Skip to main content

Our take —

This paper details a promising tool for risk stratification of patients hospitalized with COVID-19. The risk score includes 8 variables (age, sex, number of comorbidities, respiratory rate, peripheral oxygen saturation, Glasgow coma scale, urea, and C-reactive protein) that are often readily available in developed healthcare settings. The model performed well (better than 15 existing prediction tools) in development and validation cohorts, but will require validation in each new setting prior to clinical implementation.

Study design

Prospective Cohort

Study population and setting

This prospective cohort study details the development of a risk stratification tool to predict in-hospital mortality, named the 4C (Coronavirus Clinical Characterization Consortium) Mortality Score. The derivation cohort included 35,463 patients (32.2% mortality rate, median age 73, 42% female, 76% with at least one comorbidity) enrolled from February 6 – May 20, 2020, and the validation cohort included 22,361 patients (30.1% mortality, median age 76, 46% female, 77% with at least one comorbidity). Eligible participants were adult patients (18+ years) admitted to one of 260 participating hospitals in England, Scotland, and Wales who had a high likelihood of COVID-19 infection. Relevant predictors were selected prior to model development based on factors that have consistently been reported as clinically important in previous studies, that are commonly available at presentation, and that were measured on the day of hospital admission.

Summary of Main Findings

After rigorous model selection and coefficient estimation procedures, variables included in the final model (age group, sex, number of comorbidities, respiratory rate, peripheral oxygen saturation, Glasgow coma scale, urea, and C-reactive protein) were scaled to point values for the final prognostic index (the 4C Mortality Score). The 4C Mortality Score performed well in derivation and validation cohorts with good discrimination (the ability of a model to assign higher predicted risk to those who have the outcome vs. those who do not) evidenced by area under the curve (AUC) of 0.79 and 0.77, respectively, along with near perfect calibration (agreement between observed and predicted outcome risk) in both cohorts, and low brier score (average squared difference between observed and predicted outcomes; lower is better) of 0.17 in both cohorts. The authors defined 4 risk groups (low, intermediate, high, and very high) with mortality of 1.2%, 9.9%, 31.4% and 61.5% in the validation cohort, respectively. Within the validation cohort, the study also compared the 4C Mortality model to 15 previously published risk stratification scores, and demonstrated comparable, even slightly favorable, discrimination of this new model relative to all others.

Study Strengths

This was a large study and data were prospectively collected. The model uses clinical data that are routinely collected in developed country healthcare settings at the time of hospital admission. The study adheres to the TRIPOD guidelines, which set a standard for transparent reporting of model development and validation of predication models. Model performance was evaluated by several metrics, and the model performed well in the development and validation cohort and fared well in comparison to previously established prediction tools. The use of a priori variable selection, penalized regression methods to prevent overfitting, and multiple imputation of missing data reflect methodologic rigor beyond that of previous studies. The discrimination (AUC) of the model was evaluated by sex and ethnic group.

Limitations

To account for the possibility that diagnostic tests may not be universally available, the enrollment criteria did not require a diagnostic test confirming infection with SARS-CoV-2. Categorization of continuous predictors may result in loss of information and decreased predictive performance, but also enables quicker application of the tool and is arguably more clinically useful. The performance of the model was not evaluated in country or age-specific subgroups, which limits generalizability and underscores the need for the score to be validated in a new setting prior to clinical use. Additionally, some parameters in the score may not be available in resource-limited settings. Patients were followed for at least four weeks, but those without a defined outcome were considered alive. Predictive performance of the model may change with longer follow-up.

Value added

This is one of the highest quality and largest studies to develop a prediction model for COVID-19 mortality.

Our take —

This study sought to assess the extent of SARS-CoV-2 transmission among asymptomatic persons on a long-haul flight from Milan, Italy to South Korea. It found an estimated attack rate of 0.3% from infections likely obtained during the flight, and an overall prevalence of 2.3% for the disease from people who boarded the flight without symptoms. While it is impossible for them to determine for sure whether these infections were from the flight or prior, it highlights that with high adherence to passengers using high-quality masks whenever possible and social distancing by at least 6 feet, transmission can be very low during air travel.

Study design

Prospective Cohort

Study population and setting

The study objective was to assess the extent of transmission of SARS-CoV-2 during an airplane flight among both asymptomatic and symptomatic cases. On March 31, 2020, an evacuation flight from Milan, Italy, to South Korea included 310 passengers who enrolled in the study. Passengers were given N95 masks, socially distanced during preboarding only, and medical staff performed physical exams, medical interviews, and took temperature for all passengers. They identified 11 symptomatic passengers who were removed from the flight. Of the 299 passengers who arrived in South Korea, they all immediately entered medical isolation at a government facility for two weeks and were examined twice per day by medical staff via temperature checks and symptom screening. They also all received testing on their 1st and 14th day of quarantine. All crew members (n=10) and medical staff (n=8) on the flight were similarly quarantined. They also replicated a similar level of transmission in another evacuation flight of 205 passengers.

