Drowning in Coronavirus Research, Part Whatever

First, a brief statistical shoutout to the Governor.  At the Thursday briefing you said the test results showed we were moving up the epidemic curve.  They don’t, they show we are doing more testing.  Common mistake.  If you have 10,000 people newly infected on every day of a week, but on Monday you test 200 people, Tuesday you test 500 people, Wednesday you test 750 people, Thursday you test 1000 people, Friday you test 2000 people, Saturday you test 3000 people and Sunday you test 4000 people; there aren’t more people with coronavirus infections every day, there are only more tests.  That is exactly what has happened in Minnesota, we have no idea how many infected people there are because we haven’t been testing everyone.  We aren’t moving up the epidemic curve, we are moving up the testing curve.

I am trying to keep up, I really am.  Here is a somewhat important paper regarding antibodies formed as a result of coronovirus infection.  (Medrxiv Paper)   The strength and durability of antibody response is important both for recovery from the disease and prevention of reinfection and a higher percent of people with antibodies eventually limits transmissibility.  Two kinds of primary antibodies are generated during infection, IgM which tends to develop during the process of infection, and IgC, which typically develops after recovery and provides the sustained protection.  The researchers repeatedly tested 211 Chinese patients, 181 with mild or moderate disease and the remainder with more severe disease.  The amount of antibodies did not seem to be correlated with either the length of disease or the clinical outcome or course.  Those with more severe disease did have more IgC antibodies.  The encouraging suggestion is that it appears all patients developed antibodies.

People are trying very hard to figure out how many, and the characteristics of, the people who have been infected by coronavirus.  Everyone agrees that positive test counts are only picking up a fraction of the cases.  The researchers in this paper combined three methods in an attempt to estimate total infection in the United States by April 4.   (Medrxiv Paper)   As with other papers we have reported on, they used influence-like illness reporting as part of all three methods, basically using different statistical approaches to identify excess cases over typical trends and assuming that those excess cases were due to coronavirus infections.  311,000 cases were identified based on confirmed reporting.  The researchers estimate the actual number was 2.7 million to 8.3 million.  In some states they found under-reporting by 10 to 100 times.  The results are broadly consistent with other research.

Next up, another paper on the actual rates of infection in France.  This study is similar to the one I reported on earlier in the week.  (Medrxiv Paper)   The research was partly designed to validate 4 antibody assays tested by the authors.  They were then tested on 51 hospitalized patients, 209 symptomatic individuals and 200 blood donors.  Testing of pre-epidemic blood samples revealed no cross-reactivity with existing coronavirus antibodies.  The assays all performed well.  32% of the symptomatic individuals were positive for antibodies, suggesting a cause other than coronavirus for many of the symptoms.  Among blood donors, 3% were positive.  Antibody development occurred 5 to 14 days after symptoms appeared and were strongest among those with more severe disease.