key: cord-1009488-ebr7i6sn
authors: Eigner, Ulrich; Verstraeten, Thomas; Weil, John
title: Decrease in norovirus infections in Germany following COVID-19 containment measures
date: 2021-02-10
journal: J Infect
DOI: 10.1016/j.jinf.2021.02.012
sha: c28b41d6f85b60a8a66557dd78e37aef7c1b3868
doc_id: 1009488
cord_uid: ebr7i6sn

nan

Outcomes for patients with COVID-19 infection have been widely reported for the initial peak of the pandemic. 1 However, there is a lack of data describing outcomes and characteristics of readmitted patients in the resurgent peak, after corticosteroids became standard of care.

Based on data and protocols from randomised controlled trials, most international treatment guidelines recommend 6 mg dexamethasone daily (or equivalent) for up to 10 days in those hospitalised with severe COVID-19 but stopping on discharge. [2] [3] [4] [5] [6] The UK's second COVID-19 wave peaked on 9/1/2021. 7 Here we describe the characteristics of patients admitted, discharged and readmitted, due to COVID-19, to our hospital, during this second wave. We explored the relationship between clinical and biochemical variables, treatment received during a patient's first admission, and readmission risk, in relation to corticosteroid use.

We reviewed patients admitted from the community to University College Hospital (UCH) with COVID-19 as their primary diagnosis between 1st-31st December 2020. Re-attendance and readmission data were collected for patients who re-presented within 10 days following discharge from their first admission.

Data were retrospectively collected, including patient demographics, clinical data on first admission and readmission, steroid treatment and any treatment received on discharge from the first admission.

In the primary analysis, appropriate corticosteroid dosage was defined as receiving 6 mg dexamethasone daily. Statistical analysis was conducted in Stata ver. 12.1 (StataCorp). Independent data were compared using Mann-Whitney U test or t -test. Paired data were compared by Wilcoxon signed-rank and proportions by χ 2 test.

We fitted a logistic regression model to assess relationships between demographic and clinical factors and readmission risk. We conducted a sensitivity analysis considering anyone receiving a dose equivalent to 75% of 6 mg dexamethasone daily as having received steroids, using the outcome of readmission or reattendance.

The study met the NHS definition of a quality improvement project with the departmental governance lead and did not require ethical approval.

271 patients were admitted to UCH with COVID-19 in December. 25 patients were transferred from external hospitals or had nosocomially-acquired COVID-19 and 50 patients died during their first admission or remained an inpatient throughout the data collection period and were excluded from subsequent analysis. 196 patients were included in the analysis.

Median age was 58 years (IQR 47-71); 48% female; 133/196 (67.9%) had ≥1 comorbidity (as defined by the ISARIC 4C score), 8 32 (16.3%) had diabetes mellitus. Median length of stay was 4 days (IQR 2-8). 125/196 (63.8%) required oxygen of whom 30 (15.3%) required respiratory support. 124/196 (63.3%) received corticosteroids on their first admission for a median of 5 days (IQR 3-8). All patients had acceptable peripheral oxygen saturations (SpO 2 ) at discharge( ≥92% on air or within their target range). 10/196 (5.1%) were discharged with corticosteroids. 53/196 (27.0%) were followed up in a virtual clinic post-discharge.

26/196 (13.3%) patients re-attended UCH due to COVID-19, a median of 3 days (IQR 2-5) following discharge. Of these, 20 (10.2%) were readmitted. Median CRP (mg/L) rose significantly in those readmitted from 43.2 (IQR 29.4-71.6) on discharge to 91.8 (IQR 37.3-139.6) on readmission ( p = 0.021). 17/20 (85%) required oxygen and corticosteroids on readmission of whom 6 (30%) required respiratory support.

The 11/20 patients receiving steroids during their first admission, subsequently readmitted, had a shorter initial admission (median 2 days [IQR 1-3] vs 5 days [3] [4] [5] [6] [7] [8] [9] p = 0.005), received shorter courses of steroids (median 2 days [IQR 1-3] vs 5 days [3] [4] [5] [6] [7] [8] p < 0.001) and were discharged earlier in their illness course (median day 8 [IQR [6] [7] [8] [9] [10] [11] vs day 13 [IQR 9-18], p = 0.005) than those that were not readmitted. There was no difference in SpO 2 on air at discharge (95% IQR [94%-96%] for both) or in remdesivir use (27.2% vs 33.6%, p = 0.669) . Data for patients receiving inpatient corticosteroids on their first admission were quartiled based on their duration of steroids. In the first quartile, (1-3 days) readmission rates were highest at 25% ( Fig. 1 and Table 1 ). In an exploratory logistic regression analysis, only treatment with dexamethasone significantly reduced odds of readmission (OR 0.77 per day of dexamethasone 95% CI 0.61-0.92, p = 0.012). Results were similar in the sensitivity analysis considering both equivalent doses of other steroids and both re-attendance and readmission to hospital (supplementary data).

To our knowledge this is the first study to evaluate readmission rate in the recent COVID-19 wave, in the context of corticosteroid use.

https://doi.org/10.1016/j.jinf.2021.03.002 0163-4453/© 2021 The British Infection Association. Published by Elsevier Ltd. All rights reserved.

Patients admitted from the community who were discharged alive from their first admission. Excluding ITU transfers and nosocomial transmissions. Comparing characteristics of those receiving different steroid course durations by quartile ( n = 196).

Did not receive ( n = 72) 1st Quartile1-3 ( n = 40) 2nd Quartile4-5 ( n = 26) 3rd Quartile6-8 ( n = 32) 4th Quartile ≥9 ( n = 26)

Number readmitted (%) 9 (12.5) 10 (25.0) 0 (0.0) 0 (0.0) 1 ( Despite the majority meeting safe discharge criteria, the readmission rate is significant, but concordant with rates from the first wave in similar hospitals 9 . Readmitted patients presented to and were discharged from hospital earlier in their COVID-19 illness than patients who were not readmitted. They returned to hospital after a short time after reaching their illness peak, displaying a proinflammatory phenotype as evidenced by their rising CRP. Significant oxygen requirements were observed and an appreciable proportion of patients required respiratory support.

Shorter courses of steroids on first admission increased risk of being readmitted to hospital with COVID-19. Those who received 1-3 days of steroids experienced quick clinical improvement and were discharged from hospital and corticosteroids were stopped at discharge. 25% of this subgroup were readmitted.

Our data suggest that short courses of corticosteroids may not be sufficient for patients requiring hospital admission with severe COVID-19. As patients are readmitted with evidence of ongoing inflammation, it is biologically plausible that increasing corticosteroid duration would reduce the chance of deterioration postdischarge. Many hospitals have now instigated virtual follow up with daily calls. It is therefore reasonable to consider continuing a course of corticosteroids after hospital discharge as treatment can be given within these frameworks to monitor side-effects of steroids. UK national guidelines now recognise that patients may be discharged to a virtual ward where continuation of steroids may be appropriate. Our data support this.

Despite the limitations of small sample size and retrospective data collection, our data demonstrate a high readmission rate amongst patients with COVID-19 who received shorter courses of steroids. Further research is required to establish the optimal duration of steroids and how to identify patients who require ongoing steroids at discharge.

No conflicts of interests declared by an author.

Bryan Williams is supported by Health Data Research UK Better Care Catalyst Award (CFC0125) and Health Data Research UK (LOND1).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jinf.2021.03.002 . piratory syndrome coronavirus 2" or "COVID-19" or "coronavirus disease 2019" or "2019-nCoV" or "2019 novel coronavirus" and "dyslipidemia" or "hyperlipidemia" or "low-density lipoprotein" or "high-density lipoprotein" or "triglycerides" or "total cholesterol". Reference lists of eligible articles were also searched to look for additional studies. The exposure group was defined as COVID-19 patients with dyslipidemia and the control group was defined as COVID-19 patients without dyslipidemia. The outcome of interest was mortality, which was defined as mortality, death, died, nonsurvivor, fatality or deceased. All peer-reviewed articles published in English language reporting the risk factors-adjusted effect estimate on the relationship between dyslipidemia and COVID-19 mortality were eligibly selected. Accordingly, we excluded preprints, case reports, review papers, corrections, comments, animal studies and in vitro studies, studies reporting crude effect estimate, studies without sufficient data and studies reporting clinical outcomes as severe/critical illness, intensive care unit admission, invasive mechanical ventilation/intubation or composite outcomes rather than mortality. Essential information including first author, number of COVID-19 patients, gender distribution, age (mean and standard deviation (SD) or median and interquartile range (IQR)), study design, region/country, clinical outcomes, adjusted effect estimates and adjusted variables was extracted from each included study ( Table 1 ) .

We utilized Stata (version 12.1) for all statistical analyses. The pooled effect estimate and its 95% confidence interval (CI) were computed using a random-effects model. Inter-study heterogeneity was investigated using the cochrane Q test and I 2 statistic, P < 0.1 or I 2 > 50% shows a statistically significant heterogeneity. The statistical stability of the overall effects was assessed using leave-oneout sensitivity analysis. The risk of publication bias was evaluated using Begg's test. Subgroup analyses were carried out by sample size, age, male percentage, study design and effect estimate. Twotailed P -value < 0.05 was considered statistically significant.

Initial search yielded 2608 articles. After screening eligible articles according to inclusion and exclusion criteria, a total of twentyseven studies composing of 146,364 cases were enrolled into this quantitative meta-analysis. Among the included studies, twentyfour studies were retrospective, one was prospective, one was longitudinal cohort study and one was nationwide cohort study. The sample sizes across the eligible studies ranged from 98 to 35,302. There were sixteen odds ratio (OR)-reported studies, nine hazard ratio (HR)-reported studies, one risk ratio (RR)-reported study and one relative hazard (RH)-reported study.

The results of our pooled analysis are presented in Fig. 1 A, which indicates that there was no significant relationship between dyslipidemia and COVID-19 morality (pooled effect size = 1.05, 95% CI [0.99-1.12], P = 0.100; I 2 = 52.6%, random-effects model). Sensitivity analysis by deleting each study one by one demonstrated that our results were stable ( Fig. 1 B) . When we limited dyslipidemia to hyperlipidemia, there was no significant relationship between hyperlipidemia and COVID-19 mortality (pooled effect size = 1.03, 95% CI [0.95-1.12]). We still observed no significant relationship between dyslipidemia and COVID-19 mortality in the subgroup analyses by age (pooled effect size = This meta-analysis has several limitations that need to be mentioned: 1 most of the included studies were from USA, which lim- its its wider applicability of the present findings; 2 the majority of studies were retrospective, thus further well-designed studies with more prospective researches are required to verify our results; 3 although the pooled effect estimate was calculated on the basis of adjusted effects, the adjusted variables are not completely consistent across the included studies; 4 only one included study explicitly states the specific type of dyslipidemia as total cholesterol, additional studies does not explicitly states the specific type of dyslipidemia such as abnormal levels of low-density lipoprotein, high-density lipoprotein, triglycerides and total cholesterol. Further studies should focus on the relationship between specific type of dyslipidemia and COVID-19 mortality when more data are available; 5 the detailed information on medications for patients with pre-existing dyslipidemia is not available presently, thus we could not address the effects of medications on the relationship between dyslipidemia and COVID-19 mortality.

In conclusion, our current study based on adjusted effect sizes demonstrated that dyslipidemia was not significantly associated with COVID-19 mortality. Further well-designed studies with large sample sizes are warranted to confirm our findings.

The authors declare that they have no any potential conflict of interest regarding this submitted manuscript.

We read with interest the study by de Lusignan et al., who found that, among 1970,314 UK primary care patients aged ≥ 45 years, being male, increasing age, chronic disease, Black ethnicity and deprivation were associated with excess mortality during the first wave of the COVID-19 pandemic 1 . These findings highlight the unequal burden of COVID-19 across society and reflect our patient and staff experience at North Middlesex University Hospital (NMUH) 2 . NMUH is located in a socioeconomically and ethnically diverse area of London and, early in the pandemic, was identified as the second most COVID-19 pressured NHS trust in the UK 3 . During the first wave, 24 out of 26 wards were converted to COVID-19 care, intensive care capacity was doubled, and many non-acute medical services were moved offsite. Many healthcare workers (HCW) were redeployed to the frontline where they faced a potent combination of occupational and sociodemographic factors influencing COVID-19 risk.

