Significant differences exist in the availability of healthcare worker (HCW) SARS-CoV-2 testing between countries, and existing programmes focus on screening symptomatic rather than asymptomatic staff. Over a 3 week period (April 2020), 1032 asymptomatic HCWs were screened for SARS-CoV-2 in a large UK teaching hospital. Symptomatic staff and symptomatic household contacts were additionally tested. Real-time RT-PCR was used to detect viral RNA from a throat+nose self-swab. 3% of HCWs in the asymptomatic screening group tested positive for SARS-CoV-2. 17/30 (57%) were truly asymptomatic/pauci-symptomatic. 12/30 (40%) had experienced symptoms compatible with coronavirus disease 2019 (COVID-19)>7 days prior to testing, most self-isolating, returning well. Clusters of HCW infection were discovered on two independent wards. Viral genome sequencing showed that the majority of HCWs had the dominant lineage B∙1. Our data demonstrates the utility of comprehensive screening of HCWs with minimal or no symptoms. This approach will be critical for protecting patients and hospital staff.
Patients admitted to NHS hospitals are now routinely screened for SARS-CoV-2 (the virus that causes COVID-19), and isolated from other patients if necessary. Yet healthcare workers, including frontline patient-facing staff such as doctors, nurses and physiotherapists, are only tested and excluded from work if they develop symptoms of the illness.
However, there is emerging evidence that many people infected with SARS-CoV-2 never develop significant symptoms: these people will therefore be missed by ‘symptomatic-only’ testing. There is also important data showing that around half of all transmissions of SARS-CoV-2 happen before the infected individual even develops symptoms. This means that much broader testing programs are required to spot people when they are most infectious.
Rivett, Sridhar, Sparkes, Routledge et al. set out to determine what proportion of healthcare workers was infected with SARS-CoV-2 while also feeling generally healthy at the time of testing. Over 1,000 staff members at a large UK hospital who felt they were well enough to work, and did not fit the government criteria for COVID-19 infection, were tested. Amongst these, 3% were positive for SARS-CoV-2. On closer questioning, around one in five reported no symptoms, two in five very mild symptoms that they had dismissed as inconsequential, and a further two in five reported COVID-19 symptoms that had stopped more than a week previously. In parallel, healthcare workers with symptoms of COVID-19 (and their household contacts) who were self-isolating were also tested, in order to allow those without the virus to quickly return to work and bolster a stretched workforce.
Finally, the rates of infection were examined to probe how the virus could have spread through the hospital and among staff – and in particular, to understand whether rates of infection were greater among staff working in areas devoted to COVID-19 patients. Despite wearing appropriate personal protective equipment, healthcare workers in these areas were almost three times more likely to test positive than those working in areas without COVID-19 patients. However, it is not clear whether this genuinely reflects greater rates of patients passing the infection to staff. Staff may give the virus to each other, or even acquire it at home.
Overall, this work implies that hospitals need to be vigilant and introduce broad screening programmes across their workforces. It will be vital to establish such approaches before ‘lockdown’ is fully lifted, so healthcare institutions are prepared for any second peak of infections.
Despite the World Health Organisation (WHO) advocating widespread testing for SARS-CoV-2, national capacities for implementation have diverged considerably (WHO, 2020b; Our World in Data, 2020). In the UK, the strategy has been to perform SARS-CoV-2 testing for essential workers who are symptomatic themselves or have symptomatic household contacts. This approach has been exemplified by recent studies of symptomatic HCWs (Hunter et al., 2020; Keeley et al., 2020). The role of nosocomial transmission of SARS-CoV-2 is becoming increasingly recognised, accounting for 12–29% of cases in some reports (Wang et al., 2020). Importantly, data suggest that the severity and mortality risk of nosocomial transmission may be greater than for community-acquired COVID-19 (McMichael et al., 2020).
Protection of HCWs and their families from the acquisition of COVID-19 in hospitals is paramount, and underscored by rising numbers of HCW deaths nationally and internationally (Cook et al., 2020; CDC COVID-19 Response Team, 2020). In previous epidemics, HCW screening programmes have boosted morale, decreased absenteeism and potentially reduced long-term psychological sequelae (McAlonan et al., 2007). Screening also allows earlier return to work when individuals or their family members test negative (Hunter et al., 2020; Keeley et al., 2020). Another major consideration is the protection of vulnerable patients from a potentially infectious workforce (McMichael et al., 2020), particularly as social distancing is not possible whilst caring for patients. Early identification and isolation of infectious HCWs may help prevent onward transmission to patients and colleagues, and targeted infection prevention and control measures may reduce the risk of healthcare-associated outbreaks.
The clinical presentation of COVID-19 can include minimal or no symptoms (WHO, 2020a). Asymptomatic or pre-symptomatic transmission is clearly reported and is estimated to account for around half of all cases of COVID-19 (He et al., 2020). Screening approaches focussed solely on symptomatic HCWs are therefore unlikely to be adequate for suppression of nosocomial spread. Preliminary data suggests that mass screening and isolation of asymptomatic individuals can be an effective method for halting transmission in community-based settings (Day, 2020). Recent modelling has suggested that weekly testing of asymptomatic HCWs could reduce onward transmission by 16–23%, on top of isolation based on symptoms, provided results are available within 24 hr (Imperial College COVID-19 Response Team, 2020). The need for widespread adoption of an expanded screening programme for asymptomatic, as well as symptomatic HCWs, is apparent (Imperial College COVID-19 Response Team, 2020; Black et al., 2020; Gandhi et al., 2020).
Challenges to the roll-out of an expanded screening programme include the ability to increase diagnostic testing capacity, logistical issues affecting sampling and turnaround times and concerns about workforce depletion should substantial numbers of staff test positive. Here, we describe how we have dealt with these challenges and present initial findings from a comprehensive staff screening programme at Cambridge University Hospitals NHS Foundation Trust (CUHNFT). This has included systematic screening of >1000 asymptomatic HCWs in their workplace, in addition to >200 symptomatic staff or household contacts. Screening was performed using a validated real-time reverse transcription PCR (RT-PCR) assay detecting SARS-CoV-2 from combined oropharyngeal (OP) and nasopharyngeal (NP) swabs (Sridhar et al., 2020). Rapid viral sequencing of positive samples was used to further assess potential epidemiological linkage where nosocomial transmission was suspected. Our experience highlights the value of programmes targeting both symptomatic and asymptomatic staff, and will be informative for the establishment of similar programmes in the UK and globally.
