Skip to main content

Our take —

Existing research on racial and ethnic differences in COVID-19 outcomes is lacking. Though black patients make up only 31% of established patients at Ochsner Health, they comprised 70.4% of COVID-19 diagnoses, and 76.9% of all hospital admissions. However, black patients did not have an increased risk of mortality. The study highlights notable racial differences in clinical presentation, conceivably due to systematic differences in the lived experiences of African-Americans, leading to delays in care and higher prevalence of chronic conditions, but more research in this area is needed.

Study design

Retrospective Cohort

Study population and setting

This retrospective medical-record based cohort study includes 3481 patients (1030 white non-Hispanic; 2450 black non-Hispanic) who tested positive for SARS-CoV-2 on polymerase chain reaction assay at an Ochsner Health facility in New Orleans, LA between March 1 and April 11, 2020 (follow-up through May 7, 2020). Patients self-reported race and ethnic group. Those who were Hispanic, Asian, American Indian or Alaska Native, Native Hawaiian or Pacific Highlander, or who did not have a recorded race or ethnic group were excluded (n=145). The study characterized clinical presentation, hospital course, and disease outcomes among black non-Hispanic and white non-Hispanic patients with laboratory confirmed COVID-19.

Summary of Main Findings

Most of the patients included (n=2450, 70.4%) were black non-Hispanic, whereas, among 522,679 patients who received care at Ochsner Health and in the previous 12 months, only 31% identified as black non-Hispanic. Black patients were more likely than white patients to have Medicaid insurance, live in a low-income area, and more often presented with a higher prevalence of chronic comorbidities, fever, cough, or dyspnea. Compared to white patients, black patients presented with higher levels of creatinine, aspartate aminotransferase, and inflammatory markers, but had lower white cell, lymphocyte, and platelet counts. In unadjusted and adjusted analyses, black patients were more likely to be hospitalized than white patients (35.7% vs. 29.5%), but black race was not associated with mortality (unadjusted case-fatality rates 21.6% and 30.1% for black patients and white patients, respectively).

Study Strengths

This was a decently sized study at a large and integrated medical system in New Orleans, LA. Data on numerous clinical and laboratory characteristics were extracted from the medical record. Multiple imputation was used to impute missing data for BMI (n=683), venous lactate (n=268), C-reactive protein (n=287), procalcitonin (n=305), and lymphocyte (n=32). Unadjusted and adjusted analyses are consistent across several adjustment sets, indicating the robustness of results.

Limitations

The analysis was limited to one integrated-delivery health system in Louisiana and based on data available in the medical record, which is subject to data entry errors and missing data. Results may not generalize beyond the healthcare system.

Value added

This study presents a comprehensive comparison of clinical presentation and disease course between black non-Hispanic patients and white non-Hispanic patients in a major U.S. city.

Our take —

In this small, single-center study posted as a non-peer-reviewed preprint, use of convalescent plasma was associated with improved survival and reduced supplemental oxygen demands when compared to retrospectively-collected propensity-matched controls. Because this study lacked randomization and thus there may have been unobserved differences in baseline characteristics between the plasma and comparison groups, data that take advantage of random assignment of convalescent plasma, or a strong non-experimental study design, are needed to verify the findings.

Study design

Retrospective Cohort

Study population and setting

This study included 45 hospitalized COVID-19 patients from an academic medical center in New York City who were enrolled into the study between March 24 and April 8, 2020. Inclusion criteria were defined by eligibility for SARS-CoV-2 convalescent plasma under FDA emergency investigation new drug criteria (eIND). 39 of 45 eligible patients were transfused with convalescent plasma from COVID-19 patients with anti-spike antibody titers greater than or equal to 1:320. Comparison patients with similar baseline demographic and clinical characteristics were selected retrospectively from the medical records system through propensity score matching, selecting four matched comparison subjects for each transfused patient.

Summary of Main Findings

By day 14, there was no significant difference in clinical condition between plasma patients and controls (p=0.167), however, the adjusted odds ratio for worsening oxygenation was 0.86 (95% CI: 0.75-0.98, p=0.028), showing some benefit of plasma transfusion. Additionally, patients receiving convalescent plasma had increased survival when compared to controls (log-rank test: p=0.039). In an adjusted model, use of convalescent plasma was associated with improved survival in non-intubated patients (adjusted OR 0.19; 95% CI: 0.05-0.72; p=0.015) but this effect was not observed in intubated patients (adjusted OR 1.24; 95% CI: 0.33-4.67; p=0.752)

Study Strengths

There was some control of baseline confounding factors using propensity scores and exact matching.

