ABSTRACT
We estimated monthly cross-sectional seroprevalence rates of anti-nucleocapsid (anti-N) and anti-spike (anti-S) antibodies to severe acute respiratory syndrome coronavirus 2 in two U.S. nationwide studies. The nationwide blood donor seroprevalence (NBDS) study included specimens from blood donors, while the nationwide commercial laboratory seroprevalence (NCLS) study included residual serum specimens tested in commercial laboratories for reasons unrelated to the assessment of coronavirus disease 2019 infection. In September–December 2021, specimens collected from both nationwide studies were tested for anti-N antibodies. In September–October 2021, specimens from both studies within a five-state area were tested for anti-S antibodies. We used raking methods to adjust all seroprevalence estimates by the population distribution of key demographics in included states. Seroprevalence estimates of each antibody type were compared across the two studies for specimens drawn in the same U.S. states during the same time period. Our analysis revealed that over a 4-month period, national NCLS monthly anti-N estimates were 0.5–1.9 percentage points higher than NBDS estimates. In contrast, across five states during a 2-month period, NBDS anti-S estimates were 7.6 and 8.2 percentage points higher than NCLS estimates. The observed differences in seroprevalence estimates between the NBDS and NCLS studies may be attributed to variations in the characteristics of the study sample populations, particularly with respect to health status, health behaviors, and vaccination status. These differences should be considered in the interpretation of seroprevalence study results based on blood donors or commercial lab residual specimens.
IMPORTANCE
This study was the first systematic comparison between two nationwide severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) studies which estimated seroprevalence, or the proportion of the population with antibodies to the virus, using differing convenience sample populations. One study tested blood donor specimens; the other study tested specimens left over from clinical blood tests. The seroprevalence of anti-nucleocapsid and anti-spike antibodies was compared in the same states during the same months with statistical adjustments based on state demographics. Similar anti-nucleocapsid antibody seroprevalence estimates produced by two independent studies using differing convenience samples build confidence in the generalizability of their anti-nucleocapsid findings. Due to high blood donor vaccine rates, blood donor SARS-CoV-2 anti-spike antibody estimates might overestimate general population seroprevalence, an important consideration for interpreting national seroprevalence study results. Furthermore, because laboratory residuals and blood donations are two common sources of specimens for seroprevalence studies, study findings may be informative for other respiratory virus seroepidemiology studies.
KEYWORDS: SARS-CoV-2, anti-nucleocapsid antibody, anti-spike antibody, blood donors, commercial laboratory
INTRODUCTION
In July 2020, the U.S. Centers for Disease Control and Prevention (CDC) began two nationwide seroprevalence studies designed to estimate the proportion of the U.S. population with antibodies to the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The nationwide blood donor seroprevalence (NBDS) study estimates the seroprevalence of two types of antibodies: anti-nucleocapsid (anti-N) and anti-spike (anti-S) antibodies. Anti-N antibodies are produced in response to infection with SARS-CoV-2, but not to vaccination with any vaccines currently approved in the United States (1). Anti-S antibodies are produced in response to both SARS-CoV-2 infection and vaccination (2). The nationwide commercial laboratory seroprevalence (NCLS) study primarily estimated anti-N seroprevalence during the study period. However, anti-S assays were performed on specimens from some participants residing in Indiana, Maryland, New Jersey, New York, and Virginia during September–October 2021.
NBDS includes specimens from blood donors while NCLS includes residual serum specimens from persons who had blood specimens tested by commercial laboratories for clinical care unrelated to assessment of coronavirus disease 2019 (COVID-19) infection. Both studies use convenience sampling, which can limit generalizability. Compared with the general population, blood donors are more frequently non-Hispanic White and college-educated. Blood donors also tend to practice healthier behaviors (e.g., nonsmokers) (3, 4). In comparison, participants in NCLS more frequently sought clinical care involving commercial laboratories; these participants may have better health care access or greater prevalence of chronic conditions requiring frequent monitoring than the general population. The ICD-10 codes associated with NCLS specimens most frequently included health supervision visits and common chronic conditions requiring monitoring visits (e.g., hypertension and type 2 diabetes mellitus), indicating a likely predominance of specimens drawn as part of outpatient care (5).