Between 4th June and 3rd July 2020, voluntary SARS-CoV-2 antibody testing was offered to the NMUH workforce. Staff were invited to complete an online questionnaire detailing comorbidities, occupational and sociodemographic factors. Responses were anonymously linked to antibody results using occupational . Data from Public Health England, https://www.localhealth.org.uk/ . In all figures, wards were divided into even quintiles and then coloured by quintile. The values contained within each quintile are included in the quintile legends. Maps generated using the Greater London Authority mapping template, https://data.london.gov.uk/ dataset/excel-mapping-template-for-london-boroughs-and-wards . health numbers. Multivariable logistic regression was used to identify factors associated with seropositivity. Forward stepwise selection was used to determine which variables to retain in the model and checked against backward elimination. Base demographics of age, sex and ethnicity were always retained in the model. Variables tested for inclusion were underlying risk group, location of residence, Index of Multiple Deprivation (IMD) quintile, job banding, job role, workplace setting, patient interaction, HCW in household, and individual sites of work: emergency department (ED), endoscopy, estates, human resources and finance, intensive care, maternity, medicine, non-clinical, oncology, outpatient department, pathology, paediatrics, pharmacy, radiology, surgery, senior man-agement, theatres, therapies. Statistical analyses were conducted using Stata v.14.2. Total serology tests, positive tests and positivity rates were plotted according to postcode, alongside IMD score and ethnicity using Microsoft Excel. This evaluation was conducted for service improvement and did not require ethical approval according to the NHS Health Research Authority algorithm.

Of 3945 invited staff, 3285 were tested and completed the survey. Overall seropositivity was 35.7% (1173/3285), median age was 41 (IQR 31, 51) years and 72% (2369/3285) were female; White British/Irish HCW represented 23% (764/3285), while 70% (2293/3285) were from ethnic minority backgrounds, most commonly Black African (738/3285, 24%) and the Indian Subcontinent (Indian, Pakistani, Bangladeshi, Sri Lankan; 484/3285, 15%). Overall, 79% (2585/3285) reported no comorbidities, two-thirds lived within the most deprived quintiles (IMD 1, 1021/3285, 31%; IMD 2, 1024/3285, 31%). In total, 31% (1029/3285) were nursing and midwifery staff, followed by administrative (577/3285, 18%) and medical (497/3285, 15%) staff. Most worked in clinical areas (2796/3285, 85%) with patient contact (2562/3285, 78%). A third were in the lowest two NHS job bands (1177/3285, 36%).

Half the staff tested (1692/3285, 52%) lived in Enfield and Haringey Boroughs, adjacent to NMUH. Staff seropositivity rates were highest to the East of Enfield and Haringey, correspond-ing to the most deprived wards with greatest ethnic diversity ( Fig. 1 ) .

In a multivariable logistic regression model for seropositivity, ethnicity and location of residence were the sociodemographic factors reaching significance for inclusion ( Fig. 2 ) . All ethnicities had increased odds of seropositivity compared with White British/Irish staff. Black African staff were at greatest risk (Odds Ratio 2.22, 95% Confidence Interval 1.75-2.82, p < 0.001), followed by mixed ethnicity (OR 1.75, 95% CI 1.2-2.56, p = 0.004), Black Caribbean (OR 1.58, 95% CI 1.13-2.2, p = 0.007), Asian Chinese/Other (OR 1.5, 95% CI 1.1-2.05, p = 0.01). Staff who identified as White Other and from the Indian subcontinent also had increased odds of seropositivity, but this did not reach statistical significance.

In the same model, workplace setting and certain individual sites of work were the only occupational factors reaching significance for inclusion ( Fig. 2 ). All clinical staff had increased odds of seropositivity compared to non-clinical staff. The greatest risk was in COVID-19 wards not performing aerosol generating procedures (AGP) (OR 2.5, 95% CI 1.77-3.55, p < 0.001), mixed-COVID-19 (OR 2.08 95% CI 1.58-2.73, p < 0.001), non-COVID-19 wards (OR 1.81, 95% CI 1.28-2.55, p = 0.001) and ED (OR 1.72, 95% CI 1.23-2.41, p = 0.002). Staff working in AGP areas had increased odds of seropositivity (OR 1.39, 95% CI 0.99-1.94, p = 0.054), but this was of borderline significance, while outpatient areas (OR 1.34, 95% CI 0.96-1.88, p = 0.089) was not statistically significant. Medical department staff had increased odds of seropositivity (OR 1.46, 95% CI 1.17-1.83, p = 0.001) compared to other sites of work.

We found that staff at highest risk worked in non-AGP COVID-19 wards and were from minority ethnic groups, and the highest seropositivity rates mapped to the most deprived local wards. Workforce ethnic diversity and locality was striking; the majority of staff lived in adjacent boroughs and 70% were from minority ethnic backgrounds, compared to 22% in the wider NHS 4 . Similar occupational risk factors have been reported in other HCW seroprevalence surveys, but their influence in combination with sociodemographic factors on COVID-19 risk in HCWs has not been fully described 5 , 6 . One large American HCW seroprevalence study found that community and demographic factors, in particular Black ethnicity and contact with a suspected COVID-19 case, were more predictive of seropositivity than occupational factors. Similarly, we found staff seropositivity geographically mirrored COVID-19 cases among our inpatient population, mapping to local ethnically diverse and deprived areas 2 . The Health Service Journal reported a disproportionate number of NHS staff deaths among ethnic minorities 7 . Concerningly, British Medical Association surveys have found ethnic minority doctors feel less protected from COVID-19 at work than their White colleagues 8 . Furthermore, ethnic minorities are currently under-represented in national HCW surveillance studies 9 . Recent data suggest that there is significantly lower COVID-19 vaccine uptake among HCWs from minority ethnic groups and living in more deprived neighbourhoods, thus exacerbating these disparities 9 , 10 .

Further work is needed to understand the interplay of occupational and sociodemographic risk factors facing HCWs. Inequalities must be urgently addressed in order to better protect NHS staff during the ongoing pandemic.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. demic to identify contagious cases. 1 The winter peak of Covid-19 in England has seen the highest number of Covid-19 cases and hospital admissions to date, with over 30 0 0 admissions daily, and a peak of 34,015 inpatients with Covid-19. 1 Patient triage and cohorting are crucial to reducing nosocomial Covid-19, 2 but paucisymptomatic or pre-symptomatic cases limit clinical case detecting, 3 and screening with molecular diagnostics introduces delay. 4 We piloted the use of point of care antigenic testing for SARS-CoV-2 in patients admitted to hospital for rapid case detection in a period of high disease prevalence. Between December 23, 2020 and January 30 2021, patients admitted to Oxford University Hospitals NHS Foundation Trust for emergency care were tested for SARS-CoV-2 using both lateral flow device (LFD) and real-time reverse transcription Polymerase chain reaction (PCR) testing. Swabs of the nose and throat were collected by health care workers. LFD testing was performed in the admitting department by staff, using the Innova LFD. Swabs for PCR were transferred to the clinical laboratory in viral transport medium (VTM) and tested by multiplex PCR (Thermo-Fisher Taq-Path). 803 patients who had both tests performed with a maximum of 1 day between tests were included for analysis. 732/803 (91%) of patients tested had both tests on the same day. Clinical notes of patients testing positive for SARS-CoV-2 by PCR were reviewed and note made of the reported presence of symptoms of possible Covid-19 (cough, dyspnoea, fever, aguesia or anosmia) as well as admission temperature and oxygen saturation, and previous detection of SARS-CoV-2 by PCR.

Considering PCR results as the reference standard, LFDs showed high specificity ( Table 1 ) . Of 573 PCR-negative patients, 572 had a negative LFD, and 1 an invalid LFD result, i.e., specificity excluding the invalid result was 100% (exact binomial 95%CI 99.4-100%). Similarly the positive predictive value was high, among 133 patients with a positive LFD results, 133/134 (99.2%, 95%CI 95.9-99.8%) were PCR-positive, with one indeterminate PCR result in a patient testing PCR-positive 5 days later; none were PCR-negative. LFDs also had low rates of invalid results, 2/803 (0.2%).

Lateral flow testing showed modest sensitivity, and performed better in those with higher viral loads. Among all 214 SARS-CoV-2 PCR-positive patients, 133 tested positive by LFD, i.e. sensitivity was 62.4% (95% CI 55.6-69.0%), and the negative likelihood ratio was 0.38 (0.32-0.45). 80 patients were LFD-negative, PCR-positive. LFD-negative, PCR-positive individuals had lower viral loads, i.e. higher mean cycle threshold (Ct) values for the detected PCR targets (median 24, IQR 19-27), compared with LFD-positive, PCRpositive patients (median 16, IQR 13-20) ( Fig. 1 , Kruskal-Wallis p < 0.001). Sensitivity was greatest in those patients with a mean Ct < 20 (78.5%, 95% CI 71.9-85.1%) ( Table 1 ) .

On a retrospective review of patient notes, we identified at least 11/133 (8%) of LFD positive patients had no Covid-19 symptoms recorded, presenting without cough, dyspnoea, fever, anosmia, ageusia or hypoxia. Furthermore, among LFD-negative PCRpositive patients, 28/80 (35.0%) had a pre-admission SARS-CoV-2 PCR-positive swab, so 161/214 (75.2%) of patients with SARS-CoV-2 detectable by PCR could be identified by either previous results

Negative Positive or LFD at admission. The absence of either previous PCR positive swab or a positive LFD at admission had a negative likelihood ratio 0.24 (95% CI 0.19-0.31). Case identification is critical in reducing nosocomial transmission of SARS-CoV2. 2 While Ct values are not a direct measure of infectivity, they do correlate with RNA load and culture positivity and infectious dose. 5 , 6 Thus, LFD-positive patients, with higher viral loads, are most likely to represent those patients with the highest infectious risk in the healthcare environment. Additionally, in this cohort, LFDs provided incremental case detection above clinical assessment in asymptomatic adults. LFDs provide a rapid and incremental benefit to clinical triage for case finding.

The excellent specificity seen here corresponds with findings from other evaluation of LFDs, 7 as well local experience in testing asymptomatic healthcare workers, where LFDs showed a false positive rate of 0.03% compared with PCR. 8 This means a positive LFD can safely be used to triage patients to Covid-19 cohort areas for patients with confirmed infection without exposing these patients to risk of nosocomial acquisition. While a negative result cannot be used in isolation to triage a patient to a COVID-19 free area of the hospital, it does allow earlier identification of positive cases, thus relieving pressure on cohort areas for patients with unconfirmed infection status, which are often the most challenging areas in which to prevent nosocomial transmission.

Despite imperfect sensitivity, when a known COVID diagnosis was taken in account, LFDs in this population had a negative likelihood ratio of 0.24. Therefore in patients where there is a low clinical suspicion of COVID-19, a negative LFD does provide further evidence against infectious SARS-CoV-2 infection, which can also play a role in triage decisions. The sensitivity of LFDs in the emergency hospital setting is lower than that reported in previous evaluations, 7 potentially reflecting the challenges of performing LFDs in emergency department settings and the later stage of infection in patients admitted to hospital compared to those attending symptomatic community testing. We did not monitor if all tests were read after the correct time interval, nor were photographs taken of devices at reading to allow for quality assurance. Therefore, reported performance could potentially be improved.

We conclude that LFDs provide a rapid and useful case detection in an acute setting, and are thus a helpful infection control tool.

DWE declares lecture fees from Gilead outside the submitted work. No other authors have a conflict to declare

Patient screening and case reviews were undertaken as part of routine infection control in the hospital.

BCY is an NIHR Clinical Lecturer. DWE is a Robertson Foundation Fellow.

It is with great interest that we read the recent article by De Francesco et al. 1 , who reported on chlamydia pneumoniae and mycoplasma pneumoniae co-infection in patients with COVID-19. Here, we report our experience with candidaemia co-infection with COVID-19. An increased incidence of candidaemia has been noted in patients with COVID-19 and although patient characteristics, investigations and antifungal therapies have been described, 2 to our knowledge, compliance with candidaemia management bundles has not. 3 Here, we present a retrospective review of candidaemias in adult patients ( > 17 years) with PCR proven COVID-19 between 1st March 2020-31st May 2020 across six acute London hospitals. All yeasts isolated from blood cultures were identified by matrix assisted laser desorption/ionisation-time-of-flight (MALDI-TOF) mass spectroscopy (Bruker Daltonik GmbH, Bremen, Germany). Antifungal susceptibility testing was carried out using broth micro-dilution in accordance with EUCAST guidelines. 4 An episode of candidaemia was defined as blood culture growth of any Candida species.

