Skip to main content

Our take —

This was a prospective cohort study conducted in Germany. Researchers compared the immunological responses as well as local and systematic reactions between two groups: the homologous group (i.e., two doses, three weeks apart, of the same vaccine formulation [BNT162b2 by Pfizer-BioNTech]) and the heterologous group (i.e., two doses of different vaccine formulations 10-12 weeks apart [first ChAdOx1 by AstraZeneca and then BNT162b2]). As shown in other studies, participants’ B-cell immune responses were significantly higher after the 1st dose with BNT162b2 compared to ChAdOx1. However, following the second dose of BNT162b2, the heterologous group had comparable B-cell immune responses and slightly higher T-cell immune responses, versus the homologous group. Both groups experienced similar local reactions after their first and second doses. Systemic reactions were more common after the first dose of ChAdOx1 vs. BNT162b2 but less common in the heterologous group after the second dose with BNT162b2. This study highlights the laboratory evidence of efficacy and safety assurance of heterologous immunization with ChAdOx1 followed 10-12 weeks later by BNT162b2, especially for females who are more likely to have rare thromboembolic adverse event following two doses with ChAdOx1. These results are limited by a relatively small sample size and lacked an evaluation of the clinical effectiveness of heterologous immunization.

Study design

Prospective Cohort

Study population and setting

This was a prospective cohort study conducted in Germany that included healthcare workers who received either (1) homologous immunization with two doses, three weeks apart, of BNT162b2 (Pfizer-BioNTech) vaccine or (2) heterologous immunization with an initial dose of ChAdOx1-nCoV19 (AstraZeneca) vaccine followed 10-12 weeks later by BNT162B2. Baseline demographic data were collected at the study enrollment. Anti-nucleocapsid antigen was measured at enrollment to exclude participants with prior infection with SARS-CoV-2. B-Cell immunity against SARS-CoV-2 was quantified by measuring anti-receptor binding domain and anti-S1 antibodies using enzyme-linked immunosorbent assay (ELISA), and T-cell immunity by measuring INF-Gamma induced by S1 peptide antigen. High serum antibody avidity was defined as antibodies with more than 60% avidity. These B-cell and T-cell responses were compared between subsets of both groups, who were matched on age and sex. The proportion of participants who had local and systemic reactions after 1st (prime) and 2nd (booster) doses were captured by asking the participants to fill electric questionnaires on days 1, 3, 5, and 7 after every vaccination.

Summary of Main Findings

One hundred eighty-nine participants received two doses of BNT162B2 vaccine, three weeks apart (homologous boost group), and 110 participants received heterologous immunization with ChAdOx1 vaccine followed by BNT162B2 vaccine with a median time of 71 days between doses (heterologous boost group). The median ages of the groups were 35 and 38 years, respectively. The homologous group was 55% women vs. 78% in the heterologous group.

After vaccination with the first dose, 67% (95% CI: 57 – 76%) of those who received BNT162B2 had anti-S1 antibodies compared to 28% (95% CI: 18 – 40%) in the ChAdOx1 group, with a statistically significant difference. However, after the second dose with BNT162b2, all patients had anti-SARS-COV-2 S1 antibodies. Similarly, after the first dose, 95% of the homologous group had neutralizing antibodies compared to 84% in the heterologous group, with a statistically significant difference, but 100% and 99% had neutralizing antibodies after the second dose, respectively.

High serum antibody avidity was not detected after the first dose of vaccine in either group. Avidity is the overall strength of connection between an antibody and its attachment site on a disease-causing microbe. After the second dose, high serum antibody avidity was detected in 100% (95% CI: 94 – 100%) of participants in the heterologous group and 90% (95% CI 74 – 97%) in the homologous group. The median level for the interferon gamma assay was statistically higher in the heterologous group (2.25 AU) than the homologous group (1.67 AU). Across groups, there were no noticeable differences in local reactions to either the first or second dose of vaccine. However, systemic reactions were more common after the first dose of ChAdOx1 in the heterologous group vs. the first dose of BNT162b2 in the homologous group, but less common after the second dose in the heterologous group.

Study Strengths

B-cell and T-cell immune response were measured in subsets in both groups who were matched on age and sex and at comparable intervals following vaccine administration in both groups, which strengthen the comparability between the study groups. Also, data were collected prospectively which minimized the chance of information or recall bias.

Limitations

Only subsets of both cohorts had serum testing for B-cell and T-cell immune responses, and matching between both cohorts were based only on age and sex. This could have resulted in selection bias if subjects in both groups were different in a way that that could affect their immune response to the vaccine (e.g., being immunocompromised). Also, local and systemic reactions were measured by soliciting response through direct questionnaires instead of observation. Since the study was unmasked as well, participants might have been influenced to be more or less observant to their reactions, based on their prior knowledge of their specific vaccine’s side effects. Furthermore, there was not a comparison group consisting of heterologous immunization with a 3-week interval between doses, which makes it difficult to discern whether the heterologous immunization or difference in timing was responsible for these results.

