Proportions of US Blood Donors With Serological Evidence of Severe Acute Respiratory Syndrome Coronavirus 2 Infections Who Reported Survey-Based Diagnosed Infections During July 2020–December 2022 Open Access

Abstract

Graphical Abstract

Open in new tabDownload slide

The proportion of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections diagnosed by coronavirus disease 2019 (COVID-19) tests, including home antigen tests, is unknown. We detected infections among blood donors in the United States (US) by testing for nucleocapsid antibody (anti-N) seroconversion and administered a questionnaire to determine the proportion of those infections that were associated with a self-reported positive COVID-19 test. Among US blood donors with serologic evidence of SARS-CoV-2 infection who completed a survey, 47.7% reported an associated self-reported positive COVID-19 test. This proportion changed from July–December 2020 (44.9%) to July–December 2022 (54.8%). This study suggests many SARS-CoV-2 infections in adults are not diagnosed with a test.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused >1 million deaths in the United States (US) [1] and remains an important public health concern. Coronavirus disease 2019 (COVID-19) public health case surveillance has been unable to track undiagnosed and unreported infections. This is due to several factors, including individuals' lack of access to healthcare-based testing, use of home antigen tests that are not reported to public health [2, 3], and potentially a decreased public desire for testing. Furthermore, in the US, after the expiration of the COVID-19 pandemic–associated Public Health Emergency on 11 May 2023, many public health authorities stopped reporting COVID-19 case data [4]. The Centers for Disease Control and Prevention (CDC) previously recommended isolation and masking after testing positive for SARS-CoV-2 infection and currently recommends COVID-19 testing as an optional strategy to guide treatment and taking steps to avoid transmission to others [5]. However, little is known about the proportion of people with SARS-CoV-2 infections who have tested positive and how this has changed over time. This study estimated the proportion of US blood donors with serological evidence of SARS-CoV-2 infection who reported a survey-based positive COVID-19 test that coincided with their infection-induced antibody seroconversion interval.

METHODS

From June 2020 through July 2021, 2 of the largest blood collection organizations in the US, Vitalant and the American Red Cross (ARC), began testing all blood donations for SARS-CoV-2 antibodies using the Ortho VITROS anti-spike (S) total immunoglobulin (Ig) quantitative test at Creative Testing Solutions (CTS), their central blood screening laboratory. Additionally, during July 2020–December 2021, as part of a multisite cross-sectional national seroprevalence study, a random set of blood donors was tested each month. During July–December 2020, a nonreactive anti-S Total Ig result was a good indicator of no prior SARS-CoV-2 infection. In the national seroprevalence study starting in January 2021, when vaccinations had begun, reactive anti-S total Ig results of donors were reflexed for Roche anti-N total Ig testing to assess COVID-19 infection-induced versus vaccination-induced anti-S antibodies.

Beginning in January 2022, testing was limited to a longitudinal blood donor cohort that included 142 599 ARC and Vitalant donors residing throughout the continental US who had presented for donation at least twice during June 2020 through July 2021 [6]. For each donor, 1 randomly selected donation for each quarter of 2022 was tested at CTS for SARS-CoV-2 antibodies among available donations. For this cohort, the Ortho VITROS Anti-SARS-CoV-2 Total Nucleocapsid (N) Antibody test and the Ortho VITROS Anti-SARS-CoV-2 S Antibody IgG Quantitative test were used.

Due to the change in test assays used over the course of these studies, some of the total S Ig results were used in determining the seroconversion interval early in the study period, where no anti-N result was available. The methods for the use of the anti-S total Ig to determine donor seroconversion intervals in the absence of a nonreactive anti-N result have been previously described [7]. The sensitivity to detect prior SARS-CoV-2 infections of the Ortho assays among vaccinated and unvaccinated donors was previously determined to be >95% [8].

Vitalant and ARC also administered electronic surveys to the longitudinal blood donor cohort participants that included questions on SARS-CoV-2 infections and COVID-19 vaccinations. The survey was fielded as a baseline survey starting in December 2021 with follow-up surveys administered at approximately 3-month intervals through March 2023. Surveys were sent to cohort participants independent of timing of blood donations.

Evidence of infection was defined as anti-N seroconversion (with the exception for use of anti-S for the period prior to January 2021 as detailed above), and a donor's seroconversion interval was defined as the time from the date of their last anti-N–negative donation to the date of their first anti-N–positive donation.

On the electronic survey, donors reported their earliest SARS-CoV-2 infection from the beginning of the pandemic that had been confirmed using polymerase chain reaction or antigen testing or reported to have been diagnosed by a physician (which was presumed to be on the basis of a COVID-19 test). This study included seroconversion-defined infections associated with blood specimens donated from July 2020 through December 2022; self-reported positive COVID-19 tests occurring earlier than July 2020 were not included in the analysis since that was the first full month with serological testing.

