Skip to main content

Our take —

This study examined the relationship between history of BCG vaccination and SARS-CoV-2 infection prevalence among a group of 6,670 health care workers in Los Angeles County, US. Findings suggest that the BCG vaccine, which was originally developed to protect against tuberculosis, can reduce the risk of developing COVID-19 in individuals who were vaccinated in the past. Clinical trials are currently underway to examine if prospective BCG vaccination can be protective against COVID-19.

Study design

Retrospective Cohort

Study population and setting

The study included 6,679 health care workers in Cedars-Sinai Health System network around the Los Angeles County area. The correlation between COVID-19 infection and the history of BCG vaccination was assessed. SARS-CoV-2 infection was determined by measuring IgG levels in over 92% of the participants and by self-reporting of COVID-19-related symptoms in the remaining subjects. History of vaccination against meningococcal, influenza, and pneumococcal vaccination was used as control. The authors also compared the prevalence of four common preexisting conditions: hypertension, diabetes mellitus, cardiovascular diseases, and chronic obstructive pulmonary disease among the participants in the study.

Summary of Main Findings

Less than one third (30%) of the participants were BCG vaccinated with an average age of 43 years compared to 40 years in the non-vaccinated group. The rate of self-reported COVID-19 diagnosis and RT-PCR positivity were 1.9% and 1% in the BCG vaccinated group compared to 2.9% and 1.7% in the control group, respectively. In addition, anti-SARS-CoV-2 IgG antibodies were detected in 2.7% in BCG vaccinated participants and in 3.8% of the control group. Participants in the BCG vaccinated group reported higher rates of all these four co-morbidities compared to those in the control group, excluding the possibility that the differences in COVID-19 incidence between both groups reflect variations in the prevalence of preexisting comorbidities.

Study Strengths

The current study has the advantage of including a relatively large number of participants. As the participants in this study were health care workers, the self-reported medical history is more reliable compared to similar studies in which participants were recruited from the general population. In addition, the high rate of seropositivity in this cohort allowed robust testing of the hypothesis.

Limitations

Most of the limitations are related to the nature of retrospective studies including the reliability of the self-reporting medical history. In addition, there were some statistically significant and potentially confounding differences between the BCG vaccinated and not-vaccinated groups including age, sex and ethnicity. The study did not include the timeline of BCG vaccination in the participants. Finally, the mechanism of how BCG vaccination could reduce viral infection is unknown.

Value added

This study showed that in individuals who have received the safe and inexpensive BCG vaccine in the past may have a reduced risk of SARS-COV-2 infection.

Our take —

This study of a contained outbreak on a US aircraft carrier in which 1,271 of 4,779 crew members (27%) were infected with SARS-CoV-2, 77% of whom were asymptomatic at the time of diagnosis and 45% of whom never had symptoms. The rapid spread was facilitated by close-quarters conditions and by asymptomatic and pre-symptomatic infected crew members. Though this outbreak provided a unique opportunity to study SARS-CoV-2 epidemiology in a predominantly young, healthy, working-age population, the findings from this military population may not be generalizable to civilians.

Study design

Retrospective Cohort

Study population and setting

Clinical and demographic data, including results of testing by real-time reverse-transcriptase polymerase chain reaction (rRT-PCR), were obtained for all 4,779 crew members on U.S.S. Theodore Roosevelt following a COVID-19 outbreak between March 23 and May 18, 2020, infecting 1,271 persons aboard. All crew members were followed up for a minimum of 10 weeks, regardless of test results or the absence of symptoms.

Summary of Main Findings

SARS-CoV-2 spread quickly among the crew of the U.S.S. Theodore Roosevelt, with transmission facilitated by close-quarters conditions and by asymptomatic and pre-symptomatic infected crew members. 43% of those who tested positive for the virus never had symptoms. Over the course of the outbreak, 1,271 crew members (26.6% of the crew) tested positive for evidence of SARS-CoV-2 infection by rRT-PCR testing, with suspected illness in an additional 60 crew members. Among the crew members with laboratory-confirmed infection, 76.9% (978 of 1271) had no symptoms at the time that they tested positive and 55.0% had symptoms develop at any time during the clinical course. Among the 1,331 crew members with suspected or confirmed Covid-19, 23 (1.7%) were hospitalized, 4 (0.3%) received intensive care, and 1 died.