Eleven patients with concurrent candidaemia and PCR-proven COVID-19 were identified during the study period; ten were male (90.9%), mean age 62 (33-77 years). Underlying comorbidities were predominantly cardiovascular (10/11). Two patients were immunosuppressed (see Table 1 ), but neutropenia was not identified.

Ten patients (90.9%) were admitted to an intensive care unit (ICU) prior to their candidaemia diagnosis. Of the ICU patients ( n = 10), all were intubated and ventilated, had an intravascular and urinary catheter and received inotropes. The non-ICU patient also had a urinary catheter. Nine (90%) of the ICU patients received haemofiltration. None of the patients in our cohort received total parenteral nutrition.

All eleven patients received broad-spectrum antibacterials. One patient received prior antifungal treatment in hospital with topical clotrimazole and oral terbinafine for a tinea infection.

The average number of days from PCR-proven COVID-19 to candidaemia was 14.8 days and from ICU admission to candidaemia, 15.5 days (range 6-24 days). Seven out of eleven candidaemias (63.6%) were C.albican s, two (18.2%) C.parapsilosis , one (9.1%) C. glabrata and one (9.1%) C.dubliniensis . All isolates were fluconazole susceptible, except one ( Candida glabrata ), which showed intermediate susceptibility, although the patient was successfully treated with azole therapy through dose-optimisation.

An echinocandin was commenced for ten patients, as per local guidelines, pending susceptibility testing. One patient died prior to blood culture positivity and treatment. Four out of ten (40%) patients were switched to fluconazole to complete treatment. In Table 1 Characteristics of patients with concurrent COVID-19 and candidaemia. Table 1 ). Intravascular catheters were removed for nine out of ten patients (90%), the last patient dying prior to candidaemia notification. Seven out of nine patients had line tips sent for culture; two were positive for yeasts. One line tip confirmed an identical Candida spp., and hence constituted a line infection, but no further identification was available for the second.

Four patients had prior colonization with yeasts; one with the same species as their candidaemia, no further identification was available for the remaining three. Five patients were not colonized and two had an unknown status following transfer from other secondary care providers, developing candidaemia shortly after transfer.

In concordance with Mastrangelo et al., 1 there was a high 30day mortality of 54.4% (6/11) in our patient cohort. The four surviving patients (36.6%) were discharged; average total length of stay 58 days (range 31-78 days). One patient was stepped down after nine weeks in ICU but remained an inpatient until the end of our study period.

Given the high mortality rate, it is important to identify and address modifiable risk factors in an attempt to prevent the occurrence of candidaemia. Firstly, all our patients received broad-spectrum antibacterials, a recognized risk factor for candidaemia. 5 , 6 A recent study from Hughes et al. 7 demonstrated a low frequency (3.2%) of early bacterial co-infection in patients hospitalized with COVID-19, suggesting early broad-spectrum antibacterials may not be warranted. Hence, antimicrobial stewardship initiatives to review unnecessary antibacterial use remain important.

Secondly, intravascular catheters are a well-recognised risk factor for candidaemia 5 and over 90% of our patients had these. The incidence of candidaemia observed warrants further consideration, and whilst not compared to pre-COVID-19 incidence, 2 may potentially reflect pandemic unique challenges. Examples include increased ICU capacity, redeployment of less-experienced staff to ICU, challenges to aseptic technique with personal protective equipment (PPE), and patients requiring re-positioning to improve oxygenation, thus increasing possibility of line displacement/contamination. Improved aseptic intravascular catheter training focusing on PPE may be beneficial.

In addition, although we were unable to identify urinary catheters as a source in our cohort, they are a recognized risk factor for candidaemia 6 and all patients in our cohort had these.

One patient died prior to candidaemia notification. Time to blood culture positivity may be delayed, particularly for nonalbicans candidaemias, 8 and delay in treatment is known to increase mortality, 9 therefore, non-culture-based diagnostics such as galactomannan antigen and BDG should be combine with clinical data to aid diagnosis. 10 54.4% ( n = 6) of the patients were tested for BDG, and of those, 50% ( n = 3) were positive. Although not possible to demonstrate in this patient cohort, an early positive BDG may herald invasive fungal infection, enabling timely initiation of empirical antifungal therapy.

Guidelines for management of candidaemia recommend a care bundle, including repeat blood cultures at 48 h, echocardiogram, and fundoscopy to identify disseminated infection. In our cohort, only 54.5% (6/11) of patients had repeat blood cultures within 48 h, 72.7% (8/11) an echocardiogram and only 9.1% (1/11) fundoscopy. COVID-19 infection control concerns, patient positioning and PPE, with resultant challenges to ophthalmic examination, may account for the poor fundoscopy compliance, adding further weight to the need for COVID-19 specific practical training.

To conclude, during the ongoing COVID-19 pandemic it remains important to consider modifiable risk factors for candidaemia, nonculture based diagnositics to aid early diagnosis, as well as adherence to established treatment bundles.

Ethical approval was not required for this service evaluation and audit of practice.

SD, AR and NM designed the study methodology. SD, TE and XG collated the data. SD drafted the initial manuscript with all authors contributing significantly to revising this for submission. All authors agreed on the final version for submission to the journal.

This research did not receive any grant from funding agencies in the public or commercial sectors.

EC has been paid for consultancy fees by bioMerieux. SH reports personal fees from Pfizer and Shionogi. DAJ holds share options in Pulmocide Ltd and has received research grants from Pulmocide Ltd, Gilead Sciences, Astellas and Pfizer. He has received speaker and consultancy fees from Astra-Zeneca, Pfizer, Gilead, and Astellas. 

We read with interest the recent systematic review by Fricke et al., showing that the number and positivity rate of influenza cases have decreased in result of non-pharmaceutic interventions targeted at the COVID-19 pandemic. 1 Similarly, a report in the United States had shown that the incidences of acute otitis media and streptococcal pharyngitis decreased, while gonorrhea increased during quarantine. 2 These studies show that COVID-19 containment measures and the overall behavioral changes in the communities are likely to have an effect in the transmission and/or reporting of other infections. We here show the results of Norovirus (NoV) surveillance data in Germany, and describe the effect of the containment measures taken in the context of the COVID-19 pandemic on the number and rate of NoV-positive tests in Germany.

NoV is the leading cause of acute gastroenteritis (AGE) globally across all age groups, causing an estimated 18% of all diarrheal disease cases worldwide, 3 and over 20 0,0 0 0 deaths every year. 4 In Germany, NoV is notifiable to the Robert Koch Institute, which has registered nearly 10 0,0 0 0 cases of NoV notified infections every year since 2010 ( https://www.rki.de/DE/ Content/Infekt/EpidBull/Archiv/2020/24/Art _ 01.html ). NoV hospital-izations account for 11-16% of all AGE hospitalizations, and they show a seasonal distribution with a peak from December-March each year. 5 NoV diagnosis is related to reimbursement rates for gastroenteritis hospitalization in Germany, providing a strong incentive to test for NoV in the hospital setting. Taking advantage of the routine testing in one of the largest commercial laboratories, a surveillance study was designed to provide up-to-date evidence on the occurrence of NoV across all ages, circulating genotypes and co-infections in Germany.

The 2020 COVID-19 pandemic has triggered the implementation of different containment measures across the globe. Germany reported the first cases in late January 6 and responded by implementing community mitigation and mobility restriction measures since the first COVID-related deaths were reported in March. 7 From mid-March until early May, schools and bars were closed, borders with neighboring countries were controlled, travel was restricted, and general social distancing measures were adopted. With the mandatory use of masks in place since the end of April, reopening was gradual, but several limitations were still in place by the date this article was submitted. 7 This study is based on a larger prospective, laboratory-based surveillance study on NoV infection. The surveillance relies upon results from all clinical specimens tested for NoV and other enteropathogens submitted to the Limbach Laboratory (MVZ Dr. Limbach & Kollegen GbR, Heidelberg MVZ), from patients of all ages and all genders. The Limbach Laboratory tests samples referred by hospitals across all of Germany. These samples can originate from all age groups, all types of inpatients across various departments as well as outpatients whose specimens are sent to the laboratory for testing. Here we report data from the start of the surveillance (February 2018) to December 2020. Data collected includes aggregated totals of samples tested for NoV and PCR testing results. Data is provided per setting (inpatients or outpatients) and per age group, although we only report totals for all ages. To analyze trends in NoV gastroenteritis, we summarized monthly positivity rates of NoV per setting. The proportions of positives were estimated with exact 95% confidence intervals using R v.4.0.2. 8 From February 2018 to December 2020, 31,765 specimens were tested for NoV and other enteropathogens. Most specimens (27,795, 87.5%) were collected from hospitalized patients (reasons for hospitalization are unknown). Of all specimens tested, 3970 (12.5%) yielded positive PCR results for NoV. The overall percentage of NoV-positive specimens was similar in the hospital (12.4%) and outpatient (12.8%) settings. Fig. 1 shows the monthly distribution of stool samples tested for NoV, and the NoV-positive proportion among hospitalized patients. The distribution among outpatients was similar, but with larger confidence intervals due to the smaller sample size (data not shown). As expected, a strong seasonal effect is visible in the proportion of NoV-positive specimens in the years 2018 and 2019, with the proportion of NoV positives increasing from November/December until March. In these two years, the lowest proportion of NoV-positive tests were observed in July (4.3%) and August (4.8%) 2019 ( Fig. 1 ) . The overall number of tests performed in the years 2018 and 2019 shows a weaker seasonal pattern, with the number of tests never below 600 per month.

In 2020, the percentage of NoV-positive specimens decreased sharply after January, reaching near 0% as of May and continuing around 0% thereafter. The total number of samples tested for NoV also decreased from February to May, but never went below 400 per month.

The surveillance data suggest a significant impact of the COVID-19 control measures on the NoV positivity rates among stool samples from patients hospitalized in Germany after January 2020. As previously reported for Germany, 5 the number of NoV hospitalizations typically starts to increase in November-December, until the peak is reached in January-March. In the 2019-2020 season, the peak in January was followed by a steep decrease in the number of tests and proportion of NoV-positives. The months of February to May 2020 have registered a significantly lower proportion of NoV-positive specimens than in previous years, until they almost disappeared from May onwards.

Starting in March 2020, Germany has adopted several measures to contain the COVID pandemic, including closure of schools, bars and large events. At the same time, the population adopted preventive behaviors such as social distancing measures and the use of hand sanitizer. It is safe to assume that these measures could result in a decrease of other infections. NoV is highly contagious via the fecal-oral route, through contaminated hands or by consumption of contaminated food and water, giving rise to frequent outbreaks in institutions or restaurants. 9 The sharp decrease in the proportion of NoV-positive cases observed in 2020 is likely related to the closure of schools, restaurants and other institutions, as well as of other containment measures. Behavioral changes preceding the containment measures, which only became effective in March, may have caused the decrease in the NoV positivity rates to start already in February, triggered by reports from Italy or Spain. In fact, the number of cases exploded in Italy from February 22nd. On March 10th, when Germany reported its first two COVID-related deaths, Italy already counted 464 deaths and over 90 0 0 cases, and Spain had over 20 0 0 reported cases. 6 Our study has some limitations. We do not possess clinical information on the patients whose samples were tested. Though this is unlikely, we cannot exclude that patients infected with norovirus presented less frequently to medical care than other AGE patients in 2020. Our analyses are ecological in nature and we can also not exclude that there is a natural decline of NoV circulation, independent of the COVID-19 control measures. However, since NoV is known to be transmitted primarily via person-to-person contacts, it is fair to assume the control measures have played a major role in this decline. Finally, it is unknown whether NoV incidence will return to pre-COVID-19 values once behavioral restrictions are relaxed.

This study shows that NoV infections have become less frequent in Germany since the beginning of the COVID-19 epidemic in Europe. Whereas this is a positive observation from a public health perspective, it also has a significant impact on NoV vaccine development programs. Several NoV vaccines are under development, with one vaccine entering Phase II trials. 10 The COVID-19 pandemic hinders the conduct of these trials by the logistical challenges in enrolling and following up subjects, and Phase III trials are unlikely to show efficiency due to the current low NoV positivity rates. It is unknown how NoV infections will evolve once containment measures are loosened.

Ulrich Eigner and Thomas Verstraeten report consulting grants from Takeda Pharmaceuticals AG to Labor Limbach and P95. John Weil is an employee of Takeda Pharmaceuticals International AG. The COVID-19 pandemic, an unprecedented event for current generations of physicians, has stricken hard on society. 1 There is a significant lack of effective drugs for stopping viral replication. Lopinavir/ritonavir (LPV/r) is a well-known combination used in patients with HIV which was included in the arsenal against SARS-CoV-2 early in the pandemic. 2 Its use in COVID-19 was based on inconsistent results from experimental and clinical research that was mostly done while investigating other β-coronaviruses (SARS and MERS).