Summary of Main Findings

Of the 310 passengers total, 11 had symptoms and were barred from boarding, and another 7 cases who did board later tested positive during the quarantine, for a total prevalence of 5.8% who attempted to board or boarded the flight. Of those who did board, 6 had a confirmed positive result on day 1 and transferred to a hospital. None of them had underlying conditions or comorbidities. On day 14, another person with no underlying disease also tested positive with mild symptoms (coughing, runny nose, muscle aches) and transferred to a hospital, for a likely attack rate of 0.3%. When mapping where passengers sat during the flight, 4 of the 6 patients sat in the same row, though the reasons this occurred were not stated in the paper, and the person who was positive of Day 14 shared a bathroom with an asymptomatic case who tested positive on Day 1, which the authors hypothesize may have been a point of indirect transmission. The attack rate of new infection likely due to in-flight exposure for this cohort is 0.3% (1 of 293). In their replication, of 205 passengers, they found 4 positive cases, for a similar period prevalence of 2.0% and likely attack rate of 0.5% (1/202) among the among susceptible passengers.

Study Strengths

The study mapped the seating positions of people on the flight, which helps to identify potential points of contact where transmission may have occurred. They also used a replication dataset which showed a similar level of transmission on another long-haul flight from Milan, Italy to South Korea on April 3, 2020. Their testing protocol allowed them to identify asymptomatic cases who were otherwise missed via symptom screening. Also, by including the medical staff and crew in their monitoring and testing, they were able to show that transmission did not occur from the staff to the passengers, but rather new transmission was more likely to be passenger-to-passenger.

Limitations

Because individuals were not tested before they boarded the flight, authors were unable to determine whether the asymptomatic cases were exposed and infected during the flight or before. It is unlikely that all 6 who tested positive on day 1 developed infection from the flight due to the incubation period of disease. Additionally, while the authors hypothesize that the bathroom may have been an indirect point of transmission, it is impossible to know if this was the case, or whether this additional case had been infected just prior to the flight as well.

Value added

This paper shows that there are very low rates of transmission on long-haul flights with pre-flight screening by trained medical professionals and high-quality mask use at nearly all times.

Our take —

In this population-based cohort study of individuals with Type 1 and Type 2 diabetes in England, several clinical and demographic factors were associated with COVID-19 mortality. In addition to established risk factors such as age, sex, race/ethnicity, and socioecnomic status, this study highlights renal impairment, elevated HbA1c, and BMI as factors associated with COVID-19 mortality among diabetics. Complications of diabetes may impact COVID-19 outcomes, but the degree to which they can be modified in the context of the current pandemic to improve outcomes among diabetes remains unknown.

Study design

Prospective Cohort

Study population and setting

This population-based cohort study in England explored associations between COVID-19 mortality and several risk factors among individuals with Type 1 and Type 2 diabetes. The study included 98% of general practices in England using data linked via UK National Health Service numbers from the National Diabetes Audit, Hospital Episode Statistics, and Office for National Statistics. The National Diabetes Audit data were used to quantify deaths from all causes during the first 19 weeks of each year 2017-2020, and also provided age, socioeconomic deprivation, ethnicity, region, and duration of diabetes. The study population used in this analyses were individuals with Type 1 or Type 2 diabetes from the last full installment (Jan 1, 2018, to March, 31 2019) who were alive on February 16, 2020. Timing and source of data collection for considered risk factors were variable: clinical data on HbA1c, systolic blood pressure, total cholesterol, estimated glomerular filtration rate (eGFR), and prescription history (antihypertensives and statins) were taken from the most recent recorded measurement in 2019; BMI and tobacco use were based on most recent measurement between January 1, 2017 and December 31, 2019, and medical history (previous myocardial infarction, heart failure, stroke) were identified between April 1, 2017 and December 31, 2019.