Value added

This is the first study to examine B-cell and T-cell immune responses as well as local and systematic reactions between heterologous (ChAdOx1 then BNT162b2) and homologous (BNT162b2 for both doses) immunization groups. Since females have been reported to be more prone to rare but serious thromboembolic events following ChAdOx1 vaccination, this study provides strong laboratory data for the efficacy and safety of heterologous immunization, at least at the short run.

Our take —

This research letter reported on the findings from a prospective study of 109 symptomatic pregnant people under investigation for COVID-19 in the United States who enrolled in the PRIORITY study before 14 weeks of gestation. The study found that miscarriage following symptoms of illness among participants with a positive SARS-CoV-2 PCR test (n=94) was similar to those with a negative SARS-CoV-2 PCR test (n=19, ~6-7%). While these findings, like much of the available evidence on COVID-19 in early pregnancy, are limited by a small sample size, potential selection bias, and incomplete adjustment for confounding factors, it is reassuring that none of the studies to date have implicated COVID-19 as a risk factor for miscarriage.

Study design

Prospective Cohort

Study population and setting

This study is a sub-analysis that included data from 109 women (13 years or older) enrolled in a prospective cohort of COVID-19 during pregnancy during the United States (Pregnancy CoRonavIrus Outcomes RegIsTrY — PRIORITY) before 14 weeks gestation. Participants could self-enroll or be referred from prenatal clinicians across the United States. Participants were screened and consented by phone. After enrollment, participants completed questionnaires weekly for four weeks following enrollment, and the authors calculated the crude cumulative incidence of pregnancy loss before 20 weeks gestation.

Summary of Main Findings

All 109 participants (mean age 31 years, 30% Latinx, mean gestational age between 9 and 10 weeks) were symptomatic, however, 94 tested positive for COVID-19 and 15 tested negative. Participants who tested positive were more likely to identify as Latinx (35.1% versus 20%) and report more previous pregnancies and living children (mean gravida 2.68, parity 1.02 versus gradiva 1.87, parity 0.47). There were no notable differences in pregnancy loss in those who tested positive or negative for COVID-19. Six of 94 participants with COVID-19 experienced pregnancy loss (6.4%, 95% CI 2.4, 13.4%) versus 1 of 15 participants who tested negative for COVID-19 (6.7%, 95% CI 0.1%, 31.9%).

Study Strengths

This study collected data prospectively, limiting the impact of recall bias.

Limitations

This US-based study only included 109 participants in total, with only 15 in the comparison group. Given that the PRIORITY study enrolled participants partially via self-enrollment and the authors did not report the enrollment breakdown in this study, it is possible that only the most health-conscious pregnant people enrolled earlier than 14 weeks gestation (for example, only 1 participant reported current smoking). If more health-conscious individuals are also more likely to have better health outcomes (pregnancy loss in this study, 6.4%, is lower than the pre-pandemic known miscarriage rate of 10%, though this risk is not constant over gestational age and most women enrolled after the period of highest miscarriage risk), then these findings may not apply to the overall pregnant population in the US. Finally, the comparison of miscarriage between groups was descriptive and did not adjust for potential confounders including gestational age, previous pregnancy loss, maternal age, COVID-19 severity, and medication or drug use.

Value added

This study included pregnant persons with symptomatic COVID-19 disease primarily within the first trimester, a group for whom limited outcome data has been published. This study sought to examine a question of high clinical relevance for people interested in pursuing pregnancy during the COVID-19 pandemic.

Our take —

This study measured IgM, IgG, and IgA antibodies in blood samples from randomly selected households across the Cordoba department in Colombia and calculated an estimated overall SARS-CoV-2 seroprevalence of 41% during the Fall of 2020 — ranging from 36% in people ages 50-59 years to 56% in adolescents ages 10-14 years. The infection fatality ratio among age-stratified groups was highest among people ages 70+ years (1.5%) and lowest for adolescents ages 10-14 years (0.001%). The overall case fatality ratio was 5.6% — ranging for 26.5% in people ages 70+ years to 0.1% among adolescents. The study time period used to estimate seroprevalence did not span the overall pandemic period from which they drew their mortality data, which may have biased these estimates. However, these data point to a high burden of infection in this population, across all age groups. This evidence also points to a low risk of mortality but frequent risk of infection among children.

Study design

Prospective Cohort

Study population and setting

The study sought to describe the infection fatality rate and estimate seroprevalence of SARS-CoV-2 in a department in Colombia. The Cordoba department in Colombia was selected for this study, spanning 6 cities. All individuals >9 years of age at the time were considered eligible, and households were randomly selected from the catchment area. All COVID-19 deaths in these cities reported between March 2020 and October 9th 2020, were included in the analysis. Seroprevalence was estimated by blood samples for IgM, IgG, and IgA antibodies using dual recognition ELISA, collected between August and September 2020. The case-fatality rate (CFR) and infection-fatality rate (IFR) were calculated using the reported deaths, reported detected COVID-19 cases, and population size by age-group. IFR was the number of COVID-19 deaths divided by the proportion of seropositive infections from the study, and the CFR was calculated by the number of COVID-19 deaths divided by the proportion of reported COVID-19 cases.