Study Strengths

The controlled setting allowed for uniform surveillance, treatment, and access to care, and full access to clinical and demographic data for all crew members. Symptom surveillance was rigorous, utilizing both in-person health screenings and self-reporting through a digital symptom checker. All testing was done by rRT-PCR.

Limitations

Data collection from the early days of the outbreak was limited by the quality of records, and may therefore underestimate symptomatic cases, or misestimate onset dates. Observations within a military population, being younger, healthier, and with equal access to healthcare, may not be fully generalizable to civilians.

Value added

The outbreak of Covid-19 on the U.S.S. Theodore Roosevelt provided an unusual opportunity to assess an outbreak in a predominantly young, healthy, working-age population. A majority of infected crew members did not note symptoms at the time of diagnosis, and crew members with unusual or atypical symptoms were less likely to have considered themselves to be infected. This suggests that non-symptomatic or mildly symptomatic crew members played an important role in the rapid spread of the outbreak, much as young adults with asymptomatic infection appear to contribute to spread in civilian populations

Our take —

Anecdotal evidence on “Long COVID” suggests it is a pervasive problem, but its prevalence, etiology, and clinical extent have not been well researched. This study examined follow-up data an average of 54 days after hospital discharge among 384 COVID-19 patients. More than half of patients reported persistent fatigue and breathlessness, and about one-third reported persistent cough. Additionally, some participants still had abnormal laboratory values and chest x-ray results. These results highlight a number of persistent symptoms, but should be considered as preliminary and descriptive, given that pre-COVID data were not available and participants may not be representative of all those infected with SARS-CoV-2.

Study design

Retrospective Cohort

Study population and setting

This brief report detailed follow-up data collected on 384 patients (mean age: 60 years, 62% male, 43% from Black, Asian, or other minority ethnic background) with prior COVID-19 (diagnosed by nasopharyngeal swab). The population included patients who were discharged from one of three London hospitals between April and June 2020. Most data were collected four to six weeks after discharge by phone, including self-reported physical and psychological symptom burden and trajectory, but patients who had abnormal blood tests or chest x-ray findings at discharge were invited to return in person for re-evaluation.

Summary of Main Findings

Of the 384 participants included, median length of hospital stay was 6.5 days, 14.5% required ICU care, and 7.1% were intubated. At follow-up (median 54 days after discharge), 53% of participants reported persistent breathlessness, 34% reported persistent cough, 69% reported persistent fatigue, but the intensity for each of these symptoms were reported as improving in 81%, 75%, and 80% of participants, respectively. All laboratory values improved over time, though 7% (of n=247) had lymphopenia, 30% (of n=229) still had elevated d-dimer levels, and 10% (of n=190) had elevated C-reactive protein. Of 244 patients with follow-up chest x-rays, 9% showed significant deterioration.

Study Strengths

The study collected a number of clinical parameters during and following COVID-19 hospitalization.

Limitations

Only 80% of eligible participants were included in the study, and sub-analyses were restricted to smaller and variably sized populations; it is unclear how results might be different if the full eligible population were included. The study population included only patients who were hospitalized, though few had a prolonged stay in the ICU, so the results may not be generalizable to mild cases (not requiring hospitalization) or severe cases. All symptom trajectories were self-reported, which may lead to either an over- or underestimation of symptom resolution. Follow-up data were compared to admission and pre-discharge data, not pre-COVID data, so it is unclear whether COVID-19 is responsible for the persistently elevated laboratory parameters in some patients.

Value added

This is the first published report of symptom burden, laboratory markers, and chest imaging in the weeks following hospitalization with COVID-19.