Between 6th and 24th April 2020, 1,270 HCWs in CUHNFT and their symptomatic household contacts were swabbed and tested for SARS-CoV-2 by real-time RT-PCR. The median age of the HCWs was 34; 71% were female and 29% male. The technical RT-PCR failure rate was 2/1,270 (0∙2% see Materials and methods); these were excluded from the ‘Tested’ population for further analysis. Ultimately, 5% (n = 61) of swabs were SARS-CoV-2 positive. 21 individuals underwent repeat testing for a variety of reasons, including evolving symptoms (n = 3) and scoring ‘medium’ probability on clinical COVID-19 criteria (Tables 1–2) (n = 11). All remained SARS-CoV-2 negative. Turn around time from sample collection to resulting was 12–36 hr; this varied according to the time samples were obtained.
Table 3 outlines the total number of SARS-CoV-2 tests performed in each screening group (HCW asymptomatic, HCW symptomatic, and HCW symptomatic household contact) categorised according to the ward with the highest anticipated risk of exposure to COVID-19 (‘red’, high; ‘amber’, medium; ‘green’, low; Tables 4–5). In total, 31/1,032 (3%) of those tested in the HCW asymptomatic screening group tested SARS-CoV-2 positive. In comparison, 30/221 (14%) tested positive when HCW symptomatic and HCW symptomatic household contact screening groups were combined. As expected, symptomatic HCWs and their household contacts were significantly more likely to test positive than HCWs from the asymptomatic screening group (p<0∙0001, Fisher’s exact test). HCWs working in ‘red’ or ‘amber’ wards were significantly more likely to test positive than those working in ‘green’ wards (p=0∙0042, Fisher’s exact test).
All users of FFP3 masks underwent routine fit-testing prior to usage. Cleaning and re-use of masks, theatre caps, gloves, aprons or gowns was actively discouraged. Cleaning and re-use of eye protection was permitted for certain types of goggles and visors, as specified in the hospital’s PPE protocol. Single-use eye protection was in use in most Scenario 1 and 2 areas, and was not cleaned and re-used. All non-invasive ventilation or use of high-flow nasal oxygen on laboratory-confirmed or clinically suspected COVID-19 patients was performed in negative-pressure (−5 pascals) side rooms, with 10 air changes per hour and use of Scenario 2 PPE. All other aerosol generating procedures were undertaken with Scenario 2 PPE precautions, in negative- or neutral- pressure facilities. General clinical areas underwent a minimum of 6 air changes per hour, but all critical care areas underwent a minimum of 10 air changes per hour as a matter of routine. Surgical operating theatres routinely underwent a minimum of 25 air changes per hour.
Viral loads varied between individuals, potentially reflecting the nature of the sampling site. However, for individuals testing positive for SARS-CoV-2, viral loads were significantly lower for those in the HCW asymptomatic screening group than in those tested due to the presence of symptoms (Figure 1). For the HCW symptomatic and HCW symptomatic contact screening groups, viral loads did not correlate with duration of symptoms or with clinical criteria risk score (Figure 1—figure supplement 1 and data not shown).
Each individual in the HCW asymptomatic screening group was contacted by telephone to establish a clinical history, and COVID-19 probability criteria (Table 1) were retrospectively applied to categorise any symptoms in the month prior to testing (Figure 2). One HCW could not be contacted to obtain further history. Individuals captured by the HCW asymptomatic screening group were generally asymptomatic at the time of screening, however could be divided into three sub-groups: (i) HCWs with no symptoms at all, (ii) HCWs with (chiefly low-to-medium COVID-19 probability) symptoms commencing ≤7 days prior to screening and (iii) HCWs with (typically high COVID-19 probability) symptoms commencing >7 days prior to screening (Figure 2). 9/12 (75%) individuals with symptom onset >7 days previously had appropriately self-isolated and then returned to work. One individual with no symptoms at the time of swabbing subsequently developed symptoms prior to being contacted with their positive result. Overall, 5/1032 (0.5%) individuals in the asymptomatic screening group were identified as truly asymptomatic carriers of SARS-CoV-2, and 1/1032 (0.1%) was identified as pre-symptomatic. Box 1 shows illustrative clinical vignettes.
Self-isolation instructions were as described in Table 2.
Case 1: Completely asymptomatic. HCW1 had recently worked on four wards (two ‘green’, two ‘amber’). Upon testing positive, she reported no symptoms over the preceding three weeks, and was requested to go home and self-isolate immediately. HCW1 lived with her partner who had no suggestive symptoms. Upon follow-up telephone consultation 14 days after the test, HCW1 had not developed any significant symptoms, suggesting true asymptomatic infection.
Case 2: Pre-symptomatic. HCW2 was swabbed whilst asymptomatic, testing positive. When telephoned with the result, she reported a cough, fever and headache starting within the last 24 hr and was advised to self-isolate from the time of onset of symptoms (Table 2). Her partner, also a HCW, was symptomatic and had been confirmed as SARS-CoV-2 positive 2 days previously, suggesting likely transmission of infection to HCW2.
Case 3: Low clinical probability of COVID HCW3 developed mild self-limiting pharyngitis three days prior to screening and continued to work in the absence of cough or fever. She had been working in’ green’ areas of the hospital, due to a background history of asthma. Self-isolation commenced from the time of the positive test. HCW3’s only contact outside the hospital, her housemate, was well. On follow-up telephone consultation, HCW3’s mild symptoms had fully resolved, with no development of fever or persistent cough, suggesting pauci-symptomatic infection.
Case 4: Medium clinical probability of COVID HCW4 experienced anosmia, nausea and headache three days prior to screening, and continued to work in the absence of cough or fever. Self-isolation commenced from the time of the positive test. One son had experienced a mild cough ~3 weeks prior to HCW4’s test, however her partner and other son were completely asymptomatic. Upon follow-up telephone consultation 10 days after the test, HCW4’s mild symptoms had not progressed, but had not yet resolved.
Case 5: High clinical probability of COVID. HCW5 had previously self-isolated, and did not repeat this in the presence of new high-probability symptoms six days before screening. Self-isolation commenced from the date of the new symptoms with the caveat that they should be completely well for 48 hr prior to return to work. All household contacts were well. However, another close colleague working on the same ward had also tested positive, suggesting potential transmission between HCWs on that ward.
