Decline in SARS-CoV-2 Antibodies After Mild Infection Among Frontline Health Care Personnel in a Multistate Hospital Network — 12 States, April–August 2020

1Most persons infected with SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19), develop virus-specific antibodies within several weeks, but antibody titers might decline over time. Understanding the timeline of antibody decline is important for interpreting SARS-CoV-2 serology results. Serum specimens were collected from a convenience sample of frontline health care personnel at 13 hospitals and tested for antibodies to SARS-CoV-2 during April 3–June 19, 2020, and again approximately 60 days later to assess this timeline. The percentage of participants who experienced seroreversion, defined as an antibody signal-to-threshold ratio >1.0 at baseline and <1.0 at the follow-up visit, was assessed. Overall, 194 (6.0%) of 3,248 participants had detectable antibodies to SARS-CoV-2 at baseline. Upon repeat testing approximately 60 days later (range = 50–91 days), 146 (93.6%) of 156 participants experienced a decline in antibody response indicated by a lower signal-to-threshold ratio at the follow-up visit, compared with the baseline visit, and 44 (28.2%) experienced seroreversion. Participants with higher initial antibody responses were more likely to have antibodies detected at the follow-up test than were those who had a lower initial antibody response. Whether decay in these antibodies increases risk for reinfection and disease remains unanswered. However, these results suggest that serology testing at a single time point is likely to underestimate the number of persons with previous SARS-CoV-2 infection, and a negative serologic test result might not reliably exclude prior infection.

2Once infected with SARS-CoV-2, most persons develop virus-specific antibodies within 2–3 weeks. Serology tests are now being used widely in seroprevalence studies to understand patterns of viral spread, cumulative incidence of SARS-CoV-2 infection, and pandemic trajectory. Further, serologic testing has been proposed as a way to identify persons who might have developed immunity through a previous infection. Understanding how rapidly SARS-CoV-2 antibody levels decline after seroconversion is critical for interpreting serology results. A limited number of studies have found declines in SARS-CoV-2 antibody levels over time, but the frequency and timing of seroreversion (the decline in antibody levels below the positivity threshold after initial seroconversion) remains largely unknown.

3The Influenza Vaccine Effectiveness in the Critically Ill (IVY) Network, a collaboration of academic medical centers in the United States that studies influenza and COVID-19, enrolled a convenience sample of frontline health care personnel at 13 centers in 12 states,† with a target of 250 participants per center. Health care personnel were eligible if they reported regular direct contact with COVID-19 patients and worked in the emergency department, intensive care unit, or other hospital-based unit that cared for patients with COVID-19. Participants underwent two study visits: a baseline visit (conducted April 3–June 19, 2020) and a follow-up visit approximately 60 days after the baseline visit. At both visits, blood was collected for SARS-CoV-2 antibody testing, and participants were questioned about demographic characteristics, underlying medical conditions, signs or symptoms of an acute viral infection from February 1, 2020, until the visit date,§ and any previous SARS-CoV-2 testing (e.g., reverse transcription–polymerase chain reaction [RT-PCR]) for acute infection. Blood specimens collected at the baseline and follow-up visits were tested for SARS-CoV-2 antibodies at CDC using an enzyme-linked immunosorbent assay (ELISA) against the extracellular domain of the SARS-CoV-2 spike protein. The assay detects all SARS-CoV-2 immunoglobulin (Ig) types (IgA, IgM, or IgG). Specimens were considered reactive with a signal-to-threshold ratio >1.0 at a background corrected serum dilution of 1:100, with higher ratios indicating higher antibody titers. The assay has a sensitivity estimated at 96% and specificity at 99%.

4The change in signal-to-threshold ratio between the baseline visit and follow-up visit was quantified, and the percentage of participants who experienced seroreversion was reported. Logistic regression was used to evaluate the association between baseline signal-to-threshold value and seroreversion, adjusting for age, sex, race/ethnicity, number of days between the baseline and follow-up visit, and presence of one or more chronic medical condition. Analyses were conducted using Stata (version 16; StataCorp). The project was determined to be nonresearch public health surveillance by participating institutions and CDC and was conducted consistent with applicable federal law and CDC policy.¶

5Among 3,248 health care personnel, 194 (6.0%) had antibodies to SARS-CoV-2 at the baseline visit. Among these, 156 (80.4%) returned for the follow-up visit around 60 days later (range = 50–91 days). Among these 156 participants with a positive baseline serology and follow-up antibody testing performed, median age was 38 years (interquartile range [IQR] = 30–48 years), 94 (60.3%) were female, and 108 (69.2%) reported one or more symptoms of an acute infection consistent with COVID-19 between February 1, 2020 and the baseline visit. Among the 108 participants who reported symptoms, the median interval between symptom onset and baseline serology testing was 30 days (IQR = 19–40 days). Participants who reported symptoms of an acute viral illness since February had higher baseline signal-to-threshold ratios (median = 3.6; IQR = 3.1–3.9) than did those who did not report symptoms (median = 2.5; IQR = 1.5 to 3.6) (p<0.001). Among these 156 participants, 72 (46.2%) reported past RT-PCR testing for SARS-CoV-2, 46 (63.9%) of whom had positive test results; no hospitalizations were reported.