Summary of Main Findings

From February 16 to May 11, 2020, there were 1604 and 36,291 deaths recorded among people with Type 1 (population: 264,390, mean age: 46.6 years) and Type 2 diabetes (population: 2,874,020, mean age: 67.5 years), respectively. Of these, 464/1604 and 10,525/36,291 had COVID-19 included as either the primary underlying cause or a secondary cause of death on the death certificate. Among individuals with Type 1 and Type 2 diabetes, male sex, older age, non-white ethnicity, socioeconomic deprivation, lower eGFR (renal impairment), previous stroke, and previous heart failure were associated with increased risk for death in adjusted survival models. Higher levels of HbA1c were associated with death among individuals with Type 1 and Type 2 diabetes, but the association was more pronounced for individuals with Type 2 diabetes. BMI seemed to have a U-shaped relationship with COVID-19 mortality; individuals with low and high BMI (relative to BMI 25-29 kg/m^2) had increased risk of death. At the peak of the outbreak in April, 2020, about 3500 additional deaths per week occurred in people with diabetes, the majority of which listed COVID-19 on the death certificate.

Study Strengths

This was a large population-based study in England with detailed data on individuals with diabetes. The study population includes almost all people with Type 1 or Type 2 diabetes in England. Death records from previous years were used to estimate excess mortality to COVID-19 in this population.

Limitations

Medical history, clinical data, and behavioral factors were based on data collected prior to 2020 and in some cases as early as January 2017. Missing data for individual risk factors ranged from 0-30%; it seems data were analyzed with “missing data” as a column but this isn’t clear in the multivariable analysis, and it’s possible persons with missing data were excluded from adjusted analyses. Many continuous risk factors were categorized or dichotomized, which may result in incomplete control of confounding. Additionally, we caution against interpreting the effect estimates as causal, given none of the risk factors were evaluated as the main exposure of interest, there were no pre-specified hypotheses, and co-adjusting for the entire set of other variables of interest does not accurately consider confounding or eliminate the possibility of confounding due to unknown or unmeasured variables. Results should be considered as descriptive rather than causal.

Value added

This was a large population-based study seeking to explore several risk factors for COVID-19 mortality among persons with diabetes, and estimate the excess mortality from COVID-19 compared to mortality rates of previous years.

Our take —

This population-based cohort study of more than 61 million people in England demonstrated that one third of in-hospital COVID-19 deaths between March 1 and May 11, 2020 were among people with diabetes, and that the odds of in-hospital death from COVID-19 were 2.9 and 1.8 times higher among those with Type 1 and Type 2 diabetes, respectively, relative to those without diabetes, independent of sociodemographic factors and cardiovascular comorbidities. Although the analyses excluded the significant proportion of COVID-19-related deaths occurring in nursing home settings and did not adjust for known confounders like BMI and high blood pressure, they nevertheless provide among the strongest evidence to date on the absolute and relative risks of COVID-19 mortality among people with diabetes in a high-income country with universal health care.

Study design

Prospective Cohort

Study population and setting

This national population-based study evaluated the association between diabetes (Type 1, Type 2, and other types) and in-hospital death from COVID-19 in England from March 1 to May 11, 2020. Data were ascertained through the linkage of four National Health Services (NHS) electronic health records datasets, including the COVID-19 Patient Notification System. The study sample included 61,414,470 people (mean age 41 years [standard deviation (SD) 23]; 50% female) who were alive on February 16, 2020 and registered with a general practice in England, nearly the total population.

Summary of Main Findings

The prevalence of Type 1 diabetes in the population was 0.4% (n=263,830; mean age 47 years [SD 20]; 43% female), the prevalence of Type 2 diabetes was 4.7% (n=2,864,670; mean age= 67 years [SD 13]; 44% female), and 0.1% had other types of diabetes (n=41,750; mean age 39.5 years [SD 23]; 46% female). By May 11, 2020, 23,698 people with COVID-19 had died in hospitals in England, of whom 31% had type 2,1.5% had Type 1 diabetes, and 0.1% had other types of diabetes. Compared to those without documented diabetes, the odds of death from COVID-19 was 2.9 times higher for those with Type 1 diabetes (aOR=2.86; 95% CI: 2.58-3.18 ) and 1.8 times higher for those with Type 2 diabetes (aOR= 1.80; 95% CI 1.75-1.86), independent of age, sex, socioeconomic status, ethnicity, region, and significant CVD comorbidities (coronary heart disease, cerebrovascular disease, and heart failure).

Study Strengths

This study utilized NHS data, providing a study sample size of more than 61 million and including nearly the entire national population. This large sample provided the power to estimate COVID-19 mortality risk associated with diabetes by type and among population subgroups.

Limitations

COVID-19-related deaths in nursing homes were not counted as outcomes in this study and given the higher proportion of people with diabetes in these settings, their exclusion may underestimate the relative odds of death associated with diabetes. Conversely, the associations reported between diabetes and COVID-19 mortality may be overestimated because the multivariable analyses did not adjust for several potential differences known to be positively associated with both COVID-19 mortality and diabetes, such as body mass index, high blood pressure, tobacco smoking, and kidney disease. Because the denominator for estimates was the total population rather than cases or infections, the estimates may reflect both differential risks of infection and risks of death.

