Skip to main content

Our take —

A population-based serosurvey was conducted using random household-based sampling in Kinshasa, Democratic Republic of Congo from October 2020 to November 2020. Of those participating in the study, 16.6% tested positive for SARS-CoV-2 antibodies, suggesting a substantial undercounting of the true number of SARS-CoV-2 infections compared to the number of cases reported by the national health surveillance system. Findings should be interpreted with some caution as close to half of those eligible were not included in the final analyses, potentially under- or overestimating the prevalence, depending on whether those at home were more or less likely to be previously infected.

Study design

Cross-Sectional

Study population and setting

Between October 22, 2020 and November 8, 2020, a cross-sectional, household-based serosurvey was conducted to assess the prevalence of SARS-CoV-2 IgG antibodies in Kinshasa, Democratic Republic of the Congo. A sampling frame was constructed using health divisions of the city; random sampling was done based on a three-stage probability proportional-to-size sampling strategy. Venous blood samples were collected from all available participants and a Luminex-based assay was used to detect IgG antibodies against both SARS-CoV-2 nucleocapsid and spike proteins. A positive serology was based on reactivity to both SARS-CoV-2 proteins. Using a smartphone application, participants answered questions regarding their household members, symptoms, socioeconomic status, and behaviors. The seroprevalence estimate was weighted and age-standardized based on demographic data.

Summary of Main Findings

Among 1,233 participants from 292 households, the weighted, age-standardized estimate of seroprevalence was 16.6% (95% CI: 14.0-19.5). An additional 17.1% were considered “indeterminate,” as their test was shown to be reactive to one of the two SARS-CoV-2 proteins. Nearly three of every four participants shared a common yard space (72.2%, 890/1,233), as opposed to living in a single family home, and over half did not have access to handwashing at home (54.2%, 668/1,233). Based on the measured prevalence, the authors estimated that there had been a total of 2,426,406 infections in Kinshasa , which would indicate that for every one case identified through the health surveillance system, there were 292 cases that were not diagnosed.

Study Strengths

The use of random probability proportional-to-size sampling is a strength because this study likely represents the population-based prevalence better than a study conducted in a specific population group.

Limitations

A total of 2,400 individuals were eligible for the study. Of these, 1,607 were present at the time of the enrollment and 1,233 were included in the final analysis. Therefore, the proportion of eligible participants included in the final analysis was 51%. By only enrolling such a limited proportion of all eligible individuals, this study may not be representative of the entire population of Kinshasa. As a result, the true population-based prevalence may be biased downward or upward, depending on whether those previously infected were more or likely to enroll; such bias would also influence the calculated ratio of reported cases to true infections.

Value added

This is one of the first studies to assess seroprevalence in the African context in the general population. Past serosurveys have primarily been conducted among specific population groups (e.g. healthcare workers). This study helps shed light on the discrepancy between reported cases and the true number of cases.

Our take —

Using surveillance data paired with questionnaires from 622 individuals tested for SARS-CoV-2 infection at a research institute in Luanda, Angola, 14.3% tested positive. The odds of infection were higher among older individuals and those living in rural areas. Despite low data availability, this is one of the first studies to shed light on the characteristics associated with SARS-CoV-2 infection in sub-Saharan Africa, and Angola more specifically.

Study design

Cross-Sectional

Study population and setting

In this study, the associations between sociodemographic and SARS-CoV-2 test results were examined among 622 individuals at a research institute in Luanda, Angola (Instituto Nacional de Investigacão em Sau´de) between January and September 2020. Sociodemographic characteristics were collected using a surveillance questionnaire. Testing was performed because of suspected COVID-19, exposure to someone infected with SARS-CoV-2, or travel to a place with active transmission. Testing was performed using smears or swabs from the upper respiratory tract. Screening and confirmation were performed using RT-PCR assay.