A number of randomized clinical trials have observed no benefit of LPV/r beyond the standard of care. [3] [4] [5] However, voices have been raised against interpreting these results as grounds from definitively ruling out LPV/r since some of these studies lacked statistical power, reported encouraging outcomes in secondary endpoints, and included patients with a prolonged period of symptoms before initiation of treatment. 6 , 7 Indeed, there may be a subpopulation of COVID-19 patients -notably those early in the course of the infection -for whom LPV/r may improve their prognosis. In a recent report, Klement-Frutos et al. describe a favorable outcome of patient with COVID-19 after beginning of LPV/r on day 9 of symptoms. 8 Therefore, we aimed to assess the efficacy of LPV/r in a large, multicenter cohort of patients, with special interest in those who received treatment soon after the onset of symptoms.

This work belongs to the SEMI-COVID-19 Registry, which is an ongoing, nationwide, retrospective, anonymized cohort of consecutive adult patients hospitalized in Spain for microbiologically confirmed COVID-19. 9 The Registry was approved by the Ethics Committees of the participating centers, and included data on over 300 variables. The primary endpoint was raw-in hospital mortality at 30 days from admission. Patients were considered to have been ✩ A complete list of the SEMI-COVID-19 Network members is provided in the Appendix. treated with LPV/r if they had received at least one dose of the drug. Common dosage of LPV/r was 40 0/10 0 mg bid.

In order to mitigate the effects of possible confounding variables in a non-randomized assessment of treatment with LPV/r, propensity score (PS) was performed. The propensity of receiving LPV/r was estimated using a logistic regression model that included confounding variables which could have affected treatment choice or outcomes as independent variables. The nearest neighbor method with a caliper of 0.1 as used in PS matching and standardized mean differences (SMD) were calculated to evaluate adequacy of propensity matching. Both conditional logit and mixed effects logistic regressions were performed. Furthermore, univariate and multivariate logistic regression models were fitted in order to estimate the treatment effect using all data, as a sensitivity analysis. Multiple imputation was used to handle missing data and model estimates and standard errors were calculated using Rubin's rules. 10 Statistical analyses were performed using R software (v.3.6.2).

As of June 1, 2020, the Registry included 9,594 cases, of which 8,553 met the inclusion criteria (Suppl. Fig. 1 ). Fifty-seven percent were men, median age was 69 years (IQR 56-79), and half of subjects (50.2%) had high blood pressure. Patients were admitted after a median time since symptoms onset of 7 days (IQR 4-9), with median SaO 2 /FiO 2 ratio of 376 (IQR 300-452), C-reactive protein of 58 mg/L (IQR 19-123), and lymphocytes of 940 cells/μL. LPV/r was administered to 5,396 patients (63%) after a median of 0 days since admission (IQR 0-1). Table 1 shows that LPV/r was more likely to be prescribed to patients who presented with more severe clinical condition, including presence of fever, cough, radiological infiltrates (91.7% vs 80.4% p < 0.001) and a lower SaO 2 /FiO 2 ratio. On the other hand, LPV/r was less frequent among at-risk subjects in whom toxicity may be more likely: elderly patients presenting with altered mental status, dementia, or other debilitating baseline conditions, as well as patients on immunosuppressive drugs and pregnant women.

Overall, 1,509 patients died (17.6%). The univariate parameters predicting mortality is shown in Table 2 . A PS allowed for comparing two cohorts with similar values on the parameters associated with the prescription of LPV/r ( Table 1 ). Most parameters were adequately matched according to SMD, although some variables had SMD values > 0.02 (Suppl. Table 1 ): In this matched cohort, the adjusted odds ratio (aOR) for mortality for the use of LPV/r was 0.932 (95CI 0.799-1.087; p > 0.05) according to both conditional and mixed effects logistic models.

Of the 6,099 patients who were admitted to hospital within 8 days since onset of symptoms (median time to admission since onset of symptoms 5 days [IQR 3-7]), LPV/r was prescribed to 3,377 (55%). Variables associated with the use of LPV/r were similar to those observed in the cohort as a whole (Suppl. Table  2 ). In a propensity score matching carried out on this subset of patients, early use of LPV/r was not associated with a lower mortality (conditional logistic regression: aOR 1.110 (95CI 0.944-1.300; p = 0.245); mixed effects logistic regression: aOR 1.105 (95CI 0.944-1.300; p = 0.272)).

Consistent with previous studies, our analysis found no overall benefit to the use of LPV/r. [3] [4] [5] We have focused on patients who received the antiviral at an earlier stage in the hope of finding greater activity. Indeed, in other viral diseases, the administration of antiviral drugs must be done as soon as possible in order to have a clinically significant activity. 11 Of note, patients included in Rao's clinical trial had a median duration of symptoms of 13 days (IQR 11-16) 4 , and those recruited in the RECOVERY trial presented after 8 days of disease (IQR 4-12). 3 In our sub-analysis, the median duration was 5 days (IQR 3-7), thus allowing us to perform a evaluation on patients who were indeed at a very early stage of disease. However, results were again disappointing, and add another nail in the coffin of LPV/r when considering its use for COVID-19. Our study has some limitations. First, it has the biases inherent to retrospective observational studies. Also, despite the fact that the number of patients included allowed us to perform PS matching, which may have reasonably controlled for many of these biases, the balance of some parameters was not perfect according to SMD values. Still, as a multicenter study involving a large number of hospitals, it has the strength of being rooted in real-life practice, far from strict inclusion and exclusion criteria of clinical tri-als. Second, we have used mortality as a primary endpoint, as others have done, but we cannot rule out any benefits of LPV/r that would have emerged had we analyzed softer outcomes, such as time to improvement or disease duration, as suggested by the report of Klement-Frutos et al. 8 Finally, our analysis has used data from COVID-19 first wave in Spain, when efficacy of corticosteroids or other drugs was not yet proved. Thus, our analysis is not adjusted for these treatments. However, these therapies, at least during the first wave of the pandemic, have usually been reserved for severe patients, and thus may be surrogate predictors of unfavorable progress.

In conclusion, we have analyzed a large, multicenter cohort of patients with COVID-19 and have not found any benefits to administering LPV/r, even when it was administered within the first 8 days of symptoms. Our results discourage its use in SARS-CoV-2 infection.

This work was supported by the Spanish Society of Internal Medicine ( SEMI).

This study is supported by the Spanish Society of Internal Medicine ( SEMI). We gratefully acknowledge all the investigators who participate in the SEMI-COVID-19 Registry. We also thank the SEMI-COVID-19 Registry Coordinating Center, S&H Medical Science Service, for their quality control data, logistic and administrative support. We are also indebted to Claire Conrad for reviewing the English manuscript, and to Ipek Guler for her assistance with the statistical analysis.

Supplementary material associated with this article can be found, in the online version, at doi: 10 1 In a previous observational cohort study, we established that an early 4-day treatment combining corticosteroid (prednisolone dose equivalent, 1.25 mg/kg/24 h) and furosemide (80 mg/day) was effective in reducing the need for mechanical ventilation and overall mortality (OR, 0.35 [0.11-1.01]; P = 0.04) in non-critically ill COVID-19 patients. 2 The GRECCO-19 randomized trial suggested a benefit of colchicine in preventing clinical deterioration in hospitalized noncritically ill COVID-19 patients. 3 Similarly, an observational cohort study reported that salicylate treatment was associated with reduction in intensive care unit (ICU) and mechanical ventilation requirements in hospitalized COVID-19 patients, although inhospital death was not significantly modified. 4 Moreover, prophylactic or intermediate-dose anticoagulation was highly recommended in hospitalized COVID-19 patients who are at high-risk of venous thromboembolic events (VTE). 5 Specifically, direct oral anticoagulant use was shown to be associated with improved outcome. 6 Based on the data discussed above and the pathophysiology of COVID-19 and its complications, i.e. thrombosis, inflammation and congestion, we hypothesized that a five-drug regimen consisting in a 5-day course of 1 mg/kg/day prednisone, 80 mg/day furosemide, 75 mg/day salicylate, colchicine (1 mg loading dose followed by 0.5 mg one hour later then 0.5 mg every 8 h as recommended to treat acute gout) 7 and direct anti-Xa inhibitor such as rivaroxaban or apixaban would optimally mitigate COVID-19-attributed mortality. To address the effectiveness of this five-drug regimen, we designed an observational cohort study (COrtiCoid-Aspirin-Anticoagulant-Colchicine-LAsix R , the COCAA-COLA study) including all successive non-critically ill COVID-19 patients requiring > 1 L/min-oxygen and admitted to our ward between 2020/01/09 and 2020/11/30 (during the second wave in France). Patients who did not receive this regimen were treated with dexamethasone (6 mg once daily for up to 10 days) 8 and low-molecular weight heparin (control group). All patients received standard of care, i.e. oxygen with flow adapted to oximetry, proton pump inhibitor, antibiotics, insulin, potassium supplementation and loperamide if needed. No antiviral or additional immunomodulatory therapy was used due to the absence of clearly demonstrated benefit. Systematic chest computed tomography angiography was performed on admission if not contra-indicated. Anticoagulants (direct anti-Xa inhibitor in the five-drug regimen-treated patients or lowmolecular weight heparin in the others) were administered at pro-phylactic dose with the exception of patients exhibiting VTE or plasma d -dimer ≥50 0 0 ng/mL (a threshold predicting increased VTE risk in COVID-19 patients) 9 who were administered anticoagulants at therapeutic dose. Usual monitoring including pulse oximetry, electrocardiogram, finger blood sugar and daily routine chemical tests was provided.

The primary composite endpoint was requirement of high-flow oxygen therapy, non-invasive or invasive mechanical ventilation (corresponding to care escalation from ward to ICU) or 28-day mortality. The 4C Mortality Score, a risk stratification score for hospitalized COVID-19 patients, was used to predict in-hospital mortality. 10 Data were expressed as median [25th-75th percentiles] or percentages. Univariate comparisons were performed using Mann-Whitney or Fisher exact tests, as appropriate. A multivariate logistic regression model was tested with the five-drug regimen as explanatory variable and adjustment for independent covariates (gender, age, body-mass index and comorbidities) to explain the outcome. Odds ratios (OR) and their 95%CI were determined. Stratified categorical data were compared using Cochran-Mantel-Haenszel tests. P -values ≤0.05 were considered significant. Analyses were preformed using the R4.0 environment.

We included sixty-eight patients (age, 66years [54-75]; male/female sex-ratio, 3.5; body-mass index, 27 kg/m ² [24-30]; hypertension, 46%; diabetes mellitus, 44%; cardiovascular disease, 29%; chronic lung disease, 3%). Twenty-eight patients (41%) re- Fig. 1 . Impact of the prednisone/furosemide/colchicine/salicylate/direct anti-Xa inhibitor regimen in the different patient subgroups defined according to age (using the median value, 66.5 years, as threshold), gender, presence of diabetes mellitus, serum brain natriuretic peptide (BNP; threshold at 100 ng/mL) and troponin levels (threshold at 16 ng/mL). Odds ratio (OR) and their 95%-confidence intervals were determined . (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) ceived the five drug-therapy regimen whereas forty (59%) were included in the control group. Based on the 4C Mortality Score (10 [8] [9] [10] [11] [12] ), predicted mortality on admission was ∼30%. No significant differences were observed between the groups regarding the clinical and biological characteristics and the predicted mortality ( Table 1 ) . Noteworthy, 4/40 control patients (10%) at risk of cardiogenic pulmonary edema (serum brain natriuretic peptide (BNP) ≥100 ng/mL) received furosemide.

Among patients receiving the five-drug regimen, the incidence of primary composite endpoint was lower than in the control group (OR = 0.097 [0.001-0.48], P = 0.0009). Multivariate analysis confirmed the significant effect of the five-drug regimen on outcome after adjustment for independent covariates, including age, body-mass index, 4C Mortality Score, high serum BNP level and high white blood cell count (OR = 0.043 [0.0053-0.21], P = 0.0005). The model was significant compared to a model without the fivedrug regimen ( P < 0.0 0 0 01). Additionally, patient subgroups were analyzed following stratification by age (using the median value as threshold), gender and risk factors including diabetes, elevated BNP (threshold, 100 ng/ml) and troponin levels (threshold, 16 ng/mL; Fig. 1 ). Remarkably, the five-drug regimen was associated with a significant reduction in primary composite endpoint in males only. Additionally, there was a stronger and more significant protective effect of our regimen in patients with elevated-BNP (OR = 0.0 [0.0-0.47], P = 0.007) than in low-BNP patients (OR = 0.17 [0.02-0.94], P = 0.03). Thus, the primary composite endpoint was improved in elevated-versus low-BNP patients ( P = 0.0 0 03). We observed no remarkable adverse effects attributed to the five-drug regimen except mild colchicine-related diarrhea (21%) resolved with loperamide.

