Skip to main content

Our take —

In this brief report from Geneva, Switzerland, at least one-third of non-hospitalized COVID-19 patients reported persistent symptoms at an average of 43 days after diagnosis via telephone interviews. The most common persistent symptoms were fatigue, difficulty breathing, and loss of taste or smell. This study adds to the accumulating evidence base on long-term COVID-19 symptoms, though studies that follow larger groups of people over longer durations are needed to enhance our understanding of the natural history of this disease.

Study design

Case Series

Study population and setting

The authors described longitudinal COVID-19 symptoms among 669 symptomatic ambulatory patients (60% female, mean age 43 years) who were initially diagnosed at a single hospital in Geneva, Switzerland from March 18 to May 15, 2020. Patients who were not hospitalized were referred to ambulatory care centers for remote follow-up. Patients were telephoned every 48 hours for the first two days after diagnosis, then once from 30 to 45 days after diagnosis. Standardized interviews regarding self-reported symptoms were performed at each telephone contact. The authors considered two-day intervals for analysis during the first 10 days of follow-up.

Summary of Main Findings

Of all patients testing positive during the study period, 22% were hospitalized and 703 patients were enrolled in remote follow-up. Of the 669 eligible patients, 25% were healthcare workers and 31% had at least one underlying comorbidity. The number of patients reached during the first 10 days diminished due to hospitalization, patients declining follow-up, and clinical recovery (e.g., 376 of 669 patients [56%] were interviewed 9-10 days after diagnosis). Forty patients (6%) were hospitalized during follow-up. The authors attempted to reach all 669 patients during the day 30-45 interval; 510 patients (76%) were interviewed at a mean of 43 days after diagnosis. During this 30-45 day interval, 32% of the original cohort of 669 patients were reached and reported at least one persistent symptom, while 24% of patients could not be reached. The most commonly reported persistent symptoms were fatigue, dyspnea, and loss of taste or smell.

Study Strengths

A reasonably large sample of outpatients provided information on symptoms at least 4 weeks following COVID-19 diagnosis.

Limitations

Those enrolled in remote follow-up represented a small proportion (16%) of all non-hospitalized symptomatic patients testing positive for SARS-CoV-2 infection in Geneva during the study period, and may not be representative of the larger target population. For example, included patients may have been more likely to initially experience dyspnea or loss of taste or smell, making them more concerned than other patients about longer-term effects. Detailed information about patient demographics and comorbidities were not reported, making it difficult to apply these data to other patient populations. Loss to follow-up was considerable, so the prevalence of persistent symptoms more than 30 days after diagnosis must be regarded as a lower bound. Finally, symptom prevalence was assessed via self-report, which is subject to possible over-or under-ascertainment.

Value added

This is one of the few studies to date assessing symptoms among non-hospitalized COVID-19 patients, particularly beyond 30 days after symptom onset.

Our take —

The study sought to examine whether SARS-CoV-2 infections occurred in archived oropharyngeal samples from the Measles and Rubella network obtained from September to December 2019. They found one sample tested positive obtained December 5, 2019, from a 4-year-old boy who did not report a history of travel. While there may be some sample quality issues that biased findings towards the null in other samples, this finding suggests that SARS-CoV-2 was circulating in the population months before the first reported clinical case on February 21, 2020.

Study design

Case Series

Study population and setting

In order to better understand when SARS-CoV-2 infections first spread to Italy, patients identified through the Measles and Rubella Network and suspected of having measles but testing negative for measles during September 2019 to February 2020 were included in the study. Oropharyngeal swabs from 39 patients were tested for SARS-CoV-2 RNA using an in-house PCR assay for the SARS-CoV-2 spike protein.

Summary of Main Findings

Patients tested ranged in age from 8 months to 73 years with a mean of 19.9 years. From these 39 samples, one swab tested positive in a 4-year-old boy who resided nearby Milan with no reported travel history. On November 21, 2020, the child began have symptoms of cough and rhinitis. On November 30, the child developed respiratory symptoms and vomiting which developed into a measles-like rash on December 1. The swab was obtained on December 5, 14 days after symptom onset.

Study Strengths

The study strength was in the use of archived samples with detailed clinical and travel history. In addition, the partial viral sequence recovered from the one positive sample provides further confidence in the validity of the PCR result.

