Skip to main content

Our take —

This small, cross-sectional study used cardiac magnetic resonance imaging to determine whether or not competitive athletes had findings consistent with myocarditis (heart muscle inflammation) after a mild or asymptomatic SARS-CoV-2 infection. While the study found evidence of heart muscle inflammation in four of 26 participants, it is unclear if these findings are because of SARS-CoV-2 infection, or the clinical significance of the findings given the absence of post-viral symptoms or blood markers of heart disease among participants. This study does not present clear evidence to guide when athletes should return to activity after SARS-CoV-2 infection or to use cardiac MRI as a screening test in asymptomatic individuals.

Study design

Cross-Sectional

Study population and setting

This study includes 26 college athletes at the Ohio State University (Columbus, Ohio, United States) with a history of testing positive for SARS-CoV-2. Twelve of the 26 athletes reported mild symptoms, and the other 14 were asymptomatic. None of the included participants were hospitalized or required COVID-19-specific antiviral therapy. This study did not include any athletes who did not have a diagnosis of SARS-CoV-2, or any non-athletes. All participants received cardiac magnetic resonance imaging used to detect myocardial inflammation from 11 to 53 days after their positive SARS-CoV-2 test.

Summary of Main Findings

This study reported four cases of cardiac magnetic resonance imaging (MRI) findings consistent with myocarditis (inflammation of the heart muscle) findings among 26 college athletes with a history of SARS-CoV-2 infection, none of whom have symptoms of myocarditis or blood markers of heart muscle damage. It did not include cardiac MRIs from non-athlete controls or controls who did not have SARS-CoV-2. The results of this study suggest the need for more research into the prevalence and prognostic impact of abnormal CMR patterns in otherwise healthy individuals compared to those seen in competitive athletes, and a prospective study that compares baseline cardiac MRIs prior to and after infection with SARS-CoV-2.

Study Strengths

This study includes participants in an understudied population in the SARS-CoV-2 pandemic.

Limitations

The lack of cardiac MRI prior to SARS-CoV-2 infection limits our ability to determine whether these cases of myocarditis are caused by SARS-CoV-2. Furthermore, the lack of information on athletes who did not contract SARS-CoV-2 makes it difficult to compare the characteristics of cardiac MRI patterns over time in competitive athletes. The cardiac MRIs were also not performed a consistent number of days after a positive SARS-CoV-2 test, which makes it difficult to compare participants with otherwise similar characteristics to each other. Finally, this study only includes 26 participants, four of whom had cardiac MRI characteristics of myocarditis, which makes it difficult to extrapolate these findings to all athletes, especially in the absence of a control group that was not infected with SARS-CoV-2.

Value added

This study uses cardiac MRIs to visualize the hearts of competitive athletes, which could have implications for return-to-play for athletes who test positive for SARS-CoV-2 pending further investigation into baseline cardiac MRI findings in this population.

Our take —

This study presents information on the prevalence of SARS-CoV-2 infection among staff and residents of nursing homes in West Virginia following an executive order from the governor requiring universal testing in all nursing homes in the state. Overall, 0.1% of residents (11/8911) and 0.2% of staff (31/13,687) had positive results for SARS-CoV-2. Of the 42 cases identified among residents and staff, 19% were asymptomatic. While over 98% of residents and staff were tested, it is important to note that testing was done to detect current infections, thus any past infections may have been missed. The implementation of universal testing and infection prevention protocols likely prevented continuous transmissions and larger outbreaks of COVID-19 among staff and residents of West Virginia nursing homes.

Study design

Cross-Sectional

Study population and setting

On April 17th 2020, the governor of West Virginia issued an executive order for the state’s Bureau for Public Health to coordinate universal testing for SARS-CoV-2 among residents and staff in all 123 nursing homes in the state, regardless of symptoms. This order was issued following 307 reported COVID-19 cases among residents and staff in 7 West Virginia nursing homes from March 17 to April 16, 2020. From April 21st, nasal/nasopharyngeal swabs were collected from nursing home residents and tested using real-time reverse transcription PCR. An outbreak was defined as ≥ 2 more lab-confirmed cases of COVID-19 within 14 days among staff or residents of a facility. In facilities with active outbreaks, all persons who had never been tested and those who had previously tested negative received testing. In accordance with CDC guidelines, residents with positive test results were isolated in private rooms, and staff with positive results were required to isolate at home. Staff who tested negative but had were close contacts of residents/other staff who tested positive were required to quarantine at home for 14 days since last exposure. After implementation of universal testing, staff and residents were screened daily, and anyone with signs/symptoms of COVID-19 was tested.