Summary of Main Findings

The study found that there were 1,618 COVID-19-related deaths in the department during the study period. There were a reported 661,963 cases of COVID-19 among a population of 1,554,788 people. The study sampled 2,224 individuals in total. The seroprevalence was estimated at 40.8% (n=998 of 2,447 sampled), and ranged from 58% among adolescents 10 to 14 years old after adjustment (n=25), to 35% among people 50 to 59 years old after adjustment (n=60). The highest infection fatality ratio was estimated at 1.5% (95% CI: 1.4 – 1.6%) among people aged 70 years and above, while adolescents aged 10 to 14 years had the lowest at 0.001% (95% CI: 0.0003 – 0.005%). The overall estimated IFR was 0.24% (95% CI: 0.23 – 0.25%) after age adjustment. This was lower than the CFR, which was estimated at 5.6% (95% CI: 5.3 – 5.9%) overall, and ranged from 26.5% among people 70+ years old (95% CI: 25.0 – 28.0%) to 0.1% among 10 to 14 year old adolescents (95% CI: 0.006 – 0.6%).

Study Strengths

The study used seroprevalence data to estimate the infection fatality rate, rather than only examining among cases of disease for the case-fatality rate. The study also included age-stratified estimates to highlight the changes in these rates. They also used 3 types of immoglobulin as indicative of seropositivity, which will likely capture individuals who may not show an immune response on all measures simultaneously. Notably, they study found a very high seroprevalence overall among those sampled (40.8%).

Limitations

The study was limited by using demographic data reflecting the 2020 population, as opposed to the 2021 population. Similarly, they included deaths from the entire pandemic (March to October 2020), while their seroprevalence data was only derived from August to September 2020. Therefore, these estimates may not be accurate, with an inflated number of deaths due to the longer ascertainment period compared to the seroprevalence period. Additionally, individuals with prior infections may not reflect a detectable antibody response depending on the timing of infection and their immune system, which would reduce the number of seropositive individuals and inflate the estimated IFR. Finally, the model specification for their confidence intervals was not clearly explained, and it is challenging to determine if they used valid statistical methods and sample weighting for their study design, and it is not possible to estimate how biased the results may be from this.

Value added

The study reports a high seroprevalence of SARS-CoV-2, including among children, in South America.

Our take —

The highly transmissible Delta variant (B.1.617.2) of SARS-CoV-2, which was first identified from cases in India, has rapidly spread across the globe, replacing the Alpha variant (B.1.1.7) as the dominant lineage in Scotland by the end of May 2021. This study describes how the rise of the Delta variant in Scotland impacted case demographics, hospitalization rates, and vaccine effectiveness. Delta variant infection occurred more frequently among those who were young and affluent. Hospitalization risk was doubled for Delta variant cases when compared to Alpha variant cases, especially among persons with 5 or more comorbidities. Both the Pfizer- BioNTech and Oxford-AstraZeneca vaccines remained effective against breakthrough infection and hospitalization, though this effectiveness was reduced against the Delta variant as compared to the Alpha variant. This study demonstrates that the Delta variant of SARS-CoV-2 poses an increased risk to public health when compared to the Alpha variant, particularly for unvaccinated individuals.

Study design

Retrospective Cohort

Study population and setting

In late spring of 2021, the Delta variant (B.1.617.2) quickly spread across the globe, replacing the Alpha variant (B.1.1.7) as the dominant lineage in Scotland. This study used EAVE II surveillance data collected in Scotland between April 1 and June 6, 2021, to explore the impact of the Delta variant on risk of hospitalization and vaccine effectiveness. The Alpha variant genome contains a small deletion (residues 69-70) that prevents detection of the S gene with routine RT-qPCR diagnostic testing. In contrast, sequencing analysis demonstrated that Delta variant cases were 99% S gene positive and represented 97% of all S gene positive cases during the study period, allowing S-gene positivity to serve as a proxy for Delta variant detection. Cox regression was used to compare time to hospital admission by S gene status. Test-negative analysis was used to determine vaccine effectiveness against SARS-CoV-2 breakthrough infection as a function of S gene status, after adjusting for covariates.

Summary of Main Findings

During the study period, there were 19,543 observed cases of SARS-CoV-2 and 377 hospitalizations for COVID-19; 7,723 (39.5%) cases and 134 (35.5%) hospital admissions were attributed to infection with the Delta variant. Delta variant cases were associated with younger age (5-9 years) and socioeconomic affluence. COVID-19-related hospitalizations were also more frequent among Delta variant cases (HR: 1.85; 95% CI: 1.39-2.47) when compared to Alpha variant cases, after adjusting for comorbidities and demographic/temporal variables. Risk of hospitalization was greatest among individuals with 5 or more comorbidities. Both the Pfizer-BioNTech and Oxford-AstraZeneca vaccines were protective against breakthrough infection of both variants, but with reduced effectiveness against the Delta variant (Pfizer BioNTech: Alpha—92%, Delta—79%; Oxford AstraZeneca: Alpha—73%, Delta—60%).

Study Strengths

This study analyzed the impact of the SARS-CoV-2 Delta variant on hospitalization rates and vaccine effectiveness using a large, comprehensive dataset. Analyses were adjusted for several covariates that may skew the observed associations between variant status and outcomes.