Limitations

The study limitations are the length of time since the samples were collected, as degradation may have occurred in these samples which would bias the study toward null findings, and type of specimen collected from patients – oropharyngeal swabs – which are suboptimal for SARS-CoV-2 testing. Additionally, similar to the patient who had a positive swab, other patients may have had delays between symptom onset and sample collection, again biasing toward null findings. Finally, the study was not able to recover a full length genome sequence from the positive sample, precluding interpretations of the origin of the infecting virus.

Value added

This study shows that SARS-CoV-2 infections were occurring in Italy nearly 3 months before the first reported case on February 21, 2020.

Our take —

This study highlights the risk of indoor dining, even when following COVID-19 safety protocols. While the index case (case A) was identified first, evidence from anemometer and CCTV readings show that most likely case B transmitted SARS-CoV-2 infection to case A and case C, each sitting 6.5m and 4.8m away from case B respectively. While it is not possible to confirm causality, using the Epidemic Investigation Support System which tracks COVID-19 cases throughout the country, this was the only exposure overlap between these cases where transmission likely occurred. Phylogenetic evidence supports transmission between these cases. This shows the importance of air flow management in indoor spaces and the risk for droplet transmission across longer distances than previously anticipated.

Study design

Case Series

Study population and setting

Following reports on June 17, 2020, that an index COVID-19 case in Jeonju, Korea, transmitted COVID-19 to two individuals 6.5m away (21 feet) in an enclosed space during 5 minutes of exposure in an air-conditioned restaurant, the Korea Disease Control and Prevention Agency began an epidemiological and environmental investigation. Patients were interviewed for sociodemographic and exposure data, medical facility use history and credit card use. Closed circuit television images of the exposure incident were reviewed, and cell phone location data were also used to assess exposure. Nasopharyngeal specimens of cases and close contacts were tested using real-time reverse transcription polymerase chain reaction (rRT-PCR) and genome sequencing was conducted. They also conducted an environmental analysis, which included recording dining table locations, timeline, type of air conditioning, distance between individuals, and the movement route of the index case and other people in the restaurant. Air speed of the air conditioning was measured with a portable anemometer, and 39 samples were taken from the environment to test for SARS-CoV-2 and analyzed with rRT-PCR.

Summary of Main Findings

The first identified case (case A) first developed symptoms on June 16, 2020, with a probable exposure as ascertained from their personal interview between June 2 and 15. Using the Epidemic Investigation Support Symptom (EISS), only one individual (case B) out of 538 domestic cases at the time was identified as having overlap with the index case based on the tracking map, at restaurant A on June 12. Using CCTV, the investigators concluded they spent at total of 5 minutes in proximity to one another. Among 13 close contacts of case A that had to quarantine for 14 days, one additional case was confirmed on June 20. Additionally, case B seems to have spread COVID-19 to case A and case C (two people total) at the restaurant. The investigators concluded this because case B typically lives in Daejeon City and only visited Jeonju on June 12th.The attack rate was 15.4% overall (2/13). Of the 39 environmental samples collected 11 days following the sentinel event, all tested negative by rRT-PCR. In the restaurant, the index case and case B were seated 6.5m from one another, and case B and case C were sitting 4.8m from one another. The anemometer measured a maximum airflow of 1.0m/sec between case A and B and 1.2 m/sec between case B and C. The study posits that the indoor air flow may have transmitted droplets from case B to case A and C based on the direction of the air conditioner flow.

Study Strengths

The study used a wide range of data sources, from the EISS, to the anemometer readings, to CCTV images, etc. By having a relatively low country-wide incidence and strong testing and contact tracing program, they were able to track each of these individuals and identify the potential contacts and modes of exposure with a high degree of certainty.

Limitations

The notable limitation was that the environmental samples collected were done so 11 days after the initial investigation began, which may have resulted in a negative finding at the time of collection. The study also did not engage in more sophisticated modeling of airflow dynamics, and cannot reflect the totality of factors (e.g., the impact of people opening doors, the impact of server movement throughout the restaurant), and therefore they cannot determine the causality of air flow dynamics in the restaurant. Finally, the study concluded that case B likely was the index case that transmitted the virus to case A and C, however that was determined by logical reasoning of their overlap in the restaurant compared to case B living in another city, but not because case B tested positive before cases A or C, or some other empirical evidence. Therefore, this conclusion relies heavily on assumptions that may not hold in the real world.