For the HCW asymptomatic screening group, nineteen wards were identified for systematic priority screening as part of hospital-wide surveillance. Two further areas were specifically targeted for screening due to unusually high staff sickness rates (ward F), or concerns about appropriate PPE usage (ward Q) (Figure 3). Interestingly, in line with findings in the total HCW population, a significantly greater proportion of HCWs working on ‘red’ wards compared to HCWs working on ‘green’ wards tested positive as part of the asymptomatic screening programme (‘green’ 6/310 vs ‘red’ 19/372; p=0.0389, Fisher’s exact test). The proportion of HCW with a positive test was significantly higher on Ward F than on other wards categorised as ‘green’ clinical areas (ward F 4/43 vs other ‘green’ wards 2/267; p=0.0040, Fisher’s exact test). Likewise, amongst wards in the ‘red’ areas, ward Q showed significantly higher rates of positive HCW test results (ward Q 7/37 vs other ‘red’ wards 12/335; p=0.0011, Fisher’s exact test).
Ward F is an elderly care ward, designated as a ‘green’ area with Scenario 0 PPE (Tables 4–5), with a high proportion of COVID-19 vulnerable patients due to age and comorbidity. 4/43 (9%) ward staff tested positive for SARS-CoV-2. In addition, two staff members on this ward tested positive in the HCW symptomatic/symptomatic contact screening groups. All positive HCWs were requested to self-isolate, the ward was closed to admissions and escalated to Scenario 1 PPE (Table 5). Reactive screening of a further 18 ward F staff identified an additional three positive asymptomatic HCWs (Figure 4). Sequence analysis indicated that 6/9 samples from HCW who worked on ward F belonged to SARS-CoV-2 lineage B∙1 (currently known to be circulating in at least 43 countries [Rambaut et al., 2020]), with a further two that belonged to B1∙7 and one that belonged to B2∙1. This suggests more than two introductions of SARS-CoV-2 into the HCW population on ward F (Figure 4—figure supplements 1–2, Table 6). It was subsequently found that two further staff members from ward F had previously been admitted to hospital with severe COVID-19 infection.
Ward Q is a general medical ward designated as a ‘red’ clinical area for the care of COVID-19 positive patients, with a Scenario 1 PPE protocol (Tables 4–5). Here, 7/37 (19%) ward staff tested positive for SARS-CoV-2. In addition, one staff member tested positive as part of the HCW symptomatic screening group, within the same period as ward surveillance. Reactive screening of a further five staff working on Ward Q uncovered one additional infection. 4/4 sequenced viruses were of the B∙1 lineage (Figure 4—figure supplements 1–2, Table 6; other isolates could not be sequenced due to a sample CT value >30). All positive HCWs were requested to self-isolate, and infection control and PPE reviews were undertaken to ensure that environmental cleaning and PPE donning/doffing practices were compliant with hospital protocol. Staff training and education was provided to address observed instances of incorrect infection control or PPE practice.
Ward O, a ‘red’ medical ward, had similar numbers of asymptomatic HCWs screened as ward F, and a similar positivity rate (4/44; 9%). This ward was listed for further cluster investigation after the study ended, however incorrect PPE usage was not noted during the study period.
Through the rapid establishment of an expanded HCW SARS-CoV-2 screening programme, we discovered that 31/1,032 (3%) of HCWs tested positive for SARS-CoV-2 in the absence of symptoms. Of 30 individuals from this asymptomatic screening group studied in more depth, 6/30 (20%) had not experienced any symptoms at the time of their test. 1/6 became symptomatic suggesting that the true asymptomatic carriage rate was 5/1,032 (0.5%). 11/30 (37%) had experienced mild symptoms prior to testing. Whilst temporally associated, it cannot be assumed that these symptoms necessarily resulted from COVID-19. These proportions are difficult to contextualise due to paucity of point-prevalence data from asymptomatic individuals in similar healthcare settings or the wider community. For contrast, 60% of asymptomatic residents in a recent study tested positive in the midst of a care home outbreak (Arons et al., 2020). Regardless of the proportion, however, many secondary and tertiary hospital-acquired infections were undoubtedly prevented by identifying and isolating these SARS-CoV-2 positive HCWs.
12/30 (40%) individuals from the HCW asymptomatic screening group reported symptoms > 7 days prior to testing, and the majority experiencing symptoms consistent with a high probability of COVID-19 had appropriately self-isolated during that period. Patients with COVID-19 can remain SARS-CoV-2 PCR positive for a median of 20 days (IQR 17–24) after symptom onset (Zhou et al., 2020), and the limited data available suggest viable virus is not shed beyond eight days (Wölfel et al., 2020). A pragmatic approach was taken to allowing individuals to remain at work, where the HCW had experienced high probability symptoms starting >7 days and ≤1 month prior to their test and had been well for the preceding 48 hr. This approach was based on the following: low seasonal incidence of alternative viral causes of high COVID-19 probability symptoms in the UK (Public Health England, 2018), the high potential for SARS-CoV-2 exposure during the pandemic and the potential for prolonged, non-infectious shedding of viral RNA (Zhou et al., 2020; Wölfel et al., 2020). For other individuals, we applied standard national guidelines requiring isolation for seven days from the point of testing (UK Government, 2020b). However, for HCW developing symptoms after a positive swab, isolation was extended for seven days from symptom onset.
Our data clearly demonstrate that focusing solely on the testing of individuals fitting a strict clinical case definition for COVID-19 will inevitably miss asymptomatic and pauci-symptomatic disease. This is of particular importance in the presence of falling numbers of community COVID-19 cases, as hospitals will become potential epicentres of local outbreaks. Therefore, we suggest that in the setting of limited testing capacity, a high priority should be given to a reactive asymptomatic screening programme that responds in real-time to HCW sickness trends, or (to add precision) incidence of positive tests by area. The value of this approach is illustrated by our detection of a cluster of cases in ward F, where the potential for uncontrolled staff-to-staff or staff-to-patient transmission could have led to substantial morbidity and mortality in a particularly vulnerable patient group. As SARS-CoV-2 testing capacity increases, rolling programmes of serial screening for asymptomatic staff in all areas of the hospital is recommended, with the frequency of screening being dictated by anticipated probability of infection. The utility of this approach in care-homes and other essential institutions should also be explored, as should serial screening of long-term inpatients.
The early success of our programme relied upon substantial collaborative efforts between a diverse range of local stakeholders. Similar collaborations will likely play a key role in the rapid, de novo development of comprehensive screening programmes elsewhere. The full benefits of enhanced HCW screening are critically dependent upon rapid availability of results. A key success of our programme has been bespoke optimisation of sampling and laboratory workflows enabling same-day resulting, whilst minimising disruption to hospital processes by avoiding travel to off-site testing facilities. Rapid turnaround for testing and sequencing is vital in enabling timely response to localised infection clusters, as is the maintenance of reserve capacity to allow urgent, reactive investigations.