Limitations

The sample size of the study was small. The study recruited patients from a single academic hospital center in an urban location, limiting generalizability of findings. Receipt of convalescent plasma was not assigned randomly, raising concerns about observed and unobserved differences in baseline factors between groups. Matching controlled for some observed characteristics but it is unknown what other unobserved factors may have related to receipt of plasma and outcomes and thus potentially biased the study results.

Value added

This is the largest study of convalescent plasma to date and warrants further investigation with prospective randomized controlled trials.

Our take —

This contact tracing study in China showed that 22.3% (63 of 279) of cases testing positive were asymptomatic, and that asymptomatic cases experienced a notable prevalence (46.0%, n=29) of abnormal chest computed tomography scans (CTs). Those with abnormal CTs had longer time to hospital discharge compared to cases with normal CTs. However, this study was relatively small, and all patients received antiviral treatment upon admission, which may have decreased the time until hospital discharge (two negative RT-PCR tests) and impact the generalizability of these findings to other asymptomatic case populations.

Study design

Retrospective Cohort

Study population and setting

This study in Chonqing Three Gorges Center Hospital and Wanzhou District Hospital, identified 279 cases diagnosed with SARS-CoV-2 infection from January to March 2020 via contact tracing and testing, and 63 (22.6%) of those had no symptoms at diagnosis or during hospitalization.

Summary of Main Findings

The study found 63 cases who tested positive for SARS-CoV-2 infection via RT-PCR, but who did not report symptoms. Of these, 29 had abnormal chest CT findings. Those with abnormal chest computed tomography scans (CT) were more likely to be older and male than those with normal chest CT. Those with abnormal chest CT also had longer median time between initial positive test and discharge after two negative tests (13.0 days) compared to those with normal chest CT (10.4 days), but no major differences in laboratory blood testing outcomes.

Study Strengths

The study was able to identify asymptomatic cases via contact tracing and testing and follow their clinical outcomes during their observation periods at these local hospitals. The chest CT data were available for all asymptomatic patients, potentially showing more severe disease despite lack of symptoms. There were also important clinical data available about comorbid conditions and potential exposure history that adds to this study’s strengths.

Limitations

The study had a small sample size overall, with only 63 asymptomatic cases. Because they are asymptomatic, also, they may be less likely to be identified in a timely manner, and therefore the study may lack precision in estimates of their infectious period. Also, per Chinese standard protocol, all patients received antiviral treatments upon admission. Therefore, these results may not reflect the progression of disease among asymptomatic cases without this intervention, and may not be generalizable to other populations.

Value added

This is an early study of asymptomatic cases with other clinical data, including comorbid conditions, chest radiology, and laboratory serology.

Our take —

Coagulopathy has been identified as an important manifestation of COVID-19. This fairly small study of ischemic stroke patients showed greater stroke severity and higher mortality among those who were infected with SARS-CoV-2 relative to those who were not. More than half of stroke patients with COVID-19 had strokes that were diagnosed after hospitalization, indicating the need for further study of anticoagulation treatment for stroke prevention in COVID-19 patients.

Study design

Retrospective Cohort

Study population and setting

This study included all patients with ischemic stroke who were hospitalized in one of three stroke centers in New York City between March 15 and April 19, 2020. All stroke patients discharged from two of the centers between March 15 and April 19, 2019 were used as historical controls. Patients with both stroke and laboratory-confirmed SARS-CoV-2 infection (n=32, median age 63 years, 72% male) were compared to contemporaneous stroke patients without SARS-CoV-2 infection (n=70, median age 70 years, 52% male) and to historical controls (n=80, median age 69 years, 45% male). Patient characteristics were extracted from medical records.