Value added

This case series shows the risk of indoor dining even with distancing and during very short periods of exposure.

Our take —

Among 106,543 patients discharged at one of the 865 US hospitals in the Premier Health Care Database between March and July 2020, 9% were readmitted to the same hospital within two months. Factors associated with readmission included older age, several chronic conditions, and discharge to a skilled nursing facility or home health. These data may be helpful to allocate healthcare resources for follow-up care of COVID-19 patients. However, reliance on electronic medical records and heterogenous readmission reasons underscore the need for more research to understand the long term-sequelae, including hospital readmissions, of COVID-19.

Study design

Case series; Retrospective Cohort

Study population and setting

The Premier Healthcare Database, which collects data from 865 US hospitals, was used to estimate readmission rates among COVID-19 patients who were hospitalized between March and July 2020 with follow-up through August 2020. COVID-19 diagnoses were obtained from ICD-10-CM codes, and readmission for COVID-19 or other health complications that occurred within two months of the initial hospitalization were considered. Chronic conditions were identified from ICD-10-CM during or before the initial hospitalization, and COVID-19 disease severity was defined by hospital billing records (ICU admission, invasive mechanical ventilation, or noninvasive mechanical ventilation).

Summary of Main Findings

Of the 126,137 patients who were hospitalized for COVID-19 between March and June 2020, 106,543 were discharged alive, and among those, 9,504 (9%) were readmitted to the same hospital in the following two months (median time to readmission: 8 days, IQR: 3-20), and 1.6% of people had multiple readmissions. In multivariable analyses, readmission was more common for individuals with older age (age >65), white race, specific chronic conditions (COPD, heart failure, diabetes, and chronic kidney disease), a history of hospitalization in the three months prior to the admission for COVID-19, and who were discharged to a skilled nursing facility or home health organization. The most frequent discharge diagnoses for readmissions were infectious and parasitic diseases (45%), circulatory (11%), and digestive (7%)

Study Strengths

The sample size was large, and the study included a large number of sites from throughout the US. The multivariable analysis adjusted for a number of important confounding factors, including prior hospitalizations.

Limitations

There is no referent group, so it is not clear how the rate of readmission compares to readmission from other viruses; based on quick literature review, between 8 and 27% following hospitalization for severe pneumonia among Medicare enrollees. ICD-10-CM codes were used to define COVID-19 diagnosis and chronic conditions, which is subject to misclassification, likely under-reporting. Only patients readmitted to the same hospital were considered as readmissions, which may have underestimated the overall rate of readmission. More than half of discharge diagnoses from the hospital readmission were for reasons other than parasitic or infectious diseases (including COVID-19), and the study did not distinguish between readmissions based on their likely relationship (direct or indirect) to COVID-19.

Value added

This was a large, multi-center study, estimating the rate of readmission among patients discharged after hospitalization with COVID-19.

Our take —

This case series of neonates in the United Kingdom identified between March 1 and April 30, 2020, suggests that, at a population-level, SARS-CoV-2 infections are rare among neonates (~5.6 per 100,000 live births), even when COVID-19 cases are rising rapidly in a population. Clinical complications in neonates were higher than in older infants and children; however, neonates were only tested for SARS-CoV-2 infection in cases of suspected maternal infection, which may have overestimated the amount of severe disease in this study population.

Study design

Case Series

Study population and setting

This case series described characteristics and outcomes of neonates (aged 0-28 days) with confirmed SARS-CoV-2 infection who received inpatient care between March 1 and April 29, 2020 in the United Kingdom (UK). Neonates with a positive SARS-CoV-2 test were identified through the ongoing British Paediatric Surveillance Unit (BPSU), a surveillance program to study rare neonatal and pediatric diseases, as well as through data from Public Health England, Health Protection Scotland, the Paediatric Intensive Care Audit Network, and the UK Obstetric Surveillance System. All BPSU case reports were accompanied by a data collection form and BPSU staff manually collected data on non-BPSU reported cases. Severe disease met two of the following criteria: 1) one of temperature > 37.5C, apnea, cough, rapid breathing, respiratory distress, supplemental oxygen, poor feeding, diarrhea, or vomiting; 2) any of low white cell count, low lymphocyte count, or high c-reactive protein; and 3) abnormal chest x-ray. They presented descriptive statistics for characteristics of interest and calculated the incidence of neonates who received hospital care with COVID-19 using complete national birth data from the United Kingdom from 2018 (the most recent available full year data).