Our take —

This large, national, retrospective cohort study demonstrated a significantly higher risk of psychiatric sequelae among COVID-19 survivors, especially anxiety disorders, insomnia, and dementia for those above 65 years old, compared to six control health events. The study also suggested that having a pre-existing psychiatric diagnosis is an independent risk factor for being diagnosed with COVID-19. However, these results should be interpreted with caution given possible selection bias and residual confounding.

Study design

Retrospective Cohort

Study population and setting

This retrospective cohort study assessed whether a diagnosis of COVID-19 was associated with subsequent psychiatric diagnoses, and whether patients with pre-existing psychiatric diagnoses were at higher risk of being diagnosed with COVID-19. The study used the TriNetX Research Network, a network of electronic health records, including 69.8 million patients across 54 health-care organizations in the US. Diagnostic categorizations were determined using ICD-10 codes. Incidence of psychiatric sequelae following COVID-19 diagnosis was compared (after propensity score matching) to six control health events – influenza (n=26,497), another respiratory tract infection (n=44,775), skin infection (n=38,977), cholelithiasis (n=19,733), urolithiasis (n=28,827), and fracture of a large bone (37,841). For all cohorts, the study population included individuals over 10 years of age, diagnosed with the specified health event between January 20 and August 1, 2020. Incident psychiatric illness was considered as the first psychiatric diagnosis over a period of 14 to 90 days after diagnosis with COVID-19 or the control health event. To evaluate whether individuals with psychiatric diagnoses were at increased risk of developing COVID-19 (psychiatric antecedent), incidence of COVID-19 was compared between two propensity score-matched cohorts (n=1,729,837 per cohort). One group included all patients older than 18 years with a recorded psychiatric diagnosis in the previous year (Jan 21, 2019 to Jan 20, 2020) and the other group included patients with no history of psychiatric diagnosis but who did make a health-care visit in the same period.

Summary of Main Findings

Regarding psychiatric sequelae, a diagnosis of COVID-19 was associated with increased incidence of a first psychiatric diagnosis compared with all six control health events (hazard ratios [HR] between 1.58-2.24, all p<0.0001). Among psychiatric diagnoses, anxiety disorders were most common, and patients diagnosed with COVID-19 were more likely to be diagnosed with anxiety disorders than any other cohort (incidence among COVID-19 4.7%, 95% CI: 4.2-5.3%; hazard ratios (HR) 1.59-2.62, all p<0.0001). Insomnia (incidence 1.9%, 95% CI: 1.2-2.2%; HR 1.85-3.29, all p<0.0001) and dementia among patients ≥65 years (incidence 1.6%, 95% CI: 1.2-2.1%; HR 1.89-3.18, all p<0.005) were also associated with COVID-19 diagnosis, compared to all other control conditions. Regarding psychiatric antecedents, having a psychiatric diagnosis in the year before the COVID-19 outbreak was associated with a significantly increased risk of being diagnosed with COVID-19 (RR 1.65, 95% CI 1.59-1.71) compared with a cohort without a psychiatric diagnosis. These trends were robust to all sensitivity analyses.

Study Strengths

This study included a large and representative cohort of patients cared for across 54 U.S. health care centers, increasing the generalizability across the United States. Propensity matching accounted for some potential confounders, and the use of multiple controls and sensitivity analyses increased confidence in the robustness of trends.

Limitations

As this study used electronic medical record data from early in the COVID-19 pandemic when coding for diagnoses of COVID-19 changed rapidly and reliable testing was not widespread, there is potential for patient selection bias toward those with severe disease and classic presentation. There is potential for diagnostic bias as clinicians may be more likely to diagnose psychiatric illness after a COVID-19 diagnosis than control events due to differences in the nature or extent of assessments. Patients accessing care at out-of-network locations may not be captured in this dataset which may underestimate incidence figures. As the prevalence of undiagnosed patients with COVID-19 has fluctuated with testing capacity throughout the pandemic, and this study is unable to assess undiagnosed patients, outcome generalizability may be temporally limited. This study was not randomized, and residual confounding may exist despite propensity score matching. The authors do not mention missing data, and it is assumed missing data were excluded, which may additionally bias the results.