Summary of Main Findings

From April 21 to May 8, 2020, universal testing was conducted in all 123 nursing homes in West Virginia. A total of 42 COVID-19 cases (11 residents & 31 staff members) were identified in 28 nursing homes—none of these facilities had a previous outbreak. This represented a positivity rate of 0.1% among residents who were tested (n = 8911) and 0.2% among staff members who were tested (n=13,687). Of the 42 cases, 20 were single cases from 20 facilities, and 22 cases from outbreaks at 8 facilities; 6 residents and 2 staff members were asymptomatic. Prior to universal testing, there had been 32 COVID-19 associated deaths among nursing home residents in the state; however, no COVID-19 related deaths occurred among residents in the period under consideration. A total of 1.3% (115/9,026) of residents and 1.7% (239/13,926) of staff declined testing.

Study Strengths

Statewide universal testing across nursing homes using real-time reverse transcription PCR for detection of SARS-CoV-2.

Limitations

It is possible that residents and staff who declined receiving testing differed from those who were tested, however the numbers declining were proportionately low. It is important to note that testing was done to detect current SARS-CoV-2 infections, thus any past infections may have been missed.

Value added

This study describes the use of universal testing to estimate prevalence of SARS-CoV-2 infection among staff and residents of nursing homes in West Virginia during the first wave of the COVID-19 pandemic in the US. This study provides important information about nursing homes, a type of facility whose residents are particularly vulnerable to adverse outcomes from COVID-19.

Our take —

This study conducted qualitative in-depth interviews with 30 informal settlement dwellers receiving food assistance in Tshwane, South Africa during a COVID-19 lockdown. Interviews revealed the deleterious economic, social, and health-related impacts of COVID-19 lockdown policies on low-income households and communities.

Study design

Cross-Sectional, Other

Study population and setting

The authors conducted telephone interviews with 30 men and women receiving pandemic relief assistance from humanitarian organizations operating in informal settlements in Tshwane, South Africa. Interviews were conducted during the 21-day lockdown imposed by South Africa’s government, which mandated the closure of all schools and non-essential business and required all South Africans to shelter in place if not reporting for an essential job, seeking medical care, or purchasing food or other essential items.

Summary of Main Findings

Participants reported difficulties adhering to social distancing recommendations, given overcrowded housing conditions in informal settlements. In spite of the lockdown orders, participants expressed concerns about COVID-19 transmission in informal settlements due to people continuing to congregate in public spaces. Increased strain on public infrastructure, including waste management, due to crowding within the settlement was an unintended consequence of the lockdown. Many participants were forced deeper into poverty due to lost wages from business closures. Insufficient money also exacerbated food insecurity, which some participants reported destabilized medication adherence for people living with HIV and tuberculosis. While South African schools adapted to online modalities of delivery, poor internet connectivity and lack of internet access in many households resulted in temporary school discontinuation and attrition. Lastly, increased isolation due to mobility restrictions increased sentiments of anxiety and depression.

Study Strengths

The authors implemented qualitative interviews to gain a deeper understanding of informal settlement dwellers’ experiences during the mandated national lockdown. The iterative thematic analysis of interview transcripts allowed the study team to interrogate and critically examine their interpretations of participant narratives captured in the interviews.

Limitations

Given the inductive and exploratory nature of the research, the authors could not probe or compare the experiences of informal settlement dwellers of Tshwane to those in other settings. Participants were also receiving pandemic relief from humanitarian organizations, which may render their experiences incomparable to those of other informal settlement dwellers or South Africans who did not receive or require this assistance. Additionally, the qualitative results summarized did not explicitly present comparisons in salient themes or experiences between participant strata, including gender or age.

