Read on Twitter
Here is a thread to accompany our work published today in @ScienceMagazine ( https://science.sciencemag.org/content/early/2020/12/07/science.abe9728), comparing SARS-CoV-2 IgG antibody prevalence in Manaus and São Paulo. @FAPESP @The_MRC @ItauUnibanco #TodosPelaSaúde
SARS-CoV-2 attack rates are currently high in Latin America (e.g. https://www.nature.com/articles/s41586-020-2918-0) with Brazil reporting >6.5 million cases and >175 million deaths ( https://covid.saude.gov.br/ ). The Brazilian epidemic continues to unfold (data until late Oct):
Using the network of blood banks in the REDS study ( https://www.nhlbi.nih.gov/science/recipient-epidemiology-and-donor-evaluation-study-reds-program), we are measuring anti-SARS-CoV-2 IgG antibodies in blood donors every month (check updates on http://www.caddecentre.org/ )
Our current paper focuses on Manaus, where mortality has been exceptionally high ( https://www.nytimes.com/interactive/2020/07/25/world/americas/coronavirus-brazil-amazon.html), and São Paulo, also experiencing high mortality and where the circulation of SARS-CoV-2 was first detected in Latin America ( http://doi.org/10.1590/S1678-9946202062030)
Testing approximately 1,000 residual blood donation samples each month, we were able to analyze how antibody prevalence changed between Feb and Oct, spanning the course of both local epidemics.
The main figure summarizes the results for Manaus (left) and São Paulo (right)
The main figure summarizes the results for Manaus (left) and São Paulo (right)
A striking finding was rapid antibody waning, following the epidemic peak in Manaus, where the crude prevalence halved over ~2 months. This waning has been confirmed by others working with the same IgG nucleocapsid assay (e.g., https://www.medrxiv.org/content/10.1101/2020.11.02.20224824v1 and https://doi.org/10.1101/2020.08.05.20169128)
After correcting for age-sex biases, sensitivity/specificity of the test, and antibody waning, we estimate that approximately 76% of the population in Manaus became infected by October and 29% in São Paulo (similar to a population-based study in São Paulo https://www.monitoramentocovid19.org/resultados )
These results are a warning of the potential for high COVID attack rates when mitigation strategies are not effective and highlight the importance of face coverings, social distancing and hand washing to stop the spread of infection, which will be reinforced by vaccination.
There are some limitations to note:
First, do blood donors represent the wider population? Children (<16 y.o.) and older adults (>70 y.o.) cannot donate, and donors are likely to differ from the wider population in other ways, including health literacy and attitudes.
First, do blood donors represent the wider population? Children (<16 y.o.) and older adults (>70 y.o.) cannot donate, and donors are likely to differ from the wider population in other ways, including health literacy and attitudes.
Second, although specificity is known to be high, the sensitivity of the Abbott IgG antibody test is difficult to assess for blood donor screening. This is due to spectrum bias ( https://doi.org/10.1136/bmj.m4262). We obtained estimates of assay sensitivity in a cohort of mild cases.
Third, due to antibody waning, through time seroprevalence increasingly underestimates the attack rate. Further characterization of antibody dynamics in mild and asymptomatic cases are required to understand and better correct for this phenomenon.
All the data and code available @datadryad here: https://doi.org/10.5061/dryad.c59zw3r5n