Summary of Main Findings

Among 118,347 live births in the UK between March 1 and April 30, 2020, including more than 300 mothers with confirmed SARS-CoV-2 infection, a total of 89 neonates with SARS-CoV-2 infection were identified in UK hospitals, resulting in , an incidence of 5.6 (95% CI 4.3-7.1) per 100,000 live birth. Twenty-eight of the neonates were diagnosed with severe disease (incidence 2.4 (95% CI 1.6-3.4) per 100,000 live births). SARS-CoV-2 incidence was higher among Black, Asian, or other minority ethnic group neonates (11.1/100,000; 95% CI 7.4-15.9), than white neonates (4.6/100,000; 95% CI 3.2-6.4). A majority (n=34; 52%) had a close contact with COVID-19 symptoms and seventeen (26%) were born to a mother who tested positive for SARS-CoV-2 within seven days of birth (before or after). There were two cases of suspected vertical transmission and 8 cases of suspected healthcare-acquired infection. The most common symptoms included temperature > 37.5C, poor feeding, or vomiting; seven (11%) neonates were asymptomatic and only tested because of their mother’s symptoms. Most (42, 64%) received care in the postnatal ward or pediatric unit, 20 (30%) received care in a neonatal unit, and four (6%) received care in the neonatal intensive care unit. Although 33% (n=22) required respiratory support, three requiring mechanical ventilation, none died from SARS-CoV-2, although one died from another cause.

Study Strengths

This study leveraged population-level surveillance data and additional data sources to maximize the identification of neonatal cases of SARS-CoV-2 infection in the United Kingdom.

Limitations

This study used total population live births in their denominator and therefore did not calculate the incidence of neonatal SARS-CoV-2 infection among mothers with COVID-19. Furthermore, the surveillance system relied on health infrastructure that was strained by a global pandemic, which may have led to an underascertainment of cases and an underestimate of the neonatal SARS-CoV-2 population estimate. This may have been exacerbated by not testing for asymptomatic neonates except in cases of suspected maternal infection, which would also likely overestimate the complication rate of SARS-CoV-2 in neonates, as the results may only capture those with severe disease. Finally, the lack of a comparison group, control for potential confounding variables, longitudinal follow-up and this study’s small sample size precludes inferences about causality or risk factors and means it’s possible these findings may not generalize to all neonates with milder or less severe SARS-CoV-2 infection.

Value added

This study provided a population-level estimate and characterization of SARS-CoV-2 infections among neonates in the United Kingdom.

Our take —

This study was published as a preprint and thus was not yet peer reviewed. In September 2020, a case of SARS-CoV-2 re-infection was observed in a healthy, Belgian healthcare worker in her 30s, despite a documented neutralizing antibody immune response following the first infection in March 2020. While both the primary and secondary infections were symptomatic and relatively mild, the second infection was characterized by a faster immune response and milder, less prolonged infection. Differences in the genetic sequences of the two infections, often unavailable amongst COVID-19 cases, support that these were two distinct infections. Further, at the time of the second infection, neutralizing antibodies were not detected in the nasopharyngeal swab of the health care worker, and it is likely that she was the transmission link between three other genetically linked cases within the hospital ward who had no contact with one another. These findings should be interpreted as an initial piece of evidence to help us understand how long protective immunity may last, as well as provide insights into the potential for transmission among reinfection cases to others, despite initial development of neutralizing antibodies. To date there have been 26 cases of re-infection clearly documented through genetic testing of both infections, each case providing new insights into the COVID-19 pandemic.

Study design

Case Series

Study population and setting

This study describes a case of re-infection in a healthcare worker in Belgium in September 2020, along with the associated cases in a hospital-based outbreak in an internal medicine ward. Whole genome sequencing was conducted to distinguish between persistent infection and re-infection in the re-infected case, as well as to establish chains of transmission in the outbreak.