Value added

Though several studies have identified diabetes as a risk factor for COVID-19 severity and mortality, this is the first study to assess risks separately for Type 1 and Type 2 diabetes, and to quantify absolute and relative risks of in-hospital COVID-19 mortality by diabetes type for England overall, and within specific age, sex, and racial/ethnic sub-populations.

Our take —

The study objective was to describe the secondary attack rate of COVID-19 in Guangzhou, China based on exposure setting. The study found that household transmission carried the highest risk, with 10.3% secondary infection rate, while public transport had the lowest with 0.1% secondary infection rate. Severity of symptoms was also related to secondary attack rates. The study may not be generalizable to other settings because of prevention measures in China, including that people are quickly moved to hospitals for monitoring after testing positive. Still, these data show that the exposure setting is important to assessing the risk of secondary infection, and that those with severe symptoms are more likely to transmit the virus compared to asymptomatic cases.

Study design

Prospective Cohort

Study population and setting

The study objective was to identify the amount of secondary cases of COVID-19 in Guangzhou, China identified through contact tracing of index cases. Index cases were identified by the Guangzhou CDC from December 28, 2019 to March 5, 2020. From the 391 index cases (244 in Guangzhou and 147 from other areas) identified, 3410 close contacts were also identified. The study collected data about exposure setting, defined as household, public transportation, health care settings, entertainment venues or workplaces, and multiple settings considered as more than one of the other four categories. Contacts were quarantined for 14 days from last contact and followed up prospectively through April 6, 2020, with symptom monitoring conducted morning and evening, with RT-PCR testing for SARS-CoV-2.

Summary of Main Findings

Of the 3410 close contacts, 127 (4.0%) were infected as either asymptomatic or symptomatic cases, and ended quarantine after a median of 2 days (IQR: 1 to 5 days) largely due to being transferred to hospitals soon after being identified as a case. Household exposure had the highest risk for secondary infection, with 10.3% (95% CI: 8.5 to 12.2%), while the lowest risk exposure settings were those on public transportation, with 0.1% infection risk (95% CI: 0.00 – 0.4%). Asymptomatic cases had a secondary attack rate of 0.3% (95% CI: 0.0 to 1.0%), while there was a dose-response relationship with increasing attack rates for those with mild vs. moderate vs. severe cases respectively. Those with severe case contacts had the highest attack rate with 6.2% (95% CI: 3.2 to 9.1%).

Study Strengths

The study identified a large number of close contacts and tested each of them regularly, which likely identified asymptomatic cases to better estimate the attack rate. The study also collected a number of exposure variables to understand different contact settings, which is important for policymakers making recommendations. The study also identified contacts and cases through a robust surveillance system, which identified people through surveillance testing, healthcare facility testing, and close contact testing in order to identify many index cases during the study period.

Limitations

The study identified a small number of secondary cases, which means that the models had wide confidence intervals and may have imprecise point estimates. Additionally, because a number of infection control efforts had been implemented, including moving positive cases to hospital settings for monitoring, these results are unlikely to be generalizable to other settings (e.g., the median duration of quarantine was 2 days because people were moved to hospitals so quickly).

Value added

This study estimates the secondary attack rate among a range of exposure settings, using robust surveillance and contact tracing methods.

Our take —

This study, which was a preprint and had not yet been peer reviewed included two analyses of patients enrolled in the Expanded Access Program (EAP) for convalescent plasma infusions in the US: one analysis of all 35,322 patients to date and another of 3,082 patients who had data on antibody titers. The study reported that mortality was lower among those who received transfusions within 3 days of severe COVID-19 diagnosis as compared a transfusion 4 or more days following a diagnosis. The study also reported that mortality was lower among individuals who received high versus low titer units. However, the cause of these differences in mortality may not be due to plasma therapy as the patients treated changed over time – were younger, healthier, less likely to have complications, and more likely to receive remdesivir later in the study – and the titer of antibodies transfused increased and the time to treatment also decreased over time. This study highlights the need for randomized controlled studies to determine if there is a benefit for convalescent plasma for COVID-19.