Summary of Main Findings

Overall, 14.3% (89/622) tested positive for SARS-CoV-2. There was no difference between men and women, but older individuals and those from rural areas had a greater burden of disease. For those 60 years and older, the odds of infection was 23.3 times higher than that of those less than 10 years old (adjusted OR: 23.3; 95% CI: 4.83, 112), however estimates by age group were unstable due to small sample sizes within age sub-groups. Still, the odds of infection was higher for every age group compared with those less than 10 years old. Those living outside of Luanda had 7 times the odds of infection compared to those living in Luanda (aOR: 7.4; 95% CI:1.64, 33.4).

Study Strengths

This study was able to link sociodemographic results from a routine survey with regular testing from a research institute.

Limitations

It is noted that 622 individuals were analyzed as part of this study out of the 16,028 individuals tested during this time period, but there is no reference to how these individuals were sampled or to whom these results might apply. From the results, it appears as though 622 individuals completed the survey. There is no discussion on how these individuals may be different from those who did not complete the survey, potentially biasing the results of characteristics identified to be associated with a positive SARS-CoV-2 test.

Value added

This is one of the few studies to examine characteristics of those infected with SARS-CoV-2 in sub-Saharan Africa.

Our take —

This study, which included a prospective smartphone application cohort (United States, United Kingdom, and Sweden, total n = 400,750) and a cross-sectional Facebook survey (United States, n = 1,344,996), assessed SARS-CoV-2 testing, symptoms, and severity in 18 to 44 year-old women who also reported their pregnancy status from March to June 2020. Although they found that pregnant women were more likely to be hospitalized than non-pregnant women, they did not find a difference in self-reported symptom severity among hospitalized pregnant versus non-pregnant women. It is difficult to assess how clinical caution during pregnancy may have impacted the difference in reported hospitalization rates despite similar test positivity rates in women regardless of pregnancy status, especially without data on vitals or laboratory values on admission or outcomes after hospitalization (intensive care, death, etc.). Further, digital research recruitment tools may systematically leave out lower income women who may also be more likely to have more severe health outcomes due to underlying comorbidities and/or limited access to healthcare, making it difficult to generalize the findings to all reproductive aged women.

Study design

Cross-Sectional, Prospective Cohort

Study population and setting

This study, which included a prospective cohort and a cross-sectional survey to replicate cohort findings, assessed SARS-CoV-2 severity in reproductive age women (18-44 years old) who reported their pregnancy status from March to June 2020. The prospective cohort included women from the United States, United Kingdom, and Sweden (400,750 women in total) who used a smartphone application for a median of 18 days (Interquartile Range: 6, 34) between March 24 and June 7, 2020 to record symptoms associated with COVID-19. The cross-sectional replication dataset included 1,344,996 women surveyed through Facebook (participants were selected via a sampling procedure designed to achieve a representative sample of Facebook’s United States active user base), which asked about COVID-19 symptoms in the preceding 24 hours between April 6 and June 7, 2020. Participants in both prospective and cross-sectional samples were classified based on self-report according to their pregnancy status (pregnant vs. not pregnant), SARS-CoV-2 test results, COVID-19 symptoms, and hospitalization status. They used a bootstrapped train-test procedure in the prospective cohort to classify non-pregnant participants who declined to report their test results as suspected test positive based on self-reported symptoms. They assessed for differences in reported symptoms among women by pregnancy status, test positivity, and hospitalization status in both datasets. Finally, they created a severity index among hospitalized participants that was a weighted sum of symptoms present at hospital presentation.