Value added

This is among the first published studies to use qualitative methods to document the impact of COVID-related lockdowns on low-income individuals living in informal settlements, particularly in the African setting.

Our take —

This study, available as a preprint and not yet peer reviewed, described how the Wayne County Medical Examiner office (WCMEO) in Michigan implemented a surveillance strategy testing 821 decedents (i.e. deceased persons) for SARS-CoV-2 infection. Testing took place from March 16 to July 10, 2020 through nasopharyngeal swabs. Overall, 8.8% tested positive for SARS-CoV-2, and Black race and older age (above 40 years in the study) were associated with a positive test result. Among suspected cases, 20% tested positive, while among randomly selected cases not suspected to be associated with COVID-19, 5% tested positive. As testing decedents in a medical examiner’s office may be a means to supplement existing surveillance data, it may not reliably reflect infection rates for a given catchment area, but may offer insights into COVID-19 associated deaths that were not previously identified.

Study design

Cross-Sectional

Study population and setting

The Wayne County Medical Examiner office (WCMEO) serving Wayne and Monroe countries in Michigan, piloted a surveillance strategy in which they tested 821 decedents (i.e. deceased persons) for SARS-CoV-2 infection. Decedents were flagged to be tested if they had one or more of the following: probable or confirmed COVID-19 diagnosis, symptoms (e.g. fever, shortness of breath, sneezing, coughing, chest pain, or body aches), recent travel, or contacts with probable or confirmed COVID-19 diagnosis or symptoms. WCMEO staff also tested randomly selected decedents who were not flagged. Testing took place from March 16 to July 10, 2020 through nasopharyngeal swabs. Investigators compared the socio-demographics between those who tested positive and those who did not as well as compared flagged decedents vs. non-flagged decedents.

Summary of Main Findings

Among 821 decedents tested, 230 (28%) were flagged by the COVID-19 checklist and 591 (72%) were randomly selected. The mean age among all decedents was 46. Overall, 72/821 (8.8%) tested positive for SARS-CoV-2. Those who tested positive for SARS-CoV-2 were older (mean age 53 vs. 45 among negatives, p<0.001) and Black (89% vs. 51% among negatives, p<0.001). Those who were flagged to be tested were also older on average compared to decedents who were not flagged (mean age 49 vs. 45, respectively, p<0.001). Decedents flagged by the COVID-19 checklist were more likely (20% vs. 5%, p<0.001) to test positive for SARS-CoV-2.

Study Strengths

A strength of this study is that the method of testing decedents for SARS-CoV-2 infection is a means to document cases that were asymptomatic.

Limitations

The authors try to make the case that this tool could be a potential “early warning sign” for an outbreak if it is used to supplement other forms of surveillance; however, this is unlikely to be the case as infections such as SARS-CoV-2 and influenza may already spread to epidemic levels by the time cases are documented among decedents given the known lags between rising cases and deaths. Finally, it is unclear if the cause of death for any of these decedents was due to SARS-CoV-2 infection directly as most of those who were positive had a manner of death listed as “natural, other”. The sensitivity of PCR testing to identify SARS-CoV-2 viral RNA postmortem is uncertain, so it is possible that some true deaths from COVID-19 may have been missed with postmortem testing.

Value added

This study is one of the first to test decedents as a means of SARS-CoV-2 surveillance.

Our take —

This study found that adherence to self-isolation measures among Japanese workers experiencing cold-like symptoms was low despite guidelines to self-isolate if ill. Availability of work-from-home options, paid sick leave, and home delivery options of groceries and other necessities are likely needed to increase the practice of self-isolation, particularly among working populations with mild disease.

Study design

Cross-Sectional

Study population and setting

Since February 17, 2020, individuals experiencing cold-like symptoms are recommended by the Japanese Society of Travel and Health to self-isolate for 7 days after symptom onset. The authors investigated the practice of such self-isolation through an internet survey of 1,226 Japanese workers recruited by a Japanese Internet research service company. Study participants were asked (yes/no) if at any point they had experienced “cold-like symptoms” since February 17, 2020 and if yes, if they had maintained self-isolation. The survey also collected information regarding their employment and sociodemographic factors.