Study design

Prospective Cohort

Study population and setting

The US government authorized an expanded access program (EAP) nationally using an open-label protocol for patients 18 years and older hospitalized with confirmed severe COVID-19 to assess mortality rates after administering convalescent plasma donated from COVID survivors. The Mayo Clinic served as the main research body. A total of 1,959 sites registered, and 1,809 sites had at least one patient receive a transfusion from April 4 to July 4 2020. Across all study sites, 47,047 patients enrolled, of whom 36,226 received COVID-19 convalescent plasma. Compatible plasma was transfused intravenously in units of ~200mL, and additional units administered if justified based on clinical indication. Each site reported participant demographic characteristics, clinical presentation and medications received, and the number of days since COVID-19 diagnosis at the time of receiving convalescent plasma. For those who died, the date of death was reported; those without a reported death were assumed alive. A subset of participants had residual specimens saved for blood banking quality assurance; when available these were tested for SARS-CoV-2 neutralizing antibody levels. Analysis was stratified by month of transfusion – April, May, or June, 2020 – due to a marked temporal reduction in overall mortality from COVID-19 from May to July. All analyses were limited to patients who received a single unit of convalescent plasma from a single donor. This study sought to answer the following two questions: (1) was mortality lower with a shorter time to transfusion from diagnosis of COVID-19; and (2) was mortality lower with transfusion of a higher antibody titer unit of convalescent plasma. Analysis was conducted by calculating unadjusted mortality by month of transfusion and using Cox regression models to adjust for month of transfusion, gender, race, age, and clinical condition on presentation, mechanical ventilation, and receipt of hydroxychloroquine, remdesivir, or steroids prior to transfusion.

Summary of Main Findings

Of the 36,226 who received convalescent plasma, 35,322 patients were included in the analysis of time to transfusion from hospitalization and death and 3082 in the analysis of antibody titer. Overall all-cause 7 and 30-day mortality among individuals who received convalescent plasma within 3 days of hospitalization declined from 13.4% and 30.4% in April, 2020 to 6.1% and 20.2%, respectively, in June, 2020. Multiple changes in patient population, time to hospitalization, and care occurred over this time including shifts in age, gender, weight, race, clinical status, and use of remdesivir, an agent with demonstrated improved outcomes in two RCTs. For example, 49.9% of patients were receiving mechanical ventilation prior to convalescent plasma infusion in April compared to 16.4% in June while 4.7% of patients in April received remdesivir compared to 46.3% in June.

Analysis 1: Since April, there was a difference in the 7 and 30-day all-cause mortality for patients who received transfusion within 3 days of COVID-19 diagnosis compared to 4 or more days from COVID-19 diagnosis. Seven-day mortality was 8.7% and 11.9% for within 3 days and 4 or more days, respectively, while 30-day mortality was 21.6% and 26.7% for within 3 days and 4 or more days, respectively. Improved survival with earlier transfusion held over time-period and with adjusted modeling to account for age, sex, disease severity, and receipt of hydroxychloroquine, remdesivir, or steroids prior to transfusion. There were notable changes in transfusion practice with 22.7% of transfusions within 3 days of COVID-19 diagnosis in April compared to 54.3% in June.

Analysis 2: Among a subset of participants with available antibody titers (n=3082), there was evidence of a dose-response relationship between low-, medium-, or high-levels of neutralizing antibodies and mortality. For low and high titer transfusions, mortality was 13.7% and 8.9% at 7 days and 29.9% and 22.3% at 30 days, respectively. The difference in mortality by antibody titer held with stratification by transfusion month and days from diagnosis to transfusion. In addition, there appeared to be a consistent dose response in mortality from low, medium, and high antibody units. Notably, there were considerable differences in the populations receiving low and high titer transfusion; for example, during April, 11% of transfusions were high titer compared to 19% in June.

Study Strengths

The study team collected several potential clinical confounders, including disease severity at the time of transfusion and patient age that were included in an adjusted analysis and may have reduced some of the confounding in their estimates. Including the antibody titer analysis added a comparison that may have been less prone to confounding over time and by facility than the time from COVID-19 diagnosis analysis. However, even antibody titers changed over time, with a higher proportion of high titer units during the overall lower mortality month of June.

Limitations

The study did not include a comparator group that did not receive convalescent plasma which would have provided a much more robust control group to assess therapeutic benefit. In the absence of this, the researchers sought to test if patients who received convalescent plasma earlier in the disease course and with a higher dose of neutralizing antibodies had lower mortality; however, data on the timing of transfusion in the course of the illness are presented relative to COVID-19 diagnosis rather than relative to disease onset or day of hospitalization. If the time from onset of symptoms to diagnosis varied over the course of the pandemic, this could be another source of variation between time periods in the study. Substantial differences in patients receiving earlier or later convalescent plasma or higher titer plasma may have caused the observed differences, including disease severity a characteristic that used very broad categories. It appears that the cut-offs for ‘low’ and ‘high’ antibody titers were determined after the analysis started; setting those thresholds a priori would have improved the methods. The endpoint for the study was either death or no report of death rather than recovery; it’s unclear if the patients who did not die actually recovered within 30 days. Given the multiple possible confounders over time, adjusting for these in the analysis are important, but the variables adjusted for in the analysis were limited; it is plausible that patients with longer delay to intervention were those who had more severe disease. There was also no information provided on why these patients chose to receive the treatment such that the potential for selection bias could not be assessed. Finally, patients who received the intervention after 4 days had to survive that long to receive the intervention, leaving the potential for further bias.