Summary of Main Findings

In the prospective cohort, 14,049 (3.5%) of the 400,750 participants reported they were pregnant. While pregnant women were more likely to be tested (8% versus 6.1%), there were similar percentages of positive tests in pregnant and non-pregnant women (0.6% and 0.7%, respectively), positive tests and hospitalizations (0.07% and 0.09%, respectively), and suspected positive tests and hospitalizations (0.15% and 0.16%, respectively). Finally, non-pregnant women (6.7%) were more likely to have symptoms that classified them as suspected positive cases than pregnant women (4.5%). This pattern held in the cross-sectional replication dataset in receipt of testing (2.7% of pregnant women, 2.4% in non-pregnant women), and test positivity (0.4% for both groups), suspected positives (3% of pregnant women, 4% in non-pregnant women), which included 41,796 (3.1%) pregnant women of 1,344,966 participants. In the cross-sectional replication dataset, hospitalized pregnant women were more likely to report positive tests (0.09%) than hospitalized non-pregnant women (0.03% and 0.12%). Hospitalized pregnant women with COVID-19 were more likely to report abdominal pain and less likely to report delirium and skipped meals than hospitalized non-pregnant women with COVID-19 in both cohorts. Non-hospitalized pregnant women were less likely to report diarrhea than non-hospitalized non-pregnant women in both cohorts. Finally, symptom severity scores among hospitalized participants were not statistically different by pregnancy status in either cohort.

Study Strengths

This study employed large sample sizes and included participants in several countries.

Limitations

Both arms likely recruited participants who are more health conscious than the general population and are likely healthier and potentially at lower risk of adverse COVID-19 sequelae because of their baseline health status and therefore not representative to all women of reproductive age. Data from lower income women or other women who are less digitally connected are not represented; if their results differ due to underlying comorbidities that may affect COVID-19 severity, these results may not fully generalize to all pregnant women. Furthermore, the symptom severity score is difficult to interpret outside of the context of this study. It is also difficult to comment on whether differences in reported hospitalizations among pregnant and non-pregnant women was due to increased concern among physicians (who might be more likely to admit pregnant women with COVID-19) or concerning differences in vital signs or laboratory values.

Value added

This study included a large sample of women, including a substantial number of pregnant women, demonstrating the potential of using smartphone applications and social media to conduct broad-based symptom tracking during a pandemic and trying to understand COVID-19 severity in pregnant women.

Our take —

This study sought to examine COVID-19 incidence, case fatality, and testing capacity and positivity across the continent of Africa, with both by-country and by-region analyses. They found 2,763,421 cases across 55 countries between February 14 and December 31, 2020 and a case fatality of 2.4%. South Africa was particularly hard hit, with the highest percentage of cases (38.3%) and highest number of deaths (43.3%). The test per case ratio across the continent ranged from 9 to 12% throughout the pandemic suggesting insufficient testing capacity. The study was limited by what data were reported to the Africa CDC, and therefore countries that used rapid antigen tests or symptomatic-only testing regimens may had false negatives or reduced cases compared to those actually with the disease in the country. This study is the first to offer a high-level overview of COVID-19 with data from 55 countries in Africa.

Study design

Cross-Sectional

Study population and setting

The study sought to describe COVID-19 infections and mortality across Africa between February 14 and December 31, 2020. The study used data from the Africa CDC’s event-based surveillance data and publicly available public health data from 55 African Union member states and 5 regions. The outcomes of interest were country-specific and region-specific cumulative incidence per 100,000 people, weekly incidence per 100,000 people, case fatality ratio, testing ratio (defined as number of tests per 1 million population), and tests per case (defined as 3-week positive yield in tests conducted). Population estimates were taken from the UN Population Fund 2019 data. Active COVID-19 cases were estimated by subtracting reported deaths and recoveries from cumulative total cases reported.

Summary of Main Findings

The study identified 2,763,421 cases across 55 African Union member states. The majority (56%, N=31) of countries identified their first case between March 8 and 21, and 43% of cases were reported in the Southern region, followed by 34% in the Northern region, 12% in the Eastern region, 9% in the Western, and 3% in the Central region. Overall, 9 countries made up 82.6% of cases, with South Africa having the plurality of 38.3%, followed by Morocco (15.9%) and Tunisia (5.1%). Per 100,000 people, Cabo Verde had the highest incidence rate (1973.3/100,000), followed by South Africa (1819.6 per 100,000) and Libya (1526.4 per 100,000). 65,602 deaths were reported, with South Africa also having the most (43.3%, N=28,469). Across the continent the case fatality ratio was estimated at 2.4% and held steady since August 25. More than 26 million COVID-19 tests were conducted (19,956 tests per 1 million), with South Africa conducting the most. From March onwards, the tests per case ratio (3-week yield in positive tests) was between 9% and 12%, with heterogeneity between countries ranging from Algeria as 2.3% to Burundi with 94.1%.