There appeared to be a significantly higher incidence of HCW infections in ‘red’ compared to ‘green’ wards. Many explanations for this observation exist, and this study cannot differentiate between them. Possible explanations include transmission between patients and HCW, HCW-to-HCW transmission, variability of staff exposure outside the workplace and non-random selection of wards. It is also possible that, even over the three weeks of the study, ‘red’ wards were sampled earlier during the evolution of the epidemic when transmission was greater. Further research into these findings is clearly needed on a larger scale. Furthermore, given the clear potential for pre-symptomatic and asymptomatic transmission amongst HCWs, and data suggesting that infectivity may peak prior to symptom onset (He et al., 2020), there is a strong argument for basic PPE provision in all clinical areas.
The identification of transmission within the hospital through routine data is problematic. Hospitals are not closed systems and are subject to numerous external sources of infection. Coronaviruses generally have very low mutation rates (~10−6 per site per cycle) (Sanjuán et al., 2010), with the first reported sequence of the current pandemic only published on 12th January 2020 (GenBank, 2020). In addition, given SARS CoV-2 was only introduced into the human population in late 2019, there is at present a lack of diversity in circulating strains. However, as the pandemic unfolds and detailed epidemiological and genome sequence data from patient and HCW clusters are generated, real-time study of transmission dynamics will become an increasingly important means of informing disease control responses and rapidly confirming (or refuting) hospital acquired infection. Importantly, implementation of such a programme would require active screening and rapid sequencing of positive cases in both the HCW and patient populations. Prospective epidemiological data will also inform whether hospital staff are more likely to be infected in the community or at work, and may identify risk factors for the acquisition of infection, such as congregation in communal staff areas or inadequate access to PPE.
Our study is limited by the relatively short time-frame, a small number of positive tests and a lack of behavioural data. In particular, the absence of detailed workplace and community epidemiological data makes it difficult to draw firm conclusions with regards to hospital transmission dynamics. The low rate of observed positive tests may be partly explained by low rates of infection in the East of England in comparison with other areas of the UK (cumulative incidence 0.17%, thus far) (Public Health England, 2020). The long-term benefits of HCW screening on healthcare systems will be informed by sustained longitudinal sampling of staff in multiple locations. More comprehensive data will parametrise workforce depletion and COVID-19 transmission models. The incorporation of additional information including staffing levels, absenteeism, and changes in proportions of staff self-isolating before and after the introduction of widespread testing will better inform the impact of screening at a national and international level. Such models will be critical for optimising the impact on occupationally-acquired COVID-19, and reducing the likelihood that hospitals become hubs for sustained COVID-19 transmission.
In the absence of an efficacious vaccine, additional waves of COVID-19 are likely as social distancing rules are relaxed. Understanding how to limit hospital transmission will be vital in determining infection control policy, and retain its relevance when reliable serological testing becomes widely available. Our data suggest that the roll-out of screening programmes to include asymptomatic as well as symptomatic patient-facing staff should be a national and international priority. Our approach may also be of benefit in reducing transmission in other institutions, for example care-homes. Taken together, these measures will increase patient confidence and willingness to access healthcare services, benefiting both those with COVID-19 and non-COVID-19 disease.
Two parallel streams of entry into the testing programme were established and managed jointly by the Occupational Health and Infectious Diseases departments. The first (HCW symptomatic, and HCW symptomatic household contact screening groups) allowed any patient-facing or non-patient-facing hospital employee (HCW) to refer themselves or a household contact, including children, should they develop symptoms suggestive of COVID-19. The second (HCW asymptomatic screening group) was a rolling programme of testing for all patient-facing and non-patient-facing staff working in defined clinical areas thought to be at risk of SARS-CoV-2 transmission. Daily workforce sickness reports and trends in the results of HCW testing were monitored to enable areas of concern to be highlighted and targeted for screening and cluster analysis, in a reactive approach. High throughput clinical areas where staff might be exposed to large numbers of suspected COVID-19 patients were also prioritised for staff screening. These included the Emergency Department, the COVID-19 Assessment Unit, and a number of ‘red’ inpatient wards. Staff caring for the highest priory ‘shielding’ patients (Haematology/Oncology, Transplant medicine) were also screened, as were a representative sample of staff from ‘Amber’ and ‘Green’ areas. The personal protective equipment (PPE) worn by staff in these areas is summarised in Table 5. Inclusion into the programme was voluntary, and offered to all individuals working in a given ward during the time of sampling. Regardless of the route of entry into the programme, the process for testing and follow-up was identical. Wards were closed to external visitors.
We devised a scoring system to determine the clinical probability of COVID-19 based on symptoms from existing literature (Wang et al., 2020; Giacomelli et al., 2020; Table 1). Self-referring HCW and staff captured by daily workforce sickness reports were triaged by designated Occupational Health nurses using these criteria (Table 2). Self-isolating staff in the medium and low probability categories were prioritised for testing, since a change in the clinical management was most likely to derive from results.
Self-isolation and household quarantine advice was determined by estimating the pre-test probability of COVID-19 (high, medium or low) in those with symptoms, based on the presence or absence of typical features (Tables 1–2). Symptom history was obtained for all symptomatic HCWs at the time of self-referral, and again for all positive cases via telephone interview when results became available. All individuals who had no symptoms at the time of testing were followed up by telephone within 14 days of their result. Pauci-symptomatic individuals were defined as those with low-probability clinical COVID-19 criteria (Table 2).
Testing was primarily undertaken at temporary on-site facilities. Two ‘Pods’ (self-contained portable cabins with office, kitchen facilities, generator and toilet) were erected in close proximity both to the laboratory and main hospital. Outside space was designed to enable car and pedestrian access, and ensure ≥2 m social distancing at all times. Individuals attending on foot were given pre-prepared self-swabbing kits containing a swab, electronically labelled specimen tube, gloves and swabbing instructions contained in a zip-locked collection bag. Pods were staffed by a team of re-deployed research nurses, who facilitated self-swabbing by providing instruction as required. Scenario 1 PPE (Table 5) was worn by Pod nurses at all times. Individuals in cars were handed self-swabbing kits through the window, with samples dropped in collection bags into collection bins outside. Any children (household contacts) were brought to the pods in cars and swabbed in situ by a parent or guardian.