NBDS anti-N and anti-S results and NCLS anti-N results have been published on CDC’s COVID Data tracker, and national and jurisdiction-level estimates have differed (6). In this analysis, we aimed to compare anti-N and anti-S seroprevalence estimates from NBDS and NCLS using uniform statistical methods and restricting analysis to samples from similar geographic catchment regions during time periods when both studies used comparable antibody assays.
MATERIALS AND METHODS
Data
The NBDS and NCLS are repeated cross-sectional multi-state studies. For the NBDS, specimens were selected monthly from blood donations. For the NCLS, specimens were selected over consecutive 4-week data collection periods from blood specimens submitted to commercial laboratories. NBDS utilized specimens collected by blood donation organizations, excluding COVID-19 convalescent plasma donations, with approximately 2,000 specimens included in each month’s sample from allogeneic donors. These specimens were gathered through pseudorandom or convenience sampling strategies from blood donor regions within each organization’s catchment area. NCLS employed a convenience sample of residual serum specimens obtained from individuals of all ages, initially submitted for routine screening or clinical management from private clinical commercial laboratories with approximately 65,000–80,000 specimens included in each month’s sample. Laboratories excluded specimens associated with any requests for SARS-CoV-2 antibody testing (3, 7). Both studies included data on patient age, sex, residential ZIP code, and specimen collection date. Information on patients’ race and ethnicity was available for NBDS but not for NCLS (3, 7).
NBDS specimens were tested for anti-N antibodies using the Ortho Vitros anti-N total lg assay and for anti-S antibodies using the Ortho Vitros SARS-CoV-2 total lg assay. NCLS specimens were tested for anti-N antibodies using the Roche Elecsys Anti-SARS-CoV-2 N pan-immunoglobulin immunoassay in most U.S. jurisdictions. However, specimens from residents of five states (Indiana, Maryland, New Jersey, New York, and Virginia) were tested for anti-S antibodies, but not for anti-N antibodies, using the Roche Elecsys Anti-SARS-CoV-2 S pan-immunoglobulin immunoassay in September and October 2021. While sensitivity, specificity, and the waning of antibody detection over time can vary by antibody assay, all anti-S and anti-N assays used are direct, total antibody assays with similar characteristics (8).
NBDS and NCLS samples have differences in population and some variation in geographical catchment areas. To optimize comparability, we restricted the analysis to specimens from persons aged ≥16 years (the age of eligibility for blood donation). Specimens were grouped by NBDS study regions, which were defined by state and metropolitan area (3). Study regions were excluded if they contained <100 samples in either NBDS or NCLS for any month in the study period in order to obtain valid estimates (Fig. S1). The final data set for the anti-N seroprevalence comparison included results from 421,424 NBDS and 185,781 NCLS specimens from 42 states during September–December 2021. The data set for the anti-S seroprevalence comparison was limited to 2 months when anti-S data were available from NCLS, September and October 2021, and included results from 19,404 NBDS and 9,280 NCLS specimens from participants residing in five states.
Outcomes
We produced study-specific estimates of monthly anti-N seroprevalence at the national level during September–December 2021, and monthly anti-S seroprevalence for all five states combined during September–October 2021. The difference in anti-N and anti-S seroprevalence between NBDS and NCLS was measured as the NCLS estimate minus the NBDS estimate.
Statistical analyses
We used raking to create sample weights for both the anti-N and anti-S seroprevalence estimates to match the population distribution by key demographics using 2014–2018 state-level American Community Survey data (9). Demographic characteristics included in raking were different between NBDS (age group, sex, race, and ethnicity) and NCLS (age group, sex, and urbanicity) due to available data. Race and ethnicity data were available for blood donors and included in raking due to known differences in these demographics among blood donors compared with the general population. Details of the raking method were previously published (3, 7). Confidence intervals for percentage point differences between the two surveys were calculated using bootstrapping methods (10). We also reported anti-N and anti-S seroprevalence estimates that were not adjusted for population demographics as a comparison with the adjusted estimates. Statistical analyses were conducted in R version 4.0.4 with the survey package (11).