Study Strengths

The study used data across all African Union member countries to examine by-country and by-region differences. Given there was initial concern of expected high rates of COVID-19 in Africa, the study was able to examine heterogeneity between countries across different epidemiological metrics in order to provide a nuanced picture of the continent. The study had temporal data across the weeks, which allowed them to compare trends from early in the pandemic to later in the pandemic when restrictions and non-pharmaceutical intervention mandates had shifted or lifted. Additionally, by also recording changes in testing capacity and tests conducted, they were able to examine whether changes in incidence rates reflect the number of tests or actual spread of disease.

Limitations

The study used passive ascertainment which is subject to any limitations in the way data were collected in each country. For instance, if countries did not report testing data daily, there may be bias towards underestimating calculations of growth rate, case fatality ratio, active cases, and tests per case ratios. Different countries also may use different testing approaches, such as only symptomatic testing or asymptomatic testing as well. Other countries used rapid antigen tests which may have higher false negative results. There was also no way for the study to discern duplicate tests reported, and therefore the number of tests conducted may not reflect the number of people who received tests, and would deflate the test per case ratios. Finally, some of their numbers may be driven by specific countries, such as South Africa in the Southern region had the highest number of cases, while other countries in the region had much lower testing capacity as well as lower cases; therefore the region as a whole is reflective more of South Africa’s experience with COVID-19 than other countries.

Value added

This is the first continent-wide study of COVID-19 in Africa.

Our take —

The study, available as a preprint and thus not yet peer-reviewed, sought to describe the COVID-19-related mortality disparity among Native Americans in the US. The study found a standardized mortality ratio of 2.77 compared to white populations, and this was even higher in some states, with South Dakota having a mortality ratio of 9.7 as compared to the white population. They found that the standardized mortality ratio was highly correlated with cthe perent of Native Americans living on reservations. The study had many limitations due to its ecological study design, including use of data collected as far back as 2014, and potential underreporting of Native American race/ethnicity. Regardless, results show a high level of disparity in Native American mortality from COVID-19 compared to other racial/ethnic populations in the US.

Study design

Cross-Sectional, Ecological

Study population and setting

The study sought to describe risk factors for COVID-19 infection and related mortality among Native American/American Indian communities in the US. COVID-19-related death counts from the National Center for Health Statistics from January 1, 2020, through January 16, 2021 were used. Midyear population estimates of 2019 were drawn from the US Census Bureau data from 10 states: Arizona, California, Oklahoma, New Mexico, Washington, New York, South Dakota, Minnesota, Utah, and Mississippi. The American Community Survey (ACS) and the Behavioral Risk Factor Surveillance System (BRFSS) were used to estimate potential risk factors that may impact transmission and mortality. For the ACS, the analysis used 2014 – 2018 data to estimate the type of health insurance, income-poverty ratio, and household living arrangements. They also extracted data on frontline worker status using data from 2018. They used the BRFSS from 2011 to 2018 to estimate smoking status and health conditions including asthma, chronic obstructive pulmonary disease (COPD), kidney disease, cancer, heart disease, diabetes, and obesity. More recent ACS or BRFSS versions were not yet available. Finally, they utilized data from MultiState, which generates a rating of open-ness during the pandemic based on state policies and capacity/industry restrictions. They categorized race as non-Latino Native American (including American Indian and Alaskan Native), non-Latino white, non-Latino Black, and Latino, and using the My Tribal Area tool, integrated 2014 – 2018 ACS estimates of Native Americans living on- vs. off-reservation. They generated standardized mortality ratios compared to the 3 other racial categories overall and by state. They disaggregated this based on reservation living status, occupation, and chronic health conditions and behavioral risk factors, generating correlation estimates for each.