Limitations

The Delta variant did not become dominant in Scotland until close to the end of the study period (on May 19, 2021), so the full extent of this variant’s impact on hospitalization and vaccine breakthrough infection may not have been adequately captured. The study does not identify any mechanisms that lead to increased hospitalization or vaccine breakthrough infection, and it is possible that the Delta variant influences disease presentation and testing pattern; if the Delta variant results in fewer symptomatic cases but more severe illness among those who are symptomatic, individuals infected with this variant who get tested for SARS-CoV-2 would appear to have worse outcomes in terms of hospitalization and vaccine breakthrough infection. In calculations of vaccine effectiveness, authors admit that temporal adjustments may not adequately account for trends in vaccine uptake and variant spread. No formal statistical analysis comparing the effectiveness of the two vaccines to each other was completed due to insufficient sample size for vaccine breakthrough cases.

Value added

This study describes the impact of the SARS-CoV-2 Delta variant (B.1.617.2) on case demographics, rates of hospitalization, and vaccine effectiveness in Scotland between April 1 and June 6, 2021 using nationwide case data.

Our take —

This observational study of 75 lactating women in Spain, available as a preprint and thus not yet peer-reviewed, showed that vaccine type (Pfizer, Moderna or Astra-Zeneca) and previous SARS-CoV-2 infection both affect levels of SARS-CoV-2 specific IgG and IgA antibodies present in breastmilk. These results indicate the need for further study to see if these breastmilk antibodies could provide protection against COVID-19 to infants of vaccinated mothers via breastfeeding.

Study design

Prospective Cohort

Study population and setting

This study was a prospective, observational study conducted in Spain from January through April of 2021. 75 lactating women who were in high-priority groups for vaccination participated. 19 of these women were confirmed to have had past SARS-CoV-2 infection. Participants had either received the Pfizer mRNA vaccine, the Moderna mRNA vaccine, or the Oxford/AstraZeneca adenoviral-vectored vaccine. Breastmilk samples were collected at seven time points: pre-vaccination (0 weeks), 1 week, 2 weeks, and 3 weeks following the first dose. For the two mRNA vaccines, samples were also collected 1 week, 2 weeks, and 3-4 weeks following the second dose. 13 pre-pandemic breastmilk samples were also included in the analysis. SARS-CoV-2 S protein RBD specific IgG and IgA antibodies were measured in the breastmilk samples.

Summary of Main Findings

The authors of this study found that SARS-CoV-2 specific antibodies in the breast milk were dependent on vaccine type and previous exposure to the virus. There was a significant increase in IgG and IgA antibodies after vaccination, especially after the second dose of the mRNA vaccines studied here. Levels were higher compared to the pre-pandemic breastmilk samples. While IgG levels increased even further after a second dose of vaccine, IgA levels did not increase further. No major differences were found between the two mRNA vaccines. IgG levels increased in the breastmilk of women who had previously been infected increased after the first dose of vaccine in a similar manner to the way they would increase after a second dose of vaccine in an uninfected person. IgG levels were higher in the milk of vaccinated women compared to naturally infected women, while the opposite trend was found for IgA. Finally, anti-SARS-CoV-2 IgG antibodies persisted in the breastmilk by the final time point of the study, while IgA antibody levels decreased by 3-4 weeks after the second dose, with only 50-60% of samples testing positive for it by the final time point.

Study Strengths

Since lactating women were not included in vaccine trials, limited data is available about the vaccine in this group. Therefore, this study provides some important and novel information. The study is the first of its kind to investigate the effect of multiple different vaccines on antibody levels in breastmilk. Additionally, this study compared samples from vaccinated lactating women to pre-pandemic breastmilk samples, as well as samples from women naturally infected with SARS-CoV-2.

Limitations

The population included in this study was not very diverse, with over 95% of participants being white. The average age of participants was 34.9 +/- 3.9 years, and therefore few very young or old mothers were included, and age may affect the results. The study was of short duration, and longer-term samples will need to be collected in order to better study the persistence of the antibody response. Importantly, no data was collected after the second dose of the Oxford/AstraZeneca adenoviral-vectored vaccine. Therefore, comparisons cannot be made regarding the second dose of all three vaccines included in this study. Finally, no functional assays were performed in order to measure neutralizing activity of the antibodies found in the breastmilk samples.

Value added

Investigating the effect of COVID-19 vaccination type and previous SARS-CoV-2 infection status on the duration and levels of SARS-CoV-2 specific antibodies in breastmilk.

Our take —

This study describes the seroprevalence of SARS-CoV-2 infection in Iquitos, Peru in July and August 2020. In a baseline July rapid testing round, researchers estimated that the population seroprevalence was 70%. During the August rapid testing round, they estimated seroprevalence to be 66%. While this study may have had reduced sensitivity to detect cases due to the use of rapid testing, the results suggest that Iquitos, Peru has one of the highest seroprevalences of SARS-CoV-2 in the world. Depending on the duration of natural immunity, the seroprevalence reported here would surpass estimated herd immunity thresholds for SARS-CoV-2.

