Skip to main content

Drowning in Coronavirus Research, Part Eleven

By May 16, 2020Commentary

First up, an interesting paper which focuses on how to end the lockdown in the UK,  but in the course of that analysis estimates a much larger percent of cases in London than is shown by infection testing.   (Inf. Dis. Paper)     The authors estimated that were 200 times more cases than identified by positive test results.  If that were true, 29% of the population may have had the virus.

Another paper covers coronavirus antibody development.  (Medrxiv Paper)   You will recall that we all have likely been exposed to one or more of several commonly circulating coronavirus strains and that most of us develop some antibody response to these strains.  The authors were trying to ascertain if these pre-existing antibodies might be protective against the new strain wreaking havoc on the earth’s population.  The protein area used by the new coronavirus to gain entry to cells (the “spike” protein)  has two components, one of which is similar to other strains and one of which is fairly different.  First and importantly, almost all patients with confirmed coronavirus 19 disease had strong antibody responses.  Second, people with pre-existing coronavirus strain immunity often had antibodies effective against the new strain, particularly the more common part of the spike protein, but also other parts of the virus used for cell access.  So there does appear to be some protective effect against the new strain of coronavirus from development of antibodies against the prior strains.  In addition, another component of the immune system, T helper cells, also show cross-action against the new strain.  That could explain why many exposed persons don’t get infected, and why so many infections are asymptomatic or mild.

Yet another meta-analysis (a review of existing research, rather than an original research effort) tries to ascertain levels of asymptomatic coronavirus infection and the likelihood of transmission by asymptomatic individuals.   (Medrxiv Paper)   After screening 571 articles on studies, the authors included five.  Two were from China, so count me as dubious.  In any event, the proportion of asymptomatic cases ranged from 6% to 41%.  The lowest estimate was from the Chinese study, so see the comment above.  The underlying studies generally were only testing adults, in some cases with an average age over 70.  Like most meta-analyses, this one is only as good as those underlying studies, so not very good.

Here is another antibody study, among workers in two localities in Croatia.  (Medrxiv Paper)   The researchers tested about 1494 workers and found antibodies in 1.27%.  It wasn’t a random sample and I had a little trouble tracking infections and population levels, but it appears that even at this low level of antibody development, the number of undetected cases is multiples of reported cases.

An article in the Journal of the American Medical Association Network Open discusses the experience of bordering counties in Iowa, which had very limited mitigation of spread measures, and Illinois, which has one of the most extreme versions.   (JAMA Article)   The authors attempted to compare trends in the counties before and after the actions in their respective states.  This is not a particularly good design, largely because there are too many unknowns to ascribe anything to stay-at-home or business shutdown orders.  The authors, for example, did no actual work to analyze changes in contact patterns, nor did they look at testing patterns in the specific counties, but only in the respective states at large.   There also was a general very low level of cases, which limits the statistical validity of the analysis.  And gee, guess what, with all these limitations, they found that Iowa had 217 more cases than it would have had with a harsher lockdown.  Some of their sensitivity analyses were not even statistically significant.  And they mention near the end of the report that one of the counties in Iowa had a meatpacking plant which had an outbreak.  That kind of clustering screws your analysis.  But most importantly, the increase in cases means nothing.  Were there more hospitalizations, were there more deaths?  You can look at the Iowa and Illinois experience and the answer is obvious.  Iowa did just fine, with one-third the per capita death rate of Illinois.  Some researchers are just determined to prove that we all should be locked down indefinitely, for our own good, of course.

Leave a comment