Hybrid Immunity to SARS-CoV-2 During Pregnancy Provides More Durable Infant Antibody Responses Compared to Natural Infection Alone

Abstract

Background

Hybrid immunity (infection plus vaccination) may increase maternally derived SARS-CoV-2 antibody responses and durability versus infection alone.

Methods

Prospective cohort of pregnant participants with prior SARS-CoV-2 infection (anti-nucleocapsid IgG, RT-PCR, or antigen positive) and their infants had blood collected in pregnancy, at delivery/birth, and postpartum tested for anti-spike (anti-S) IgG and neutralizing antibodies (neutAb).

Results

Among 107 participants at enrollment, 40% were unvaccinated and 60% were vaccinated (received ≥1 dose); 102 had previous SARS-CoV-2 infection in pregnancy (median, 19 weeks’ gestation); 5 were diagnosed just prior to pregnancy (median, 8 weeks). At delivery, fewer unvaccinated participants (87% anti-S IgG+, 86% neutAb) and their infants (86% anti-S IgG+, 75% neutAb) had anti-S IgG+ or neutAb compared to vaccinated participants and their infants (100%, P ≤ .01 for all). By 3–6 months postpartum, 50% of infants of unvaccinated participants were anti-S IgG+ and 14% had neutAb, versus 100% among infants of vaccinated participants (all P < .01), with lower median antibody responses (anti-S IgG log₁₀ 1.95 vs 3.84 AU/mL, P < .01; neutAb log₁₀ 1:1.34 vs 1:3.20, P = .11).

Conclusions

In pregnant people with prior SARS-CoV-2, vaccination before delivery provided more durable maternally derived antibody responses than infection alone in infants through 6 months.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in pregnancy is associated with increased risk of severe coronavirus disease 2019 (COVID-19) disease, including intensive care unit admission and death [1–4], with even higher risk among pregnant people with comorbidities [5]. Furthermore, adverse pregnancy outcomes, including preterm birth and stillbirth, are more likely among pregnant people with SARS-CoV-2 infection [2, 6].

COVID-19 vaccination is safe in pregnancy and is encouraged, even with previous SARS-CoV-2 infection [7–9]. Infants <6 months of age are not eligible for COVID-19 vaccination, and therefore rely on maternally derived IgG antibodies (via transplacental transfer) for protection [10, 11]. Previous studies have shown that infants born to pregnant people vaccinated for COVID-19 had more persistent anti-spike (anti-S) immunoglobulin G (IgG) antibodies compared to those with infection alone [12, 13]. While immunity from natural infection and vaccination (hybrid immunity) appears to provide more durable SARS-CoV-2 antibody responses in general [14, 15], the role of hybrid immunity on durability of maternally derived functional (ie, neutralizing) antibody responses in pregnant people and their infants has not been well characterized.

We evaluated longitudinal SARS-CoV-2 anti-S IgG and neutralizing antibody (neutAb) responses in a prospective cohort of pregnant people with prior SARS-CoV-2 infection and their infants by maternal vaccine status in the Seattle, Washington metropolitan area, to evaluate the role of hybrid immunity on durability of these responses.

METHODS

Study Setting and Participants

Study setting and enrollment procedures have been previously reported [16]. Briefly, pregnant people aged ≥18 years seeking antenatal care at University of Washington (UW)-affiliated medical centers with evidence of prior SARS-CoV-2 infection by anti-nucleocapsid (anti-N) IgG positive (IgG+), reverse transcription polymerase chain reaction (RT-PCR) positive, or antigen positive during or within 6 months prior to pregnancy, identified during a seroprevalence in pregnancy study [16] or medical record review, were eligible for enrollment into the cohort study.

Study team members obtained informed consent via phone and email. Electronic medical records (EMR) were used to abstract medical and obstetric history, pregnancy/birth and postpartum outcomes, SARS-CoV-2 RT-PCR and antigen results, and COVID-19 disease severity [17]. Participants were asked to report dates of SARS-CoV-2 positive results, if known, and symptom severity through an enrollment and follow-up survey, to identify additional subsequent COVID-19 episode(s) not documented in the EMR. Reinfection was captured via follow-up phone survey and EMR abstraction. COVID-19 vaccination status was abstracted from EMR-linked Washington state immunization registry, and/or from self-report.

