A resident of rural Snohomish County, "Jean" had COVID-like symptoms in late Dec 2019 and has subsequently tested positive in a serological assay. This may have been COVID-19 infection, but it's not certain (or even likely). 1/7
There are three possibilities:
Scenario 1. False positive in the serological assay.
Scenario 2. A non-COVID infection in December, followed by an asymptomatic COVID-19 infection between Feb and April.
Scenario 3. A bona fide COVID-19 infection in December.
2/7
Scenario 1. False positive in the serological assay.
Scenario 2. A non-COVID infection in December, followed by an asymptomatic COVID-19 infection between Feb and April.
Scenario 3. A bona fide COVID-19 infection in December.
2/7
In the first case, we know that the Abbott serological assay used by @UWVirology has a reported specificity of 99.9% (medrxiv.org/content/10.110…). Thus, if prevalence of COVID is ~0% in Dec, we would have a 0.1% chance of a false positive. 3/7
In the second case, we believe that the total attack rate by May 1 in Washington State was ~2% (covid.idmod.org/data/Sustained…). If we assume that roughly 50% of infections are asymptomatic, this gives a 2% x 50% = ~1% chance of asymptomatic COVID-19 infection. 4/7
In the third case, we know from @seattleflustudy work that 0 out of 3600 acute respiratory specimens from Seattle area collected in January were COVID-19 positive. A Bayesian estimator for mean prevalence would be 0.03%. 5/7
Given these probabilities, we can compare hypotheses under equal priors. We get a 9% probability of scenario 1 (false positive in serological assay), 89% probability of scenario 2 (asymptomatic COVID-19 infection) and 2% probability of scenario 3 (COVID-19 infection in Dec). 6/7
Thus, it's possible that Jean had COVID in December, but it's much more likely to have been an asymptomatic infection in the intervening months. 7/7