The GRECCO-19 trial showed improved time to clinical deterioration in hospitalized COVID-19 patients receiving colchicine; however, the benefit relied on a narrow margin of clinical significance. 3 By adding colchicine to the recommended corticosteroid and anticoagulant, together with aspirin and furosemide, we succeeded in improving the outcome. The five drugs included in our regimen were given orally for a short course, paving the way for an outpatient treatment. Interestingly, the recent COLCORONA trial conducted in non-hospitalized COVID-19 patients supported colchicine-related benefit in reducing hospitalizations, need for mechanical ventilation and mortality. 11 Colchicine dose regimen differed between the three studies with higher cumulative colchicine doses in the GRECCO-19 (22 mg) and COLCORONA trials (16.5 mg) compared to ours (8 mg). Using the same primary composite endpoint, our five-drug regimen significantly improved prognosis in comparison to the corticosteroid/furosemide combination of our previous study 2 ( P = 0.0 0 01).

In conclusion, our data highlight the benefit and safety of an early short-course oral regimen combining prednisone/colchicine/salicylate/direct anti-Xa inhibitor/furosemide to reduce the risk of high flow oxygen need, mechanical ventilation requirement or 28-day mortality in hospitalized non-critically ill COVID-19 patients. Our preliminary observational findings should be confirmed in larger cohorts.

This study was part of the French COVID-19 cohort registry conducted by the REACTing consortium (REsearch and ACTion targeting emerging infectious diseases) and directed by INSERM (Institut national de la santé et de la recherche médicale) and ISARIC (International Severe Acute Respiratory and Emerging Infection Consortium). Our institutional ethics committee approved the study (N °, IDRCB, 2020-A00256-33; CPP, 11-20 20.02.04.68737).

J.-P.K. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

All the authors agree to publish.

The authors declare that they have no competing interests. 

Dear Editor, A significant number of patients suffering from COVID-19 infection refers illness-related symptoms several weeks or months after the acute episode. The so-called post-COVID syndrome or long COVID syndrome includes persistent symptoms that could be the result of residual inflammation, organ damage, non-specific effects from the hospitalization or prolonged ventilation, social isolation or impact on pre-existing health conditions. 1-3 Very recently, Moreno-Perez et al 2 published in Journal of Infection a large prospective series of 277 patients (66% of them hospitalized) in a mediterranean population the post-COVID syndrome at about 3 months after diagnosis was detected in 141 individuals (51%), 59% of hospitalized patients and 37% of outpatients. 2 Several series with different follow-up periods have reported the incidence of persistent symptoms ranging from about one third in outpatients and up to 90% in hospitalized patients. [1] [2] [3] [4] [5] [6] However, our clinical experience led us to suspect that patients with asthma had postcovid syndrome less frequently. In fact, in patients with preexisting asthma, a lower COVID-19 susceptibility has been reported. 7 , 8 It is possible that this lower susceptibility also affects the long-term persistence of disease symptoms. For this reason, we evaluated the persistence of symptoms attributed to COVID-19 three months after infection in a series of patients with asthma under follow-up in our department.

In the period between March 3 and December 11, 2020, a total of 2995 patients over 14 years of age tested positive for SARS-CoV-2 (determined by RT-PCR technique) in our health department, 77 of them with asthma (2.6%). Three of these 77 patients were excluded for this study (two of them due to death and one due to lack of follow-up). A total of 74 patients with asthma were periodically surveyed by telephone about their clinical evolution. Symptoms referred at three months were recorded for evaluation. The study was approved by the local institutional ethics board, that authorized the study without written individual patient consent statement due to the characteristics of the disease. Severity of asthma was established according to the prescribed therapy following international GINA recommendations (https://ginasthma.org/): mild (step 1 and 2 of GINA), moderate (step 3 and 4) or severe (step 5). Asthma was classified as allergic, eosinophilic and non-T2. The allergic group included patients with elevated IgE, a positive prick test to pneumo-allergens or seasonal asthma associated with rhinitis. Eosinophilic patients included those who were non-allergic with a blood eosinophil count of more than 300 per millilitre. Patients who did not meet these criteria were classified as non-T2 phenotype.

Of the 74 patients with asthma, 42 were females (57%). The median age was 49 years (interquartile range 34-60). Asthma was mild in 17 (23%) patients, moderate in 52 (70%) and severe in five (7%) (four receiving omalizumab and one benralizumab). Twentyfive patients with asthma (34%) were asymptomatic during SARS-CoV-2 infection, 34 (46%) developed symptoms but did not require hospital admission, and 15 (20%) were hospitalised. Fortysix patients were classified as having allergic asthma, seven as eosinophilic asthma, and 21 as non-T2 asthma. Admitted patients were five allergic, three eosinophil and seven non-T2 phenotypes.

Seven of the 74 (9.5%) patients with asthma infected by SARS-CoV-2 that were followed-up reported post-COVID syndrome at 3 months; none of the 25 asymptomatic patients, 3 of the 34 patients (8.8%) with COVID-19 that not required hospitalization (2 fatigue and one hyposmia), and 4 (27%) of hospitalized patients (cough 2 of them, and dyspnoea and fatigue one patient each symptom). If we only consider symptomatic patients at diagnosis, the prevalence of post-COVID syndrome was 14% (7 of 49). All but one of the 7 patients with post-COVID syndrome were receiving inhaled corticosteroids. Only one patient with post-COVID syndrome was classified as non-T phenotype asthma. Two patients with post-COVID syndrome had mild asthma, 4 moderate and 1 severe.

In our experience, patients with preexisting asthma have a lower prevalence of post-COVID syndrome than that reported among the totality of COVID-19 patients. In addition to the mentioned series by Moreno-Perez et al 2 other experiences with general COVID-19 population reported consistent results. In a series of 110 patients hospitalised with COVID-19, at 8-12 weeks postadmission most (74%) had persistent symptoms (notably breathlessness and excessive fatigue), 4 almost three times as many of our hospitalized asthmatic patients. (27%). In another series with 177 patients with less severe disease (6% asymptomatic, 85% with mild disease and 9% requiring hospitalization) with a median followup of 5.6 months after illness onset, one third reported at least one symptom (8.8% in our moderate cases), the most common, fatigue. 6 The reason for this lower prevalence of post-COVID syndrome in asthma could theoretically be related to immune characteristics of the patients or to treatment. In fact, in-vitro studies have shown that inhaled glucocorticoids reduce the replication of SARS-CoV-2 in airway epithelial. 9 However, almost all our symptomatic patients were receiving this drug.

Our study has several limitations. Patients, of one single centre, were followed up by telephone and no face-to-face interview was conducted. In addition, it is possible that the patients did not report mild symptoms or psychological disturbances. This implies that it is an exploratory study that needs to be confirmed.

In conclusion, we have found that our patients with asthma have a low prevalence of persistent symptoms at three months of onset of COVID-19. It seems of interest to confirm this finding in other centres and with larger samples and to analyse its possible causes.

It has recently been suggested that prior SARS-CoV-2 infection is associated with protection against symptomatic reinfection [ 1 , 2 ] . The role of protective immunity after COVID-19 has been assessed in population-based and cohort studies, where symptomatic recurrences with positive SARS-CoV-2 RT-PCR results were investigated [ 2 , 3 ] , usually lacking genomic sequencing to confirm reinfection. However, limited data are available to date about the frequency of long-term asymptomatic reinfections and/or recurrences. Because of their confirmed transmission risk [4] , asymptomatic infections also have significant epidemiologic implications in terms of public health control. To answer this question, longitudinal studies with sequential sampling following SARS-CoV-2 infection would be required, ideally including sequencing of viral genomes to discern between reinfection and disease recurrence. Recently, the Centers for Disease Control (CDC) have proposed an investigation protocol for identifying cases with a high index of suspicion for reinfection [5] , that prioritizes new detection of SARS-CoV-2 RNA ≥90 days since first infection, whether or not symptoms are present, availability of paired respiratory specimens with a RT-PCR cycle threshold (Ct) value < 33, and genomic sequencing to confirm reinfection. An acknowledged limitation of the protocol consists in the exclusion of asymptomatic or mildly symptomatic individuals who never seek testing for SARS-CoV-2. We conducted a prospective study in a cohort of patients hospitalized for microbiologicallyconfirmed COVID-19 in the first wave, who were longitudinally followed-up during a 6-month period with sequential nasopharyngeal and blood sampling. We evaluated the incidence of late reinfections and recurrences, both symptomatic and asymptomatic, and validated the CDC predictive criteria to identify late reinfections occurring in our cohort by genomic sequencing of the suspected cases. Blood and nasopharyngeal samples were obtained during hospital stay and at 1, 2 and 6 months after patients' discharge for measuring antibody levels and SARS-CoV-2 RNA. IgG antibody plasma levels against the SARS-CoV-2 internal nucleo-capsid protein (N-IgG) and the spike protein (S-IgG) (Anti-SARS-CoV-2 IgG ELISA, Euroimmun, Lubeck, Germany) were tested, and SARS-CoV-2 RNA was detected by RT-PCR (AllplexTM 2019-nCoV Assay, Seegene, Seoul, Korea) which targeted the E, RdRP, and N genes. Genome sequencing of SARS-CoV-2 was performed on nasopharyngeal samples following ARTIC amplicon sequencing protocol for MinIon version V3-Phylogenetic analysis was done using webserver Nextstrain ( https://nextstrain.org/ ), with the SARS-CoV-2 database Nextclade ( https://clades.nextstrain.org/ ). 146 patients admitted for COVID-19 were followed-up. Median age was 64 years, 88 (60.3%) were male, and 72.6% had coexisting comorbid diseases. SARS-CoV-2 shedding lasted a median (Q1-Q3) of 13 (2.2-33.8) days, median (Q1-Q3) time from illness onset to seropositivity was 12 (8-15) days, and peak S-IgG was 5.9 (0.3-7.1) absorbance/cut-off (S/CO) and peak N-IgG 4.1 (0.3-4.9) S/CO. At 1 month after discharge, 40/146 (27%) subjects tested positive for SARS-CoV-2 RNA; 15/127 (11.8%) at 2 months, and 5/134 (3.7%) at 6 months. We analyzed the 5 patients with positive RT-PCR occurring more than 90 days since first COVID-19 diagnosis ( Table 1 ) . Median (range) time from diagnosis to new detection of SARS-CoV-2 RNA was 183 (167-204) days. Cases included 3 men, with ages ranging from 44 to 73 years, and 3 of them had subjacent comorbidity. Two patients were readmitted to hospital at re-positivity, and 3 patients remained asymptomatic. Only one patient had a Ct < 33, and in the other four patients the Cts ranged from 33 to 38. Genomic sequencing was performed in 4 individuals with available paired samples. In the three patients with Ct ≥33, all of them asymptomatic, the same clade 20B was detected. In two of them, the clade showed the same hallmark single nucleotide variants. In the third patient, the follow-up sample showed two new mutations, a K374R substitution in the N gene and an A222V substitution in the S gene, probably reflecting adaptive viral changes associated to persistent infection. Genomic sequencing of the symptomatic patient with a Ct of 18 showed phylogenetically distinct genomic sequences; the first sample was member of the clade 20A, and the most recent sample was member of the clade 20B. The 3 patients with asymptomatic recurrence and the symptomatic patient with no sequencing data showed detectable antibody levels at the time of SARS-CoV-2 RNA re-positivity, ranging from 3.01 to 6.01 S/CO for S-IgG and 2.6 to 2.46 S/CO for N-IgG. The patient with symptomatic reinfection had no detectable antibody levels at the time of re-positivity.