Summary of Main Findings

During the study period, 0.9% (32/3556) of hospitalized COVID-19 patients had an imaging-confirmed ischemic stroke. Of these 32 stroke patients with SARS-CoV-2 infection, stroke was the reason for admission in 14 (44%), while the 18 (56%) had stroke diagnosed after admission for COVID-19. The median duration from COVID-19 symptom onset to stroke diagnosis was 10 days (interquartile range: 5 – 16.5 days). 14 (44%) of these patients died, 8 (25%) were discharged home or to rehabilitation, and 10 (31%) remained critically ill at the end of follow-up. Compared to both contemporaneous and historical controls, patients with COVID-19 had a higher risk of mortality (among those with outcomes, mortality was 64% in the COVID-19 group vs. 9.3% in contemporaneous controls and 6.3% in historical controls), higher National Institutes of Health Stroke Scale (NIHSS) scores, and a higher likelihood of cryptogenic stroke (i.e unknown cause).

Study Strengths

The use of historical controls helps address the possibility of misclassification of SARS-CoV-2 infection status.

Limitations

Stroke was likely under-diagnosed in COVID-19 patients, particularly those with critical illness whose symptoms may be difficult to detect. Strokes diagnosed after hospital admission may have occurred earlier, as screening was likely impractical during acute COVID-19 treatment. 30% (14/46) of contemporaneous controls were screened but not tested for SARS-CoV-2 infection; some of these patients may have had mild or asymptomatic infection, and the resulting misclassification may have diluted observed differences between the groups. Limited availability of broad imaging likely resulted in an artificially high prevalence of cryptogenic stroke type. Findings were limited to ischemic stroke and did not include other manifestations of severe coagulopathy.

Value added

This study shows dramatically higher mortality among ischemic stroke patients with COVID-19 relative to stroke patients without COVID-19, adding to the emerging literature on coagulopathy as an important non-respiratory symptom of SARS-CoV-2 infection.

Our take —

This preprint, non peer-reviewed study combined epidemiological and genomic data to provide a detailed picture of the emergence and limited onward spread of SARS-CoV-2 in Australia. The researchers show, using both epidemiological and genomic data, that most of the cases in Australia were travel-related. The use of genomic data enabled the researchers to resolve community transmissions of unknown origin and identified high-risk social areas that were responsible for community spread of SARS-CoV-2. The study demonstrates how genomics-based COVID-19 surveillance can help implement targeted interventions to control COVID-19 spread.

Study design

Retrospective Cohort

Study population and setting

This study analyzed 903 SARS-CoV-2 virus sequences and associated epidemiological data collected from 1,333 COVID-19 positive individuals in Victoria, Australia. These cases are from January 25 to April 14, 2020, representing the first 1,333 cases in the country. The purpose of the study was to demonstrate the integration of genomics-based COVID-19 surveillance into public health response.

Summary of Main Findings

Active case-finding and contact tracing identified a total of 1,333 COVID-19 laboratory confirmed cases in Victoria over the study period. 62% of these cases were from travelers returning mostly from the Americas and Europe. Sequence data from this cohort showed that the sequences were representative of the global diversity at that time, consistent with the epidemiological information. Over 10% of cases with associated epidemiological data were from an unknown source in Australia, but analysis of their sequences identified potential transmission routes. Additionally, analysis of the 903 sequences generated in this study revealed clusters of cases associated with specific social venues in Melbourne, Australia, providing genomic evidence for community transmission. The authors also examined within-patient virus diversity (in cases where multiple samples were available from the same patient) and found limited evidence for viral evolution during the course of infection.

Study Strengths

A key strength of this study is that the large number of genomes generated (903) represents a large fraction of the first cases detected in Victoria, Australia. This dense sampling allows the authors to draw more specific conclusions about transmission of COVID-19. Additionally, the authors also describe epidemiological data for a large number of the sequenced samples, and use the combination of genetic and epidemiological information to support and confirm findings. The methodology is also clear and very detailed, and all data were made publicly available.

Limitations

This study would be improved by providing more information about the published sequences used to compare to the generated sequences from Australia, as this impacts inferences related to the source of viral importations. Additionally, the within-host section of the paper is limited due to the small number of patients from whom multiple sequences were available, so additional studies are needed to fully characterize within-host evolution.