Summary of Main Findings

In this study, 2,789 COVID-19-related deaths were estimated from January 1, 2020 to January 16, 2021 among Native Americans. They estimated a crude death rate of 1.63 times that among the US white population, and a standardized mortality ratio of 2.77. This was greater than the standardized mortality ratio within the Black population (1.64) and the Latino population (1.81). Stratifying by state, they found geographic differences as well, with South Dakota having the highest standardized mortality ratio at 9.7 compared to the state’s White population, while California had the lowest at 1.6 times the mortality to the state’s White population. The standardized mortality ratios for the 10 states were correlated with increasing percentages of Native Americans living on reservations (correlation = 0.8). In their sociodemographic and behavioral correlations, they found the income-poverty ratio was highly negatively correlated with the standardized mortality ratio (-0.86).

Study Strengths

The study made use of a wide range of data to describe the health disparities impacting Native Americans, an often underreported population. They disaggregated by meaningful variables indicative of structural risks of disease, such as living on a reservation which may impact access to health services, and living in multigenerational or crowded households, and having insurance. They also examined individual-level factors, such as clinical risks through COPD and diabetes.

Limitations

The study’s primary limitation was that they used many different data sources which may have different reporting guidelines and criteria. Therefore, these results paint an overall picture of Native American health and health disparities, but do not generate individual-level estimates of risk factors and are limited to standardization by age and place alone.. They also limited their analysis to individuals reporting Native American as their only race, which likely underreports the true number of Native American people in the US. This standardization does not reflect differences in the underlying clinical health between white and Native American populations likely due to differences in access to health services and clinical care, and may be biased. They also used data from prior years going as far back as 2014 which may not reflect more recent trends in disease and social factors.

Value added

This is a large study of Native American people in the US, reflecting the health disparities they face compared to white and other racial/ethnic groups.

Our take —

This cross-sectional study examined the seroprevalence and symptom onset of SARS-CoV-2 infection among Orthodox Jewish communities in 5 US states. They found high prevalence of reported symptoms (61.0%) and high seroprevalence overall (30.1%) using antibody testing. Symptom onset was most frequent in March 2020, generally between March 9 and 31. The study’s primary limitation was the use of antibody testing which only reflects ever having been infected, and not whether onset of presumed COVID-19 symptoms actually corresponded to SARS-CoV-2 infection. Therefore, these estimates may not accurately reflect incident disease over the entire study period, but rather history of an illness and a SARS-CoV-2 infection. Further, estimates may under or overrepresent individuals based on infection severity or presumption, or community sub-group and thus should be interpreted with these caveats.

Study design

Cross Sectional

Study population and setting

The objective of this cross-sectional study was to understand the signs and symptoms, and seroprevalence, of SARS-CoV-2 in a cultural community with reported high rates of infection across 5 states in the US: New York, New Jersey, Connecticut, California, and Michigan. Participants were recruited in partnership with local non-profit and social service organizations serving Orthodox Jewish people 18 years and older. In the first stage of recruitment, which aimed to determine self-reported symptoms and infection, 12,626 individuals began the survey, 9,507 completed the it (75.3% completion) and 603 had obtained a positive PCR test (6.6%) during their illness. In the second stage of recruitment, a subset totaling 6,665 adults (70.1% response rate) had antibody testing following survey completion. Of the 6,665 in the antibody cohort, 422 (6.4%) obtained a positive PCR test during their illness and 2004 (30.1%) had a positive antibody test at the time of the study. The survey included patient demographics, symptoms of COVID-19, date of symptom onset, and whether they had been tested for SARS-CoV-2 by nasal swab.

Summary of Main Findings

In the full survey cohort, 61.0% (N=5803) of people in the survey cohort reported symptoms at any point in the study. The earliest date of symptom onset with a positive nasal swab test was on February 8, 2020 in Michigan. The median and mode dates of symptom onset occurred within the same 1-week period from March 13 to 20 across all sites. In the antibody cohort (N=6,665 individuals), 2004 individuals tested positive via antibody test (30.1%). The highest seroprevalence was in New Jersey (32.5%, N=1080), followed by New York (30.5%, N=671). As in the full cohort, most individuals within the antibody cohort reported symptom onset between March 9 and March 31, though the earliest reported date with an eventual positive antibody test was in New Jersey on December 18, 2019.