For participants identified from the seroprevalence survey, the initial SARS-CoV-2 anti-N IgG+ result was used as their first sample for the cohort study. For participants identified by SARS-CoV-2 RT-PCR and/or antigen results per EMR, their first study blood sample was collected at ≤1 month of consent. Participants were eligible for follow-up blood sample collection at up to 6 time points at 1, 2, 3, 6, and 12 months postenrollment, including delivery (maternal and cord blood). Infants were eligible for blood sample collection draws at 1–2, 2–4, 6, and 12 months of age.

Laboratory Methods

SARS-CoV-2 Anti-N and Anti-S IgG Serology

Samples were tested for SARS-CoV-2 anti-N and anti-S IgG using the Abbott Architect chemiluminescent immunoassay. Samples with Abbott index ≥1.4 were considered anti-N IgG+; samples with anti-S ≥ 50 AU/mL were considered anti-S IgG+ per manufacturer recommendations [18, 19].

SARS-CoV-2 Neutralization Assays

SARS-CoV-2 D614G pseudovirus neutralization assays developed at the Fred Hutchinson Cancer Center were validated and performed in the UW Virology Laboratory [20]. One day prior to infection, 12 500 HEK293T/ACE2 cells were seeded in 50 µL D10 complete media per well in 96-well plates. Participant blood was heat inactivated for 30 minutes at 56°C prior to creation of 10-fold to 7290-fold dilutions in duplicate with final volume of 80 µL. CV30 monoclonal antibody (Ab02019-12.1; Absolute Antibody) diluted in duplicate from 810 to 1.11 ng/mL in 80 µL was used as positive control and for quality monitoring. Samples of 60 µL diluted blood, CV30 positive control, or D10 complete media alone was mixed with 60 µL D614G pseudovirus at 1 × 10⁶–1 × 10⁷ relative light units (RLU)/mL and incubated on a rocker at 37°C for 1 hour. HEK293T/ACE2 cells were infected with 100 µL pseudovirus/CV30-blood mixture and 50–55 hours after infection, 100 µL media was removed from each well, 30 µL BrightGlo was added to each well, and luminescence was measured on a Victor Nivo Multimode Microplate Reader with 1-second integration and no attenuation. Adjusted RLUs for participant blood at each dilution were averaged and normalized to the average RLU of no-serum control wells. A 4-parameter logistic regression was used to determine 50% neutralization dose (ND₅₀), ND₈₀, and R².

Statistical Analysis

Pearson χ² tests were used to compare proportions and Wilcoxon rank sum tests to compare distributions of continuous variables. Abbott anti-S IgG (AU/mL) and neutAb (ND₅₀) results were log₁₀ transformed to reduce skewness. Median values for Abbott anti-S IgG and neutAb (ND₅₀) were only calculated among individuals with positive values.

Participants were classified as unvaccinated or vaccinated after receiving ≥1 dose of any SARS-CoV-2 vaccine. Unvaccinated status was classified as no history of an mRNA vaccine or viral vector vaccine against SARS-CoV-2; ≥ 1 vaccine was further defined as partial (1 dose of mRNA vaccine), fully vaccinated (2 doses of mRNA vaccine or 1 dose of viral vector vaccine), and boosted (≥3 doses of mRNA vaccine [or at least 1 dose mRNA vaccine plus a viral vector vaccine] or 2 doses of viral vector vaccine) [9]. Participant vaccine status was defined at blood collection; infant vaccine status was defined by maternal vaccine status at delivery.

In primary cross-sectional analyses, antibody responses were evaluated during pregnancy, delivery/birth (via cord blood), <3 months postpartum, and 3–6 months postpartum, and stratified by participant vaccine status. If 2 samples were collected during the same analytical window, the earliest available sample in pregnancy was used, while for postpartum analyses, the latest available sample within the window was used. For secondary longitudinal analyses, all available participant and infant samples were included.

Transplacental transfer ratios were calculated by dividing the log₁₀ transformed anti-S IgG (AU/mL) and neutAb (ND₅₀) of cord blood by maternal delivery result, with transplacental transfer ratio of ≥1 considered efficient transplacental transfer [13].

Correlates of anti-S+ IgG and neutAb in infants at birth and from 2 to 6 months of age (using the latest available sample if more than 1 sample collected in this timeframe) were assessed in separate univariable models using generalized linear models with Poisson family to calculate prevalence ratios (PR); variables with P < .10 in univariable analyses were included for evaluation in multivariable models. All statistical tests were 2-sided with α of .05.