Study design

Prospective Cohort

Study population and setting

This study sought to describe the seroprevalence of SARS-CoV-2 infection and estimate transmission among a population-based cohort in Iquitos, an area of Peru with one of the highest mortality rates from COVID-19. To be eligible, respondents had to have lived in Iquitos at least since SARS-CoV-2 was first identified in Peru on March 16, 2020. The study excluded people in institutions, people receiving treatment for COVID-19, people unable to give blood, and health workers or individuals living with a health worker. Participants were enrolled from July 13 to July 18, 2020, with one follow-up test between August 13 and August 18, 2020. Stratified population sampling was conducted across four levels (four districts, 40 sectors, 2,500 blocks, 90,354 households). Participants aged 13 years and older were eligible, provided guardians consented for participants aged 13 to 17 years. In total, 726 individuals were eligible and 716 participants were enrolled (99%). After one month, 621 participants were tested again (87% of 716). Seroprevalence was estimated using IgG and IgM antibodies assessed via rapid test. Surveys were also conducted to collect data on sociodemographic characteristics.

Summary of Main Findings

At baseline in July 2020, 528 of 716 (74%) participants tested positive for IgM (7%, n=50) or IgG (74%, n=526) SARS-CoV-2 antibodies. After adjusting for sampling procedures, the estimated population seroprevalence was 70% (95% CI: 67 – 73). After follow-up, 621 participants were tested, with 422 (68%) seropositive for either IgM or IgG. The adjusted seroprevalence was 66% (95% CI: 62 – 70%). The incidence of new SARS-CoV-2 seroprevalent cases was estimated at 2% (95% CI: 1 – 3%) in the 1-month follow-up period between test and retest. The test-retest positivity among participants was estimated at 65% (95% CI: 61 – 68%) among those who tested positive at baseline and one month later. Researchers found no difference in seroprevalence between men and women, and among the 619 participants, individuals in urban areas had lower incidence of new cases compared to rural areas (prevalence ratio: 0.21, 95% CI: 0.07 – 0.68).

Study Strengths

This study had data available among almost all eligible participants (99%) and a high retention rate (87%), which reduced the likelihood of selection bias. Additionally, the researchers used survey weights to estimate the population-level seroprevalence, which adds further robustness to the estimates. The study also tested for multiple immunoglobulin types (i.e., IgG and IgM) which captured individuals who would have been missed if they had used a single testing type.

Limitations

This study used rapid tests, which may be less sensitive than a polymerase chain reaction (PCR) test. Therefore, some people with previous infections may have been missed and these results would be an underestimate of the true seroprevalence of SARS-CoV-2. These findings are also impacted by waning immunity, in which individuals who initially have evidence of IgG and IgM antibodies may have reduced or even undetectable antibody levels over time. If participants had previous SARS-CoV-2 infections but undetectable antibodies, then the results presented here would also be an underestimation of true prior infections.

Value added

This study, from a major city in Peru, reports one of the highest seroprevalences of SARS-CoV-2 in the world.

Our take —

This small, prospective study from a nursing home in eastern France followed 31 residents (who had 84% vaccination coverage with BNT162b2/Pfizer) and 59 staff members (32% vaccination coverage) after one resident was diagnosed with the Beta SARS-CoV-2 variant. Vaccination reduced the likelihood of infection with the Beta SARS-CoV-2 variant and any resulting disease severity, however, infections still occurred in this population of older residents and staff members. Additional studies in larger populations will need to replicate and expand on these findings to assess how vaccination impacts the infectivity and severity of the Beta variant in nursing home populations who may not develop as robust of an immune response to the vaccine due to age. Vaccination among residents and staff remains important to reduce infection risks and severity within these vulnerable communities.

Study design

Prospective Cohort

Study population and setting

This prospective study followed 31 residents and 59 staff members at a nursing home unit in Jura, France. Residents were eligible to receive the BNT162b2 (Pfizer) mRNA vaccine starting in January 2021, and 26/31 (83.9%) were fully vaccinated prior to the start of the study. Among staff members, 16/59 (32.2%) were vaccinated. Residents and staff were followed from March 8, 2021, when a resident was diagnosed with the Beta (501Y.V2) SARS-CoV-2 variant until March 29, 2021. All residents and staff were tested at baseline and every seven days, or with the onset of symptoms, with nasopharyngeal swabs until there were no further positive test results. Some positive RT-PCR tests were sequenced with next-generation sequencing. Vaccinated individuals who tested positive for SARS-CoV-2 were also tested for nucleoprotein and receptor-binding domain spike protein antibodies with enzyme immunoassay.

Summary of Main Findings

The included residents had a mean age of 87 years (standard deviation 8.2 years) and were 64.5% female. Between the first Beta variant diagnosis on March 8, 2021 and March 29, 2021, there were 18 total residents and 11 staff members with evidence of SARS-CoV-2; the last positive test occurred on March 15, 2021. Of 10 samples sequenced from residents, all were positive for the Beta variant. Among the 26 vaccinated residents, 13 (50%) were infected, whereas all 5 (100%) non-vaccinated residents were infected. The BNT162b2 vaccine was therefore 50% (95% CI: 34-73%) efficacious against SARS-CoV-2 infection among residents. Among the vaccinated residents, who had various levels of anti-spike antibodies at diagnosis that did not correlate with disease severity, two (15.4%) had asymptomatic disease, nine (69.2%) developed mild to moderate symptoms, and two (15.4%) died after developing severe disease. Among unvaccinated residents, four (80%) developed severe disease and one (20%) died. One (5.2%) vaccinated staff member versus ten (25%) unvaccinated staff members were infected, none of whom developed severe disease.