Value added

A randomized clinical trial of plasma therapy in China and one in the Netherlands demonstrated no significant clinical benefit 28 days post onset of illness; however, these studies had limited statistical power to detect a meaningful difference. This study adds value because of the large number of patients enrolled, and showed no major safety concerns in this large group of patients. This is the first report on mortality outcomes from the national Expanded Access Program for convalescent plasma therapy from COVID-survivor donors.

Our take —

In the U.S. state of Maine, the overwhelming majority of COVID-19 contacts who participated in symptom monitoring opted to do so with an automated system using text messages or the web. Contacts who did not participate were not enumerated, making it difficult to draw broad conclusions about the effectiveness of this tool. Among all participating contacts, 12% developed COVID-19 (representing 10% of all cases in Maine during the study period), 68% of whom had household exposure; these findings are in line with estimates from contact tracing studies from South Korea and elsewhere.

Study design

Prospective Cohort, Other

Study population and setting

This study reported on 1,622 contacts (median age 29 years, 50% female) of 614 COVID-19 patients in the US state of Maine who were enrolled in an automated, web-based symptom monitoring program. Contacts were defined as anyone who was within 6 feet of an infectious contact (from 2 days before symptom onset to 10 days after symptom onset; for asymptomatic cases, the date of a positive test was used instead of symptom onset) for 15 minutes or longer. Contacts were instructed to report symptoms daily via an online questionnaire for the duration of their recommended 14-day quarantine; symptoms included cough, difficulty breathing, fever, chills, shaking with chills, muscle pain, headache, sore throat, and new loss of taste or smell. If contacts preferred not to report symptoms with the automated system, they were directly monitored by contact tracing investigators. Demographic information and symptom monitoring preferences were collected at enrollment. Case investigations were undertaken for any contact with a positive SARS-CoV-2 test result or with symptoms absent testing.

Summary of Main Findings

The vast majority (96%) of enrollees chose automated symptom monitoring over direct monitoring by public health investigators. Of those opting for automated monitoring, 60% preferred text message delivery, 21% preferred texted web link, 8% preferred telephone, and 8% preferred emailed web link. Twenty-nine percent of participants were enrolled within two days of their last contact with the index case. There were an average of 2.9 contacts per index case enrolled, and 29% of participating households had more than one enrollee. Among enrolled contacts with available data (76% for race, 63% for ethnicity), 59% were white and 39% were Black, while 4% identified as Latino or Hispanic. The primary language spoken by participants was reported as English by 80% of contacts, French by 7%, and Somali by 7%. Symptoms or a positive SARS-CoV-2 test result were reported by 231 (14%) enrollees, 190 (12%) of whom met the case definition for COVID-19. Of these 190, 127 had confirmed COVID-19, while the remaining 63 were considered probable cases. Probable and confirmed cases represented 10% of all reported cases (n=1869) in Maine during the study period. Of the 165 cases of COVID-19 among enrollees with data on source of exposure to the index case, 68% had household exposure, 18% had community exposure, and 16% had health care exposure. Four patients were hospitalized, and one died.

Study Strengths

Although the participation rate among all contacts is unknown, those who did participate provided useful data on preferences and outcomes. A sizable proportion of COVID-19 cases in Maine (10%) were identified through contact tracing and assessed via the automated monitoring program during the study period.

Limitations

Data were not available on the total number of contacts reported by index cases. Therefore, the participation rate could not be calculated, and it is not possible to draw strong conclusions about the acceptability of automated symptom monitoring. Additionally, if participation was low among contacts, the preferences, reporting behavior, and outcomes among this group of contacts may not be representative of contacts in Maine. Losses to follow-up may be under-reported, as they were not distinguished from those released from quarantine. Similarly, COVID-19 cases were likely under-reported, since SARS-CoV-2 testing was not required or administered to all contacts.

Value added

This study adds valuable (if incomplete) data on contact tracing preferences and outcomes, as few such results from the United States have been published to date.

Our take —

The findings in this study underscore the importance of a multi-pronged approach to reduce COVID-19 in the workplace, and by extension, the wider community. These approaches include testing to identify cases, contact tracing, modifications to the environment (i.e., protective barriers in workspaces), administrative (scheduling, social distancing), and use of personal protective equipment.