General estimating equations (GEE) with a Gaussian link and robust errors were used to measure rates of change in log₁₀ anti-S IgG and log₁₀ neutAb (ND₅₀) since first maternal positive result and since infant birth. GEE models were used to determine whether average monthly rates of change differed by pregnancy status at time of sample collection, trimester of infection (among people infected during pregnancy), presence of self-reported symptoms, and vaccination status at time of sample collection (for participants) or at delivery (for infants). Multivariable models were constructed with all covariates significant at P < .10 in univariable models. Analyses were performed using Stata 17.1 (StataCorp).

Ethics Statement

This study was approved by the UW Human Subjects Division and Valley Medical Center Research Oversight Committee. The activity was reviewed by the US Centers for Disease Control and Prevention (CDC) and was conducted consistent with applicable federal law and CDC policy. All participants provided informed written consent.

RESULTS

Between January 2021 and August 2022, we enrolled 107 pregnant people with prior SARS-CoV-2 infection (Figure 1). Among 107 participants, 40% were unvaccinated and 60% were vaccinated (≥1 dose) and considered to have hybrid immunity at the time of first blood sample collection. Participant characteristics were similar between participants who were vaccinated or remained unvaccinated at delivery (Table 1), except for increased vaccination among participants self-identified as of Asian race (Supplementary Table 1). Among 102 (95%) participants with initial SARS-CoV-2 infection during pregnancy, median gestational age at detection of infection was 19 weeks (interquartile range [IQR], 13–28; 23 weeks for participants identified through RT-PCR, 17 weeks for those identified through antigen tests, and 13 weeks for participants identified through an anti-N IgG+ result). Five participants (5%) had SARS-CoV-2 infection at median 8 weeks (IQR, 1–24) prior to pregnancy per SARS-CoV-2 RT-PCR positive results. Sixty-nine (64%) participants had their first blood sample collected during pregnancy (median 30 weeks’ gestational age; IQR, 19–38), 34 (32%) at delivery, and 4 (4%) postpartum (median 8 weeks postpartum; IQR, 5–11).

Pregnancy

Among 69 participants with blood samples collected in pregnancy, median time from infection (RT-PCR-positive or antigen-positive result) to first sample collection in pregnancy was 5 weeks (IQR, 4–10). At time of first sample in pregnancy, 30 (43%) participants were unvaccinated, 16 (23%) were partially vaccinated or had completed a primary series, and 23 (33%) were boosted. Overall, 84% of first blood samples collected in pregnancy were anti-N IgG+, 97% anti-S IgG+ (median log₁₀ 4.40 AU/mL), and 93% were neutAb+ (median log₁₀ neutralization 1:3.88) (Table 2).

In pregnancy, while the proportion of participants with anti-S IgG+ antibodies was similar between those with ≥1 vaccine dose and unvaccinated participants (39 of 39 [100%, 95% confidence interval {CI}, 91%–100%] vs 28 of 30 [93%, 95% CI, 78%–99%]; P = .10), vaccinated participants were more likely to have detectable neutAb (37 of 37 [100%, 95% CI, 91%–100%] vs 25 of 30 [83%, 95% CI, 65%–94%]; P = .01) (Table 2). Participants receiving ≥1 vaccine dose had higher median log₁₀ anti-S IgG+ (4.40 vs 2.72 AU/mL, P < .01) and log₁₀ neutralization (1:4.00 vs 1:2.33, P < .01) (Figure 2).

Delivery and Birth

Among 89 participants with delivery blood samples, median time from infection (RT-PCR-positive or antigen-positive result) to delivery was 17 weeks (IQR, 10–24). By delivery, 26 (29%) participants remained unvaccinated, 25 (28%) were partially vaccinated or had completed a primary series, and 38 (43%) were boosted.