Value added

The study was part of a national public health surveillance program that combined epidemiology and viral genomics analyses. There was strong concordance between epidemiological and genomic data, and most of the community acquired transmissions and of unknown source were identified using genomic data. This was useful in providing information on how community acquired transmissions occurred and how to implement appropriate interventions. The report demonstrated how integration of genomics-based COVID-19 surveillance into public health response aided targeted intervention programs by identifying high-risk areas (like large social venues in Melbourne) responsible for community transmission of SARS-CoV-2.

Our take —

In this descriptive study, health care workers (HCW) had higher rates of infection but a lower case fatality rate than the general population. HCW in general hospitals had a higher infection rate than those at specialty or community hospitals. Results of the study were likely influenced by selection bias because HCW are more likely to receive testing for COVID-19 and more likely to be exposed during early stages of the epidemic with limited access to PPE.

Study design

Retrospective Cohort

Study population and setting

This retrospective cohort study includes 2457 health care workers from 145 hospitals in Wuhan, who had laboratory-confirmed COVID-19 and received assistance from the Red Cross Humanitarian Aid Fund between January 26 and March 26, 2020. Statistics on the total population of Wuhan were gathered from the Wuhan Statistics Bureau. Case infection rate was estimated as the number of laboratory-confirmed COVID-19 cases divided by the total population of interest (i.e. health care workers or non-health care workers).

Summary of Main Findings

Among the estimated 117,100 health care workers in Wuhan, 2,457 were diagnosed with COVID-19 and received assistance from the Humanitarian Aid Fund, corresponding to a case infection rate of 2.1%. During the same period, the general (non-health care worker) population of Wuhan was close to 11 million people with 47,549 cases of COVID-19, corresponding to a case infection rate of 0.43%. The case fatality rate among health care workers was 0.69%, and among non-health care workers was 5.3%. Among health care workers, those working in general hospitals had the highest case infection rate (2.93%), whereas 0.8% and 0.5% of health care workers in specialized and community hospitals, respectively, were infected. Additionally, nurses had a slightly higher case infection rate than doctors (2.22% vs 1.92%).

Study Strengths

This was a very large and multi-site study of health care workers in Wuhan, which is an understudied area.

Limitations

The analyses were descriptive with no adjustment for potential confounding factors. The authors do not provide any details on the clinical characteristics (symptoms, comorbidities) of included patients. The authors do not address selection bias, which is likely given that health care workers are more likely to get screened for COVID-19 than the general population. In the early stages of the epidemic, access to personal protective equipment was limited, but it is unclear how this influenced the rates of infection.

Value added

This is a large study that compares COVID-19 infection rates between health care workers and the general population.

Our take —

This observational study does not support the use of hydroxychloroquine in the management of non-severe COVID-19 patients who require supplemental oxygen. The researchers found no significant differences between the treatment and control groups in terms of survival without ICU transfer at day 21, overall survival at day 21, progression to acute respiratory distress syndrome by day 21, and weaning from supplemental oxygen by day 21.

Study design

Retrospective Cohort

Study population and setting

This study included 181 patients, age 18 to 80 years, with documented SARS-CoV-2 pneumonia who required oxygen but not intensive care who were recruited from four French tertiary care centers between March 12 and March 31, 2020. Patients with severe disease, including acute respiratory distress syndrome (ARDS) or progression requiring intensive care, were excluded from the study. Participants were not randomized to “treatment” or “control.” Instead, 84 patients who received hydroxychloroquine within 48 hours of admission to hospital were classified as the treatment group and were compared with 89 patients who did not receive hydroxychloroquine. 8 patients received hydroxychloroquine after 48 hours and were excluded from the primary analysis .

Summary of Main Findings

Hydroxychloroquine use within 48 hours of admission was not associated with clinical benefit. The primary outcome was survival without ICU transfer at day 21. In the weighted analyses, 76% of the treatment group and 75% of the control group survived until day 21 without ICU transfer (weighted hazard ratio 0.9, 95% confidence interval 0.4 to 2.1). Overall survival at day 21 was 89% in the treatment group and 91% in the control group (weighted HR 1.2, 0.4 to 3.3). Survival without ARDS at day 21 was 69% in the treatment group compared with 74% in the control group (weighted HR 1.3, 0.7 to 2.6). At day 21, 82% of patients in the treatment group had been weaned from oxygen compared with 76% in the control group (weighted risk ratio 1.1, 95% confidence interval 0.9 to 1.3). Eight patients in the treatment group (10%) discontinued hydroxychloroquine due to adverse cardiac events.