Summary of Main Findings

The re-infected healthcare worker (HCW) was female, young (30s), healthy, and immunocompetent. During her initial infection in March 2020, she had a mild but prolonged infection in which she presented with cough, fever, headache, malaise and labored breathing and was away from work for a month. Testing three months after infection demonstrated that she had developed neutralizing antibodies. The re-infection in September 2020 was also symptomatic. While the HCW was symptomatic in both instances, the second episode was more mild and there was a rapid immune response (fast rise in serum IgG and neutralizing antibodies). Of the four other people (3 patients and 1 healthcare worker) involved in the hospital outbreak in September 2020, the three patients shared a viral genome sequence similar to that of the re-infected healthcare worker, while the other healthcare worker appears to have been infected by an external transmission source. The sequences collected from the re-infected healthcare worker and the three patients closely resembled Belgian sequences obtained in the summer months, while the sequence from the re-infected healthcare worker from her March infection resembled sequences circulating at that time. In September, the re-infected healthcare worker’s initial nasopharyngeal swabs contained virus but no neutralizing antibodies and she was the primary point of contact that was infected across the patients, suggesting likely onward transmission, though no transmission to close contacts outside of the hospital was documented.

Study Strengths

The use of whole genome sequencing, rather than a threshold number of days, allows for the cleaner distinction between re-infection and persistent infection (viral genomes are derived from different SARS-CoV-2 clades in the first and second episode). Also, because antibodies were measured 3-months post-initial infection, the authors were able to determine that the woman had at least initially developed neutralizing antibodies.

Limitations

Samples were not available from the HCW re-infected right before the second infection; thus it is not known whether the neutralizing antibodies previously measured had waned by this time (6 months post initial infection) or if they were present but just insufficient to prevent re-infection. Further, this is a single case and may not be representative of all patients with a primary COVID-19 diagnosis.

Value added

Description of this case can help us begin to understand how long protective immunity (even with an effective immune response) may last following a primary infection. These data also provide insights regarding the potential for transmission from re-infected cases to others.

Our take —

Among more than 400,000 US women of reproductive age with symptomatic COVID-19 in a national surveillance database, pregnant women had three times the risk of ICU admission and 1.7 times the risk of death compared to non-pregnant women of similar age. Although severe COVID-19 outcomes were rare, these findings suggest that pregnancy may be independently associated with COVID-19 severity. Unfortunately, the data were severely limited by missing pregnancy status among many women excluded from analysis, and by potentially underreported outcomes among those included; inferences about elevated risks associated with pregnancy are subject to therefore uncertainty.

Study design

Case Series

Study population and setting

This study compared COVID-19 disease severity outcomes between pregnant and non-pregnant women aged 15-44 years with symptomatic PCR-confirmed SARS-CoV-2 infections reported through national COVID-19 case surveillance or the National Notifiable Diseases Surveillance System from January 22 to October 3, 2020. The surveillance data included 1.3 million female cases aged 15-44 years, and pregnancy status was documented on a third of these case report forms (461,825). Among those with known pregnancy status, COVID-19 symptoms were documented in 409,462 case reports, yielding a study sample of 23,434 pregnant and 386,028 nonpregnant women. All outcomes and other data were ascertained from the case reports, reported by public health agencies and health care providers according to local procedures for reporting and investigating cases. The analyses incorporated all updates to these case reports submitted by October 28, 2020.

Summary of Main Findings

Pregnant women with symptomatic COVID-19 were more likely than nonpregnant women of reproductive age to be admitted to the intensive care unit (ICU) (10.5 v 3.9 per 1,000 cases), receive invasive ventilation (2.9 v 1.1 per 1,000 cases), or receive extracorporeal membrane oxygenation (ECMO) (0.7 versus 0.3 per 1,000 cases). In multivariable Poisson regression adjusted for age, race/ethnicity, and underlying comorbidities, the risk among pregnant compared to nonpregnant women was 3 times higher for ICU admission (adjusted RR [aRR] 3.0; 95% CI 2.6–3.4), 2.9 times higher for mechanical ventilation (95% CI=2.2–3.8), and 2.4 times higher for ECMO (95% CI 1.5–4.0). There were 34 deaths reported among pregnant women and 447 deaths among nonpregnant women, corresponding to an adjusted risk ratio of 1.7 (aRR 1.77 95% CI 1.2-2.4; 1.5 v 1.2 per 1,000 cases). The risk of death was highest among black women, independent of their pregnancy status.