Study design

Prospective Cohort

Study population and setting

The aim of this study was to investigate an outbreak of COVID-19 among 3,635 employees of a meat processing facility in South Dakota during the first wave of the COVID-19 pandemic in the United States. Following notification of a case of COVID-19 among employees at the facility on March 24 2020, the South Dakota Department of Health (SDDOH) led an investigation to isolate the case, and identify and quarantine contacts. A contact was defined as a person within 6 feet of the COVID-19 case for ≥ 5 minutes during the infectious period, i.e., between start of symptoms and end of isolation. On April 1, the definition of a contact was expanded to include the 48 hours before symptom onset. By April 2, 19 cases were confirmed, and enhanced testing was put in place such that any employee with signs & symptoms of COVID-19 could get a test from a local health facility. The following additional measures were also implemented: employee screening, physical barriers installed on production lines, and masks were made optional. By April 11, 369 cases were confirmed, and a phased closure of the facility began. From April 12-14, no more animals were slaughtered, only already slaughtered meat was processed for shipping, and the facility began closing departments. From April 13, masks became mandatory. From April 15, only staff responsible for maintenance, cleaning & sanitization, transportation of remaining meat, and implementation of COVID-19 prevention protocols reported to work. A case was defined as a reverse transcription–polymerase chain reaction (RT-PCR) positive test in a symptomatic individual, or a positive test in an asymptomatic person up to 2 weeks after the phased closure which began on April 12. Employees who did not work during this period of March 2 to April 25 were excluded from the analysis.

Summary of Main Findings

The facility had 3,635 employees who harvested and processed animals from March 2 to April 25, 2020. In total, 25.6% (929) of employees, and 8.7% (210 out of 2,403) of their contacts outside of work were diagnosed with COVID-19. Within the two counties where the facility is located, a total of 2,199 COVID-19 cases were identified during this period, and facility employees made up almost 42% of these cases. Median age of persons with COVID-19 was 42 years among employees, and 29 years among contacts. Overall, 4.2% of employees with COVID-19 were hospitalized (median age, 60 years), and 4.3% of contacts with COVID-19 were hospitalized (median age, 64 years). Two employees died. In the first 3 weeks of the outbreak, the proportion with COVID-19 increased 5-fold (week 1 = 0.2%, week 2 = 1.2%, and week 3 = 6.8%). There were an average of 67 cases per day in week 4, which declined to 10 cases per day within 7 days of facility closure.

Study Strengths

Use of an RT-PCR positive test to confirm COVID-19 infection.

Limitations

Enhanced testing among facility workers likely led to more case detection among employees than among contacts or community members. Given that most employees were tested for COVID-19 following onset of symptoms, the proportions in the study are likely to be underestimates of the true proportion with COVID-19 among facility employees and their contacts.

Value added

This study demonstrates the potential of COVID-19 to spread rapidly in settings where individuals are in close proximity to one another, and for extended periods. It additionally illustrates the importance of enacting COVID-19 control measures before and immediately after introduction of COVID-19 within a workplace setting.

Our take —

This study examined transmission within school and early childhood education centers before and following distance learning recommendations in New South Wales, Australia. Of 3033 COVID-19 cases in the region through May 1, 2020, 27 primary cases were identified in 25 educational settings, who had 1448 close contacts. Students and staff continued to attend while infectious for a median of 2 days, and there were 18 secondary cases identified overall across 4 settings. In these settings, staff-to-staff transmission had the highest rate (4.4% of contacts), followed by staff-to-child (1.5% of contacts), while child-to-child transmission had the lowest rate (0.3% of contacts). This study likely missed asymptomatic cases, and may underestimate the level of secondary transmission. However, it shows the low risk of transmission during school closures and other infection control policies throughout NSW.

Study design

Prospective Cohort, Other

Study population and setting

In New South Wales (NSW) Australia, the study identified children and school or early childhood education center (ECEC) staff from all confirmed COVID-19 cases in NSW. From January 25 to May 1, 2020, NSW had 3033 COVID-19 cases overall. Of these, 97 (3.2%) of the 3033 were children and 22 (0.7%) had links to an education setting (e.g. educators, ECEC staff). Because of school closures and other policies, 19 of these children attended an educational setting (school or ECEC) while infectious, and were included in the study sample. Among school staff, all 22 were identified as having attending school or ECEC during their infectious period and were also included in the study sample. Of these 41 cases identified, 27 were identified as primary cases in 15 schools and 10 ECEC settings, and using contact tracing, 1448 of their close contacts were identified, and 663 contacts were tested for COVID-19. The study was able to obtain symptom questionnaires from 288 contacts. During the study period, school attendance rates declined from 90% to 5% after distance learning recommendations were implemented March 23rd, 2020.