In addition to Pod-based testing, an outreach HCW asymptomatic screening service was developed to enable self-swabbing kits to be delivered to HCWs in their area of work, minimising disruption to the working routine of hospital staff, and maximising Pod availability for symptomatic staff. Lists of all staff working in target areas over a 24 hr period were assembled, and kits pre-prepared accordingly. Self-swabbing kits were delivered to target areas by research nurses, who trained senior nurses in the area to instruct other colleagues on safe self-swabbing technique. Kits were left in target areas for 24 hr to capture a full cycle of shift patterns, and all kits and delivery equipment were thoroughly decontaminated with 70% ethanol prior to collection. Twice daily, specimens were delivered to the laboratory for processing.
The swabbing, extraction and amplification methods for this study follow a recently validated procedure (Sridhar et al., 2020). Individuals performed a self-swab at the back of the throat followed by the nasal cavity as previously described (Our World in Data, 2020). The single dry sterile swab was immediately placed into transport medium/lysis buffer containing 4M guanidine thiocyanate to inactivate virus, and carrier RNA. This facilitated BSL2-based manual extraction of viral RNA in the presence of MS2 bacteriophage amplification control. Use of these reagents and components avoided the need for nationally employed testing kits. Real-time RT-PCR amplification was performed as previously described and results validated by confirmation of FAM amplification of the appropriate controls with threshold cycle (CT) ≤36. Lower CT values correspond to earlier detection of the viral RNA in the RT-PCR process, corresponding with a higher copy number of the viral genome. In 2/1,270 cases, RT-PCR failed to amplify the internal control and results were discarded, with HCW offered a re-test. Sequencing of positive samples was attempted on samples with a CT ≤30 using a multiplex PCR based approach (Quick et al., 2017) using the modified ARTIC v2 protocol (Quick, 2020) and v3 primer set (Artic network, 2020). Genomes were assembled using reference based assembly and the bioinformatic pipeline as described (Quick et al., 2017) using a 20x minimum coverage as a cut-off for any region of the genome and a 50.1% cut-off for calling of single nucleotide polymorphisms (SNPs). Samples were sequenced as part of the COVID-19 Genomics UK Consortium, COG-UK), a partnership of NHS organisations, academic institutions, UK public health agencies and the Wellcome Sanger Institute.
As soon as they were available, positive results were telephoned to patients by Infectious Diseases physicians, who took further details of symptomatology including timing of onset, and gave clinical advice (Table 2). Negative results were reported by Occupational Health nurses via telephone, or emailed through a secure internal email system. Advice on returning to work was given as described in Table 2. Individuals advised to self-isolate were instructed to do so in their usual place of residence. Particularly vulnerable staff or those who had more severe illness but did not require hospitalisation were offered follow-up telephone consultations. Individuals without symptoms at the time of testing were similarly followed up, to monitor for de novo symptoms. Verbal consent was gained for all results to be reported to the hospital’s infection control and health and safety teams, and to Public Health England, who received all positive and negative results as part of a daily reporting stream.
Swab result data were extracted directly from the hospital-laboratory interface software, Epic (Verona, Wisconsin, USA). Details of symptoms recorded at the time of telephone consultation were extracted manually from review of Epic clinical records. Data were collated using Microsoft Excel, and figures produced with GraphPad Prism (GraphPad Software, La Jolla, California, USA). Fisher’s exact test was used for comparison of positive rates between groups defined in the main text. Mann-Whitney testing was used to compare CT values between different categories of tested individuals. HCW samples that gave SARS CoV-2 genomes were assigned global lineages defined by Rambaut et al., 2020 using the PANGOLIN utility (O'Toole and McCrone, 2020).
As a study of healthcare-associated infections, this investigation is exempt from requiring ethical approval under Section 251 of the NHS Act 2006 (see also the NHS Health Research Authority algorithm, available at http://www.hra-decisiontools.org.uk/research/, which concludes that no formal ethical approval is required). Written consent was obtained from each HCW described in the anonymised case vignettes.
Sequencing data have been deposited in GSAID under accession codes EPI_ISL_433989-EPI_ISL_433992, EPI_ISL_434005, EPI_ISL_433489-EPI_ISL_433497. Researchers will be prompted to register and log on to the website to access the datasets (https://www.epicov.org/epi3/frontend#1f1442).
Presymptomatic SARS-CoV-2 infections and transmission in a skilled nursing facilityNew England Journal of Medicine.https://doi.org/10.1056/NEJMoa2008457
Softwareartic-ncov2019 / primer_schemesGitHub.
Characteristics of health care personnel with COVID-19 - United states, February 12-April 9, 2020MMWR. Morbidity and Mortality Weekly Report 69:477–481.https://doi.org/10.15585/mmwr.mm6915e6
WebsiteExclusive: deaths of NHS staff from covid-19 analysedAccessed April 25, 2020.
Asymptomatic transmission, the achilles’ Heel of Current Strategies to Control Covid-19New England Journal of Medicine 19:NEJMe2009758.https://doi.org/10.1056/NEJMe2009758
WebsiteWuhan seafood market pneumonia virus isolate Wuhan-Hu-1 complete genomeAccessed April 26, 2020.
Self-reported olfactory and taste disorders in SARS-CoV-2 patients: a cross-sectional studyClinical Infectious Diseases : An Official Publication of the Infectious Diseases Society of America 26:ciaa330.https://doi.org/10.1093/cid/ciaa330
WebsiteReport 16: role of testing in COVID-19 controlAccessed April 23, 2020.
Immediate and sustained psychological impact of an emerging infectious disease outbreak on health care workersThe Canadian Journal of Psychiatry 52:241–247.https://doi.org/10.1177/070674370705200406
Epidemiology of Covid-19 in a Long-Term care facility in king county, WashingtonNew England Journal of Medicine 382:2005–2011.https://doi.org/10.1056/NEJMoa2005412
WebsiteTo understand the global pandemic, we need global testing – the Our World in Data COVID-19 Testing datasetAccessed April 25, 2020.
WebsiteSurveillance of influenza and other respiratory viruses in the UK to 2019Accessed April 27, 2020.
WebsiteCoronavirus (COVID-19) in the UK 2020Accessed April 25, 2020.
WebsitenCoV-2019 sequencing protocol v2Accessed April 27, 2020.
WebsiteStay at home adviceAccessed April 25, 2020.
WebsiteCOVID-19: management of exposed healthcare workers and patients in hospital settingsAccessed April 25, 2020.
WebsiteReport of the WHO-China joint mission on coronavirus disease 2019 (COVID-19)Accessed April 25, 2020.
WebsiteCOVID–19 Strategy UpdateAccessed April 25, 2020.