Study Strengths

The study’s main strength was the large number of Orthodox Jewish people who participated in the study, allowing the researchers to examine geographic differences and describe the temporal trends in symptom onset across multiple states. They additionally noted that the date of median and mode onset were approximately 7 to 10 days following a major Jewish festival (Purim) across all sites. They also were able to accurately describe prior infection using seroprevalence measures with their antibody testing.

Limitations

The primary limitation was that the researchers only had cross-sectional data available to them reflecting ever-infection through antibody testing, and then self-reported symptoms. For instance, for individuals reporting very early symptom onset in December and January, it was not possible to determine if this was SARS-CoV-2 infection or an unrelated respiratory illness and they later became infected with SARS-CoV-2 which was then captured by the antibody test. Additionally, the study only included cases of disease where participants could participate in the community, as opposed to hospital-based data collection. Therefore, it may not have reflected cases of severe disease. There also may have been volunteer bias, with individuals suspecting they had SARS-CoV-2 being more likely to participate than others, which would inflate their estimate of seroprevalence. Additionally, the population was largely Ashkenazi Jewish with limited racial diversity, thereby reflecting primarily white Orthodox Jewish people and not Orthodox Jewish people of color.

Value added

The study is the largest to date of a tight-knit religious and cultural community that experienced high prevalence of COVID-19 during the pandemic.

Our take —

This study, available as a preprint and thus not yet peer-reviewed, investigated a COVID-19 outbreak in two wards in a large hospital in Portugal. Large scale testing was initiated after a recently discharged patient presented to the emergency department and tested positive for COVID-19. This individual was also a close contact of a current inpatient in the hospital within a non-COVID-19 ward. In total, 27 out of 102 staff (26.4%) and 21 out of 92 (22.8%) inpatients in the single non-COVID-19 ward of interest tested positive. Whole-genome sequencing and bioinformatics analysis determined that cases were part of a single origin SARS-CoV-2 variant, belonging to the COG-UK lineage B.1.1 and Nextstrain clade 20B. Most individuals were asymptomatic during the time of testing, suggesting likely silent transmission. The findings highlight the importance of periodic testing of staff and patients in healthcare settings for earlier detection of cases, and prevention and control of outbreaks.

Study design

Cross-Sectional

Study population and setting

This study investigated a COVID-19 outbreak in a large hospital in Portugal, which provides care to about 250,000 people and has about 1,500 employees. An outbreak took place in a non-COVID-19 ward (Ward B) in Summer 2020. The investigation was triggered following presentation to the emergency department by an individual who had concluded a hospital stay three days prior in another ward (Ward A). This individual was a close contact of a current inpatient in the non-COVID-19 ward (Ward B). The inpatient and a clinician (from Ward B) presented with COVID-19 like symptoms and subsequently tested positive, triggering further testing of close contacts, or persons presenting with symptoms in both wards in the 15 days prior. In total, 245 staff and patients were tested for SARS-CoV-2. Nasal and oropharyngeal swabs were collected, and RT-PCR was performed. RNA samples that were positive for SARS-CoV-2 were sent for whole-genome sequencing and bioinformatics analysis.

Summary of Main Findings

No cases were detected among 51 persons tested in Ward A. However, in Ward B, 27 out of 102 staff (26.4%) and 21 out of 92 (22.8%) inpatients tested positive. Most staff were asymptomatic at testing and all recovered. While inpatients were mostly asymptomatic at testing, mortality rate was high, 12 out of 21 died (57.1%). Results of high-quality SARS-CoV-2 genome sequences from 39 positive samples (22 staff and 17 patients) showed that all cases were part of a single origin, belonging to the COG-UK lineage B.1.1 and Nextstrain clade 20B, with spike amino acid changes D614G and L176F. The investigators hypothesized that the virus was likely introduced to the ward a few weeks before the large screening, then spread through interactions among health care workers, and interactions between health care workers and patients. Patient-to-patient transmission was very limited. Following the investigation, the hospital put in place and reinforced stricter infection prevention and control measures.