Value added

To the best of our knowledge, this study is the first to investigate psychiatric sequelae and antecedents of COVID-19 in a large cohort.

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 study compared clinical complications of SARS-CoV-2 and influenza using two well-defined cohorts of hospitalized adults from the U.S. Veterans Health Administration. Individuals hospitalized with COVID-19, who generally had fewer comorbidities, had worse clinical outcomes (ICU admission and death) and more frequent complications involving multiple organ systems (including respiratory, hematologic, neurologic, and renal complications) as compared to patients hospitalized for influenza. However, differences in clinical-ordered testing and residual confounding may account for some of the observed associations.

Study design

Retrospective cohort

Study population and setting

This study included 5,453 influenza A or B patients diagnosed from October 1, 2018 to February 1, 2020, and 3,948 COVID-19 patients diagnosed from March 1 to May 31, 2020. Patients were drawn from the nationwide U.S. Veterans Health Administration, and identified from, and data on age, sex, race/ethnicity, ICD-10-CM diagnosis codes, and medical follow-up (ICU admission, discharge, death) were extracted. The population was restricted to adults who received a laboratory-confirmed influenza or COVID-19 diagnosis during the first 14 days of hospitalization or in the 30 days preceding hospitalization. Risks of in-hospital complications were compared between influenza and COVID-19 patients.

Summary of Main Findings

Hospitalized COVID-19 patients (n=3,948, 94% male, median age 70 years) had a lower prevalence of most underlying conditions than patients hospitalized with influenza (n=5,453, 94% male, median age 69 years), but had longer hospital stays (8.6 days vs. 3 days), higher rates of ICU admission (36.5% vs. 17.6%), and higher mortality (21% vs. 3.8%). In adjusted analyses, many complications of COVID-19 were slightly more prevalent among racial and ethnic minorities (separately modeled as non-Hispanic Black or African American, Hispanic or Latino, or non-Hispanic other race vs. non-Hispanic White), including intracranial hemorrhage, cerebral ischemia/infarction, pneumonia, respiratory failure, acute respiratory distress syndrome (ARDS), acute kidney failure, dialysis, and sepsis. After adjusting for age, sex, race/ethnicity, and underlying medical conditions, hospitalized patients with COVID-19 (vs. influenza) had a statistically significant increased risk of 17 acute complications involving multiple organ systems (respiratory, cardiovascular, neurologic, renal, etc.). For example, COVID-19 patients had increased risks relative to influenza patients for ARDS (adjusted relative risk (aRR) = 19), pneumothorax (aRR=3.5), pneumonia (aRR=2), and respiratory failure (aRR=1.7), but decreased risk for asthma exacerbation, COPD exacerbation, and hypertensive crisis.

Study Strengths

This was a relatively large study. Controls were drawn from the same underlying population as cases, with the same data collection protocols and similar inclusion criteria, which helps comparability. The authors conducted sensitivity analyses comparing COVID-19 and influenza cases diagnosed during the same months (though from different years) to assess bias from seasonality.

Limitations

A main concern is whether there are differences in caring for patients with COVID-19 vs. influenza that leads providers to differentially screen for complications during hospitalization. For example, it is plausible that influenza patients receive more clinician-ordered testing for respiratory complications, whereas individuals with COVID-19 are screened more often for a broader range of respiratory and non-respiratory complications, given media and medical attention to the heterogeneous clinical presentation of COVID-19. Additionally, residual confounding is possible due to factors such as geographic location, misclassification of comorbid conditions, and variations in reporting of different complications; these could lead to either under- or over- estimation of the true associations.

Value added

Although the relative severity of influenza and COVID-19 is reasonably well understood, this study provides a detailed look at differences in the frequency of specific severe complications between the two illnesses.