6Among the 156 participants who returned for follow-up, the signal-to-threshold value for 146 (93.6%) had declined since the baseline visit, including 44 (28.2%) participants who experienced seroreversion (Table 1) (Supplementary Figure, https://stacks.cdc.gov/view/cdc/97358), with antibody levels falling below the threshold for positivity. Among 108 participants who reported previous COVID-19–compatible signs or symptoms, 21 (19.4%) seroreverted, compared with 23 (47.9%) of 48 of participants who did not report symptoms (p<0.001). Among 72 participants with previous RT-PCR testing, one (2.2%) of 46 with a positive test result versus seven (26.9%) of 26 with a negative test result seroreverted. Seroreversion occurred in 64.9% (37 of 57) of participants with a low antibody response (baseline signal-to-threshold value = 1.0–2.9) and 7.1% (seven of 99) of participants with a high antibody response (baseline signal-to-threshold value ≥3.0) (p<0.001) (Figure). A higher baseline signal-to-threshold ratio was associated with lower odds of seroreversion at the follow-up visit (adjusted odds ratio [aOR] for a 1-unit increase in signal-to-threshold ratio = 0.29; 95% CI = 0.18–0.46) (Table 2). In this model, a 10-year increase in participant age was associated with higher odds of seroreversion (aOR = 1.74; 95% CI = 1.06–2.85). Compared with non-Hispanic White participants, odds of seroreversion were lower among non-Hispanic Black participants (aOR = 0.11; 95% CI = 0.15–0.76) and Hispanic participants (aOR = 0.10; 95% CI = 0.01–0.88).

7In this study of 156 frontline U.S. health care personnel who received positive SARS-CoV-2 antibody test results in spring 2020 and returned for follow-up testing approximately 60 days later, 146 (93.6%) had a decline in antibody levels between baseline and follow-up, and 44 (28.2%) had complete seroreversion, i.e., a decline of antibody to levels below the threshold for positivity. A higher percentage of those with low baseline antibody levels seroreverted (64.9%) than did those with high baseline titers (7.1%). These results suggest that a substantial proportion of persons infected with SARS-CoV-2 might have negative serologic test results in the months following infection. This has several important implications. Cross-sectional seroprevalence studies that estimate the number of persons who have been infected with SARS-CoV-2 will likely underestimate incidence because a proportion of previously infected persons will likely serorevert and thus not be counted as having been previously infected. In addition, these results challenge the notion of using serologic testing results at an individual level to designate previous SARS-CoV-2 infection. COVID-19 convalescent plasma is widely being used as a treatment for COVID-19, including through a Food and Drug Administration Emergency Use Authorization in the United States; these results demonstrate that the optimal window for collecting convalescent plasma with high levels of SARS-CoV-2 antibodies from donors who have recovered from COVID-19 might be short because of substantial decline in antibody levels within 60 days. Whether decline in SARS-CoV-2 antibodies increases risk for reinfection and disease in humans remains unknown. Humoral immunity to primary infections from a novel virus might not be as durable or strong as that to secondary infections, but memory B-cell and T-cell responses might reduce the severity of illness with repeat exposure or infection.

8The findings in this report are subject to at least four limitations. First, the timing of the baseline serologic test relative to symptom onset was not standardized, which might affect baseline signal-to-threshold ratios, particularly for recently acquired infection in which antibody levels might still have been increasing. Second, the study population was derived from a convenience sample, which might result in nonrepresentativeness. Third, 38 (20%) participants were lost to follow-up, limiting size of the study population. Finally, misclassification of antibody status was possible; however, this was considered to be unlikely because of the high sensitivity and specificity of the ELISA.

9In this study of frontline health care personnel at 13 medical centers who received positive SARS-CoV-2 antibody test results in spring 2020, more than one quarter were seronegative approximately 60 days after testing. Because SARS-CoV-2 antibody levels might decline in a proportion of persons following primary infection, a negative serology test does not reliably exclude previous infection. These antibody declines might not equate to loss of protective immunity or increased risk for reinfection; this was not assessed in this study. Cross-sectional seroprevalence studies to evaluate population immunity are likely to underestimate rates of previous infection because antibodies appear to only be detectable for a discrete period of time following infection.

10Most persons develop virus-specific antibodies to SARS-CoV-2 after infection; however, the timeline of antibody decline over time is uncertain.

11Among 156 frontline health care personnel who had positive SARS-CoV-2 antibody test results in spring 2020, 94% experienced a decline at repeat testing approximately 60 days later, and 28% seroreverted to below the threshold of positivity. Participants with higher initial antibody responses were more likely to have antibodies detected at the follow-up test than were those who had a lower initial antibody response.

12SARS-CoV-2 antibodies decline over weeks following acute infection. Negative SARS-CoV-2 serologic results do not exclude previous infection, which has significant impacts on how serologic studies are interpreted.