Summary of Main Findings

82 out of the 1,226 participants (6.7%) indicated having had cold-like symptoms between February 17 and the day they took the survey (May 12 to 17, 2020). Of the 82 individuals, only 14 (17.1%) maintained strict isolation, defined as not leaving their homes at all or exclusively leaving for hospital visits. Most of the individuals who did not maintain strict self-isolation left their homes to shop for groceries or other necessities (70.7%) or to go to work (62.2%). Inability to work from home (OR: 4.22, 95% CI: 1.02-17.43) or being a company’s employee (as opposed to being self-employed – OR: 25.81, 95% CI: 2.23-298.31) were both indicated as being statistically significant factors in breaking the 7-day self-isolation period.

Study Strengths

The study empirically assessed the efficacy of self-isolation measures and potential barriers to following isolation guidelines among a working population. Data on type of job, impact of self-isolation on employment prospects, and reasons for lack of self-isolation were collected.

Limitations

Those who were unemployed, stay-at-home individuals, or students were excluded, limiting the generalizability of the findings. Cold-like symptoms are not necessarily indicative of COVID-19. Overall, sample size was limited. Of the 82 individuals experiencing cold-like symptoms, 24 (29.3%) indicated having violated their 7-day of self-isolation to go to “other places for different reasons”. Authors did not clarify what those other locations or reasons may be. Recall bias may also be an issue, given that the online study took place in May but participants were asked about their behaviour and their health since mid-February. Disease severity or symptom type was not assessed or correlated with propensity for self-isolation. In addition, as the authors note, study participants were recruited from a pool of individuals enrolled in a single Internet research company, potentially introducing selection bias and limiting generalizability of findings to the larger Japanese population. Only 6.7% (n=82) of the study population exhibited cold-like symptoms. Transmission of COVID-19 and isolation measures varied in the February – May, 2020 timeframe; the survey did not ask when participants experienced cold-like symptoms and subsequent self-isolation within this 3-month time frame, potentially neglecting confounding factors. Finally, authors note that the study population was recruited from a larger longitudinal study investigating personal protective measures during the COVID-19 pandemic, which may have altered the perception and implementation of participants’ self-isolation behaviour.

Value added

This study adds to the very limited evidence base on the actual practice of self-isolation among workers reporting cold-like symptoms following implementation of self-isolation guidelines to prevent COVID-19 transmission.

Our take —

Though causal relationships between individual behaviors and SARS-CoV-2 infection cannot be established in this cross-sectional survey, available as a preprint and thus not yet peer reviewed, the survey tool may be useful for future longitudinal studies to evaluate the role of individual behaviors in transmission dynamics. Mask use and social distancing were associated with lower rates of SARS-CoV-2 test positivity, while more frequent travel, especially via public transportation or to places of worship, were associated with increased SARS-CoV-2 test positivity. While plausible, these results should be considered cautiously in light of several study limitations.

Study design

Cross-Sectional

Study population and setting

The study included 1,030 individuals (median age: 43 years, 55% female) from the US state of Maryland who participated in an online survey between June 17 and June 28, 2020, shortly after Stage Two of Maryland’s phased re-opening plan began. The survey was designed to evaluate adoption of non-pharmaceutical interventions (social distancing and mask use), travel, access to SARS-CoV-2 testing, and SARS-Cov-2 test results. The survey was distributed through an online platform (Dynadata), equipped with security checks, quality verifications, and preset quotas for age, gender, race/ethnicity, and income to accrue a sample representative of the Maryland population. Participants were excluded if they were less than 18 years old, currently resided outside the state of Maryland, did not complete the survey, or did not respond to ever being tested for SARS-CoV-2 (the survey was distributed to 2,322 individuals, 1,466 responded to at least 1 survey question, and 1,030 met all inclusion criteria).

Summary of Main Findings

Sociodemographic characteristics were broadly representative of Maryland’s population. During the prior two weeks, 96% of participants left their home at least once: 92% travelled for essential services, 66% visited friends/family, 49% went to an indoor venue (bar, restaurant, salon), and 25% went to an outdoor venue (pool, beach). Practicing of social distancing increased with age, and mask use was least common among white individuals. In all, 55 participants self-reported ever testing positive for SARS-CoV-2 in the past, and in the prior 2 weeks, 62/102 participants who wanted/needed a SARS-CoV-2 test received one, of whom 18 tested positive. In multivariable analyses, more frequent use of public transportation and more frequent visits to a place of worship were strongly and positively associated with ever testing positive for SARS-CoV-2, whereas adoption of social distancing was negatively associated with a positive test.