Study Strengths

Use of reverse-transcription PCR testing to confirm COVID-19 diagnosis. Genome sequences associated with the outbreak had >88% of the genome covered by at least 10-fold.

Limitations

While investigators hypothesized about which health care workers and patients could have been the index case given the epidemiologic and phylogenetic data, this could not be definitively determined.

Value added

This study highlights the usefulness of combining epidemiologic and genomic data in improving understanding about transmission of SARS-CoV-2.

Our take —

In a survey of 742 non-remote employees working in a non-healthcare setting in the US between March and June 2020, about half (45.6%) reported occupational use of protective equipment for COVID-19 prevention (e.g., face shields, masks). Fewer than a third of participants (28.9%) reported voluntary use of protective equipment if their employers did not mandate nor prohibit it. These survey data were weighted to match the US population, however, the prevalence and effect estimates may have been biased. For example, the researchers depended on mail-based recruitment, which is most likely to reach people with stable addresses, and self-reported data, in which people may misrepresent themselves.

Study design

Cross-Sectional

Study population and setting

A sample of US adults (aged 18 and older) were recruited randomly by mail in June 2020 to participate in an online survey measuring COVID-19 precautions in the workplace. Analyses were restricted to participants who self-reported working in non-healthcare settings and in-person from March 2020 onwards. The relationship between employer provision of protective equipment for COVID-19 mitigation in the workplace (e.g., masks, face shields, other personal protective equipment) and voluntary use of protective equipment were investigated using risk differences, estimated from weighted regression models.

Summary of Main Findings

Among 742 participants retained in the analysis, half (45.6%) reported using protective equipment in the workplace—over half (55.5%) of whom were required by their employers to do so. Among those who did not use protective equipment in the workplace, a majority (77.2%) perceived not needing them in the workplace. Compared to higher-income adults, lower-income adults were less likely to report using protective equipment (22.3% vs. 48.9%) and that their employer mandated using protective equipment (22.3% vs. 27.7%), but were more likely to report being unable access protective equipment (12.6% vs. 4.5%) and were prohibited from using protective equipment (6.8% vs. 2.5%). Protective equipment was reported as being used by one quarter (28.9%) of participants whose workplaces had no policies mandating or prohibiting the use of protective equipment. Controlling for occupation type and self-reported proximity to others in the workplace, voluntary use of protective equipment was 22.3% higher among adults provided with protective equipment in the workplace relative to adults whose employers did not provide protective equipment.

Study Strengths

Investigators captured multiple response options (i.e., inability to obtain, prohibited from using, required use, provided but not required use) to measure workplace policies governing occupational provision and use of protective equipment.

Limitations

Mail-based recruitment may have oversampled adults whose experiences and behaviors are different from those who were unable to participate, potentially producing biased prevalence and effect estimates. Additionally, because the survey was fielded at a single point in time (June 2020), results may not be representative of employed adults at other points in time. Because these were self-reported data, investigators could not confirm personal protective equipment usage or workplace provision of protective equipment.

Value added

This is among the first studies to estimate the prevalence of provision and use of protective equipment for COVID-19 prevention in non-healthcare occupational settings in the United States.

Our take —

This study examined paired nasopharyngeal (NP) and saliva samples from children aged 3-18 years at community-based screening centers in Dubai, United Arab Emirates in October 2020. Conditions for testing included contact with a confirmed COVID-19 patient, presumptive symptoms, or return to school. Both symptomatic and asymptomatic students were tested. Sensitivity and specificity of saliva samples were 87.7% and 98.5% respectively, compared to RT-PCR tests of the NP samples. This suggests that saliva-based testing may be a valid alternative to NP swab testing in school-age children. However, results did not include breakdowns by age or by symptoms (vs. no symptoms) and may not be generalized to children under 3 years of age.