Our results show that late asymptomatic RT-PCR re-positivity does occur after COVID-19, even 6 months later, and does not necessarily represent new infection, despite the prolonged time interval elapsed and the negativity of subsequent RT-PCR tests since the first diagnosis. Although asymptomatic and symptomatic SARS-CoV-2 re-positivity had been reported, median time to recurrence was usually lower, around 1-2 months [ 6 , 7 ] . We found that the CDC criteria showed to satisfactorily predict reinfection, since none of patients not meeting the proposed criteria showed to be reinfected after genomic sequencing testing, while reinfection was actually confirmed in the suspected case according to criteria. Unfortunately, paired samples were not available for sequencing the viral genomes of one of the patients, who had a symptomatic repositivity with a Ct value of 36. This patient would not have been classified as a case of suspected reinfection by CDC criteria. Interestingly, confirmed recurrences were accompanied by coexisting detectable antibody levels, as it also occurred with the symptomatic suspected recurrence, while antibodies were not present in the patient with reinfection. These findings reinforce the protective role of antibodies against reinfection. Peak antibody levels after the first SARS-CoV-2 infection in patients with recurrence did not differ from average values observed in the cohort, and S-IgG levels at the time of recurrence were within the range of peak levels. Whether this could have contributed to the absence of symptoms in 3 of the 4 patients is unknown. Immune dysfunction has been implicated among the factors potentially contributing to reactivation of latent persistent virus after COVID-19 [2] . Despite the adequate antibody levels, additional immune deficits, such an insufficient cellular immune response to SARS-CoV-2, might have had a role in the delayed RT-PCR re-positivity.

Our study provides long-term data about the natural history of COVID-19. Asymptomatic recurrences are detected up to six months after COVID-19. The CDC criteria are helpful to distinguish between disease recurrence and reinfection.

This work was supported by the RD16/0 025/0 038 project as a part of the Plan Nacional Research + Development + Innovation ( R + D + I ) and cofinanced by Instituto de Salud Carlos III -Subdirección General de Evaluación y Fondo Europeo de Desarrollo Regional; Instituto de Salud Carlos III (Fondo de Investigaciones Sanitarias [grant number PI16/01,740; PI18/01,861; CM 19/00,160; CM20/00,0 6 6; COV20-00,005]). The funding agencies had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

All authors declare no conflict of interest.

Dear Editor,

The emergence of new SARS-CoV-2 variants, especially those of concerns, and their rapid dispersal emphasize the importance of active surveillance for SARS-CoV-2 variants worldwide. 1 , 2 In the morning of 28th January 2021, after 55 days without SARS-CoV-2 community transmission in Vietnam, two PCRconfirmed cases of SARS-CoV-2 infection were reported. They came from two neighboring provinces, Hai Duong (HD) and Quang Ninh (QN), in the north of Vietnam. 3 By the end of the day, 88 cases had been confirmed in these two provinces.

On the 28th of January 2021, a 28-year old man (patient 1) presented to a local district hospital in Ho Chi Minh city (HCMC) in southern Vietnam. He had just flown back from HD, where he had attended a relative's wedding party on 18th January. One of his relatives in HD tested positive for SARS-CoV-2 on the 28th January. As per the control measures in Vietnam, 4 a nasopharyngeal throat swab (NTS) was obtained from patient 1 and tested positive for SARS-CoV-2 by RT-PCR 5 on 29th January. Because the strain of the virus responsible for the outbreak in the north was unknown, we whole genome sequenced SARS-CoV-2 directly from the NTS of patient 1 using the ARTIC protocol, 6 and obtained a complete genome on 31st January. Lineage analysis using Pangolin 7 returned B.1.1.7, representing the first report of B.1.1.7 from a case of locallyacquired infection in Vietnam. 3 Contact tracing identified a total of 162 close contacts of patient 1, but none were positive for SARS-CoV-2 by RT-PCR on 4th February.

The rapid expansion of the outbreak in the north, possibly caused by variant B.1.1.7, raised concerns about a nationwide outbreak (Supplementary Figure 1A) . This prompted HCMC to conduct enhanced surveillance for SARS-CoV-2, primarily focusing on highrisk groups, including those working at Tan Son Nhat (TSN) international and domestic airport in HCMC. Subsequently, a baggage handler (patient 2) working at the airport and his brother (not working at the airport) were found positive for SARS-CoV-2 on 6th February. The next day, four co-workers (patients 3-6) of patient 2 also tested positive for SARS-CoV-2, while all contacts of patient 2 s brother were negative. At this time, the outbreak in the north had expanded to another 10 provinces/cities ( Supplementary Figure 1B) .

To dissect the epidemiological picture of the ongoing outbreak, we whole genome sequenced SARS-CoV-2 from the NTS of patients 2-6 using the ARTIC protocol. We obtained 3 complete genomes (1 from patient 2 on 8th February, and 2 from patients 3 and 4 on 10th February). All were assigned to sub-lineage A.23.1 (Pangolin).

After the detection of these six confirmed cases, contact tracing and testing detected 30 additional PCR confirmed cases, totaling 36 infected cases, including 9 members of TSN airport staff in total. 3 The remaining cases were contacts of these 9 individuals (data not shown). Two additional SARS-CoV-2 whole genomes were successfully obtained from a brother of patient 6 and one of the airport staff; all belonged to A.23.1.

The 5 obtained whole-genome sequences of sub-lineage A.23.1 clustered tightly on a phylogenetic tree, and were closely related to A.23.1 strains collected from other countries ( Fig. 1 ). Our findings suggest the TSN airport-associated cluster was caused by a single 

Demographics of the study participants and contact details between RT-PCRconfirmed cases of the TSN airport associated cluster. infected with this variant, 82 (68.9%) were identified in Brazil, five (4.2%) in Japan, 20 (16.8%) in Europe, and three (2.5%) in the USA (Supplemental Table 1 ), suggesting global prevalence.

In summary, we have confirmed the emergence of five clades over time. Consecutive analysis identified SARS-CoV-2 variant 20 J/501Y.V3 (P.1 lineage) in a patient and detected mutations that are identical to those of the original P.1 variant discovered in Brazil 7 . This is the first report of 20 J/501Y.V3 (P.1 lineage) in Kofu, Japan.

None.

This study was supported by a Grant- 

In contrast to influenza A (H1N1), influenza B infections are discounted given their limited host range, low rate of antigenic drift, low incidence, and milder disease severity than influenza A. 1 However, influenza B infection appeared to be similar to influenza A in clinical presentation, 2 and pediatric influenza B-associated mortality is greater than that of influenza A. 3 Additionally, oseltamivir has been reported to be less effective at reducing the viral response and duration of fever in outpatients with influenza B compared to those with influenza A; 4 however, high-dose oseltamivir might be more effective. 5 Despite these disparate reports, comprehensive studies comparing the characteristics and pathogenesis of influenza infections caused by A and B viruses are still lacking, particularly among severe cases who are admitted to the intensive care unit (ICU).

Neutrophils are an important component of the exaggerated inflammatory response in influenza infection; 6 however, little is known about neutrophil extracellular traps (NETs), which are expelled by the nuclear components of the cells. 7 We have recently reported the pathogenic role of NETs, and plasma NETs might be regarded as a sensitive biomarker for severe influenza A infection. 8 , 9 Therefore, it is reasonable to investigate the role of NETs in influenza B and to determine whether there are differences in the roles of NETs between influenza A and B viruses.

We included 30 influenza A and 10 influenza B virus-related ICU admissions of China-Japan Friendship Hospital from 2017 through 2018. Informed consent was obtained from all patients, and our study was approved by the Ethics Committee of the China-Japan Friendship Hospital (No. 2,001,814). Table 1 was the patients' clinical characteristics.

The levels of NETs and inflammatory mediators in blood and bronchoalveolar lavage fluid (BALF) were quantified as described in our previous study. 8 Comparisons of plasma NETs showed no significant difference in these two groups ( Fig. 1 A) . However, the NETs burden of deaths in the influenza A group was higher than that in the influenza B group ( p = 0.0193) ( Fig. 1 A) . Additionally, the influenza A group with multiple organ dysfunction syndrome (MODS) was not significantly higher than the corresponding influenza B patients ( Fig. 1 A) . Intriguingly, the NET level was specifically higher in the BALF of patients with influenza B compared to those with influenza A ( p = 0.0019) ( Fig. 1 B) . Consistent differences were also found between the pulmonary NET levels of deaths in the two groups ( p = 0.0050) ( Fig. 1 B) . Furthermore, a higher level of pulmonary NETs was observed in the influenza B patients with MODS than in the influenza A cases ( p = 0.0280) ( Fig. 1 B) .

As for the inflammatory mediators, the concentrations of circulating interleukin (IL) −7, IL-18, and interferon (IFN)-γ were higher in the influenza A group than in the influenza B group ( Fig. 1 C) . In contrast to influenza A patients, pulmonary mediators, including IFN-γ , IL-1 β, chemokine ligand (CCL) 3, CCL4, and fibroblast growth factor-2 were markedly elevated ( Fig. 1 C-D) . However, levels of IL-2, monocyte chemotactic protein-1, interferon-inducible protein-10, stem cell factor, and vascular endothelial growth factor-D in BALF were lower in the influenza B group than in the influenza A group ( Fig. 1 D-E) .

In this study, we compared the roles of NETs between severe influenza A and B patients. With indistinguishable plasma NET levels compared with influenza A patients, the pulmonary NET levels were significantly increased in influenza B patients. This finding suggests that the enhanced pulmonary NETs have pathogenic roles in influenza B infection but not in influenza A infection.

Although patients with influenza B have a relatively higher oxygen index, they are not distinguishable by clinical features from patients with influenza A, which is consistent with the previous studies. 2 To date, several studies have provided evidence for the role of NETs as a sensitive biomarker for severe influenza A infection. 9 Consequently, we further determined the levels of NETs in the lung and plasma from patients with severe influenza B and influenza A infection. The NET levels in BALF were increased in influenza B patients, although the plasma NET burden was similar, highlighting that pulmonary NET production in influenza B might induce more severe lung damage than influenza A infection. In fact, the generation of NETs systemically correlates with influenza A viral pathogenesis. 8 Hence, the pulmonary, rather than the circulating NETs seem to be pathogenic in influenza B infection but not in influenza A infection.

Patients with influenza A infection demonstrated an intense immune response, as evidenced by increased circulating inflammatory mediators. 10 In our findings, IL-7, IL-18, and IFN-γ in plasma indicated significant differences between the two groups, and followed the same trend as the plasma NET levels, suggesting a close correlation between the NET burden and the inflammatory response in circulation. As for the neutrophil chemoattractants, the circulating IL-18 was dramatically elevated in the influenza A group, indicating a pathogenic role of plasma NETs in patients with 

Dear Editor, In a recent paper, Majra et al. underlined the major role of superspreading events (SSEs) in SARS-Cov-2 transmission. 1 Heterogeneity in transmission, clustering, characterized by a small number of persons (superspreaders) responsible for the majority of the events, is a common feature of outbreaks, in particular at the early and late stages.

During the 2013-2016 Ebola outbreak in Guinea we quantified the exposures of contact persons to Ebola virus disease (EVD) patients and explored the consequences of the contacts pattern in terms of contact tracing and modeling. 2 After consent, a questionnaire detailing every exposure to EVD cases, including funerals, was passed and the number of exposures per contact person was summed. A high-risk exposure was defined as a close contact with a symptomatic EVD case or contact with body fluids or participating in a burial ritual.

Between May 2016 and September 2017, 1721 participants were enrolled in four locations (Conakry, Forécariah, Macenta, N'Zérékoré) (51.4% males, age range = [7-88 years], median age 21 years IQR [16-32]). They had made a total of 3074 contacts (exposures) with EVD cases (range = 1 to 17 exposures per person; median = 1; [IQR 1-2]). Overall, the frequency distribution of the cumulative number of exposures showed an overdispersed, aggregated, distribution: 84.3% of the participants ( n = 1451/1721) made at most two exposures to an EVD case. They were only 1.2% to report ten exposures or more. Aggregation was less pronounced in rural than in urban setting: the proportion of participants reporting less than three contacts was respectively 94.1% (and conversely 5.9% reporting three or more contacts) and 78.6% (21.4%) ( p < 10 −3 ).

Only 15.7% of the participants, at risk of being infected, concentrated three or more exposures with a large difference between the urban and rural settings (21.4% vs 5.9%). This clustering was also observed in terms of high-risk exposures with 86.2% of the participants reporting at most two high-risk exposures with again a marked difference between rural and urban settings (94.1% vs 81.6%, p < 10 −3 ). We fitted a negative binomial regression model using GAMLSS R-package with a zero-truncated distribution, people without contact being not included, by design. 3 The median number of exposures by person surveyed did not differ by rural or urban setting while the variance of the distribution was 26 times larger in urban setting than in rural one ( Table 1 and Fig. 1 ). This regression confirms the observed high degree of overdispersion of the contacts in an urban setting.