Study Strengths

This study prospectively followed the transmission of the Beta SARS-CoV-2 variant through a presumably closed population that included vaccinated and unvaccinated nursing home residents and staff, a highly vulnerable population to negative COVID-19 outcomes.

Limitations

This study is very small, making it difficult to generalize their findings to a broader population. The study also presumed a closed population, though samples from only ten of the 29 infected residents and staff members were sequenced to confirm infection with the Beta variant. It is unclear how the original resident was infected with SARS-CoV-2 and whether subsequent infections were from the original resident with the Beta variant or from exposure to contacts outside the nursing home, possibly with another variant in the untested individuals. If, for example, staff members brought a different community-acquired SARS-CoV-2 variant strain into the nursing home, that would make it more difficult to interpret the results of this study as with regard to the BNT162b2 vaccine’s efficacy against infection with the Beta variant of SARS-CoV-2.

Value added

This study suggests that BNT162b2 vaccination substantially decreased but did not eliminate Beta SARS-CoV-2 variant infectivity and severity in a nursing home population.

Our take —

This cohort study, which enrolled 2,130 pregnant people across 18 countries from March to October 2020, found an increased risk of maternal and infant morbidity and mortality among 706 participants diagnosed with COVID-19 compared with similar uninfected controls. Despite the potential for residual confounding and lack of generalizability to pregnant people prior to their third trimester, this study supports the inference that COVID-19 may increase the risk of adverse pregnancy outcomes.

Study design

Prospective Cohort

Study population and setting

This cohort study enrolled 2,130 pregnant people (1 case of laboratory or clinically-diagnosed COVID-19 to 2 sequential uninfected controls from the same clinic) across 18 countries on five continents from March to October 2020. If sequential controls were not available at enrollment, or participants were recruited at a hospital admission for delivery, sequential deliveries among pregnant people without COVID-19 were included as controls. The authors assessed the impact of a COVID-19 diagnosis on maternal and neonatal morbidity and mortality, adjusted for country, month entering study, age, history of medical comorbidities, and, in models that included preterm birth as an outcome, previous preterm birth. Primary outcomes included the maternal morbidity and mortality index (MMMI), severe neonatal morbidity index (SNMI), and severe perinatal morbidity and mortality index (SPMMI), all of which are binary composite outcomes of a variety of maternal and neonatal comorbidities and mortality. The risk of each outcome was evaluated using Poisson models with robust standard errors. Sensitivity analyses adjusted for additional potential confounders (marital status, overweight, smoking, and drug use during pregnancy status), restrictions to pregnant participants with laboratory-confirmed COVID-19, and only singleton pregnancies.

Summary of Main Findings

Of the 706 participants with a COVID-19 diagnosis, 656 (92.9%) had laboratory or radiologic confirmation, 288 (40.8%) were asymptomatic, and 214 (30.3%) had three or more symptoms. The most common symptoms were cough (34.1%), fever (28.2%) and loss of smell (17%). Infected and uninfected participants had similar mean age, mean BMI, years of education, prevalence of work outside the home, and prevalence of cohabiting or being married. Participants with COVID-19 were more likely to use recreational drugs (2.8% versus 1.4%), have a previous stillbirth or neonatal death (5.5% versus 3.3%), have diabetes (4.7% versus 1.4%), and be overweight (BMI > 25 kg/m2; 48.6% versus 40.2%). Participants without diagnosed COVID-19 were more likely to smoke (3.8% versus 2.9%) and have a previous pregnancy (70.9% versus 66.8%). Participants with diagnosed COVID-19 were at a higher risk of MMMI (adjusted risk ratio [aRR] 1.54, 95% CI: 1.33, 1.78), SNMI (aRR 2.66, 95% CI 1.69, 4.18), and SPMMI (aRR 2.14, 95% CI: 1.66, 2.75). These results were similar after sensitivity analysis.

Study Strengths

This study included a large number of women from all over the world, which increases generalizability. They also assessed maternal and infant outcomes in a prospective manner after COVID-19 diagnosis, which limits the influence of recall bias.

Limitations

The authors did not specify the pregnancy trimester for each participant, although they did note that most were in their third trimester, which limits the generalizability of their findings to pregnant individuals diagnosed with COVID-19 earlier in pregnancy. They also did not adjust for a variety of other potential confounders including, but not limited to, trimester or level where participants were enrolled (clinic versus hospital). Although they did adjust for country, there may be a great deal of variability in outcomes depending on the particular location where participants were recruited. Those decisions may lead to residual confounding that could bias their findings in unpredictable ways and, if outcomes were correlated at each recruitment site, artificially narrow confidence intervals. Finally, the authors did not address how they handled missing covariate or outcome data.