Study Strengths

Larger sample size than previous observational studies of hydroxychloroquine; propensity score weighted analysis attempts to control for observed confounding variables,

Limitations

Observational study with a larger population than previous hydroxychloroquine studies. Propensity score weighted and unweighted analyses presented.

Value added

Observational study with a larger population than previous hydroxychloroquine studies. Propensity score weighted and unweighted analyses presented.

Our take —

The authors present COVID-GRAM, a web-based application that aims to predict incident critical illness among patients hospitalized for COVID-19. Considering several limitations, including bias due to overfitting, and the need for validation in settings outside of China, the risk score is a promising point-of-care tool for identifying patients at hospital admission who are most likely (or unlikely) to develop critical illness, which can aid in decision-making about how to best allocate resources.

Study design

Retrospective cohort

Study population and setting

This study developed and validated a risk score for critical COVID-19 illness, defined as the composite of admission to the ICU, invasive ventilation, or death. The development cohort included 1590 patients (mean age 49 years, 57% male, 25% with coexisting condition) with lab-confirmed SARS-CoV-2 infection admitted to 575 hospitals in China between November 21, 2019 and January 31, 2020. The validation cohort (n=710; mean age 48 years, 54% male, 24% with coexisting condition) included data from four sources, three of which had follow-up through February 28, 2020. Seventy-two variables, including clinical signs and symptoms, imaging results, laboratory findings, demographic variables, and medical history, measured at hospital admission were considered.

Summary of Main Findings

Eight percent (n=131) of patients in the development cohort and 12% (n=87) of patients in the validation cohort developed critical illness. Using LASSO (least absolute selection and shrinkage operator) regression, the authors narrowed 72 candidate variables down to 19. The 19 variables were then put into a logistic regression model, and the 10 that remained significant (p<0.05) were included in the risk score (chest radiography abnormality, age, hemoptysis, dyspnea, unconsciousness, number of comorbidities, cancer history, neutrophil-lymphocyte ratio, lactate dehydrogenase, and direct bilirubin), which is available as a web-based tool. The mean AUC (a measure of how well the model can distinguish between persons with and without the outcome) in both the development and validation cohorts was 0.88.

Study Strengths

The study developed an easy-to-use web-based calculator using 10 variables that are typically available at the time of hospital admission. Multiple imputation was used for variables with missingness <20% The data were independently reviewed and verified by two clinicians. Validation was done in an independent and external dataset.

Limitations

The data for development and validation cohorts were from China, so the applicability of the model to populations outside of China is unknown. It is unclear whether the authors corrected for overfitting; they use the bootstrap to estimate mean AUC over 200 bootstraps, but this metric should be corrected for overfitting, and failure to do so can lead to overly optimistic predictions. Data were collected during the early stages of the pandemic in China, and hospital practices may differ from current guidelines. The authors assumed linear relationships for continuous predictors, which is unlikely.

Value added

This paper presents an easy-to-use web-based application to predict risk of critical illness among patients hospitalized for COVID-19.

Our take —

In this study from mostly metropolitan Atlanta, Black patients represented a higher-than-expected proportion of COVID-19 hospitalizations, but were not more likely to die or receive invasive mechanical ventilation than a combined group of non-Black patients. These patterns should continue to be monitored, and rigorous multivariable analyses (including socio-economic factors, for example) are warranted. As expected, younger patients and those without high-risk conditions were less likely to be admitted to the ICU, receive invasive mechanical ventilation, or die.

Study design

Retrospective cohort

Study population and setting

This convenience sample includes 305 adults with laboratory-confirmed COVID-19 who were selected sequentially from 698 adult patients hospitalized at one of eight Georgia hospitals between March 1 and March 30, 2020.

Summary of Main Findings

The median age of patients was 60 years, 50.5% were female, 83.2% were Black, and 73.8% had a condition that put them at high-risk for COVID-19. Overall, 39% of patients were admitted to an ICU, and, of those, 30.2% received invasive mechanical ventilation (IMV). Most patients (76.4%) were discharged alive, 7.9% were still hospitalized at the time of analysis, and 17.1% of patients died. In unadjusted analyses, Black patients (versus non-Black patients) were not more likely to receive IMV or to die; these results were supported by an adjusted time-to-event model with a composite outcome of death or IMV.

