Online Program

Return to main conference page

All Times EDT

Thursday, October 7
Thu, Oct 7, 2:45 PM - 4:00 PM
Speed Session

Seroprevalence of SARS-CoV-2 Antibodies: Representativeness of Nonprobability Serology Samples from Multiple Commercial Laboratories in the United States (309926)

Myrna Charles, CDC 
Ronaldo Iachan, ICF 
*Yun Kim, ICF 
Adam Lee, ICF 
Davia Moyse, ICF 

Keywords: COVID-19 seroprevalence, Sample bias, Representativeness

Measuring seroprevalence of antibodies to SARS-CoV-2 can help estimate the proportion of the population that has been infected with the virus that causes COVID-19. CDC contracted with ICF to examine the percentage of persons with SARS-CoV-2 antibodies in all 50 states, the District of Columbia, and Puerto Rico. Serology specimens were collected from individuals who visited commercial laboratories for non–COVID-19 related reasons. To assess potential bias, ICF and CDC examined the representativeness of the nonprobability sample in terms of race/ethnicity and socioeconomic status (SES) by linking county-level American Community Survey (ACS) data. Due to the absence of individual race, ethnicity, and SES information for serologic specimens, we aggregated the serology data to county-level augmented by ACS data by age, sex, race, income, and education. We analyzed the probability of being represented in the sample population, controlling for socio-demographics in logistic regression models. We provide an assessment of the representativeness of the sample population and the implications for potential biases which cannot be adjusted by the weighting process.