Summary of Main Findings

Of the 27 primary cases identified in the 25 educational settings, 15 were staff (55.6%) and 12 were children (44.4%). The median amount of time primary cases continued to attend school/ECEC was 2 days (range: 1 to 10 days). All of the primary cases had acquired the infection locally, with most exposure sources unknown, but, when known, mostly household contacts. There were 1448 close contacts monitored over the study period, and 663 (43.7%) were tested for COVID-19. Secondary transmission was identified among 4 settings (16.0%), with 3 schools and 1 ECEC, with 18 secondary cases (1.2% attack rate) identified through contact tracing in the schools/ECECs. Among 288 contacts with symptom questionnaire data, 22.6% (n=65) of them developed symptoms during the 14 day quarantine. Staff-to-staff transmission had the highest rate (4.4%), compared to staff-to-child transmission (1.5%) and child-to-staff transmission (1.0%), and child-to-child transmission (0.3%).

Study Strengths

The study identified all residents of NSW who tested positive, and determined their school attendance; while this did result in a small sample size, it reduced the potential selection bias for a representative sample of the area. They were also able to have a symptom questionnaire paired with contact tracing to understand how many people went on to develop symptoms, and how many continued to attend schools/ECECs. Using a prospective study, they also were able to temporally identify how changes in school policy (such as closures) reduced transmission in NSW.

Limitations

The study had small sample sizes, likely due to school closures which reduced transmission compared to what it would have been without these measures. Many contacts were tested after developing symptoms, so asymptomatic cases are likely still missing from this assessment. There was incomplete symptom data available for most close contacts as well, which may reduce the estimated number of secondary cases identified. Similarly, schools/ECECs defined close contacts and may have used differing definitions, which would also lead to contacts not being identified and reduced estimated case numbers. Finally, in their attack rate calculation, they limited it to individuals identified through their enhanced contact tracing, and did not include secondary cases identified through the NSW Department of Health population surveillance, which may deflate the estimated attack rate if these individuals were missed in the tracing

Value added

This is one of the largest studies that examines transmission in school and ECEC settings, including before and following school closures.

Our take —

This study of the incarcerated population in 16 Massachusetts Department of Corrections (DOC) facilities and 13 county-level systems from April 5 to July 8, 2020 found substantially higher rates of SARS-COV-2 infection among incarcerated persons compared to the general population. Specifically, the rate of SARS-COV-2 infection was 44.3 cases/1000 persons (about 2.9 times that of the Massachusetts general population & 4.8 times that of the US general population). Mitigation measures such as increased testing, contract tracing and avoiding overcrowding are urgently needed to prevent high rates of SARS-COV-2 transmission within this population, and the larger community.

Study design

Prospective Cohort

Study population and setting

This study examined infection with SARS-COV-2 among incarcerated individuals and staff in Massachusetts jails and prisons. Researchers analyzed publicly available data reported by 16 Massachusetts Department of Corrections (DOC) facilities and 13 county-level systems from April 5 to July 8, 2020. The denominator for the rates of SARS-CoV-2 infection was the baseline population at the facilities.

Summary of Main Findings

664 out of 14,987 incarcerated individuals tested positive for SARS-COV-2 by July 8, 2020. This translated to a rate of 44.3 cases/1000 persons (about 2.9 times that of the Massachusetts general population & 4.8 times that of the US general population). County facilities had lower rates of SARS-COV-2 among incarcerated persons compared to DOC facilities (35.7 vs. 52.4 cases/1000 persons). However, the proportion positive (positivity rate) was higher in county facilities than in DOC facilities (14% vs. 5%). A total of 368 confirmed cases were observed among staff across facilities, but the authors were unable to calculate case and testing rates in this group, and no reason was specified why. Incidence of SARS-COV-2 was lower among facilities that released a higher percentage of the baseline population (county jails released ~ 21% of their overall population compared to ~9% in DOC).

Study Strengths

Laboratory confirmation of SARS-COV-2, and inclusion of data from several facilities were strengths of the study.

Limitations

Authors noted high variability in reporting of SARS-COV-2 data across facilities. Little information was provided about how testing was administered, including if testing was performed at random or amongst symptomatic persons and close contacts. Differences in testing rates between county and DOC facilities (254/1000 persons vs 1093/1000 persons) likely contributed to lower rates of SARS-COV-2 cases identified in county facilities and the higher positivity rates. Thus, the actual rate of SARS-CoV-2 infection may be under or overestimated, based on who was and was not tested within this population of incarcerated individuals, especially in county jails.

Value added

This paper presents important data on the burden of SARS-COV-2 among incarcerated persons, a group at high risk for infectious disease acquisition.