Study Strengths

This report details clinical data on a decently sized, multi-site cohort of hospitalized patients in Georgia. Time-to-event analysis was used with censoring to account for patients still hospitalized and not receiving IMV.

Limitations

The selection process of the convenience sample isn’t well defined. Patients were not followed after discharge. Medical record data may differ by hospital. The multivariable model was based on a stepwise selection approach, which can lead to residual confounding. The main outcome was a composite of death or IMV. The authors do not provide a breakdown of non-Black racial/ethnic groups.

Value added

This study highlights clinical and epidemiologic characteristics of COVID-19 patients from a somewhat large cohort of hospitalized patients in metropolitan Atlanta and southern Georgia.

Our take —

This serologic validation study showed that the commercially-available Abbott SARS-CoV-2 IgG assay had optimal sensitivity 17 days post-symptom onset in a mostly hospitalized patient population, but the sensitivity of the assay remains unknown in other populations such as mild or asymptomatic cases of infection. The assay had optimal specificity in pre-COVID-19 pandemic specimens, but further validation is needed to assess whether the assay cross-reacts with other coronaviruses.  The study also demonstrates the feasibility of using the Abbott SARS-CoV-2 IgG assay to perform a community serosurvey.

Study design

Cross-Sectional; Retrospective Cohort; Other

Study population and setting

This study evaluated the test performance of the Abbott SARS-CoV-2 IgG assay, which is a chemiluminescent microparticle immunoassay for the qualitative detection of IgG antibody against the SARS-CoV-2 nucleoprotein in plasma or serum. To assess assay sensitivity, 689 excess serum samples from 125 patients in Seattle who were PCR-positive for SARS-CoV-2 infection were tested. To assess assay specificity, 1,020 de-identified serum samples from unique individuals that had excess serum originally obtained for herpes simplex serology in 2018-2019 (i.e., pre-COVID-19) were tested. In addition, the study evaluated the seroprevalence of anti-SARS-CoV-2 among 4,856 individuals who were sampled in April 2020 via the Crush the Curve program in Boise, Idaho.

Summary of Main Findings

Sensitivity of the Abbott SARS-CoV-2 IgG assay increased with time since symptom onset and from the date of PCR positivity among mostly hospitalized COVID-19 patients. For this group, the assay had 100% sensitivity at 17 days post symptom onset and 14 days from the date of PCR positivity. In pre-COVID-19 serum samples (n=1,020), assay specificity was 99.0%; only one false-positive result was observed. Specificity of the assay could be improved to 100% while maintaining sensitivity at 100% by increasing the manufacturer’s cut-off value for seropositivity. Assay results were also shown to be reproducible. Among the participants in Boise, Idaho, anti-SARS-CoV-2 seroprevalence was 1.8% overall, 2.1% in males, and 1.6% in females.

Study Strengths

The study used longitudinal sera to describe increases in antibody detection over time and assess the impact of infection duration on assay sensitivity. The sample sizes used to assess both assay sensitivity and specificity were large.

Limitations

The specimens used to evaluate test performance characteristics of the Abbott SARS-CoV-2 IgG assay were not well characterized. Most of the sera used to assess assay sensitivity were from mostly hospitalized patients who were recently infected with SARS-CoV-2. It is unclear whether the Abbott SARS-CoV-2 IgG assay would have optimal sensitivity in mild or asymptomatic cases of SARS-CoV-2 infection or recovered COVID-19 patients who are no longer PCR-positive. Additionally, the study did not include specimens from patients known to be infected with other coronaviruses, so it is unclear whether the Abbott SARS-CoV-2 IgG assay produces false-positive results due to cross-reactivity. Finally, the Crush the Curve participants were a self-selected sample and are not representative of all Boise, Idaho residents, so the seroprevalence estimates may not be reflective of the general Boise population.

Value added

This study provides key test performance characteristics for a commercially-available SARS-CoV-2 IgG antibody assay and demonstrates its utility in community serosurveys.