Are SARS-CoV-2 seroprevalence estimates biased? Pre-print with @sakitakahashi1 and @bgreenhouse1 @EPPIcenter_UCSF. Short answer: We need data on the sensitivity of serological assays to detect asymptomatic and mild infections! osf.io/y3fxt Thread 1/10
Growing evidence suggests that asymptomatic and mild SARS-CoV-2 infections, together making up over 95% of all infections, may be associated with lower antibody titers than more severe infections. 2/10 bit.ly/3dkUoLh 
Yet, many of the serological assays that are currently available have been optimized and validated using only samples from PCR+ severe and/or hospitalized patients as positive controls. 3/10 bit.ly/3cc1AIg
We performed a simple simulation to illustrate that, if antibody levels depend on disease severity, assay sensitivities determined using mostly severe infections (Panel A, sets 2 & 3) will overestimate the actual sensitivity of the assay in the general population. 4/10
Thus, estimates of the population seroprevalence obtained using these biased sensitivities would underestimate the true seroprevalence. (Panel B) The magnitude of the bias will depend on how much sensitivity varies with infection severity. 5/10
Similarly, including only recently infected individuals as positive controls may be problematic, as antibody levels are known to wane over time. This will become especially important as the pandemic progresses and more individuals will have been infected longer ago. 6/10
These results suggest that we urgently need to: (1) Quantify the extent to which the sensitivity of the assays used in ongoing serosurveillance studies varies with disease severity and over time. 7/10
(2) Conduct studies characterizing the long term kinetics of antibody responses to SARS-CoV-2 across the severity spectrum; 8/10
(3) Revisit reporting requirements for performance characteristics of SARS-CoV-2 serological assays. At minimum, the characteristics of the included validation sets should be described. 9/10
Correct interpretation of SARS-CoV-2 seroprevalence studies will not be possible until we know the sensitivity of serological assays to detect mild and asymptomatic infections and infections occurring months beforehand. 10/10