At delivery, participants with ≥1 vaccine were more likely to have anti-S IgG+ (100% [95% CI, 94%–100%] vs 87% [95% CI, 68%–97%]; P < .01) and neutAb (100% [95% CI, 91%–100%] vs 86% [95% CI, 66%–96%]; P < .01) with higher median log₁₀ anti-S IgG+ (4.40 vs 2.93 AU/mL, P < .01) and log₁₀ neutralization (1:3.95 vs 1:2.45, P < .01), compared to participants who remained unvaccinated. Similarly, cord blood from infants born to participants with ≥1 vaccine was more likely to have anti-S IgG+ (100% [95% CI, 94%–100%] vs 86% [95% CI, 64%–97%]; P < .01) and neutAb (100% [95% CI, 91%–100%] vs 75% [95% CI, 51%–91%]; P < .01), with higher log₁₀ median anti-S IgG+ (4.40 vs 2.95 AU/mL, P < .01) and log₁₀ neutralization (1:4.00 vs 1:2.37, P < .01) compared to infants born to unvaccinated participants.

Postpartum and Infancy

By 3 months postpartum, 11 (21%) of 52 participants with blood samples remained unvaccinated, 22 (42%) were partially vaccinated or had completed a primary series, and 19 (37%) were boosted. Participants with ≥1 vaccine had higher median log₁₀ anti-S IgG+ (4.40 vs 3.09 AU/mL, P < .01) and log₁₀ neutralization (1:4.00 vs 1:2.62, P < .01) compared to unvaccinated participants. Similarly, by 3 months after delivery (median 1.8 months; IQR, 1.5–2.1), infants born to participants with ≥1 vaccine prior to delivery were 2.3 times more likely to have neutAb (100% [95% CI, 81%–100%] vs 43% [95% CI, 18%–71%]; PR, 2.3 [95% CI, 1.3–4.3]; P < .01), and had higher log₁₀ median anti-S IgG+ (4.40 vs 2.09 AU/mL, P < .01), and log₁₀ neutralization (1:3.62 vs 1:1.97, P < .01).

Between 3 and 6 months postpartum, 4 (12%) of 33 participants with blood samples remained unvaccinated, and 29 (88%) had completed a primary series or were boosted. Participants with ≥1 vaccine had higher median log₁₀ anti-S IgG+ (4.40 vs 3.05 AU/mL, P < .01) and median log₁₀ neutralization (1:3.79 vs 1:2.82, P = .02) compared to unvaccinated participants. Between 3 and 6 months of age (median 4.0 months; IQR, 3.5–4.7), infants of participants with ≥1 vaccine were twice as likely to have detectable anti-S IgG+ (100% [95% CI, 83%–100%] vs 50% [95% CI, 16%–84%]; PR, 2.0 [95% CI, 1.0–4.1]; P = .05) and 7 times more likely to have neutAb (100% [95% CI, 72%–100%] vs 14% [95% CI, 4%–58%]; PR, 7.0 [95% CI, 1.1–45.3]; P < .01) with higher log₁₀ median anti-S IgG+ (3.84 vs 1.95 AU/mL, P < .01) compared to infants born to unvaccinated participants.

Transplacental Transfer

Among participants with paired delivery and cord blood samples, 67 of 73 (92%) had evidence of efficient transplacental transfer of anti-S IgG+ and 38 of 49 (78%) of neutAb (Figure 3). Vaccinated participants were more likely to have efficient transplacental transfer versus those who remained unvaccinated at delivery for both anti-S IgG (97% [95% CI, 88%–100%] vs 73% [95% CI, 45%–92%]; P < .01) and neutAb (86% [95% CI, 71%–95%] vs 54% [95% CI, 25%–81%]; P = .02).

Correlates of Detectable Anti-S IgG and neutAb in Infants at Birth and by 6 Months of Life

In multivariable analyses, infants born to participants with SARS-CoV-2 infection in the third trimester (compared to earlier in pregnancy) were less likely to have anti-S IgG+ or neutAb at birth, while maternal vaccination prior to delivery (compared to no vaccination) was associated with greater likelihood of anti-S IgG+ or neutAb at birth (Table 3). Additionally, male infants were more likely to have neutAb at birth. Persistence of maternally derived anti-S IgG+ and neutAb in infants between 2 and 6 months of age (median, 3.8 months; IQR, 3.2–4.5) was associated with maternal vaccination prior to delivery (adjusted PR, 5.0; 95% CI, 1.41–17.76; P < .01).

Longitudinal Maternal and Infant Antibody Responses

Among participants with available longitudinal samples, participants with hybrid immunity had significantly higher median anti-S IgG+ (4.40 vs 2.69, P < .001) and neutAb responses (4.00 vs 2.33, P < .001) at first sample collection compared to unvaccinated participants with infection alone. All participants, except 1, remained anti-S IgG+ and neutAb+ throughout follow-up (Supplementary Figure 1). Among infants with available longitudinal samples, those born to mothers with hybrid immunity had higher median anti-S IgG+ (4.40 vs 2.91, P < .01) and neutAb responses (3.93 vs 2.33, P < .01) at first sample collection compared to those born to participants with infection alone.