Value added

This study provides more robust evidence to support that SARS-CoV-2 infection during pregnancy leads to poor maternal and infant outcomes in a geographically diverse group of women.

Our take —

The study evaluated adverse events among pregnant people, primarily health care workers, who received an mRNA COVID-19 vaccine in the US from December 2020 to February 2021. Using v-safe, the CDC active surveillance system, they found that the most commonly reported reactions were injection site pain, fatigue, and headaches; they did not identify significant differences in vaccine reactions by pregnancy status. In the v-safe pregnancy registry, the proportions of stillbirth, preterm birth, and other adverse birth outcomes were within expected ranges for US samples from prior studies, though the v-safe sample was not representative of pregnant populations in the US and these numbers cannot be compared without accounting for other risk factors. Likewise, VAERS included few reports of adverse events associated with COVID-19 vaccines among pregnant women. Longer follow-up will be required to assess the potential for congenital anomalies and other outcomes potentially associated with vaccination during the first trimester. Despite the limitations of the early surveillance data, these preliminary results suggest that mRNA vaccines are well-tolerated and safe among pregnant people.

Study design

Prospective Cohort

Study population and setting

This study investigated the safety of the messenger RNA (mRNA) COVID-19 vaccines among pregnant individuals, using US post-vaccination surveillance data from those receiving a first or second dose from December 14, 2020 through February 28, 2021. The first study sample was drawn from the CDC v-safe COVID-19 vaccine surveillance system, a new smartphone-based text message and online survey application. Participation in v-safe was voluntary; individuals could register to participate after getting a COVID-19 vaccine. A subset of pregnant v-safe participants was contacted by phone for recruitment into the v-safe pregnancy registry if they were 18 years or older and received the vaccine during pregnancy or periconception (defined as 30 days before the last menstrual period until 14 days after). Registry participants completed a telephone survey on sociodemographic, medical and obstetric history, pregnancy complications, and birth outcomes. Medical records were also obtained for obstetric and pediatric care; infants were followed until age 3 months. Finally, data from the Vaccine Adverse Event Reporting System (VAERS), a longstanding passive surveillance system, was used to describe the number and types of adverse events reported following an mRNA COVID-19 vaccine where pregnancy status was also noted. Descriptive statistics were reported for pregnant persons in each of the three study samples, overall and stratified by vaccine and/or age group.

Summary of Main Findings

In total, 35,691 v-safe participants (61% aged 25-34 years; 76% non-Hispanic white) reported vaccination with either Pfizer-BioNTech or Moderna mRNA vaccines during pregnancy or the periconception period, of whom 16,982 and 12,273 reported a reaction one day following their first and second dose, respectively. The most common reaction was pain around the injection site (after dose 2, 91.9%), followed by fatigue (after dose 2, 71.5%), headache (after dose 2, 55.4%), and body aches (after dose 2, 54.1%). The relative frequency of these reactions was similar among non-pregnant female v-safe participants aged 18-54 years. The v-safe pregnancy registry (65% aged 25-34 years; 94% health care workers) included individuals who were vaccinated in all trimesters of pregnancy (25.7% third trimester, 43.3% second trimester, 28.6% first trimester, and 2.3% periconception). Pregnancy-related outcomes were assessed among the 827 individuals whose pregnancy ended in a live birth or pregnancy loss by the time of analysis. Spontaneous abortion occurred in 104 (12.6%), primarily during the first trimester (96 of 104, 92%). There was one stillbirth. There were 712 pregnancies that ended in the live birth of 724 neonates, including multiples; most received dose 1 in their third trimester.  Preterm birth occurred among 60 of 636 neonates born to those vaccinated before 37 weeks gestation (9.4%) and major congenital anomalies were reported among 16 of 724 neonates (2.2%). No neonatal deaths were reported. In the VAERS, there were 221 reports of adverse events after the COVID-19 vaccine among pregnant people, of which 66 (29.9%) involved pregnancy or neonatal outcomes. The most frequently reported pregnancy-related adverse event was spontaneous abortion (46 of 221, 20.1% of adverse events reported) and there were three reports of stillbirth. No congenital anomalies were reported in VAERS.

Study Strengths

The study triangulated data from several sources, including the new v-safe active surveillance system, the associated v-safe pregnancy registry, and the VAERS to rapidly disseminate early data on adverse events among pregnant persons. The v-safe pregnancy registry followed individuals prospectively to assess pregnancy and perinatal outcomes, and confirmed outcomes through medical records. Data was available from both the Pfizer-BioNTech and Moderna vaccines, giving insight into both mRNA vaccines and enabling comparisons of adverse events between the two.

