Skip to main content

Drowning in Coronavirus Research, Part 85

By September 9, 2020Commentary

Ok, I have to lead off with a piece of research that has a surprising conclusion; wearing a mask was actually associated with an increase in the likelihood of infection.  (SSRN Study)   I just want to start by reiterating that 1) I don’t care what people do in regard to masks; 2) I don’t think they make a difference either way; 3) survey data can be very different from what people actually do, that is a well-understood research problem; and 4) I get a schadenfreudic thrill when this kind of result comes out, thinking of how it must make the mask nuts squirm in agony.  The research was done in Vermont and collected survey information from about 1700 primary care patients, around 450 of whom also did PCR testing and an antibody survey.  The overall positivity rate was 2.2%.  The number of contacts with adults and seniors was associated with an increase in infections.  Occupation, living in an apartment, and, wait for it, wearing a facial mask outside of work were associated with an increased number of contacts and thus probability of infection.  Only a small number of people were positive, so while the results are statistically significant, some results are still based on low incidence.

And this paper further deals with issues about testing results.  (Euro. Paper)   The research dealt with the period of infectiousness and the correlation with PCR testing threshholds.  The researchers were attempting to bridge the gap between a PCR positive test and the patient actually being infectious.  This is a very important and well-done study.  Samples from 425 symptomatic cases were taken and the cycle number of the PCR test was compared with the ability of the sample to culture viable virus.  The culture test takes time but is the gold standard for infectiousness.  At a cycle time or number of 30, only 50% of the samples from positive PCR tests cultured viable virus.  The percent dropped rapidly at higher cycle times with less than a third viable at a cycle number of 32.  The peak of culture viability was for samples taken on the day of symptom onset and infectiousness was very low after a week from symptom onset.  Asymptomatic and pre-symptomatic cases had similar results of culturing viable virus as did symptomatic ones.   Now if we only knew how much of a dose of the virus was needed to cause an infection.

This demonstrates yet another example of how “experts” have really failed us during the epidemic.  (CFR Paper)   The paper examines congressional testimony by NIH experts (think Dr. Fauci) who said that CV-19’s case fatality rate would be ten times that of influenza.  Turns out that these experts could not distinguish between a case fatality rate and an infection fatality rate for influenza or CV-19.  The case fatality rate refers only to known, detected cases.  The infection fatality rate, which can be substantially lower, is based on estimates of all infections, including undetected ones.  The IFR of flu is about .1%.  The CFR is a much higher 2% to 3%.  Both measures are looking similar to what will be ultimately be the numbers for CV-19.  (You have to adjust the CV-19 CFR both for the much greater testing than is done for flu, but also for the very liberal attribution of deaths to CV-19.  I believe the apples to apples comparison would show flu to be far more lethal.)  But the expert (Dr. only wear a mask when you are on camera) mistakenly used flu’s IFR as its CFR and then said CV-19 had a ten times higher CFR than flu, contributing to the hysteria and panic.   The author discusses several other issues related to government reactions to the epidemic.  Pretty damning stuff.

This is a large contact tracing study from Taiwan.  (JAMA Article)   There were 100 index patients who had 2761 contacts, and 22 secondary cases.  Most of these secondary cases occurred within 5 days of symptom onset in the index case.  Household and non-household family contacts were much more likely to become infected.  The secondary cases were also more likely among older people.  None of the index cases was under age 11.  None of the asymptomatic index cases were responsible for any secondary cases.

Here is an interesting illustration of the problems with infection testing, using an example of testing Parliament in the UK.  (CEBM Blog)

One more study on testing.  (Medrxiv Paper)   This study is coming at the problem from the other end, false negatives.  It claims that the limit of detection for many PCR tests is too high to detect a number of cases.  Although they use a formula to convert PCR test results to viral load, what they don’t address is what was dealt with in the sample culturing paper summarized above, whether or not those “low” positive cases and associated low viral loads actually represent someone who is infectious, and only culturing can determine that.   Their formula said that viral loads vary widely, from very low to very high.

Join the discussion 3 Comments

  • SteveD says:

    ‘The IFR of flu is about 0.1%. The CFR is a much higher 2% to 3%. Both measures are looking similar to what will be ultimately be the numbers for CV-19.

    According to the same CDC, the IFR of Covid19 is 0.26%. (2.6 times that of influenza) Also IFR can and probably does vary with location and change over time.

    CFR is also a function of how much and broadly you test (the denominator). The majority of people with influenza are never tested for influenza. You cannot logically compare case fatality rates of influenza to Covid-19, since we test far more extensively for Covid-19. If we tested that extensively for influenza (everyone and their dog), there would be far more cases (but probably not that many more deaths) and thus you would have a substantially lower CFR.

    Which also means CFR is not a particularly good measure of anything. (but IFR which is a good measure, is always an estimate)

  • SteveD says:

    Wearing facial masks was not associated with a higher infection rate in this study. There was no significant difference at p < 0.5, merely a trend. Also the number of people positive for CoVid was only 10 which was compared to a very large number of individuals negative for CoVid. Ten is probably too low to make any conclusions. Their final statement that behavior changes need to be studied is of course correct.

    • Kevin Roche says:

      Yeah, I think the authors made a stronger statement than the analysis supports. They were making an inference more related to contact levels than masks and as you point out, small numbers. But given that people keeping pushing masks and saying they will reduce transmission by 25% or more, I can’t resist any opportunity to tweak them. More on that shortly.

Leave a comment