Study Strengths

This is the largest study of COVID-19 severity among pregnant women in the US, and the surveillance data captures cases outside of hospital settings. Importantly, these data include a comparison group of nonpregnant women of reproductive age with whom to directly compare COVID-19 outcomes by pregnancy status. The large sample size enabled the researchers to explore subgroup analyses by age group and race/ethnicity.

Limitations

The primary limitation of this study was the high proportion of missing data (including data on outcomes, pregnancy status, underlying comorbidities, and race/ethnicity) and the variable data quality within the passive surveillance system. The majority (65%) of COVID-19 cases among women aged 15-44 years had unknown pregnancy status and were excluded from analysis; if these women differed systematically from those with known pregnancy status, results could be substantially biased. Resource constraints and heterogenous procedures likely limited outcome ascertainment, particularly for outcomes which may occur weeks after the case was originally reported. The assumption that missing outcome data (25-70%) were non-outcomes likely underreported outcomes and may have biased the associations. The inferences about the risk of death were most susceptible to this misclassification bias given their small numbers. The authors do not comment on whether obstetric settings might be more likely to report outcome data than non-obstetric settings, which would inflate risks associated with pregnancy. Though the analyses adjusted for important variables affecting disease severity, there may be residual confounding in the association between pregnancy and disease severity due to misclassification of underlying comorbidities (e.g. 50% missing) and other covariates in the case reports. Finally, it is possible that health care providers may have had a lower threshold for interventions in symptomatic pregnant women (e.g., ICU admission) relative to non-pregnant women with otherwise similar clinical presentations.

Value added

This large study included a comparison group of nonpregnant women with COVID-19 and adds to the limited evidence on whether COVID-19 severity differs by pregnancy status, much of which has been conducted in hospital samples and without an appropriate nonpregnant comparison group.

Our take —

In March 2020, there was an outbreak of SARS-CoV-2 at a nursing home in the Netherlands. Despite an elevated secondary attack rate among church attendees and the timing of the outbreak following a church service, detailed epidemiologic investigation and whole genome sequencing revealed that the outbreak was likely due to many different, complex introductions into the community, including but not limited to a church service. Sequencing of positive samples can be a helpful tool for understanding the sources of transmission, and insights gained from these studies can be used to create an evidence-based prevention strategy for future outbreaks.

Study design

Case Series

Study population and setting

In early March 2020, an outbreak of SARS-CoV-2 was detected at a nursing home in the Netherlands. This outbreak occurred following an on-site church service. The goal of this study was to retrospectively create chains of transmission resulting from this outbreak and to evaluate the role of the church service in onward transmission. Symptomatic residents and healthcare workers of the nursing home as of March 14, 2020 were tested for SARS-CoV-2, and those testing positive prior to April 15 were included in these analyses. Non-residents who attended the church service were asked to complete a questionnaire. Contact tracing data were made available from the public health service (date of symptom onset, testing date, church attendance etc.) and mitigation measures and a map of the facility were provided by the nursing home. Whole genome sequencing was conducted on positive SARS-CoV-2 samples and compared with samples from the same area.

Summary of Main Findings

A total of 39 people attended on-site church service on March 8, 2020, including 13 residents and 26 non-residents. 77% (30/39) developed symptoms following the service, and 14/39 tested positive for SARS-CoV-2 (36%). Those attending the service were together for approximately 70 minutes (50-minute service with singing and eating and 20-minute coffee). Between March 8 and April 15, 21% (62/300) of residents and 5% (30/640) of healthcare workers tested positive for SARS-CoV-2. Among those residents who tested positive, 34% died (21/62). Among residents, the attack rate for churchgoers was 85% (11/13) and for non-churchgoers was 18% (51/287). Sequencing revealed that 7 of the churchgoers were infected with distinct viruses and that there were at least 17 different introductions of SARS-CoV-2 in the nursing home, suggesting that the outbreak was more complex than simply a seeding caused by the church event.