Study Strengths

Survey distribution methods allowed for rapid and secure data collection. The study population appeared to be broadly representative of the Maryland population with respect to demographic variables and self-reported SARS-CoV-2 positivity rates counts in Maryland. Some sensitivity analyses were conducted.

Limitations

Self-selection into the study population and the requirement of participants to have an internet connection limit the generalizability of the results. Additionally, because data were collected cross-sectionally, it cannot be established that self-reported behaviors preceded SARS-CoV-2 test results. For example, some individuals who previously tested positive and recovered from SARS-CoV-2 may believe they are immune, thus traveling more often and practicing less social distancing and mask use. Unadjusted sensitivity analyses restricted to self-reported SARS-CoV-2 test positivity in the previous two weeks demonstrated somewhat similar results in the main analysis, but the protective association between social distancing and test positivity was not observed. However, even in these analyses, temporal relationships cannot be established. Self-selection into the study and self-reporting of behaviors and test results may introduce bias, underestimating the association between behaviors and infection rates if individuals with perceived “riskier” behaviors are less likely to participate, inaccurately over-report non-pharmaceutical interventions, or are less likely to report testing positive.

Value added

In this study, a rapid and cost-efficient online survey tool was used to evaluate the association between social distancing, mask use, and SARS-CoV-2 test results.

Our take —

Younger respondents, compared to older respondents, perceived higher risks for SARS-CoV-2 infection, death, and hospitalization due to COVID-19 for all age groups. Those that perceived greater average COVID-19 risks for others were more likely to favor stricter lockdowns. Those that perceived greater individual COVID-19 risks were likely to self-report behaviors that reduced contact with others.

Study design

Cross-Sectional, Other

Study population and setting

The authors conducted an online survey of 1,522 Americans between May 6 and 13, 2020 to assess individual perceptions of risk related to COVID-19. The authors recruited respondents based on previously selected quotas for demographic variables (sex, age, household income, geographic region, and race). Respondents exhibiting misunderstandings of ratios and proportions were screened out. Respondents were asked to consider 1,000 people similar to themselves (based on age, zip code, sex, income, race, etc.), and to estimate the number of people who would contract COVID-19 in the next nine weeks, and how many would be hospitalized or die; they were then asked how these risks differed by age, sex, and race. Respondents were also asked to estimate the risk of non-COVID health issues (hospitalization, death, heart attack or cancer) amongst people that were similar to themselves. Participants were also asked a range of questions about their attitudes and personal behaviors related to COVID-19.

Summary of Main Findings

Compared to older participants, younger respondents (18-34 years) estimated higher risks for infection, hospitalization, and death due to COVID-19 for people similar to themselves and for those older than themselves. The authors performed a similar analysis using an international dataset and found similar results, with older respondents estimating lower risks of contracting, hospitalization, and death from COVID-19 compared to those who were younger. Individuals who were more likely to estimate greater COVID-19 risks were also likely to estimate greater non-COVID-19 health risks (e.g. risk of heart attack, cancer). Finally, individuals who perceived higher COVID-19 related risks for themselves were less likely to go out and attend routine doctor’s appointments and fill prescriptions, suggesting that individual risk perception correlated with individual behavior. On average, those that perceived greater risks for others were more likely to prefer stricter or longer lockdown measures, suggesting that individual beliefs about average risks inform policy preferences.

Study Strengths

The authors recruited and collected data from a large, diverse sample of the US population. The authors observed broadly similar results in a publicly available global database. This suggests that the observed age-related differences in risk perception may not be specific to the US.