Our take —

This study sought to estimate the transmission rate and risk factors for SARS-CoV-2 infection among close contacts of all COVID-19 cases in Singapore. Using detailed questionnaires and contact-tracing methods, they identified 1114 index cases and 7770 contacts from January 23 to April 3, 2020. They found that household contacts had the highest clinical attack rate (5.9%) compared to work contacts (1.3%) and social contacts (1.3%), though after accounting for estimates asymptomatic cases, estimates of secondary infection rate increased to 11% for household contacts, 4% for social contacts, and 5% for work contacts. Sharing a bedroom with an index case and speaking (regardless of masks or being indoors) with an index case for 30+ minutes were associated with transmission. This study included all people in Singapore, representing an important contribution to population-based estimates of transmission.

Study design

Retrospective Cohort

Study population and setting

The study objective was to estimate the transmission risks of COVID-19 given close household contacts. The study was conducted from January 23 to April 3, 2020, and included contact-tracing of all confirmed COVID-19 cases in Singapore. Household contacts were defined as living with the index case, while non-household contacts were defined as having contact for 30 minutes or more within 2 meters of the index case (indoors or outdoors, regardless of mask use). Contacts were all required to quarantine for 14 days and complete symptom monitoring 3 times per day via telephone and received PCR testing for SARS-CoV-2 if they reported symptoms. If they did not report symptoms or received a negative PCR test after reporting symptoms, they were eligible to also receive a serological test and complete further risk assessments. The study used Bayesian models to estimate asymptomatic cases and missed diagnoses.

Summary of Main Findings

From January 23 to April 3, 2020, there were 1114 index cases identified, and 7770 close contacts traced. These included 1863 household contacts, 2319 who were contacts from work, and 3588 contacts from social gatherings. Using a symptoms-based testing strategy, 188 (2.42%) secondary cases were identified among contacts, which, when limited to those with complete data available, reflected a clinical attack rate of 5.9% (5% CI: 4.9 – 7.1%) for household contacts, 1.3% for work contacts (0.9 – 1.9%), and 1.3% (1.0 – 1.7%) for social contacts. Bayesian modeling coupled with serological testing of all individuals who completed the quarantine without symptoms or a positive PCR test estimated that 62% (95% CrI: 55 – 69%) of COVID diagnoses were missed under this testing strategy, and 36% (95% CrI: 27 – 45%) of infected individuals were asymptomatic. Among participants who completed a detailed risk assessment, sharing a bedroom (OR: 5.38, 95% CI: 1.82 – 15.84), and being spoken to by an infected person for 30+ minutes (OR: 3.07, 95% CI: 1.55 – 6.08) were associated with transmission, while meal-sharing, lavatory co-usage, and indirect contact were not significantly associated with infection.

Study Strengths

The study was able to model and estimate the proportion of missed diagnoses and asymptomatic infections, despite not actively testing all? asymptomatic cases. These modeled estimates allow us to estimate the true infection rate that may be occurring. Additionally, because cases were legally required to quarantine and report symptoms, there was a high rate of participation and proportion of respondents with complete data. Using PCR testing on all symptomatic contacts increased the validity of these results by minimizing false negatives and positives and providing a high degree of accuracy, especially when coupled with the subset of serological testing. Their risk assessment also included 70-items to pinpoint specific types of exposure that may have occurred, adding granularity to this data.

Limitations

As the study noted and estimated, a number of cases were likely missed under this testing schema, which highlights the importance of testing regardless of symptom-status, in order to not miss asymptomatic cases. Singapore had a number of strict infection protocols that individuals were legally required to adhere to, and therefore these results may not be generalizable to other contexts where prevention measures are less common. For instance, it is not standard medical protocol that everyone who tests positive for SARS-CoV-2 infection receives inpatient hospital care, which will likely reduce the rate of transmission.

Value added

This study examined household and non-household contacts among all Singapore COVID-19 cases.

Our take —

Researchers enrolled 333 persons in the United States with fitness tracking devices (e.g., FitBits) who had COVID-like symptoms and received a SARS-CoV-2 test. Variations in sleep and activity, alongside self-reported symptoms can suggest a possible positive COVID-19 case and may make individuals more aware of possible infective status prior to confirmed testing. However, it is also possible that participants changed their lifestyles as a result of learning that they had been exposed to a person with COVID-19 or that they were positive. The methods in this paper do not explicitly account for either of these possibilities and therefore, should be interpreted with caution.