RESULTS
The characteristics of the participants differed between the NBDS and NCLS. For both anti-N and anti-S results, participants in the NCLS were older than the participants in the NBDS (Fig. S2 and S3), and a higher proportion of participants were women (≥60% in NCLS compared to 48%–49% in NBDS). In addition, participants in the NBDS were predominantly non-Hispanic White (85%–86%).
From September to December 2021, the anti-N seroprevalence estimates increased from 25.2% (95% confidence interval [CI]: 24.8%–25.5%) to 30.1% (95% CI: 29.7%–30.4%) for NBDS and from 27.1% (95% CI: 26.4%–27.7%) to 31.4% (95% CI: 30.8%–32.0%) for NCLS (Fig. 1; Table 1). The monthly difference in national anti-N seroprevalence estimates between the two studies ranged between 0.5 (95% CI: −0.1 to 1.1) and 1.9 (95% CI: 1.1–2.6) percentage points.
Fig 1.
Infection-induced (anti-nucleocapsid) seroprevalence in U.S. regions, September–December 2021 (A) and combined infection-induced and vaccine-induced (anti-spike) seroprevalence in five U.S. states, September–October 2021 (B) by the NCLS study and the NBDS study.
TABLE 1.
Infection-induced (anti-N) seroprevalence in U.S. regions, September–December 2021 and combined infection-induced and vaccine-induced (anti-S) seroprevalence in five U.S. states, September–October 2021 by the NCLS study and the NBDS studya
| Month | Anti-N seroprevalence (national) | Anti-S seroprevalence (5 states combined) | ||||
|---|---|---|---|---|---|---|
| NCLS | NBDS | Percentage point differenceb | NCLS | NBDS | Percentage point differenceb | |
| September | 27.1% (26.4%, 27.7%) | 25.2% (24.8%, 25.5%) | 1.9 (1.1, 2.6) | 85.7% (84.2%, 87.2%) | 93.9% (93.2%, 94.5%) | −8.2 (−9.8, −6.7) |
| October | 28.4% (27.8%, 28.9%) | 27.4% (27.0%, 27.7%) | 1.0 (0.4, 1.7) | 86.9% (85.4%, 88.3%) | 94.5% (93.9%, 95.1%) | −7.6 (−9.1, −5.6) |
| November | 29.5% (29.1%, 30.0%) | 29.0% (28.7%, 29.4%) | 0.5 (−0.1, 1.1) | |||
| December | 31.4% (30.8%, 32.0%) | 30.1% (29.7%, 30.4%) | 1.3 (0.5, 2.0) | |||
Means and 95% confidence intervals (in parentheses) are reported.
Percentage point difference was calculated as the seroprevalence estimated from NCLS minus the seroprevalence estimated from NBDS.
From September to October 2021, the anti-S seroprevalence estimates were higher in the NBDS than in the NCLS for all five states combined and increased from September to October in both studies. The monthly difference among these five states slightly declined from 8.2 (95% CI: 6.7–9.8) percentage points in September to 7.6 (95% CI: 5.6–9.1) percentage points in October.
The adjusted and unadjusted estimates for both anti-N and anti-S seroprevalence were similar for both studies (Table S1).
DISCUSSION
NBDS and NCLS are the only ongoing national studies monitoring the proportion of the U.S. population with antibodies to SARS-CoV-2. A comparison revealed similar but slightly higher anti-N national seroprevalence estimates for the NCLS whereas the NBDS had higher anti-S seroprevalence.
Differences in estimates of anti-N seroprevalence between the two studies were minimal at the national level, but the NCLS anti-N seroprevalence estimates were higher than the NBDS estimates. This might be because blood donors are required to go through a health screening which leads to the eligible donors being healthier than non-eligible individuals (12); a person with persistent fatigue or other symptoms consistent with post-COVID-19 conditions may be less likely to be included in this population. In addition, individuals who donate blood tend to have higher educational levels (3), which are correlated with an increased likelihood of engaging in protective behaviors (e.g., social distancing, mask-wearing, vaccination) that reduce the risk of COVID-19 infection (13).