Limitations

The authors did not report the number of potential participants who were screened out due to incorrect answers or were ineligible due to demographic quotas. When participants were asked to consider a group of people similar to themselves, it is not clear which social identifiers participants used. Based on the data analysis, the authors assumed that age is a primary social identifier. However, given that participants were asked to consider multiple social identifiers at once (race, age, sex, etc.), it is unclear if analysis by a single variable (i.e. age) is justified. Recruitment via social media platforms or phone may bias the sample in ways not captured by basic demographic variables. Furthermore, very little detail was provided on how the data were analyzed, making the statistical methodology hard to assess. Two potentially important but unmeasured determinants of risk perception that may be correlated with age (therefore potential confounders of the observed association) are political views/affiliations and personal knowledge of someone who had COVID-19.

Value added

How COVID-19 risk perceptions vary by subpopulation has not been widely researched, but could inform how social and behavioral change interventions are adapted for different demographic groups in the US.

Our take —

This was a population-based study that aimed to highlight epidemiological and clinical features of the COVID-19 epidemic in Brazil. Investigators found that the estimated basic reproductive number in Brazil was 3.1, and that areas with higher income per capita were more likely to have cases versus higher numbers of acute respiratory infections of unknown causes in lower income areas. The finding that higher income areas were more likely to have cases differs from other reports of COVID-19 being more prevalent in lower income communities; however, this could be due to greater access to testing in these areas. Overall, the study can be a useful tool for those wanting to improve surveillance and reporting of COVID-19 in Brazil.

Study design

Cross-Sectional

Study population and setting

This article presents epidemiologic and clinical features of the COVID-19 epidemic in Brazil up to May 31, 2020. This included estimating the basic reproductive number (R0) for Brazil and key municipalities, comparing these estimates with European countries, and identifying demographic factors associated with infection. Investigators also explored the association between socioeconomic status and COVID-19 geographic distribution.

Summary of Main Findings

Investigators analyzed 514,200 COVID-19 cases that were reported by 75% of municipalities across Brazil’s five regions. The most common symptoms among cases were cough, fever, and shortness of breath. The R0 for Brazil was 3.1 (95% Bayesian credible interval (BCI) 2.4-5.5). Investigators found this was higher, yet had overlapping 95% BCI intervals with European countries (Spain: 2.6, France: 2.5, United Kingdom: 2.6, and Italy: 2.5). Investigators found that census tracts with a higher per capita income were more likely to have cases of COVID-19, whereas census tracts with a lower per capita income were more likely to have cases of severe acute respiratory infections of unknown causes.

Study Strengths

The study used multiple surveillance reporting systems and has a large population-based sample.

Limitations

There was only individual level data for two of their surveillance reporting systems (i.e. REDCap and SIVEP-Gripe). Thus, investigators could not make individual level inferences concerning the association between socioeconomic status and COVID-19 diagnosis.

Value added

This is a detailed documentation of the nature of the COVID-19 epidemic in Brazil and can be useful in future efforts to improve surveillance and reporting in Brazil.

Our take —

This study used contact tracing data conducted through the Chinese Centers for Disease Control. The overall attack rate was low (<1%); however, this study provides information that is not well documented (i.e. comparing secondary attack rates based on spatial positioning and travel time among passengers on a train). Though this is one of few studies to make such comparisons, the findings are only among symptomatic cases and do not represent subsequent infections among asymptomatic individuals.

Study design

Cross-Sectional

Study population and setting

The Chinese and local Centers for Disease Control and Prevention conducted contact tracing for 2,568 confirmed COVID-19 cases who traveled from December 19, 2019 to March 6, 2020 on the G train, which is the mostly widely used train in China and accounts for over 60% of China’s rail passengers. The Chinese government started issuing lockdowns and public transit travel restrictions in different cities, starting with Wuhan, China and other hot spots in the Hubei province starting January 23, 2020. Researchers aimed to examine spatial distance (ranging from seats A (window seat), B, C (aisle seat), D (aisle seat), and F (window seat) within each row of the train) of contacts and duration of travel time in relation to viral transmission risk. Someone was a contact of a case if they had traveled within a three-row seat distance (either in front or behind) and 5 columns of an index case within 14 days of the case’s symptom onset. Investigators calculated the attack rate for different seat locations adjusting for co-travel time in order to explore individual level transmission risks on public transportation.