Study design

Prospective Cohort, Retrospective Cohort

Study population and setting

This analysis was restricted to 333 people with fitness trackers (Fitbit, Apple HealthKit, Google Fit) in the United States who reported COVID-like symptoms, sought testing for SARS-CoV-2, and were enrolled in the study between March 25 and June 7, 2020. The authors evaluated whether the addition of sensor data (i.e., changes in resting heart rate, sleep, and physical activities) to symptoms could improve detection of people who reported a positive test result for SARS-CoV-2. Authors evaluated the ability of each model to correctly classify people with and without the disease using receiver operating curves (ROC) and the area under the curve (AUC). AUC is a common metric for simultaneously considering the sensitivity and specificity of tests that attempt to diagnosis a disease.

Summary of Main Findings

Fifty-four (16.2%) participants reported being positive for SARS-CoV-2. Symptomatic people who reported receiving a positive test result were significantly more likely to get more sleep and take fewer steps daily than those who reported a negative result. Combining both sensor and symptom data resulted in an AUC of 0.80, which was significantly better than self-reported symptom or sensor data alone. An AUC of 0.8 suggests an 80% chance that the test will correctly distinguish an infected from a non-infected patient

Study Strengths

This study drew participants from across the United States and gathered data from objective, real-time fitness trackers worn by participants.

Limitations

The most pressing limitation of this study is whether participants changed their sleeping and activity as a result of receiving their diagnosis. If participants altered their habits after their diagnosis, then the proposed diagnostic method (i.e., detecting changes in sensor data) would not be useful in predicting infection. The authors acknowledge that participation in this study was limited to persons with fitness trackers, who are likely not representative of the general population.

Value added

People diagnosed with COVID-19 may have COVID-specific symptoms and change their sleeping and physical activity habits, as measured by a fitness tracker. Fitness trackers and self-reporting may offer new ways to potentially identify COVID-19 positive individuals before RT-PCR testing is done.

Our take —

This population-based retrospective study from Denmark utilized existing health records, blood group information, and SARS-CoV-2 testing data to suggest a moderate difference in risk of SARS-CoV-2 infection among those with blood groups O (decreased risk) and A (increased risk). However, the strength of the blood group and SARS-CoV-2 susceptibility relationship remains unclear. Furthermore, as the prevalence of blood group types can differ significantly by genetic ancestry or ethnicity, additional work is needed to characterize the risk of blood group antigens on SARS-CoV-2 risk in non-European populations.

Study design

Retrospective Cohort

Study population and setting

Leveraging electronic health record data from all Danish patients and other centralized databases, this retrospective cohort analysis explored the ABO and RhD blood group distributions among 473,654 individuals tested by real-time polymerase chain reaction (PCR) for SARS-CoV-2 who had blood group information available (out of 841,327 individuals tested between 2/27/2020 – 7/30/2020). These mostly symptomatic (74%) individuals were compared with a reference group of 2,204,742 non-tested individuals from across Denmark who had blood group information available. In addition to PCR-based testing results, demographic and COVID-19 clinical outcomes including death were explored for associations with blood group.

Summary of Main Findings

There were no significant differences observed between RhD or ABO blood groups for hospitalization or death due to COVID-19. However, individuals with blood group O were less likely to have a positive SARS-CoV-2 test (RR 0.87) and those with blood group A were more likely to have a positive SARS-CoV-2 test (RR 1.09) overall. These analyses did not adjust for the changing testing strategy throughout the early course of the pandemic in Denmark.

Study Strengths

A largely ethnically homogenous population may minimize potential bias introduced from differences in distributions of blood types by ethnicity, which may also be associated with exposure to the virus or health outcomes due to socioeconomic histories . Further, Denmark has free and universal healthcare services which are linked to a set of exceptional centralized data repositories used to conduct the analysis . Given the large sample size, this study was well powered to detect differences in COVID-19 testing positivity among those with blood groups A, B, and O.