In contrast, the NBDS anti-S seroprevalence estimates were higher than the NCLS estimates for all the states combined, most likely reflecting a higher vaccination rate among blood donors compared with the general population. NBDS self-reported vaccination rate estimates (defined as ≥1 dose of a COVID vaccine) were higher than CDC data for the general population (80% vs 67%, 13 percentage point difference in September 2021; 81% vs 69%, 12 percentage point difference in October 2021) (14). This reflects the difference in the NBDS and NCLS anti-S estimates, suggesting that NCLS anti-S estimates and vaccine acceptance among patients included in NCLS may be more similar to the general population.
This analysis has at least four main limitations. First, restricting the analysis to specific catchment areas and dates increased the comparability between studies, but might decrease the generalizability of findings. Second, race and ethnicity data were not available from commercial laboratories, but these characteristics are known to influence access to medical care, insurance coverage, prevalence of chronic diseases, and ability to miss work (15) due to systemic inequities. Because the racial and ethnic composition of the NCLS sample may differ from the general population, the inability to adjust for these variables estimates may impact comparisons to NBDS estimates. Third, the sensitivity of the anti-S assays may differ after infection with viral SARS-CoV-2 variants, particularly among the unvaccinated or those without previous infection with early variants. Because the nucleocapsid antigen is more conserved, the sensitivity of anti-N assays is less likely to be impacted. A study has shown high concordance between the Roche and Ortho anti-S assays used in these studies, and the theoretical impact of this on the comparison of these studies should be minimal (8). Finally, anti-S assays were available for both the NBDS and NLCS only for a small set of geographic regions in a limited time period. Therefore, these differences in anti-S between studies may not be generalizable to other time periods or geographic regions during the pandemic.
The similar anti-N seroprevalence estimates produced by two independent studies using differing convenience samples build confidence in the generalizability of their anti-N findings. As demonstrated in this analysis, NBDS anti-S estimates might overestimate anti-S seroprevalence in the general population because of high vaccine rates in blood donors, which is an important consideration in the interpretation of national seroprevalence study results. Furthermore, two of the most common sources of convenience samples are blood donations and laboratory residual specimens collected from patients presenting for medical care. Because the NBDS and NCLS had overlapping populations and time periods of study, this comparison of SARS-CoV-2 seroprevalence between the two studies may be informative for other respiratory virus seroepidemiology studies.
ACKNOWLEDGMENTS
The authors would like to thank Kevin Berney at the U.S. Centers for Disease Control and Prevention for creating the maps of the geographic coverage of the samples.
S.-Y.Z.K. served as lead analyst and author of this publication. E.N. conducted background research and writing for this publication. T.J.B. contributed to the introduction and discussion portion of this publication. K.E.N.C. and J.M.J. both served as senior authors who guided the research questions, result interpretation, and writing for this publication.
The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Contributor Information
Kristie E. N. Clarke, Email: vhz9@cdc.gov.
Gabriel I. Parra, US Food and Drug Administration, Silver Spring, Maryland, USA
ETHICS APPROVAL
CDC ethics officials reviewed the study protocols of both the NBDS and NCLS to ensure activities were conducted in compliance with applicable federal law and CDC policy [see, e.g., 45 C.F.R. part 46, 21 C.F.R. part 56; 42 U.S.C. §241(d); 5 U.S.C. §552 a; 44 U.S.C. §3501 et seq.].
SUPPLEMENTAL MATERIAL
The following material is available online at https://doi.org/10.1128/spectrum.00123-24.
Fig. S1, S2, and S3; Table S1.
ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.
REFERENCES
- 1. Centers for Disease Control and Prevention . 2020. Multistate assessment of SARS-CoV-2 seroprevalence in blood donors. Available from: https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/blood-bank-serosurvey.html. Retrieved 14 Jul 2022.
- 2. Centers for Disease Control and Prevention . Commercial laboratory seroprevalence surveys. Available from: https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/commercial-lab-surveys.html. Retrieved 14 Jul 2022.