The aggregated distribution of the cumulative number of exposures means that a small number of the participants made many exposures and that the majority of the contact persons were exposed only once or twice. The median number of contacts did not differ but the associated variance was dramatically larger in urban setting. These trends are obviously driven by the population density and social closeness in large cities.

Although our study concerns contact persons who did not develop the disease after exposition, and may not be representative of the whole exposed population to Ebola virus, this clustering in exposure has implications for backwards contact tracing by targeting the surveillance toward the core of contact persons who made the greatest number of exposures. Indeed, another report on contacts from Kindia and Forecariah in Guinea suggested to stratify contacts-persons to focus on those most at risk. 4 Risk level assessment should include not only the closeness of exposures but also their number. While these contact patterns do not concern the infected population, the EVD source cases potentially transmitting the infection, it provides an additional evidence of the heterogeneity in contact rate, with a high degree of clustering following a power-law, close to the 80/20 of the Pareto principle, frequently observed in life sciences, human behavior and infectious diseases. 5 The main limitation of our study relies on the retrospective and declarative nature of the data and the likely recall bias. Recalling the number of exposures was however robust in our study since the network of contact persons was initially identified by the survivors regularly followed by our research staff. In addition, we showed in a previous study that the seropositivity against Ebola virus among participants was correlated with the declaration of high-risk contacts. 2 Taking into account sources of heterogeneity in models of transmission, beyond basic compartmental models assuming homogeneous mixing of the population, in which everyone shares the same epidemiological profile, and no stochastic effects, could substantially affect the modeling of the transmission dynamics and the elimination threshold to achieve herd immunity. 6 Living settings as well as age-dependent incidence, infectiousness, genetic features, human behaviors, occupation or spatial patterns are all a source of heterogeneity. 7 Our findings argue to account at least for urban/rural heterogeneity and an unobserved heterogeneity, representing other sources of individual variability, in modeling transmission. Host heterogeneity is best incorporated in networkbased models but SIR compartmental models as well could account for heterogeneity. 8 Surveys informing matrix of social contacts, "who might infect who", are still largely lacking, including in sub-Saharan Africa, limiting our ability to account for the heterogeneity in modeling transmission. However, mobile phone identification and social networks activities provide means to approach contact behavior. 9 In the case of the current SARS-CoV-2 pandemic, screening and communication strategies targeting potential superspreaders, such as connected people whose occupation implies a high frequency of contact, and SSEs, could be a cost-effective strategy when Re is close to 1, decreasing, at the late stage of an outbreak. In sub-Saharan Africa where the indirect effects of general restrictions on the fragile economies, health system, immunization coverage, access to foods, are at the forefront, this targeted strategies could save more lives than a blanket strategy. 10 imally cooked and the association with listeriosis is being increasingly recognised 10 .

In summary, the current dietary advice from the NHS for the UK pregnant woman 3 recommends avoiding eating unpasteurised dairy products including milk, soft cheeses (brie, camembert), chilled ready-to-eat foods like prepacked sandwiches and pâté. Targeted health education including safe food preparation practises at home among ethnic minority mothers is urgently needed not typically eating currently highlighted listeria risk foods.

We declare no competing interests.

Characteristics of hospitalized COVID-19 patients discharged and experiencing same-hospital readmission -United States

The RECOVERY Collaborative Group Dexamethasone in hospitalized patients with covid-19 -preliminary report

Angiotensin converting enzyme genotypes and mortality from COVID-19: an ecological study

Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy

Clinical characterization of patients with COVID-19 in primary care in catalonia: retrospective observational study

Risk factors associated with in-hospital mortality in a US national sample of patients with COVID-19

Machine learning prediction for mortality of patients diagnosed with COVID-19: a nationwide Korean cohort study

Characteristics and prognosis of COVID-19 in patients with COPD

Association of sex, age, and comorbidities with mortality in COVID-19 patients: a systematic review and meta-analysis

The association of hypertension with the severity and mortality of COVID-19 patients: evidence based on adjusted effect estimates

Autoimmune diseases are independently associated with COVID-19 severity: evidence based on adjusted effect estimates

Disparities in the excess risk of mortality in the first wave of COVID-19: cross sectional study of the English sentinel network

Letter in response to 'Modelling SARS-CoV2 spread in London: approaches to lift the lockdown' local experience, national questions. How local is local enough?

Revealed: The hospitals facing the most pressure to meet coronavirus demand

Seroprevalence of SARS-CoV-2 antibodies in healthcare workers at a London NHS Trust

Differential occupational risks to healthcare workers from SARS-CoV-2 observed during a prospective observational study

Exclusive: deaths of NHS staff from covid-19 analysed

Covid-19: ethnic minority doctors feel more pressured and less protected than white colleagues, survey finds

Effectiveness of BNT162b2 mRNA vaccine against infection and COVID-19 vaccine coverage in healthcare workers in England

Association of demographic and occupational factors with SARS-CoV-2 vaccine uptake in a multi-ethnic UK healthcare workforce: a rapid real-world analysis

A highly effective reverse-transcription loop-mediated isothermal amplification (RT-LAMP) assay for the rapid detection of SARS-CoV-2 infection

Reducing nosocomial transmission of COVID-19: implementation of a COVID-19 triage system

Virology, transmission, and pathogenesis of SARS-CoV-2

Put to the test: use of rapid testing technologies for Covid-19

Correlation between 3790 qPCR positives samples and positive cell cultures including 1941 SARS-CoV-2 isolates

Predicting infectious SARS-CoV-2 from diagnostic samples

COVID-19: rapid antigen detection for SARS-CoV-@ by lateral flow assay: a national systemic evaluation for mass-testing. medRxiv

Home-based SARS-CoV-2 lateral flow antigen testing in hospital workers

Co-infection of Chlamydia pneumoniae and Mycoplasma pneumoniae with SARS-CoV-2 is associated with more severe features

COVID-BioB Study Group. Candidemia in COVID-19 patients: incidence and characteristics in a prospective cohort compared to historical non-COVID-19 controls

ESCMID * guideline for the diagnosis and management of Candida diseases 2012: non-neutropenic adult patients

Evaluation of blood stream infections by Candida in three tertiary hospitals in Salvador, Brazil: a case-control study

Evaluation of species distribution and risk factors of candidemia: a multicenter case-control study

Bacterial and fungal coinfection among hospitalized patients with COVID-19: a retrospective cohort study in a UK secondary-care setting

Time to positive culture and identification for Candida blood stream infections

Delaying the empiric treatment of Candida bloodstream infection until positive blood culture results are obtained: a potential risk factor for hospital mortality

Clinical performance of the (1, 3)-β-D -glucan assay in early diagnosis of nosocomial Candida bloodstream infections

Impact of non-pharmaceutical interventions targeted at COVID-19 pandemic on influenza burden -a systematic review

Impact of Covid-19 quarantine and school cancelation on other common infectious diseases

Global prevalence of norovirus in cases of gastroenteritis: a systematic review and meta-analysis

Global economic burden of norovirus gastroenteritis

Norovirus gastroenteritis among hospitalized patients

Emerging COVID-19 success story: Germany's strong enabling environment 2020

R: A language and environment for statistical computing

Viral gastroenteritis

Norovirus vaccine: priorities for future research and development

How will country-based mitigation measures influence the course of the COVID-19 epidemic?

Potential therapeutic options for COVID-19: current status, challenges, and future perspectives

Lopinavir-ritonavir in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial

A trial of lopinavir-ritonavir in adults hospitalized with severe covid-19

WHO Solidarity Trial Consortium Repurposed antiviral drugs for Covid-19 -interim WHO solidarity trial results

A trial of Lopinavir-Ritonavir in Covid-19

A trial of Lopinavir-Ritonavir in Covid-19

Early administration of ritonavir-boosted lopinavir could prevent severe COVID-19

Clinical characteristics of patients hospitalized with COVID-19 in Spain: results from the SEMI-COVID-19 registry

Multiple Imputation for Nonresponse in Surveys John Wi

Dolin Raphael Principles and practice of infectious diseases

WHO Rapid Evidence Appraisal for COVID-19 Therapies (REACT) Working Group Association between administration of systemic corticosteroids and mortality among critically Ill patients with COVID-19: a meta-analysis

Early short-course corticosteroids and furosemide combination to treat non-critically ill COVID-19 patients: an observational cohort study

Effect of colchicine vs standard care on cardiac and inflammatory biomarkers and clinical outcomes in patients hospitalized with coronavirus disease 2019: the GRECCO-19 randomized clinical trial

Aspirin use is associated with decreased mechanical ventilation, ICU admission, and in-hospital mortality in hospitalized patients with COVID-19

Impact of oral anticoagulation on clinical outcomes of COVID-19: a nationwide cohort study of hospitalized patients in Germany

Anticoagulation in hospitalized patients with COVID-19

updated EULAR evidence-based recommendations for the management of gout

RECOVERY Collaborative Group Dexamethasone in hospitalized patients with COVID-19 -preliminary report

Thrombosis in hospitalized patients with COVID-19 in a New York city health system

Risk stratification of patients admitted to hospital with COVID-19 using the ISARIC WHO Clinical Characterization Protocol: development and validation of the 4C Mortality Score

Efficacy of colchicine in non-hospitalized patients with COVID-19

post-COVID" syndrome: how deep is the damage

Post-acute COVID-19 Syndrome. Incidence and risk factors: a Mediterranean cohort study

Characterizing long COVID in an international Cohort: 7 months of symptoms and their impact

Patient outcomes after hospitalisation with COVID-19 and implications for followup: results from a prospective UK cohort

Postdischarge symptoms and rehabilitation needs in survivors of COVID-19 infection: a cross-sectional evaluation

Sequelae in adults at 6 months after COVID-19 infection

Asthma and COPD in hospitalized COVID-19 patients

COVID-19 susceptibility in bronchial asthma

Prior SARS-CoV-2 infection is associated with protection against symptomatic reinfection

Recurrent COVID-19 including evidence of reinfection and enhanced severity in thirty Brazilian healthcare workers

Assessment of protection against reinfection with SARS-CoV-2 among 4 million PCR-tested individuals in Denmark in 2020: a population-level observational study

SARS-CoV-2 Transmission From People Without COVID-19 Symptoms

Common Investigation Protocol for Investigating Suspected SARS-CoV-2 Reinfection

Is recurrence possible in coronavirus disease 2019 (COVID-19)? Case series and systematic review of literature

The Possibility and Cause of Relapse After Previously Recovering From COVID-19: a Systematic Review

Introduction of Brazilian SARS-CoV-2 484 K.V2 related variants into the UK

CoV-2 lineage A variant (A.23.1) with altered spike has emerged and is dominating the current Uganda epidemic

gov.vn, an official website of the Vietnamese Ministry of Health providing update information about COVID-19

COVID-19 control in Vietnam

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

A dynamic nomenclature proposal for SARS-CoV-2 lineages to assist genomic epidemiology

The natural history and transmission potential of asymptomatic SARS-CoV-2 infection

During the ongoing evolution of SARS-CoV-2, newly emerging lineages are likely to be circulating in the human population and genomic surveillance will be important for evaluating the emergence, spread, vaccine efficacy, and transmissibility of these lineages. Currently, an emergent D614G mutation in the spike glycoprotein of SARS-CoV-2 is prevalent globally 4 . More recently, new emerging lineages with spike protein mutations have been discovered in the United Kingdom (B.1.1.7 lineage, 20I/501Y.V1, also named VOC 202,012/01) 5 , South Africa (B.1.351 lineage, 20H/501Y.V2) 6 , and Brazil (P.1 lineage, 20 J/501Y.V3) 7 , 8 . All of these lineages have a N501Y mutation in the receptor binding domain (RBD), which directly binds to the angiotensin converting enzyme 2 (ACE2) receptor of the host cell, contributing to increased transmissibility. Both B.1.351 and P.1 lineages also have additional K417N/T and E484K mutations. K417N/T, E484K, and N501Y confer reduced neutralizing activity of convalescent and mRNA vaccineelicited serum 3 . To determine the genomic characteristics of the SARS-CoV-2 variant identified in the Kofu, Japan, we started whole genome sequencing analysis using the Ion Torrent Genexus System

This patient was a 46-year-old man who entered our hospital in early February 2021 with a fever at 38.9 °C and with a history of staying in Brazil. RT-qPCR indicated a high viral load (7.1 log 10 /μL) and low cycle threshold (Ct) value of 15. The patient had displayed mild symptoms upon returning to Japan 4 days earlier and was admitted to another hospital. However