Limitations

While the authors attempted to account for potential confounding due to ancestry-related differences in blood group type prevalence through an adjusted analysis, ethnic population-specific data on the counts of non-Western individuals are not provided (e.g., stratified results) and would be needed to confirm the authors findings. Temporally stratified results are not provided, but coupled with ethnicity-specific testing counts may be of potential importance given the testing limitations during these stages of the pandemic. We stress the importance of ethnicity-specific results because many non-European populations have a lower prevalence of O group antigen, and non-western individuals were disproportionately identified as cases (18%) compared to comprising 9% of the Danish reference population.

Value added

Blood group type has been associated with the susceptibility to multiple infectious diseases and this study provides important preliminary findings within the Danish population related to the potential association between blood groups A (increased risk) and O (decreased risk) and the risk of SARS-CoV-2 infection.

Our take —

Mortality risks among 5,121 hospitalized patients with COVID-19 declined each month in New York City from March through August 2020. This study, which adjusted for patient demographic and clinical characteristics, provides support that improved survival was not only because of patient characteristics. However, the study could not evaluate any specific reasons for the improvement (possibilities include improved pharmaceutical treatment, greater clinical experience, lower patient volume, and changing admission criteria). Results may not be generalizable to settings with a different trajectory of cases than New York City, since hospitals were overwhelmed with COVID-19 cases in the early spring of 2020.

Study design

Retrospective cohort

Study population and setting

This study estimated changes by month in COVID-19 mortality risk among 5,121 hospitalized patients from a single health system in New York City from March 1 to August 31, 2020. Included patients were 18 years and older, with laboratory-confirmed SARS-CoV-2 infection; 229 hospitalizations (4.4%) were repeated hospital stays from 208 patients. Mortality was defined as death in hospital or discharge to hospice. Data were taken from electronic medical records. A multivariable logistic regression model (with a range of demographic and clinical variables including age, sex, race, BMI, comorbidities, admission oxygen saturation, admission d-dimer concentrations, and admission CRP concentrations) were used to calculate monthly adjusted mortality rates. Average marginal effects for each month were estimated with a second model that included month as a covariate.

Summary of Main Findings

The majority of hospitalizations (53%) occurred from late March to mid-April 2020, and 79% of hospitalizations occurred by the end of April. The median duration of hospital stay for those who died or were discharged to hospice was 8 days. Over time, the median age, the proportion of males, and the proportion with comorbidities decreased. Adjusted mortality declined in each successive month from 25.6% in March to 7.6% in August. The average marginal effect of each month was increasingly negative relative to March, reaching -18.2% in August. Restricting the study population to those with at least 3 days in hospital produced similar results; restricting to those with a principal diagnosis of COVID-19, sepsis, or respiratory disease produced similar but attenuated results.

Study Strengths

This straightforward analysis used estimates of average marginal effect, which have intuitive interpretations (i.e., % change in probability of death for a given month). Follow-up was complete on the vast majority of patients.

Limitations

Changing criteria for hospitalization over time could bias results; for example, if patients with less severe disease were admitted in later months, outcomes would appear more favorable. The clinical covariates in multivariable models may not have sufficiently captured patients’ mortality risk status. It is not clear how covariates measuring age, race/ethnicity, BMI, and smoking history were defined and modeled (e.g., categorical vs. continuous). Comorbidities were assessed via binary indicator variables (e.g., yes vs. no) and models are thus subject to residual confounding. The study population is from a single health system, which limits generalizability– for example, New York’s hospital system was particularly overwhelmed during late March and April, and this burden may have contributed to higher mortality during those months.

Value added

Declines in mortality among hospitalized patients with COVID-19 have been observed in different settings, but this is the one of the first studies to adjust for patient characteristics and clinical parameters, providing evidence that the mortality risk decline in New York City is not simply due to a differing mix of patients (e.g., younger with fewer comorbidities).