About these surveys
CDC is working with commercial laboratories to conduct large-scale geographic seroprevalence surveys to estimate the percentage of people who were previously infected with SARS-CoV-2, the virus that causes COVID-19 disease. The strategy involves working with state, local, territorial, academic, and commercial partners to better understand COVID-19 in the United States using serology (antibody) testing for surveillance (“seroprevalence surveys” or “serosurveys”). For the surveys, de-identified clinical blood samples are tested for antibodies to SARS-CoV-2.
Initial Ten-Site Commercial Laboratory Seroprevalence Survey
The commercial laboratory seroprevalence survey used blood samples collected from 10 US sites during March to July 2020. These sites were Connecticut, Louisiana, Minnesota, Missouri, New York City, Philadelphia, San Francisco, South Florida, Utah, and Western Washington State. The survey included blood specimens tested for reasons unrelated to COVID-19, such as for routine medical care or a sick visit. CDC tested about 1,800 samples collected from each of these 10 sites, approximately every 3–4 weeks.
Nationwide Commercial Laboratory Seroprevalence Survey
In July 2020, CDC expanded the seroprevalence survey to include commercial laboratories across 50 US states, Washington, DC, and Puerto Rico. The survey uses blood samples submitted to commercial laboratories for reasons unrelated to COVID-19, such as routine medical care or a sick visit.
This survey aims to collect 50,000 blood samples every 2 weeks for a total of more than 1.2 million samples until August 2021. Seroprevalence estimates for both sexes and specific age groups might not be available for some sites because too few blood samples were collected. The results of the nationwide survey cannot be compared directly with results from the initial 10-site commercial laboratory seroprevalence survey because the laboratories from which samples were collected are different, the SARS-CoV-2 serology tests used in each study vary, and the geographic distribution of the two study populations changed. Researchers are investigating what percentage of blood specimens tested have antibodies against SARS-CoV-2, how this varies across geographic areas and age groups, and whether the percentage of SARS-CoV-2 infections has changed over time during the survey.
Weighted seroprevalence estimates with 95% confidence intervals are calculated for each site every 2 weeks. Seroprevalence estimates are adjusted (weighted) to match age and sex distribution to the sample population within each site and then standardized to the age and sex distribution obtained from the American Community Surveyexternal icon population data for each site. This approach accounts for several sources of variability including biases in sample collection, the demographics of each site, and accuracy of different serology tests (i.e., sensitivity and specificity). This variability is reported as 95% confidence intervals.
Interpreting Serology Results from These Surveys
These surveys have limitations to consider when interpreting the results.
The surveys aim to collect specimens nationwide, although results might not represent the geographic and demographic distribution of the population. Blood samples for the study were not chosen randomly and might not be representative of the US population.
People who have blood taken for routine medical care or sick visits might not represent people in the general population because of differences in their overall health, their disease exposure risk, because they sought health care and had a blood test, or because of their immune response to SARS-CoV-2 infection.
Seroprevalence estimates for age and sex might not be available for all places. This can occur when there are too few samples to calculate the estimates for a specific age or sex group. Some results could be false-positive results (the test result is positive, but the person does not really have antibodies to SARS-CoV-2), or false-negative results (the person has antibodies to SARS-CoV-2, but the test doesn’t detect them). False-positive results are more likely to change the survey results if it is an area where the percentage of people previously infected is relatively low. This might cause results to estimate that more people are infected in the community than actually are.
Results from seroprevalence surveys should not be interpreted to mean that people who have tested positive for having SARS-CoV-2 antibodies are immune. We do not know whether having SARS-CoV-2 antibodies provides protection against getting infected again. Other studies are planned to learn more about SARS-COV-2 antibodies, including how long they last, whether they provide protection against getting infected again, and if people get infected again, whether having antibodies can make that illness milder.
While some seroprevalence surveys study risk factors for infection, such as a person’s occupation or underlying health conditions, this seroprevalence survey was not designed to be able to provide that information.
Finally, other seroprevalence surveys are designed to show how long antibodies last in people’s bodies following infection. This survey was not designed to provide that information.