Study Strengths

Because the samples were sequenced and detailed comparisons made to samples in the same area, it was possible to determine that the church event was not the only source of the observed outbreak.

Limitations

Limited data were provided on the characteristics of the resident churchgoers and the resident non churchgoers. These data would have been helpful in understanding why resident churchgoers experienced a higher attack rate than resident non-churchgoers.

Value added

This study provides further evidence that complex, evidence-based mitigation strategies are needed to combat outbreaks for SARS-CoV-2. In this case, eliminating the church service alone would not have been enough to prevent an outbreak as transmission within the nursing during this time was already widespread.

Our take —

The degree of immunity acquired by individuals after SARS-CoV-2 infection, and how it changes over time, is a matter of considerable public health concern. In contrast to a previous study showing rapidly waning levels of antibodies that target a different SARS-CoV-2 antigen among patients with mild symptoms, this study from New York City shows that neutralizing antibodies to SARS-CoV-2 spike protein diminished over time but remained at high levels up to 3 months after symptom onset in a group of patients with mostly mild illness. The patient population was small and not well described, so it is unclear if these results apply to the broader population of those infected with SARS-CoV-2. Important questions remain about how varying levels of SARS-CoV-2 antibody translate to protection from infection.

Study design

Case series; prospective cohort; other

Study population and setting

This paper reports on three related studies: 1) Enzyme linked immunoassay (ELISA) testing for IgG antibodies to the SARS-CoV-2 spike protein among 72,401 individuals with laboratory-confirmed or suspected infection approximately 30 days after symptom onset from the Mount Sinai Health System in New York City, from March to October 6, 2020; 2) testing of 120 samples that had a known ELISA titer for neutralization of SARS-CoV-2 using a quantitative neutralization assay; and 3) longitudinal screening of 121 patients for IgG antibodies to the SARS-CoV-2 spike protein at two additional time points (approximately 82 and 148 days after symptom onset) after the initial screening (approximately 30 days after symptom onset).

Summary of Main Findings

Less than 5% of all individuals screened required hospitalization or emergency room evaluation. Of those screened, 30,082 (42%) tested positive for detectable antibodies to the SARS-CoV-2 spike protein at a titer of 1:80 or higher. Most of those testing positive had moderate-to-high titers (defined as 1:320 or higher): 2.3% had a titer of 1:80, 4.8% of 1:160, 22.5% of 1:320, 31.8% of 1:960, and 38.6% of 1:2880. Neutralizing titers significantly correlated with ELISA titers (Spearman’s r=0.87). Half of sera with spike-binding titers between 1:80 and 1:160 had neutralizing activity, while 90% of sera with 1:320 titers and all those with 1:960 titers or above had neutralizing activity. Among the 121 individuals sampled over a total of three time points (at an average of 30, 82, and 148 days post-symptom onset), the geometric mean titers (GMT) declined from 764 to 690 to 404. Among those with a 1:320 titer or lower, antibody titers increased on average at the second time point, followed by a decrease at the third time point. Three individuals with initially low titers (1:80) lost reactivity, one at the second time point and two at the third. The correlation between neutralizing and ELISA titers remained high at the third time point (r=0.79).

Study Strengths

This study examined antibodies to the spike protein of SARS-CoV-2, which are likely to be more relevant to immunity than antibodies to nucleoprotein. Antibody titers were measured in a large number of individuals using an assay with high accuracy in a validation panel. Longitudinal analyses of antibody titers considered two time points after the initial assay, allowing for discernment of nonlinearities in trajectory.

Limitations

Patient demographic and clinical characteristics were not reported, which makes it difficult to interpret how these data apply to any particular group of individuals. Of particular concern is the lack of data on COVID-19 clinical severity in either the larger study population or in the two substudies. Thus, while it appears that these studies were conducted on primarily mild cases of COVID-19, this study did not assess any relationship between disease severity and antibody response. Also, the timing of initial antibody screening relative to symptom onset or date of potential exposure was not characterized in the larger study population and there was variation in the timing of antibody assays in the longitudinal study. Not all individuals screened for antibodies were tested for SARS-CoV-2 infection, which means an unknown number of people who had been infected with SARS-CoV-2 may have tested negative for antibodies. The size of the population in the longitudinal study (n=121) and the lack of reported demographic or clinical characteristics makes it difficult to generalize the results.