Study design

Cross-Sectional

Study population and setting

Between October 1 and 23, 2020, 476 population-based children (age 3-18 years) were recruited from Dubai Health Authority community-based screening centers in Dubai, United Arab Emirates. Conditions for testing included contact with confirmed COVID-19 patient, presumptive symptoms, or return to school. Paired saliva and nasopharyngeal (NP) swab samples were taken by a trained healthcare person, and RT-PCR was used for SARS-CoV-2 detection.

Summary of Main Findings

The prevalence of COVID-19 diagnosed by NP swab was 16.7%, and 15.9% by saliva. The sensitivity and specificity of using the saliva sample was 87.7% (95% CI 78.5-93.9) and 98.5% (95% CI 96.8-99.5), respectively. Positive and negative predictive values were 92.2% (95% CI 84.2-96.3) and 97.6% (95%CI 95.7-98.6), respectively. Concordance between NP and saliva was not different by age or gender.

Study Strengths

This was a prospective study where a large cohort of population-based, school-aged children, both symptomatic and asymptomatic were recruited. Comparable results were observed between NP and saliva samples.

Limitations

No further age breakdown was used to determine if saliva testing is useful in younger vs older children, or in those under 3 years old. Likewise, the authors did not report whether NP and saliva concordance were different in asymptomatic vs symptomatic children.

Value added

This study was one of the largest to look at saliva and NP sampling for COVID-19 detection in population-based children (3-18 years).

Our take —

This study evaluated the Panbio SARS-CoV-2 Rapid Test Device (Abbott) among symptomatic children aged 0-17 years in a pediatric emergency department in Madrid, Spain between September 25 and October 14, 2020. The rapid antigen-based test was found to be sensitive (78%) in detecting SARS-CoV-2 infection in symptomatic children compared to the reference standard of RT-qPCR. In children with negative rapid testing results, confirmatory RT-qPCR testing may be recommended to rule out false negatives. These results are limited to symptomatic children, and the reported sensitivity may not be generalizable to the asymptomatic general population.

Study design

Cross-Sectional

Study population and setting

Between September 25 and October 14, 2020, paired nasopharyngeal samples were collected from 440 symptomatic children age 0-17 years (median age 3 years) in the pediatric emergency department of Hospital Universitario La Paz in Madrid, Spain. Results of the Panbio SARS-CoV-2 Rapid Test Device (Abbott) were compared to results from reference standard RT-qPCR.

Summary of Main Findings

In this study population of symptomatic children, the prevalence of COVID-19 was 4%. Compared to RT-qPCR testing, the rapid antigen testing yielded a sensitivity of 78% (95% CI: 52–93%) and specificity of 100% (95% CI: 99–100%). The rapid antigen testing missed 4/18 (22%) cases. This suggests that the sensitivity of rapid testing may be weaker for children, compared to adults.

Study Strengths

This study was one of the first and largest studies to date examining the accuracy of SARS-CoV-2 rapid antigen testing in symptomatic children in a real-world, emergency clinical setting.

Limitations

Although this was a symptomatic population recruited from the emergency department, the prevalence of COVID-19 infection was low. This led to a low number of positive cases and limited statistical power to evaluate the accuracy of the rapid test for identifying positive COVID-19 cases. Additionally, the study included mostly children of younger ages (median age 3, interquartile range 1-7) and did not evaluate potential differences in rapid test sensitivity for younger versus older children. The study was also only comprised of symptomatic children so generalizability of these findings to the asymptomatic, general population may be limited.

Value added

This study provides evidence that the Panbio SARS-CoV-2 Rapid Test may have lower sensitivity for detecting COVID-19 infection in symptomatic children than in adults, relative to the gold standard of RT-qPCR. Although the rapid test may be used to quickly identify SARS-CoV-2 in symptomatic children, negative results may need to be further confirmed with RT-qPCR-based testing.