For both anti-S IgG and neutAb, rate of decline was significantly faster during pregnancy (P = .001 and P < .001) and among those with hybrid immunity at the time of sample collection (P < .001 and P = .06) (Supplementary Table 2). Anti-S IgG and neutAb declined more rapidly in infants born to pregnant people who were vaccinated before delivery (P < .001 and P = .09) (Supplementary Table 2). While the rate of decline in antibody levels was more rapid among pregnant people with hybrid immunity and their infants, antibody levels remained higher than among participants with infection alone and their infants due to significantly higher initial set point at the time of first sample collection among participants with hybrid immunity. Infant anti-S IgG declined faster in those whose pregnant parent reported symptoms of COVID-19 (P = .007) and who were infected during the second trimester of pregnancy (P = .05) compared to those infected during the first trimester; however, despite this decline, levels remained high due to the initial set points at time of first sample collection among vaccinated participants.

Four participants had confirmed reinfection during the study, 2 during pregnancy and 2 in the postpartum period, both of whom were fully vaccinated 5 months prior to reinfection and were 9 months and 12 months from first confirmed infections, respectively. All 4 participants had an increase in both anti-S and neutAb (Supplementary Figure 1). One infant born to an unvaccinated participant had documented SARS-CoV-2 infection during follow-up at 7 months of age, with 1 sample obtained after infection, showing an increase from their cord anti-S IgG results (Figure 4). Sensitivity analyses excluding participants with reinfection and infant infection from our longitudinal models produced similar results. No infants received COVID-19 vaccination prior to any study sample collection.

DISCUSSION

In a cohort of people with SARS-CoV-2 infection during or immediately before pregnancy and their infants, maternal hybrid immunity from infection plus vaccination was associated with greater likelihood of having detectable anti-S IgG and neutAb at delivery for both themselves and their infants. Infants born to pregnant persons with hybrid immunity had a longer duration of detectable immune responses compared to natural infection alone. Additionally, vaccinated participants with prior infection and their infants had higher antibody concentrations (both anti-S IgG+ and neutAb+) through 6 months postpartum compared to those with natural infection alone.

As of 30 June 2023, vaccination against SARS-CoV-2 with ≥1 dose among the US pregnant population reached 74%, with only 14% having received the bivalent booster dose [21, 22]. In the United States, infants ≥6 months of age became eligible for vaccination against SARS-CoV-2 starting on 18 June 2022 [23, 24]. In our study, we found natural infection alone did not provide persistent antibody responses in infants through 6 months of age and transplacental transfer of maternally derived antibodies after SARS-CoV-2 infection in pregnancy was more efficient in participants with hybrid immunity. We noted substantially reduced durability and magnitude of both anti-S, and perhaps more importantly, neutAb responses among infants born to pregnant persons who remained unvaccinated by delivery. Strikingly, all infants born to participants with SARS-CoV-2 infection in pregnancy with at least 1 vaccine prior to delivery retained anti-S and neutAb up to 6 months of life (up to 7.5 months for those with additional samples). In contrast, only 50% of infants born to unvaccinated pregnant people maintained anti-S IgG+ status and 14% retained functional neutAb+ during this vulnerable period prior to their own vaccine eligibility at 6 months of age. Infants born to mothers with hybrid immunity at birth with anti-S and/or neutAb+ up to 6 months of age also had higher antibody responses, although the significance of greater responses in terms of protection of infant infection is unknown. Previous studies have described similar findings of continued presence of anti-S IgG in infants born to vaccinated pregnant people, including those with recovered infection and predelivery vaccination, with waning antibodies before 6 months in infants of unvaccinated birth parents with prior infection alone [12, 25]. Our study extends these findings with more frequent sampling, longer duration of follow-up (up to 12 months postpartum), and longitudinal neutAb assessment. Prior evaluations of neutAb have focused primarily on cross-sectional transplacental transfer and not persistence [26]. In one of the few studies evaluating neutAb, shorter duration between maternal infection and delivery and infant male sex were associated with lower transplacental transfer of SARS-CoV-2 neutAb. Intriguingly, transfer of neutAb to SAR-CoV-2 was inferior to other pathogens, including cytomegalovirus, measles, rubella, and varicella zoster virus [26]. In our study we found maternal vaccination prior to delivery was associated with presence of anti-S IgG and neutAb at birth and persistence through 6 months of life, while infant male sex was associated with having neutAb at birth. A shorter time between infection and delivery (ie, having infection in a later trimester) was associated with lower likelihood of anti-S IgG Ab+ or neutAb+ at birth. Maternal vaccination prior to delivery was the strongest predictor of maternally derived antibody persistence through 6 months of age.