Jos WM van der MeerSenior and Reviewing Editor; Radboud University Medical Centre, Netherlands
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Your paper nicely demonstrates the importance of systematic and comprehensive testing of coronavirus infection in healthcare workers. Asymptomatic carriers, albeit a minority, are detected this way, and their contribution to transmission of the virus in the hospital setting can be estimated.
Your submission has been evaluated by a Deputy Editor and a group of Senior Editors, and they considered your work would qualify for fast track. The paper was assigned to me to judge the quality of the manuscript, of the reviews, and the revisions carried out based on the reviews. I must say that I was impressed by the quality and the novelty of the revised paper. Also the seven reviewers had provided very positive, constructive and complete reviews.https://doi.org/10.7554/eLife.58728.sa1
[Editors' note: we include below the reviews that the authors received from another journal, along with the authors’ responses.]
Comments to the author:
1) Section “Research in context”: Abstracts were screened and judged for relevance, but there is no indication of the criteria for this relevance assessment. This should be described.
Many thanks for pointing out this omission. Amended, the line now reads “Abstracts were manually screened and judged for potential relevance. If the manuscript discussed HCW testing for SARS-CoV-2 the contents of the paper were reviewed and, where appropriate, referenced.”
2) Please define HCW, does this include people such as chaplains, for example?
Thank you for querying this. We have now provided a definition of HCW for the purposes of this study in the first paragraph of the methods section of the revised manuscript to improve clarity.
3) What about the other hospital staff who are also near patients but not giving treatment/care? For example, cleaners, porters, and people delivering meal trays? They could also potentially transmit the virus to patients… why was the study limited to HCWs? Later in the paper you use the term "patient-facing", better to get terms all organized earlier in the paper (see my comment 2 above).
Please see our response to point 2 for clarification of the range of HCW that were offered testing. No symptomatic patient-facing or non-patient-facing staff were excluded from testing. No patient-facing or non-patient facing staff working in areas targeted by the asymptomatic screening arm of the programme were excluded from testing. We have altered the wording of the discussion to avoid confusion with regards to this.
4) Abstract methods: "throat/nose self-swab" is not clear and it seems the "/" should be a "+"
We have corrected this as suggested.
5) The methods should be clearer to state that household contacts included adults and children (I had to hunt for the latter detail in the supplement).
We have corrected this as suggested.
6) Are you stating that written informed consent was only obtained from people in the case vignettes rather than everyone in the study? Even if research ethics committee approval was not required for this study in the UK, it would be ETHICAL to get written informed consent from all participants because HCWs are a vulnerable population. Also, you need this formal consent process in order to explain to them the details about data sharing, a huge issue because the UK is still following the GDPR until Dec 31st.
This was run as a service evaluation through occupational health, HCW agreed to be tested. No personal identifiable data has been shared or disclosed so GDPR does not apply.
7) For self-isolation, were people sent home or somewhere else?
Individuals were requested to self-isolate in their usual residence. We have added this information for clarity to the main text.(See SI, paragraph “Results reporting”)
8) What do you mean by "PPE reviews were undertaken"? Was this to ask them if they had PPE, enough PPE, the proper PPE, if they were re-using PPE? If you found troubling answers, did you alert anyone?
We have elaborated on the nature of PPE reviews and interventions enacted in the event of improper PPE use in the revised manuscript. No instances of inadequate provision of equipment were observed, so training and education on correct use of the equipment provided was the only means of intervening.
9) This sentence is confusing: "Given the low seasonal incidence of alternative viral causes of high COVID-19 probability symptoms in the UK26, and the high potential for SARS-CoV-2 hospital exposure during the pandemic, we took the pragmatic approach to allow continued work by these individuals, as long as they had been entirely well for the preceding 48 hours." But these were people who you knew were positive? 48hrs of no symptoms is not enough? Shouldn't they be negative before you allow them to work? Symptoms can wax and wane.
Many thanks for pointing out the clarity needed here. This is a key issue with testing currently asymptomatic HCWs. Given the current evidence suggests that viable, infectious virus is rarely shed longer than 8 days post-onset of symptoms (Wolfel et al., 2020) and that viral RNA can continue to be shed for up to 37 days and probably beyond (Zhou et al., 2020), we took the approach that these HCWs (provided they were currently well) did not pose a significant infectious risk and could continue to work. In essence, we felt that a positive test in this group should not cause us to deviate from the current symptom-based isolation advice. The exception to this rule concerned those who worked with bone marrow transplant recipients, where we followed recent national guidance requiring a negative test prior to a return to work.
Given the long period of post-symptomatic RNA shedding, as screening of asymptomatic HCWs is expanded, clear, national, guidelines will be required to deal with this scenario. Prolonged isolation of HCWs, possibly for many weeks after their illness will have a major impact on the workforce, potentially with no benefit towards infection control.
For clarity, the sentence now reads:
“This approach was based on the following: low seasonal incidence of alternative viral causes of high COVID-19 probability symptoms in the UK26, the high potential for SARSCoV-2 exposure during the pandemic and the potential for prolonged, non-infectious shedding of viral RNA.”
10) Regarding Table S4, tell us more clearly if there was a connection re PPE scenario and your clinical findings. Also, refer back to my comment 8, maybe the scenario did not actually match to the actuality of PPE access/use in the ward (the PPE review)? I need more clarity here. I only see a brief mention about Ward Q for Figure 3.
Thank you for this suggestion. We have now added a more detailed description of the hospital’s routine COVID-19 infection control and PPE practices as a caption to Table
S4 of the revised manuscript. We have also added more detail about ward Q (See Table S2 (previously Table S4) and Results section).
11) Did any of the wards allow external/community visitors?
A line has been added to the methods section highlighting that wards were closed to external visitors.
This is a timely and important manuscript reporting a SARS-CoV-2 surveillance program in healthcare workers (HCWs) at a teaching hospital in the UK. It is very important that they report test-positive HCWs that are not showing clinical symptoms. These results should be considered by all jurisdictions that are designing and implementing SARS-CoV-2 testing protocols for HCWs.
Many thanks to this reviewer for the complimentary assessment of our work.
1) Overarching comment: it is unfortunate that the authors did not appear to investigate exposures of the positive HCW cases in this study. It appears that they conducted telephone interviews to record information on clinical symptoms, but not potential exposures. Without exposure information, it is difficult to know the degree to which positive HCWs were infected at work versus from community or household exposure. Preliminary data reported by the WHO from China suggests that the majority of HCW cases may be from household, not occupational exposure. A recent study from the US CDC suggests a higher proportion of cases in HCWs were in individuals that only had known exposure to SARS-CoV-2-postive cases, but it is only a small portion of HCW cases that may not be representative of the broader US situation.