In this analysis, the self-reported diagnosis proportion was defined as the number of reported positive COVID-19 tests that occurred within anti-N seroconversion intervals divided by the total number of anti-N seroconversion intervals. Donors reporting a positive COVID-19 test preceding their seroconversion interval (potentially due to there being 2 distinct infections) were excluded from this study. Donors who also had their last date of survey participation occurring prior to the end of their seroconversion interval were excluded from this analysis (since the self-reported COVID-19 tests could be any date within the seroconversion interval, their last date of survey participation needed to be on or after the seroconversion intervals).

To investigate the temporal changes in the reported diagnosis proportion, the midpoint dates for all identified seroconversion intervals (n = 33 651) were calculated and the proportions of those with a corresponding self-reported positive COVID-19 test (n = 16 062) were categorized into 5 half-year periods (July–December 2020 through July–December 2022), based on the midpoint date.

The reported diagnosis proportions were calculated by donor age, sex, race and ethnicity, self-reported vaccination status at the time of donation (defined as ever having received a COVID-19 vaccine), US Census region, and urbanicity (using the 2013 National Center for Health Statistics urban/rural classification of counties [9]). For donors who received their first COVID-19 vaccine during a seroconversion interval, the order of vaccination and infection could not be inferred; although these donors were included in the overall analysis, they were excluded in the analysis stratified by vaccination status at time of infection (Table 1).

The overall study protocol was approved by the ARC, University of California, San Francisco, Advarra, and Westat institutional review boards. Because all donors provided voluntary, informed consent for use of their de-identified data and residual blood samples from routine blood donations for research, the study was deemed nonhuman subjects research. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy (eg, 45 Code of Federal Regulations [CFR] part 46; 21 CFR part 56; 42 US Code [USC] 241[d]; 5 USC 552a; and 44 USC 3501).

Chi-squared tests of association were performed to compare the proportion of donors reporting a positive COVID-19 test by demographic group (age, sex, race and ethnicity, self-reported vaccination status, region, and urbanicity) and calendar time (5 half-year periods). All statistical analyses were done using Statistical Analysis Software (SAS, version 9.4).

RESULTS

The study sample included 33 651 repeat donors with serologic evidence of an anti-N seroconversion that occurred between July–December 2020 and July–December 2022 (Table 1). The majority were urban-dwelling (77.6%), non-Hispanic White (91.9%), female (54.5%), aged <65 years (69.3%), and residing mainly in the Midwest (35.6%) and West (29.9%) Census regions of the US. Nearly half of donors reported that they were vaccinated (47.0%) before the beginning date of their anti-N seroconversion interval, with an additional 9.5% indicating that their vaccination occurred during their seroconversion interval.

A total of 16 062 (47.7%) of the study donors reported a positive COVID-19 test that occurred during the anti-N seroconversion interval (Table 1, Supplementary Figure). This proportion of donors reporting a positive test was higher in younger age groups (16–29 years: 53.5%; 30–49 years: 50.9%; 50–64 years: 47.8%) than among those aged ≥65 years (44.5%) (P < .0001). Higher proportions were also observed among donors residing in the West (51.4%) and Northeast (51.7%) census regions compared with donors residing in the South (47.1%) and Midwest (43.4%) (P < .0001). The observed proportions were highest among Hispanics (52.1%) and non-Hispanic Asians (52.3%) and lowest among non-Hispanic Blacks (35.8%) (P < .0001), significantly higher among females (49.7%) than among males (45.3%) (P < .0001), higher among donors who were vaccinated prior to their seroconversion interval (51.0%) than among unvaccinated donors (44.7%) (P < .0001), and higher among donors living in urban areas (49.5%) than those living in rural areas (41.5%) (P < .0001).

The overall proportions of reported diagnoses among serologically identified infected donors changed gradually from July–December 2020 (44.9%) to July–December 2022 (54.8%) (Table 1 and Figure 1; P < .0001). This pattern was generally observed across all donor demographic groups.

Figure 1.

Proportion of blood donors with documented infection-induced antibody seroconversions who reported a corresponding severe acute respiratory syndrome coronavirus 2–positive swab over time by age, vaccination rate, and urbanicity of residence. Evidence of infection was defined as presence of anti-nucleocapsid (N) antibodies beginning in January 2021 and the presence of anti-spike (S) antibodies prior to January 2021 because coronavirus disease 2019 (COVID-19) vaccine use was minimal prior to January 2021. A seroconversion interval was defined as the time from the date of their donation without infection-induced antibodies to the date of their first donation with infection-induced antibodies. The proportion was calculated as the number of reported positive COVID-19 tests that occurred within a donor's seroconversion interval divided by the total number of seroconversion intervals.