Summary of Main Findings

The overall attack rate among contacts was 0.32% (234/72,093) (95% CI: 0.29, 0.37). The attack rate for those seated in seat A from was 0.28% (95% CI: 0.21, 0.39), from seat B: 0.41% (95% CI: 0.31, 0.54), from seat C: 0.34% (95% CI: 0.26, 0.45), seat D: 0.34% (95% CI: 0.26, 0.45) and seat F: 0.27% (95% CI: 0.20%, 0.36%). Attack rates were different depending on the row a contact was sitting. Contacts in the same row as an index case had an attack rate of 1.5%, which is 10 times higher than the attack rate of being 1 or 2 rows away from the index case. Travel time was associated with the attack rate as there was an increase by 0.15% per hour of travel time, though this relationship strengthened after a contact traveled for more than 4 hours.

Study Strengths

This study provides useful surveillance data that is usually not available in other settings (i.e. information concerning potential contacts of confirmed cases from train travel). The date of symptom onset and date of diagnosis were available for all cases, making this a rich data source for contact tracing analyses.

Limitations

Those without symptoms were not tested, and thus results represent an underestimation of travel risk since asymptomatic infections are not represented. Considering that the majority of cases have mild or no symptoms, it is unclear if the data represent the true infectivity of cases during train travel. Additionally, although there are separate attack rates within each row depending on which seat a contact was sitting in (i.e. A vs. B vs. C, etc.), the results do not account for how close a contact was to a given case (i.e. if a contact was sitting in seat A and was sitting next to the case versus if the contact was sitting in seat A and sitting several seats away from a given case).

Value added

This study provides information concerning the likelihood of contracting SARS-CoV-2 virus while traveling on a train. The evidence can be useful in informing people’s decisions when trying to travel or informing safety guidelines to reduce transmission on trains.

Our take —

This is a contact tracing study based in Trento, Italy, published as a preprint and thus not yet peer reviewed. There were 2,812 cases and 6,690 contacts. There was an overall secondary attack rate of 13.3%, with the highest secondary attack rate occurring among contacts over the age of 75 years. However, index cases who were between the age of 0-14 years had the highest percentage of contacts who became infected (22%). There was no routine testing of contacts, so most of them were identified as a case through being symptomatic; thus, there is likely an underrepresentation of the secondary attack rates. However, the finding that the youngest age group of index cases had the highest transmission among their contacts is helpful to note when policy makers are making decisions on reopening schools.

Study design

Cross-Sectional, Ecological

Study population and setting

The provincial agency for health services (APSS) in Trento, Italy conducted contact tracing from March to April, 2020 using a contact tracing website. Data on cases was provided by the central local health unit database while data on contacts of cases was collected by telephone interviews contact tracers from each local health district. A contact was defined as anyone who had contact with a confirmed or probable case within 48 hours prior or 14 days after symptom onset.

Summary of Main Findings

There were 2,812 reported cases, with almost half having up to three contacts each, for a total of 6,690 contacts (890 of whom developed symptoms). Prior to the lockdown on March 10, 2020, (consisting of shutting down schools, universities, and businesses except for grocery stores, pharmacies, and newsstands), the majority of contacts were non-cohabitating family or friends (~37%); however, after March 10, the majority of contacts became household contacts (67%). Ultimately, household contacts comprised 56% of all contacts and non-cohabitating family or friends comprised 27%. The secondary attack rate steadily increased with age (e.g. 18.9% among those 75 year and older vs. 8.4% among those 0-14 years). However, the youngest age group (0-14 years) were more likely to spread infection than any other age group, as 22% of their contacts became infected.

Study Strengths

This study has a large sample of cases and respective contacts. The contact tracing website also provides a centralized resource for data on cases and contacts that can be helpful for future analyses.

Limitations

Classifying a contact as a case was determinant on being symptomatic and having an epidemiological link as there was no routine testing conducted among contacts. Thus, the study is likely reporting an underrepresentation of how many contacts became infected (especially among younger age groups as these groups are more likely to exhibit mild to no symptoms).

Value added

As schools are opening up in the United States and other countries, the fact that secondary infection was more likely to occur in the youngest age group in this study suggests a potential for high levels of transmission both in schools and households if there are not protocols in place to reduce transmission while children are in school.