Skip to main content

Drowning in Coronavirus Research, Part 50

By July 27, 2020Commentary

Catching up on a bunch of research in a couple of posts.

The original paper that tied in with questions I and others had about how on cruise ships and in general so many exposed people appeared not to get infected, and why even among those who got infected there were so many asymptomatic and mild cases, was the Gomes et al paper in May, which hypothesized that people were ignoring variable in exposure, susceptibility and infectiousness, and that if those were taken into account, population immunity would be achieved at infection levels far lower than those commonly asserted, as low as 10% to 20% of the population.  They have followed up with a second paper.  (Medrxiv Paper)  This one is getting more attention.   In it the group tinkered with their model and applied it to actual epidemic results in several countries–Belgium, England, Portugal and Spain.  Basically the premise is that there is demonstrated variation in exposure, susceptibility and infectiousness and that those more susceptible get infected earlier, leading to faster population immunity.   A far better explanation of the paper is given by Nicholas Lewis here (Lewis Post)  Mr. Lewis had built his own model based on variability in susceptibility and infectiousness early on, and it accurately predicted the outcome of the epidemic in Sweden for example.

We all know what life is like under a lockdown.  Here is a survey from the Netherlands and Belgium.  (Medrxiv Paper)   Over 2000 respondents from each country completed the survey after 8 weeks of lockdown.  Generally not happy campers.  Among other notable findings, 26% said they had avoided needed health care and very large percents reported absenteeism and presenteeism from work.

This paper describes an extensive contact tracing effort in China.  (Medrxiv Paper)   Since China has been very opaque about the extent of the epidemic in that country, we are appropriately cautious about anything coming from researchers there.  The authors examined 1178 infected persons and 15,648 contacts.  159 of the infected person cases were asymptomatic. Note that the infected people were not necessarily index cases, many of them were obviously infected by someone else, so the contact tracing was backward and forward from the likely date of exposure.   Looking at clusters of cases, infectiousness was estimated to peak about 2 days before symptom onset, and most transmission events had occurred by 7 days after symptoms started.  Middle-aged and older adults were more likely to generate cases than children.  Infectious did not appear to vary by age, but susceptibility did.  Household contacts had the highest risk of transmission.  Presymptomatic transmission may have accounted for as many as 62% of cases, while asymptomatic transmission was rare.  Although the effect was small, transmissibility decreased with the number of contacts, which is an odd finding.  While the authors warned about a potential role of children in transmission, they noted that their data actually had so few cases among children that transmission was hard to track.  In fact, there was an extremely low percent of cases among children compared to their proportion in the population.

Here is a study on using cell phone data to track people’s behavior changes near the start of the epidemic in Germany.  (Medrxiv Paper)   The authors built an epidemic model which attempted to incorporate this mobility data.  One interesting finding was that people had changed their behavior before any governmental lockdowns.  Hmmm, maybe you can trust the population?   Naaaah, we need 50 mini-dictators telling us all what to do.  Also interesting, a great example of how modeling is generally misunderstood and worthless.  They attempted to ascertain the effect of certain mitigation measures, such as mask wearing.  They put in parameters that told the model that mask wearing would reduce cases, so what do you think the model told them in response?  That masks reduced cases, isn’t that amazing.  But even then, the model didn’t having mask-wearing result in much in the way of reductions.

Here is another little tidbit related to masks.   The paper discusses an outbreak at a meatpacking plant in Germany.  (SSRN Paper)  Notwithstanding mandatory masking, apparently due to air conditioning and other factors, a number of workers became infected, even though they were 8 meters or more apart from the index case.

We know how poorly the models forecast cases and deaths from coronavirus at the outset of the epidemic.  Well they weren’t any better in forecasting what would happen when the lockdowns were lifted, as this article describes.  (AIER Article)  The models assumed far higher case and death rebounds than have occurred.

And what would any set of research summaries be without more evidence of damage from lockdowns.  This paper describes what the authors called a “brutal” 30% reduction in hospital admissions for heart attacks.  That is not a good thing, because it means most of the missing cases are people who had a heart attack likely died rather than go to the hospital.  (SSRN Paper)   The data was collected from 21 health systems in France, comparing admissions in the four weeks before and after the lockdown was instituted in that country.

Leave a comment