ABSTRACT
Background: Understanding the cumulative incidence of SARS-CoV-2 infections in the United States has been limited by asymptomatic infections, waning antibodies after natural infection, incomplete case ascertainment and reporting, and limited representative samples. We conducted a probability survey of US households to measure SARS-CoV-2 infection and immune response and to estimate the cumulative incidence of SARS-CoV-2 infection. Methods: A multistage random sample of US postal addresses were mailed a kit to self-collect an anterior nares swab and a dried blood spot (DBS) sample from August to December 2020. Specimens were tested by EUA-approved PCR and serology tests. Weighted estimates of antibody prevalence, together with historical patterns of antibody waning, were used to estimate the cumulative incidence of SARS-CoV-2 infections, the diagnosed fraction, and infection fatality ratio (IFR). Weighted estimates were used to calculate prevalence ratios comparing demographic, geographic, and clinical subgroups. Results: 37,056 kits were mailed to sampled US households. Overall, 5,666 surveys were completed by December 8, 2020;of these, 4,654 also returned a DBS specimen with a valid antibody result. Overall participation rate was 11.8%. We estimated 39,421,841 (95% credible interval (CrI): 33,759,801-43,958,068) total infections by October 30, 2020, an estimated diagnosed fraction of 17% (95% Crl: 15-21%) and an estimated IFR of 0.64% (95% CrI: 0.58-0.75%). Daily seroprevalence peaked by Sept 2020 and remained stable through November 2020 due to a balance of waning antibodies and new infections (Figure). Non- Hispanic Black (PR: 2.2;95% confidence interval (CI): 1.2-4.0) and Hispanic (PR: 3.1, CI: 1.8-5.3) respondents were more likely than White non-Hispanic to have laboratory evidence of prior SARS-CoV-2 infection. Prevalence was also higher among those living in metropolitan areas (PR vs non-metropolitan areas: 2.5, CI: 1.3-5.0) and among those reporting cold or flu symptoms (PR: 2.6, CI: 1.6-4.1) or loss of taste or smell (PR: 12.8, CI: 8.5-19.4) since January 1, 2020. Conclusion: We report the results of the first national probability sample of US households to assess the prevalence of antibodies to SARS-CoV-2 and cumulative incidence. As of October 30, 2020, about 1 in 8 US residents aged ≥18 years had been infected with SARS-CoV-2, and about 1 in 6 of those had been diagnosed. Household-based probability surveys provide a minimally biased benchmark to characterize epidemic dynamics.
ABSTRACT
Background: Developing representative estimates of COVID-19 vaccine acceptance will be essential to public health planning as the vaccine supply moves towards sufficiency in meeting initial levels of demand. We conducted a national probability household survey to assess vaccine willingness and history of SARS-CoV-2 infection based on antibody response. Methods: Study materials were sent to an address-based sample frame that includes nearly all residential addresses in the US. Participants completed a behavioral survey and dried blood spot (DBS) specimen collection for SARS-CoV-2 antibody testing during the study period, August 9 - December 8, 2020. Vaccine willingness was measured with a 5-point Likert scale item with responses ranging from “Very unlikely” to “Very likely.” Sample weights were calculated and applied to descriptive statistics and prevalence ratios (PR). We categorized persons as either Ig negative, Ig positive and aware of prior COVID-19 infection, or Ig positive and unaware of prior COVID-19 infection. Results: A total of 4,654 respondents completed the survey and had a valid antibody test result, representing 242,875,582 US adults. Overall, a substantial proportion, 32% (76,967,749 adults), were unsure or unwilling to receive a COVID-19 vaccine. Many groups at increased risk for SARS-CoV-2 had higher proportions unsure or unwilling, including Black (46%) relative to White (30%, p<.001) race, persons working outside home (38%) relative to at home (21%, p<.001), and smokers (44%) relative to nonsmokers (29%, p<.001) (Table 1). Dissonance between transmission risk and vaccine willingness was also observed in biologic data. Persons Ig positive (previously infected) and unaware of their status had a higher point estimate of unwillingness to be vaccinated (39%) than persons Ig negative with no history of infection (31%, p=.28). Overall, we estimate 12% (29,241,030 adults) were very unlikely to be vaccinated, 7% (15,729,748) were somewhat unlikely, 13% (31,996,971) were unsure, 19% (44,958,518) were likely, and 50% (119,820,865) were very likely. Conclusion: In the first national probability survey with biomarker data, we demonstrated that many groups with higher risk for COVID-19 infection had lower willingness to take a COVID-19 vaccine. This finding is in accordance with pre-existing fault-lines of inequity in our society. Substantial vaccine uptake promotion is needed, and should be targeted to address inequities correlated with vaccine willingness.