Our study has several strengths and some limitations. Prior studies have described longitudinal anti-N and anti-S IgG responses in pregnancy and transplacental transfer in the setting of infection or vaccine, although few evaluated hybrid immunity [25]. This is one of the largest longitudinal evaluations of people with prior SARS-CoV-2 infection in pregnancy and their infants, and one of few that evaluated both binding (anti-S) as well as functional (neutAb) antibody responses by vaccine status. Because of the long duration of the study, reinfection occurred and was captured. Antibody response duration was similar even when participants with reinfections were excluded from the analysis. While both anti-S IgG responses and neutAb responses are elicited from either natural infection or vaccination, neutAb titers are likely more closely associated with protection from subsequent infection and severe outcomes [26, 27]. The assay used to evaluate neutralization was based on the more ancestral SARS-CoV-2 D614G spike protein, which allowed a standardized, validated basis for comparison for neutralizing antibodies and can be adjusted for protection against circulating variants [28].

Our study enrolled participants with initial SARS-CoV-2 infection prior to availability of bivalent vaccines, potentially limiting generalizability to the current era of vaccine boosters targeting variants of concern. The number of infants with samples available at 3–6 months of age was limited, particularly among infants born to unvaccinated individuals; however, we were able to identify significant differences in antibody responses by maternal birth vaccine status.

Although we limited primary analyses to cross-sectional comparisons at key clinically relevant time points of pregnancy, delivery/birth, and early/late infancy, we incorporated all available samples in secondary longitudinal analyses. Additionally, we focused our primary comparisons between pregnant people with hybrid immunity who received ≥1 vaccine and those who remained unvaccinated, and due to limited sample size did not have power in most cases to detect significant differences by vaccine states further stratified by different vaccine uptake (ie, partial, full, boosted). Given realities of high global infection rates and inequitable vaccine distribution, one study strength is our ability to illustrate durability of maternally derived antibodies in the setting of natural infection and ≥1 vaccine dose.

CONCLUSIONS

We found natural infection SARS-CoV-2 in pregnancy without vaccination, while associated with appreciable anti-S IgG+ and neutAb in pregnant people, did not assure efficient transplacental transfer of maternally derived antibodies or long-term protection to infants through 6 months of age. In our study, hybrid immunity during pregnancy provided both universal anti-S IgG+ and neutAb to infants through at least 6 months of age. These data further support vaccination in pregnancy prior to delivery, including after infection, to ensure protection for both pregnant people and their infants [8, 9].

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. The authors thank our site partnerships within the University of Washington (UW) system, clinics, and in-patient obstetric units. The health care staff within these sites helped ensure appropriate samples were collected. We also acknowledge the Research Testing Services team and the general laboratory study staff at the UW for performing study assays. Above all, our sincerest thanks to the study participants.

Author contributions. S. M. L., A. K., J. A. E., and A. L. D. designed the study. S. M. L., M. C. A., J. N. E., and A. L. D. developed the study protocol. S. M. L. and A. L. D. are the principal investigators. S. M. L., E. A. W., B. A. R., and A. L. D. were responsible for the statistical design of the study. S. S. S., A. L. G., E. A. G., S. R. B., I. S. A., and A. P.-O. performed and managed samples for laboratory testing. E. A. W., M. C. A., and J. N. E. managed data collection and cleaning. S. M. L., E. A. W., M. C. A., and A. L. D. performed the data analysis and developed the tables and figures. S. M. L., E. A. W., and M. C. A. wrote the initial draft of the manuscript. All authors read and approved the manuscript.

Financial support. This work was supported by the Centers for Disease Control and Prevention (grant number 75D301-20-C09610) and Merck & Co, Inc (grant number MISP 60429).

References

Author notes