Many thanks for this important point. However, in order to draw strong conclusions about the predominant sources of HCW infection, an entirely new study would be required. This would be likely to require visits to and sampling of individual households, taking detailed epidemiological data about all members of the household, recurrent visits to follow up development of symptoms and sequencing-based studies of any positive household isolates. In Cambridge UK, many HCWs live in shared accommodation, often with multiple other HCW. This can further complicate conclusions as to work vs household acquisition of infection. An additional concern is that given the wide spectrum of clinical features from asymptomatic – paucisymptomatic – truly symptomatic, even with extremely detailed epidemiological information and sequencing, the direction of transmission may still be difficult to disentangle.
Anecdotally, several HCWs did mention household contacts who also had symptoms or who had previously tested positive for SARS-CoV2. We have included some of this data in the case vignettes for illustrative purposes (see Table 2).
2) I feel this study missed this opportunity to collect these exposure data from a captive patient population. It is not a fatal flaw, just an unfortunate circumstance. They do highlight the future need for this work in the discussion. The study claims to have identified SARS-CoV-2 transmission within the hospital, and the results, particularly sequencing analysis, support this to some degree. However, the story would be stronger with epidemiologic reporting of exposure information of positive HCWs. I think this limitation needs to be more clearly stated in the discussion. The authors rightly state that strain identification through sequencing is not a discriminatory method to pin-point viral sources of infection, given the relatively low mutation rate of coronaviruses. It is not clear from the study design/reported methods if tested household contacts were the source of or recipient of the virus to/from the positive HCWs. This is another related piece of information that could have been teased out through epidemiologic investigation.
Thank you for this suggestion. We have added to the discussion to highlight to this study limitation.
3) Abstract Methods: please include the location of the hospital (city, country) to direct the reader to the population of interest for the study beyond HCWs in a teaching hospital.
We have added these details.
4) Abstract Findings: is there room to include the sequencing analysis of the symptomatic HCW screening group and symptomatic household contacts?
Many thanks for this suggestion. Unfortunately due to limits on the word count in the Abstract, and our concern to emphasise the nuanced nature of interpretation of the sequencing data, we decided to limit discussion of this data in the Abstract. We state:
“Viral genome sequencing showed that the majority of HCWs had the dominant lineage B.1.”
We discuss this data in more detail in the remainder of the manuscript.
5) Methods: please clarify how the two parallel streams of testing are integrated. The results make reference to "RT-PCR failure rate" that is presumably tied to this, but it is never explained (main text or supplementary info).
Many thanks for this suggestion; we have updated the methods section as requested to describe entry into and subsequent integration of the two parallel streams of testing
(See “Staff screening protocols”). We also define “RT-PCR failure” as “failed internal control”.
6) There are a number of pieces of "Methods" presented in the results – some specific examples listed below.
Many thanks for pointing these out; we have addressed all concerns below
7) Methods/Results: risk of anticipated exposure to COVID-19 (red, amber, green) – definitions of what constitutes a high, medium and low risk ward should be provided in the main text (of the Methods section, not results), or at least referred to Table S3 in the Supplementary Materials. The sentence in Staff Screening Protocols alludes to these risk areas, but this must be made explicit to the reader. It would also be nice if Table S3 could provide concrete examples of types of wards/hospital service areas that fell into the different criteria. I realize that part of the criteria consider whether or not the area has held a confirmed SARS-CoV-2, but presumably some types of wards/services are always at higher risk and would play into this categorization. Table S4 gets more into these details, but still is not a concrete description – just general terms like "areas where aerosol-generating procedures are carried out". The reason I feel this is important is to provide some guidance on the types of hospital wards/services that are at higher risk, not just those called "red" or "amber" based on the definitions. Maybe this is better provided in the descriptive results – the types of wards/services falling into each risk category based on the definitions.
We have now included examples of the service areas that fall into red/amber/green categories in Table S1 (previous Table S3), since this table additionally includes details of PPE used in each area. We have explicitly referenced Tables S1-S2
(previously Table S3-S4) in the Methods section as requested.
8) Results – please define what is meant by "RT-PCR failure rate" – it is not clear to the reader – please define in the methods. This might relate to the parallel streams of testing, but this is not clear from the main text and must be explained.
As detailed above, we have now explained this in the methods section.
9) Results text referring to Figure 1—figure supplement 1: please provide the linear regression coefficient and P-value for this lack of correlation between CT values and days since symptom onset in this analysis. Please provide a brief description of the linear regression in the Supplementary Info (how was the linearity of a continuous predictor checked with the outcome? How were model assumptions checked?
Many thanks for this suggestion. We note that reviewer 6 also asked about the appropriateness of the use of a linear regression model to analyse these data. Reviewer 4 asked for further details about the relationship between symptom group (asymptomatic/early symptomatic/late-post symptomatic) and Ct values.
We agree that the underlying assumptions of normality in the linear regression model may not apply to our data. A better approach would be a non-parametric analysis of between-group differences. (see Figure 1—figure supplement 1).
We have therefore re-analysed the data originally presented in Figure 1—figure supplement 1, grouping positive individuals according to days since symptom onset in line with the definitions we used in the text (symptomatic ≤7 days, symptomatic >7 days). We have presented these data using a similar layout to Figure 1. Between the early- and late- symptomatic groups we did not find a relationship between symptom duration and Ct value using nonparametric testing (Mann-Whitney test).
Note that this reanalysis does not change our original findings, of a lack of a clear relationship between Ct value and symptom duration.
The reviewer may also wish to note our response to reviewer 4 who asked us to assess any relationship between our symptom-based risk classification (Low/Medium/High risk) and Ct value for individuals testing positive. We were unable to detect a difference between risk groups, but were underpowered to test for this due to the relatively small number of individuals in the medium/low risk groups. We have made an adjustment to the manuscript to mention this, without over-interpreting these findings.
10) Results section (Three subgroups of…): the sentences presenting statistical results in the last third of this paragraph are very confusing. I had great difficulty interpreting them. They could be better presented in a Table.
We agree with the reviewer that these findings were reported in a confusing manner. We have simplified the description of these data in the text and improved the crossreferencing to Figure 2.
11) This links to comments on Figure 3, which would be better to include in this table than a figure. It would allow for inclusion of 95% CIs, which I strongly recommend, and would avoid the confusion of the number of tests scale from Figure 3 that makes it look like most wards has zero test positives. It also avoids any issue with red-green colour blindness.
