Skip to main content

Drowning in Coronavirus Research, Part 23

By June 17, 2020Commentary

This is another antibody survey from Brazil.  (Medrxiv Paper)   The country has a series of these random surveys going and this one is from a state in the country and covers some large and smaller cities.  About 5900 individuals were tested in the middle of May.  In the larger cities, the positive rate was 2.1% and in the smaller ones it was .26%.  People who lived with people who tested positive were much more likely to also be positive.

This seroprevalence survey comes from Spain.  (Medrxiv Paper)    The survey looked at both a random sample, and patients in the primary care setting who might have had symptoms but did not have a positive infection test.  311 people participated in the random sample part and 5.47% had a positive test for antibodies.  Of the 644 symptomatic patients, 38.5% had a positive antibody test.

Transmission of coronavirus in Germany was analyzed in this paper.  (Medrxiv Paper)   Looking at Hamburg, the researchers used contact tracing and viral genome variability to analyze the transmission pattern.  The initial case was a traveler returning from Italy, who then infected 2 contacts, but only 2 out of 132 contacts, and so on.  A number of the subsequent infections resulted from within household transmission.

Many states have unfortunately used their own models to guide public policy.  Connecticut had one, described here.  (Medrxiv Paper)   It was slightly more sophisticated than the current version of the Minnesota model in its design.  The model was county based, which is good for avoiding one size fits all results.  It used asymptomatic, mild and severe buckets and had only the severe cases needing hospitalization or dying.  They used a separate compartment for congregate living settings, which is really smart.  They varied the percent of asymptomatics in the scenarios.  They did appear to use a pretty standard contact model but adjusted some aspects of the contacts in a way that seemed realistic.  They used a relatively high initial transmission rate.  And as they calibrated the model to actual results, it looks good initially, but then has large peaks that seem unrealistic.

This article is a good summary of what is known about immunology and this strain of coronavirus.  (Cell Paper)   It is long and detailed and demonstrates how the virus attempts to thwart the immune system, and what its vulnerabilities are.

Mask use has some uncertainty of scientific evidence.  This study in Health Affairs purports to show a clear benefit.  (HA Article)  The author used a before and after study design of mandates on wearing masks in public places in several states that had mask mandates versus those that did not and said mandated mask wearing caused a one or two percentage point reduction in case growth rates.  The study, however, doesn’t say anything about actual compliance with mask mandates, or about how many people in non-mandate states were wearing them.  In the absence of that data, and an unclear attempt to control for all other relevant variables, including the mandating of other social distancing requirements at the same time, the analysis does not seem to clearly suggest that the decline wasn’t going to occur anyway or that it could actually be attributed to mask wearing.

Join the discussion One Comment

  • Matt R says:

    Re: The Mask Study, would you be willing to add a bit more commentary about how the 1-2% “benefit” sits both in the context of the confidence interval (was it AT Least 1-2%, sitting outside of the confidence interval, for example) and also how the 1-2% compares to the growth percentage? In otherwords, if it moves the growth rate from 5% down to 3%…that’s a 20% reduction, wow! If it moves it from 75% down to 73% that’s only a 3% reduction… So basically curious if you can tell us a bit more about both statistical and practical significance (although you do a great job of pointing out that is all moot given the poor study design/ lack of control).

Leave a comment