Open in new tabDownload slide

Among vaccinated donors, the reported diagnosis proportion increased from 47.9% in July–December 2021 to 56.0% during July–December 2022 (P < .0001). Conversely, the reported diagnosis proportions among serologically identified infected unvaccinated donors decreased from 50.3% in July–December 2021 to 37.8% in July–December 2022 (Table 1 and Figure 1; P < .0001).

DISCUSSION

Approximately half of first SARS-CoV-2 infections identified by anti-N seroconversion among blood donors were associated with a self-reported positive COVID-19 test. From July–December 2020 through July–December 2022, the proportion of infections associated with a positive COVID-19 test increased across all donor demographic groups. These changes in the proportions of diagnosed infections were modest, ranging from 45.0% to 55.0%. In contrast, the proportion decreased among unvaccinated donors, reaching 37.8% for the July–December 2022 period.

The findings of this study are comparable with previous US studies. A cohort study in Los Angeles found that 56% of anti-N–positive adults did not have a previous diagnosis of COVID-19 [10]. US National Health and Nutrition Examination Survey data from 2021–2022 showed that 43.7% of respondents with serologic evidence of prior SARS-CoV-2 infection reported that they never had COVID-19 [11].

Demographic groups may differ in their testing behaviors. Younger donors may have been more likely to test because of work requirements compared with older donors. A previous study found older people were less aware of where to find COVID-19 tests and less able to identify symptoms that warrant testing for COVID-19 [12]. Older adults are at increased risk for severe illness with COVID-19, and CDC recommends older adults to receive antiviral treatment after a positive SARS-CoV-2 test [13]. However, older blood donors in this study had lower rates of infections identified by positive tests compared with younger adults, suggesting that additional testing among older adults is needed. Additionally in this study, non-Hispanic Black donors and those residing in rural areas were less likely to report a positive test during their time of seroconversion, potentially reflecting differences in access to testing [14, 15]. This study suggests that vaccinated donors test more frequently for COVID-19 when they have respiratory symptoms than unvaccinated donors, likely reflecting differences in attitudes toward COVID-19 between vaccinated and unvaccinated people [16].

This study has several limitations. First, positive COVID-19 tests and vaccination history were self-reported by survey respondents and were not subject to any validation, making recall bias likely. Second, multiple infections during a seroconversion interval could result in an overestimate of infections associated with a positive swab, but this was likely minimal. Third, this analysis focused on first-time infections during 2020–2022. The proportion of infections that are diagnosed by COVID-19 tests in subsequent years may differ. Fourth, different tests were used to test for prior infection over the course of the study. However, the 2 anti-N antibody assays have been found to have similar qualitative and quantitative results [17]. Fifth, not all infections result in detectable anti-N antibodies, and anti-N antibodies may wane. The presence of S antibodies at the time of infection may blunt the N antibody response and >95% of blood donors had S antibodies by the end of 2022 [6]. However, when using the direct total Ig assays used in this study, prior studies have shown that >90% of specimens collected >1 year after infection have anti-N detectable antibodies, including in those with prior vaccination [8]. Sixth, this study was conducted among blood donors, who may have testing practices that differ from the general population, limiting generalizability.

This study suggests that many adults with SARS-CoV-2 infections may be unaware they are infected with SARS-CoV-2 during the period they are infectious. This limits the impact of relying on positive testing to identify people who need to take precautions to prevent further spread. These results support the updated CDC respiratory virus recommendations that isolation and other prevention strategies be based primarily on symptoms rather than a positive diagnostic test [5]. Importantly, this study suggests that older adults who qualify for COVID-19 antiviral treatment may benefit from additional testing to identify those who should be treated.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors gratefully acknowledge CDC reviewers, whose comments greatly improved the manuscript. The contributions of numerous supervisory, laboratory, and data management staff, including Hasan Sulaeman, Brendan Balasko, Jahnavi Bhaskar, Patricia Villaflor, Kaya Duncan, Zhanna Kaidarova, Anh (Paul) Nguyen, Marjorie D. Bravo, Edward P. Notari, James Haynes, Jamel Groves, Gary Holley, Rebecca Fink, Phillip Williamson, Sherri Cyrus, Valerie Green, Athena Nguyen, Dave Kovach, Chloe Byrne, Daishia Hall, Tatum Fenner, Katherine Elmore, Nicole Whitaker, Ali Corrales, Jordyn Cordero, Andrea Caro, Traianka Eftimova, and Melissa Briggs-Hagen, are also acknowledged with gratitude.

Author contributions. A. A.:. Analysis, writing, and review. E. G., D. J. W., and C. A. H.: Analysis and review. M. S., R. V. F., and B. R. S.: Review. P. S.: Sample coordination and review. M. C. L.: Sample and testing coordination and review. M. B. and J. M. J.: Review and writing.

Financial support. This work was supported by research contracts from the Centers for Disease Control and Prevention (contract number 75D30120C08170).

References

Author notes