17th May 2020
Serological studies I’m ignoring: Chelsea, Dutch, Danish, Iran etc.
These studies don’t meet the most basic tests
We did extensive internal validation at MGH of this assay before we used it in the field. We've tested 377 negative controls – only 2 false positives. Neg ctrls included many samples in which we documented common coronavirus antibodies & samples with other respiratory illnesses
— Tyler E Miller, MD, PhD (@TylerEMiller) April 22, 2020
A small study in the Boston suburb of Chelsea has found the highest prevalence of antibodies so far. Prompted by the striking number of COVID-19 patients from Chelsea colleagues had seen, Massachusetts General Hospital pathologists John Iafrate and Vivek Naranbhai quickly organized a local serology survey. Within 2 days, they collected blood samples from 200 passersby on a street corner. That evening, they processed the samples—and shared the results with a reporter from The Boston Globe. Sixty-three were positive—31.5%. The result carries several large caveats. The team used a test whose maker, BioMedomics, says it has a specificity of only about 90%, though Iafrate says MGH’s own validation tests found a specificity of higher than 99.5%. And pedestrians on a single corner “aren’t a representative sample” of the town, Naranbhai acknowledges. [Source]
IFR: 0.08 but only for people below 70.
- Its sample is skewed because for the blood donor issue
- It excludes those over 70, although the prevalence of disease (not deaths) is the focus, so that should not affect results too much.
Another serology study, in the Netherlands, produced a similar figure for antibody prevalence that was revealed in the country’s House of Representatives on 16 April. Hans Zaaijer, a virologist at Sanquin, the Dutch national blood bank, who helped lead the study, says the team used a commercial test, which “consistently shows superior results” in validation studies, but didn’t provide more details. The results made it clear that the country was not yet near the “herd immunity” that some had hoped for. Nevertheless, the government said on 21 April that it would start to lift some restrictions in the coming weeks, opening elementary schools and allowing children’s sports teams to practice. [Source]
Sample is excessively skewed.
In a population-based cluster random sampling design through phone call invitation, a total of 196 household including 552 subjects agreed to participate in this study.
What about children?
Households are not necessarily a useful way to estimate broader infection.