches the threshold of detection. That is the cycle threshold. Obviously the more cycles required to get to that level, the lower the initial amount of DNA. Which means a 40 cycle test will detect much lower levels of DNA than a 35 cycle threshold. In fact this is such a large difference in sensitivity that we probably can’t compare US case numbers to European case numbers.
It’s the same deal for RT-PCR which measures covid19 RNA but in this case the initial cycle converts RNA to DNA by reverse transcriptase after which regular DNA PCR is performed.
Thanks SteveD. Couldn’t have said it better myself.