Value added

This study provides some of the strongest evidence to date regarding the persistence of neutralizing antibodies to SARS-CoV-2 over time, particularly among those without severe disease. This has implications for both pandemic planning and vaccine development.

Our take —

Among 2,475 pregnant women in Brazil with indications of severe COVID-19, a large proportion (24%) either died, was admitted to the ICU, or required mechanical ventilation. In addition to comorbidities known to be associated with adverse COVID-19 outcomes, Black ethnicity and having poor access to health care were risk factors for poor outcomes. The study highlights how social vulnerability and delays in care can lead to severe COVID-19 outcomes among pregnant women, but it is unclear how risks may be different relative to non-pregnant women of a similar age and comorbidity profile.

Study design

Case series

Study population and setting

This study used a Brazilian national surveillance system for acute respiratory distress syndrome (ARDS) to identify 2,475 pregnant and postpartum women with a diagnosis of ARDS attributed to COVID-19 (defined by any laboratory, clinical, or epidemiological criteria) as of July 14, 2020. The surveillance system defined ARDS as the presence of dyspnea, respiratory distress, or oxygen saturation <95% in combination with a flu-like symptom and did not assess lung edema. The authors evaluated risk factors for the composite endpoint of death, admission to the ICU, or mechanical ventilation. Demographic variables, clinical variables, and variables related to health care access were abstracted from the database. Multivariable logistic regression was used to test associations between potential risk factors and the outcome.

Summary of Main Findings

The median age of pregnant or postpartum women with COVID-19-related ARDS was 30 years; 20% were older than 35 years, and 28% had at least one comorbidity. Among all 2,475 women, 590 (24%) had the composite adverse outcome, and 204 (8%) died. Of those who died, 40% were never admitted to an ICU. Multivariable logistic regression was restricted to the 2,184 women with complete data. From this analysis, being postpartum at ARDS notification (OR 2.43, 1.94 to 3.05), obesity (OR 2.12, 1.38 to 3.27), diabetes (OR 1.66, 1.18 to 2.35), Black ethnicity (OR 1.61, 1.06 to 2.44), or age greater than 35 years (OR 1.39, 95% CI: 1.08 to 1.78) was associated with experiencing the adverse outcome. Additionally, several variables related to health care access were associated with the outcome: living in a peri-urban area (OR 3.58, 1.16 to 11.04); living in an area without Family Health Strategy, a program providing primary care through community health workers (OR 2.77, 1.15 to 6.69); and living more than 100 km away from the hospital providing ARDS notification (OR 1.83, 1.23 to 2.73). Results were broadly similar when the outcome was restricted to ICU admission.

Study Strengths

This study drew from a national surveillance system and included a large number of pregnant and postpartum women; it was able to consider several variables related to health care access.

Limitations

No comparison is available with non-pregnant women with a similar age and comorbidity profile, so it is not possible to determine whether pregnancy conferred an elevated risk of adverse outcome of COVID-19 in this setting. No data were presented on the gestational age at COVID-19 diagnosis, and outcomes may depend on the timing of SARS-CoV-2 infection relative to pregnancy. Without data on the causes of maternal deaths or independent risk factors for mortality such as caesarean delivery (which is particularly common in Brazil), it is not possible to attribute deaths to COVID-19 with certainty. No data were available on whether women had caesarean deliveries, which are particularly common in Brazil; caesarean sections may increase the likelihood of adverse outcomes. Data for at least one variable were missing for a large proportion of women in multivariable logistic regression; these women may differ meaningfully from the included women. The ARDS definition for mandatory disease notification in Brazil was less specific and includes less severe cases than what is typically used to define ARDS; the inferences are therefore more applicable to pregnant or postpartum women with moderate to severe COVID-19. The definition of ethnicity (i.e., as distinct from skin color) was not described. The authors did note that ethnicity was a marker of social vulnerability; however, a more detailed analysis is warranted to discern how race, ethnicity, income, and other potential markers of vulnerability interact to affect COVID-19 outcomes for women in this region.

Value added

This study based on nationwide surveillance data in Brazil provides one of the largest studies to date on COVID-19 outcomes among pregnant women, and includes data related to poor access to health care.