Limitations

Due to the short study period, most pregnancies were ongoing at the time of analysis; therefore, the data on pregnancy-related and neonatal outcomes was limited to a smaller sample primarily vaccinated in the third trimester. Assessment of congenital anomalies will require longer follow-up to observe outcomes among those vaccinated during the first trimester of pregnancy during organ development. Since the study collected data during a timeframe when vaccine availability was limited to prioritized groups, the source populations were not representative of the pregnant population in the US and the v-safe sample was further self-selected to those who proactively registered online. There is an overrepresentation of non-Hispanic white respondents and health care workers among the v-safe samples, and the observed frequency of negative pregnancy and neonatal outcomes among those in the registry may not be generalizable to populations with more risk factors for these outcomes. The study lacked a comparison group of pregnant women with similar risk profiles who did not receive the vaccine, and the authors acknowledged that the external estimates from prior studies are not directly comparable. Finally, the adverse events reported through VAERS are difficult to interpret given the selectivity of reporting.

Value added

This is the largest study with real-world data examining adverse events among pregnant people who received an mRNA vaccine, filling an important knowledge gap for this population, which was excluded from the initial vaccine trials.

Our take —

Despite considerable attention paid to long-term symptoms of COVID-19, few studies have used a well-defined control group of people without COVID-19. Of 323 SARS-CoV-2 seropositive and 1072 seronegative healthcare workers in Stockholm who were enrolled in April-May 2020 and followed through January 2021, the prevalence of self-reported moderate to severe symptoms lasting for more than 8 months after enrollment was markedly higher among seropositive individuals (15% vs. 3%). Nine percent of seropositive participants reported anosmia (loss of smell) lasting at least 8 months. These findings provide strong evidence that long-term symptoms are associated with SARS-CoV-2 infection, as they were more common than those observed in a well-defined reference group, though results should be interpreted cautiously given the reliance on self-reported data.

Study design

Retrospective Cohort; Prospective Cohort

Study population and setting

This longitudinal study included healthcare workers at Danderyd Hospital in Stockholm, Sweden who were enrolled between April 15 and May 8, 2020. Blood samples, which were obtained at baseline and prospectively collected every four months thereafter, were analyzed for SARS-CoV-2 IgG antibodies, reflecting previous infection. Participants reported demographics, COVID-19 symptoms, and COVID-19 severity at baseline and again after 8 months of follow-up, when they described the duration of symptoms (<2 months, 2 months or more, 4 months or more, and 8 months or more) and severity (mild, moderate, severe) for 23 predefined symptoms. Individuals with detectable SARS-CoV-2 IgG antibodies at baseline who reported severe symptoms were excluded, as were individuals who were initially seronegative but converted during follow-up. Among the 2149 participants initially enrolled, only 1395 were included in the analysis, including 323 who were seropositive for SARS-CoV-2 IgG antibodies at baseline and 1072 who were seronegative. Exclusion reasons were: severe symptoms reported at baseline (n=50), seroconverted during follow-up (n=404), and did not complete 8 month follow-up (n=300; 20 seropositive and 280 seronegative).

Summary of Main Findings

Seropositive and seronegative individuals were relatively similar with regard to age (median age: 43 years vs. 47 years), sex (84% female vs. 83% female), and prevalence of underlying chronic diseases (22% vs. 24%). Overall, 26%, 21%, and 15% of seropositive individuals reported at least 1 moderate to severe symptom lasting at least two, four, or eight months, respectively, compared to only 9%, 7% and 3% of seronegative individuals. Seropositive individuals had 4.4 times the risk of having moderate to severe long-term symptoms lasting at least 8 months, compared to seronegative individuals (RR=4.4, 95% CI: 2.9-6.7). For seropositive individuals (vs. seronegative individuals), the most common moderate or severe symptoms at 8 months or later were anosmia (loss of smell; 9% vs. 0.1%), fatigue (4% vs. 1.5%), ageusia (loss of taste; 3.7% vs. 0.1%), and dyspnea (shortness of breath; 1.9% vs. 0.3%). Seropositive patients (vs. seronegative patients) were additionally more likely to report moderate-to-marked disruptions to work life (8% vs. 4%), social life (15% vs. 6%), and home life (12% vs. 5%).

Study Strengths

This study longitudinally followed individuals for more than 8 months, regardless of SARS-CoV-2 infection status, allowing for estimation of the excess population-level prevalence of long-term symptoms due to COVID-19.

Limitations

Symptom duration and severity during follow-up were retrospectively self-reported, which may have resulted in a higher reporting of more severe symptoms among people who knew they previously had COVID-19. It is additionally possible that some individuals in the seronegative group had had COVID-19 but did not have detectable SARS-CoV-2 IgG antibodies during follow-up, resulting in misclassification of their exposure. The resultant bias would likely be towards the null, leading to stronger associations between SARS-CoV-2 seropositivity and long-term symptoms, unless individuals with falsely negative antibody tests had milder symptoms. Additionally, 6% of seropositive individuals and 20% of seronegative individuals did not complete follow-up, which may have resulted in some selection bias. Similarly, differential participation may have induced bias in an unpredictable direction; for example, if seronegative individuals with long-term symptoms due to other conditions were more likely to participate than seronegative individuals without any long-term symptoms, this could bias the results toward the null. All analyses were unadjusted, and despite comparability between seropositive and seronegative individuals on several baseline characteristics, residual confounding is likely and caution is warranted interpreting the relative risks as causal.

Value added

This is one of the first studies to describe long-term outcomes of COVID-19 that included a well-defined control group with similar follow-up.