In the RBD of the spike protein, three mutations (K417T, E484K and N501Y) were identified. These results indicated that we had identified a variant related to 20 J/501Y.V3 (P.1 lineage) in Japan

1 lineage) samples sequenced have 33-40 mutations compared with original strain reported from

Introduction of the South African SARS-CoV-2 variant 501Y.V2 into the UK

No evidence for increased transmissibility from recurrent mutations in SARS-CoV-2

mRNA vaccine-elicited antibodies to SARS-CoV-2 and circulating variants

Tracking Changes in SARS-CoV-2 Spike: evidence that D614G Increases Infectivity of the COVID-19 Virus

European Centre for Disease Prevention and Control: Rapid increase of a SARS-CoV-2 variant with multiple spike protein mutations observed in the United Kingdom

Emergence and rapid spread of a new severe acute respiratory syndrome-related coronavirus 2 (SARS-CoV-2) lineage with multiple spike mutations in South Africa

Genomic characterisation of an emergent SARS-CoV-2 lineage in Manaus: preliminary findings

Phylogenetic relationship of SARS-CoV-2 sequences from Amazonas with emerging Brazilian variants harboring mutations E484K and N501Y in the Spike protein

Pooling RT-qPCR testing for SARS-CoV-2 in 10 0 0 individuals of healthy and infection-suspected patients

Prospective study of 1,308 nasopharyngeal swabs from 1,033 patients using the LUMIPULSE SARS-CoV-2 antigen test: comparison with RT-qPCR

Fatal influenza B infections: time to reexamine influenza research priorities

Comparing clinical characteristics between hospitalized adults with laboratoryconfirmed influenza A and B virus infection

Hospitalization for Influenza A Versus B

A comparison of the effectiveness of oseltamivir for the treatment of influenza A and influenza B: a Japanese multicenter study of the 20 03-20 04 and 20 04-20 05 influenza seasons

A prospective intervention study on higher-dose oseltamivir treatment in adults hospitalized with influenza a and B infections

H5N1 and 1918 pandemic influenza virus infection results in early and excessive infiltration of macrophages and neutrophils in the lungs of mice

Neutrophil extracellular traps kill bacteria

High level of neutrophil extracellular traps correlates with poor prognosis of severe Influenza A infection

Circulating rather than alveolar extracellular deoxyribonucleic acid levels predict outcomes in Influenza

Clinical aspects and cytokine response in severe H1N1 influenza a virus infection

SARS-CoV-2 (COVID-19) superspreader events

Prevalence of infection among asymptomatic and paucisymptomatic contact persons exposed to Ebola virus in Guinea: a retrospective, cross-sectional observational study

Generalized additive models for location scale and shape (GAMLSS) in R

Contact tracing activities during the Ebola virus disease epidemic in Kindia and Faranah, Guinea

Heterogeneities in the transmission of infectious agents: implications for the design of control programs

Characterizing the transmission dynamics and control of ebola virus disease

Infectious diseases of humans: dynamics and control

When individual behaviour matters: homogeneous and network models in epidemiology

Modeling infectious disease dynamics in the complex landscape of global health

Potential impact of the COVID-19 pandemic on HIV, tuberculosis, and malaria in low-income and middle-income countries: a modelling study

we observed 3 Covid-19 . Clinical presentation was typical for listeriosis, Supplementary table. Presentation of the three most common clonal complexes (CC1, CC2, and CC6) of L. monocytogenes did not differ, however, numbers were small. Food exposures with significantly lower odds of illness among ethnic minorities versus British cases between

0.0010), carrots (OR 2.40, p -value 0.0017), parsley (OR 2.40, p -value 0.017), and consuming Kosher/Halal foods (OR 11.86, p -value < 0.0 0 01) had higher odds between ethnic minorities versus British cases (Supplementary table). Mothers from ethnic minorities reported eating equally vegetables (OR 1.31, p -value 0.51), and salads (OR 0.86, p -value 0.70) than British. Ethnic minority cases stored loose meat (mean 1.57 days) longer than British cases since purchase (mean 1.30 days, t -test p -value = 0.093). Ethnic supermarkets selling non-British foods (OR 14.32, p -value < 0ethnic minorities. Multivariate model

Neighbourhood unemployment and other socio-demographic predictors of emergency hospitalisation for infectious intestinal disease in England: a longitudinal ecological study

The epidemiology of listeriosis in pregnant women and children in New Zealand from 1997 to 2016: an observational study

Differences among incidence rates of invasive listeriosis in the US FoodNet population by age, sex, race/ethnicity, and pregnancy status

Mortality risk factors for listeriosis -a 10 year review of non-pregnancy associated cases in England

A priori and a posteriori dietary patterns in women of childbearing age in the UK

LiSEQ -whole-genome sequencing of a cross-sectional survey of Listeria monocytogenes in ready-to-eat foods and human clinical cases in

International MLST database for Listeria monocytogenes

Emergence of pregnancy-related listeriosis amongst ethnic minorities in England and Wales

We would like to thank Li Shi, Ying Wang, Jian Wu, Peihua Zhang, Yang Li and Wenwei Xiao (All are from Department of Epidemiology, School of Public Health, Zhengzhou University) for their kind help in searching articles and collecting data, and valuable suggestions for data analysis.

Haiyan Yang and Yadong Wang designed the study. Hongjie Hou and Jie Xu performed literature search. Hongjie Hou and Haiyan Yang performed data extraction. Xuan Liang, Haiyan Yang, Hongjie Hou and Jie Xu performed statistical analyses. Haiyan Yang, Hongjie Hou and Yadong Wang wrote and reviewed the manuscript. All the authors approved the final version of the manuscript.

This study was supported by grants from National Natural Science Foundation of China (grant number 81973105), Key Scientific Research Project of Henan Institution of Higher Education (grant number 21A330 0 08) and Joint Construction Project of Henan Medical Science and Technology Research Plan (grant number LHGJ20190679). The funders have no role in the data collection, data analysis, preparation of manuscript and decision to submission.

The authors are grateful to Dr. Daniela Bertsch, Ulrike Betz and Melissa Kolb for their contribution to the study in the field. The authors acknowledge Ana Goios for medical writing and editorial support, and Anirudh Tomer for data analysis support (both affiliated to P95 Epidemiology and Pharmacovigilance, Leuven, Belgium).

This work was supported by Takeda Vaccines, Inc. The sponsor provided some input on the study design of this study. The authors had responsibility for the submission of this manuscript for publication. We are indebted to Ms Le Kim Thanh, Lam Anh Nguyet and the Molecular Diagnostic Group of the Hospital for Tropical Diseases for their logistic/laboratory support. We thank the patients for their participations in this study,

We thank the researchers who deposited the SARS-CoV-2 sequencing data in the GISAID. We also thank Masato Kondo, Ryota Tanaka, and Kazuo Sakai (Thermo Fisher Scientific) for technical help, all of the medical and ancillary hospital staff.

We thank the patients for their involvement in the study.

NNZ and FZS were responsible for the experimental design. LLZ and YHT collected the patient samples and analyzed the data. LLZ and YZ conducted the experiments and performed statistical analyses. NNZ and LLZ performed the experiments, analyzed the data, and wrote the manuscript. HZ and QYZ assisted with designing the experiments and interpreting the data. FZS supervised the process and revised the manuscript. All authors have approved the manuscript for submission.

This work was supported by the National Natural Science Foundation of China (grant numbers 820 0 0 090 and 82003575 ); PhD Research Foundation of the Affiliated Hospital of Jining Medical University (grant numbers 2020-BS-003 and 2021-BS-008 ); and Natural National Natural Science Foundation of Jining Medical University (grant number JYP2019KJ26 ).

We are grateful to the participants who agreed to respond to the survey and we thank the members of the Contactebogui study group for their contribution.This study was funded by the Ebola French Task Force, the "Institut National de la Santé et de la Recherche Médicale/REACTing", the "Institut de Recherche pour le Développement", and « MUSE/Université de Montpellier, France (ANR_16-IDEX-0 0 06)".

We thank Mike Harte and Thomas Thackray for excellent data management assistance. No separate funding was received, this study was carried as part of routine work in Public Health England.

 Figure 2 ).Since its first detection in Rwanda in October 2020, as of 19th March 2021, A.23.1 has been reported in 23countries worldwide. 9 Notably, recently, A.23.1 has emerged and become a predominant sub-lineage circulating in Kampala, Uganda. 2 Viruses of A.23.1 carry four defining mutations in spike protein (F157L, V367F, Q613H and P681R). Of these, Q613H is predicted to be biologically equivalent to the D614G, which emerged in early 2020, and has been shown to increase the transmisibility. As a consequnence, A.23.1 is now listed as one of the five variants (B.1.1.7, P1, B.1.351, and B.1.525) to be tracked globally. 9 The turn-around time from RT-PCR diagnosis to SARS-CoV-2 lineage determination by whole-genome sequencing was between 1.5-3 days. This was achievable because of pre-existing sequencing infrastructure and expertise, and helped by the low prevalence of SARS-CoV-2 in Vietnam. The sequencing findings were critical to informing rapid public health responses in HCMC. Indeed, the detection of the B.1.1.7 variant in the north led to enhanced surveillance in the south and the detection of the TSN airport cluster, which may otherwise have gone unnoticed.Active surveillance for SARS-CoV-2 variants has been applied in developed countries since the beginning of the pandemic. 10 It is now one of the top priorities of the WHO. However, success stories from a resource-constrained setting like Vietnam remain uncommon. Thus, enhancing the sequencing capacity in these recognized hotspots of pathogen emergence is of vital importance for both the global COVID-19 research agenda and the control of future emerging infections.In summary, while our findings have expanded the geographic distributions of B.1.1.7 and A.23.1 variants, the data emphasize the importance of active surveillance for SARS-CoV-2 worldwide. The sequencing capacity in low-and middle-income countries must be strengthened to address the challenges of the ongoing COVID-19 pandemic and future emerging infections.

Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.jinf.2021.03.017 .

Supplementary material associated with this article can be found, in the online version, at doi: 10 severe influenza A. However, with a significantly increased BALF NET burden in influenza B patients, the IL-18 level was not elevated in the influenza B group, suggesting that other neutrophil chemoattractants in the lung might be potential NET-related mediators associated with influenza B infection. Moreover, the inflammatory mediator levels in BALF showed a similar trend with pulmonary NETs, suggesting a correlation between the pulmonary NETs and the robustness of the local pulmonary inflammation. Collectively, the pathogenic role of NETs in BALF is likely attributable to their roles as inducers of inflammation, which ultimately leads to lung damage following influenza B infection.Limitations of our study include the small sample size, the single-center experience, and failure identification of the lineage of influenza B, which might have distinct pathogenicity.In conclusion, in contrast to influenza A, in which circulating NETs are predictors of poor outcomes, influenza B infection could induce an enhanced production of pulmonary NETs, which may play a pathogenic role. Thus, targeting pulmonary NETs might be an innovative therapeutic approach for influenza B infection.

A recent paper in this journal by Rose et al. described the increasing hospitalizations due to infectious intestinal disease among vulnerable population groups in the UK, including those from ethnic minorities or unemployed. 1 Pregnancy related listeriosis is a severe illness for the unborn and newly delivered infant. Ethnic minorities may have higher incidence as in New Zealand and the USA. 2 , 3 In England, guidance on food consumption during pregnancy to avoid listeriosis is available through universal maternity care and NHS website. 4 The majority of cases of listeriosis occur amongst non-pregnant individuals. 5 Risk food consumption amongst pregnant women has not been recently investigated. 6 We characterized listeriosis among pregnant women from the national Listeria surveillance database in England between 2005 and 2020 based on sampling dates. Population birth and deprivation (Index of Multiple Deprivation) postcode data was derived from Office for National Statistics, ONS. 7 A case of pregnancy associated listeriosis was defined microbiologically confirmed Listeria monocytogenes infection in a mother or her undelivered or newly delivered infant. Responses were classified into two nonnationality based ethnicity categories as British versus "ethnic minority" (other white or any other ethnicities). Of British population, 78.7% are white British, 6.2% other white, 8.0% Asian, 3.5% any Black, 3.7% others. The standard proportions of exposures were calculated with odds ratios and adjusted for deprivation (Stata, v15/16). All culture confirmed L. monocytogenes were tested by whole genome sequence analysis since December 2015 to 2020,

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jinf.2021.03.024 .