- 3. Jones JM, Stone M, Sulaeman H, Fink RV, Dave H, Levy ME, Di Germanio C, Green V, Notari E, Saa P, et al. 2021. Estimated US infection- and vaccine-induced SARS-CoV-2 seroprevalence based on blood donations, July 2020-May 2021. JAMA 326:1400–1409. doi: 10.1001/jama.2021.15161 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Patel EU, Bloch EM, Grabowski MK, Goel R, Lokhandwala PM, Brunker PAR, White JL, Shaz B, Ness PM, Tobian AAR. 2019. Sociodemographic and behavioral characteristics associated with blood donation in the United States: a population-based study, September 2019. Transfusion 59:2899–2907. doi: 10.1111/trf.15415 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Clarke KEN, Kim Y, Jones J, Lee A, Deng Y, Nycz E, Iachan R, Gundlapalli AV, MacNeil A, Hall A. 2023. Pediatric infection-induced SARS-CoV-2 seroprevalence increases and seroprevalence by type of clinical care—September 2021 to February 2022. J Infect Dis 227:364–370. doi: 10.1093/infdis/jiac423 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Centers for Disease Control and Prevention . COVID data tracker: antibody seroprevalence. Available from: https://covid.cdc.gov/covid-data-tracker/#antibody-seroprevalence. Retrieved 4 Jan 2024.
- 7. Bajema KL, Wiegand RE, Cuffe K, Patel SV, Iachan R, Lim T, Lee A, Moyse D, Havers FP, Harding L, Fry AM, Hall AJ, Martin K, Biel M, Deng Y, Meyer WA, Mathur M, Kyle T, Gundlapalli AV, Thornburg NJ, Petersen LR, Edens C. 2021. Estimated SARS-CoV-2 seroprevalence in the US as of September 2020. JAMA Intern Med 181:450–460. doi: 10.1001/jamainternmed.2020.7976 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Sulaeman H, Grebe E, Dave H, McCann L, Di Germanio C, Sanghavi A, Sclar V, Bougie DW, Chatelain G, Biggerstaff BJ, Jones JM, Thornburg NJ, Kleinman S, Stone M, Busch MP. 2023. Evaluation of ortho VITROS and roche elecsys S and NC immunoassays for SARS-CoV-2 serosurveillance applications. Microbiol Spectr 11:e0323422. doi: 10.1128/spectrum.03234-22 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. United States Census Bureau . American community survey data. Available from: https://www.census.gov/programs-surveys/acs/data.html. Retrieved 25 Apr 2022.
- 10. Efron B. 1979. Bootstrap methods: another look at the jackknife. Ann Statist 7. doi: 10.1214/aos/1176344552 [DOI] [Google Scholar]
- 11. Lumley T. 2004. Analysis of complex survey samples. J Stat Soft 9:1–9. doi: 10.18637/jss.v009.i08 [DOI] [Google Scholar]
- 12. Atsma F, Veldhuizen I, Verbeek A, de Kort W, de Vegt F. 2011. Healthy donor effect: its magnitude in health research among blood donors. Transfusion 51:1820–1828. doi: 10.1111/j.1537-2995.2010.03055.x [DOI] [PubMed] [Google Scholar]
- 13. Wise T, Zbozinek TD, Michelini G, Hagan CC, Mobbs D. 2020. Changes in risk perception and self-reported protective behaviour during the first week of the COVID-19 pandemic in the United States. R Soc Open Sci 7:200742. doi: 10.1098/rsos.200742 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Busch MP, Stramer SL, Stone M, Yu EA, Grebe E, Notari E, Saa P, Ferg R, Manrique IM, Weil N, Fink RV, Levy ME, Green V, Cyrus S, Williamson PC, Haynes J, Groves J, Krysztof D, Custer B, Kleinman S, Biggerstaff BJ, Opsomer JD, Jones JM. 2022. Population-weighted seroprevalence from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, vaccination, and hybrid immunity among US blood donations from January to December 2021. Clin Infect Dis 75:S254–S263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Cha AE, Cohen RA. 2020. Demographic variation in health insurance coverage: United States. National Center for Health Statistics (U.S.). doi: 10.15620/cdc:113097. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Fig. S1, S2, and S3; Table S1.