We thank the reviewer for this suggestion. For clarity, we have added a supplementary table (now Table S5) that also shows the data from this figure.
12) Results: define PPE for the first time.
We have corrected this omission.
13) Results: what kind of ward was Ward Q?
Thank you for querying this. We have provided further detail on the nature of ward Q in the revised manuscript.
14) Table 1: title caption should be changed to indicate that these results include total number of tests performed and number/proportions of positive results. These proportions should include measures of dispersion (preferably a 95% confidence intervals).
We thank the reviewer for this suggestion, and have amended the caption for Table 1.
However, we do not agree that the proportions should include measures of dispersion – this table gives a descriptive account of the number/proportions of positive results we observed. Please also see our reply to reviewer 6, where we justify our use of Fisher’s exact test as opposed to the use of confidence intervals for binomial proportions.(see
15) Table 2: clinical vignettes. While these are somewhat interesting, they could provide more information, particularly about possible SARS-CoV-2 exposures. For each one, I wanted to know what their occupational versus home/community exposure risks were. Without this, I don't find this section as informative. It appears from the study that detailed exposure information was not sought from the study participants. This is an unfortunate omission as this information is vitally important to understand the true occupational transmission of SARS-CoV-2 to HCWs. The sequence analysis suggests commonality, but also variability in that Ward F has multiple viral strains.
Many thanks for this suggestion. We have revised this section to make it more informative in general (as also detailed above in response to “overarching comment”), in particular giving more detail about potential exposures (see Table 2).
16) Table 3: positive tests by symptom-based probability. The title of this table should refer the reader to Table S2 for the definitions of the clinical probability categories. This table should also include measures of dispersion (95% CIs).
We have adjusted the title of this table to refer to Table S4 (previously S2) as suggested. Since the proportions given show the distribution of the positive test results within each group according to clinical risk group and not the percentage of all test results for each screening group (i.e. note that the row totals sum to 100%) we feel it is not appropriate to apply measures of dispersion. We have amended the figure legend to clarify this.
17) Figure 1: CT should be defined.
We have corrected this omission.
18) Figure 2: Please clarify this figure – probably better presented as a table rather than the pie charts. This would allow inclusion of 95% CIs and avoid any red/green colour confusion for anyone with color blindness. Further, the info presented in the center pie chart is described will in the text (including the categorization), but is not well-described in the figure caption (and it is confusing). If kept as a pie chart (not my preference), it should be made clear that the left and right pie charts for the proportions in high/medium/low are for either ii or iii, respectively, with labels or description in the figure. Second, what to the proportions in the center pie chart represent? It looks like the proportion of all positive tests.
Many thanks for this suggestion. We have updated the colours in the pie chart to avoid difficulty for anyone with colour blindness. We have removed the confusing information about the central pie chart and referenced the main text, and clarified other information highlighted above in the figure legend. We did not think it appropriate to add 95% CIs, since the numbers described are actual numbers of HCW. (See Figure 2 and Results section.)
19) Figure 4: what is the scale on the Y-axis – number of positive tests? What is the "reactive screening group"? This term is partially defined in the Results but should be defined in the figure for clarity.
We have amended the Y axis label on Figure 4 to clarify that this is the number of tests returning positive for SARS-CoV-2. In line with the reviewer recommendation, we have expanded the figure legend to clarify the meaning of each group.(see Figure 4)
20) Supplementary Material: Tables S1, S2, S3 and S4 need titles to better orient the reader. They are referred to and are important to the case and risk definitions in the main text. One cannot go to the tables and understand what they include without Table titles without inferring it from the text. For example, Table S1 refers to Major and Minor symptoms, but this is not evident from looking at the table itself.
Thank you for this suggestion. We have added titles to Tables S1-S4 to improve clarity.
21) Table S3: the main text refers to the alignment of risk categorization of red = high, amber = medium, and green = low risk in the Results section. This makes sense, but in Table S3 (again, requires a title), should explicitly list that red = high, etc.
Thank you for this suggestion. We have added labels to the columns.
22) Table S5 caption: this should include details as to which isolates were included (e.g., all isolates from the study? All HCW isolates? What about household exposure isolates?).
Thank you for this suggestion. We have added updated the Table S5 caption to clarify this. Isolates from all HCWs and household contacts in the study are included in the table.
23) Figure 4—figure supplement 1: the colours are too similar to be able to distinguish the phylogenetic relationships of virus sequences from different wards. I know this is supplementary material and not in the main text, but this is very interesting and I am not able to clearly tell them apart. Maybe each could be given a number that is added to the figure, or some other way to distinguish them.
Many thanks for this helpful suggestion. We have now added ward letters to the phylogenetic tree to aid clarity. (see Figure 4—figure supplement 1)
1) Abstract: percentages are quoted to spurious accuracy. Of the 31, one could not be followed-up. This should, I suggest, be reported in Abstract and the relevant % is then 12/30 or 40%. Likewise, 75% is in fact 9/12 (say so) and provide numerator/denominator for 55% (currently reported as 54.8%). Likewise, 15% (not 15.4%) and 4/52 (8%).
We have adjusted the manuscript according to these suggestions.
2) Introduction: perhaps disconcerting that asymptomatic HCW onward transmission accounts for "only" 16-23% on top of isolation based on symptoms. Presumably depends on PPE-provision – might be worth mentioning this?
The Imperial College, London, UK report from which this estimate derives from does not specifically mention the role of PPE-provision in their model. They do however mention the importance of 24 hour turnaround for results. The sentence has been amended so it now reads “weekly testing of asymptomatic HCWs could reduce onward transmission by 1623% on top of isolation based on symptoms, provided results are available within 24 hours.”
3) Results: since 21 individuals underwent repeat testing but reasons sum to 3+11, fewer than 21. Please clarify.
We apologise for the lack of clarity here. 21 individuals underwent retesting. There was a range of reasons, including the two specifically cited in the text (evolving symptoms and “medium” clinical COVID-19 probability). The other ten individuals included those originally screened due to symptoms then re-captured as part of the asymptomatic ward screening programme. We have clarified the text as requested.
4) Results. p12, last para, sentence 1: terse – do you mean ward (green, amber, red) or household contact or what (presumably not that location of swabbing was material; but throat versus nasophyngeal could be meant – tho' I doubt it)... please rephrase.
Many thanks for spotting this omission – we have added that this means “ward”.
5) Results: I would mention FIRST that one HCW could not be contacted to obtain further history and then provide % out of 30 who were re-contacted, ie 20%, 40%, 40%.
Many thanks for this suggestion – we have updated all data and figures as suggested (throughout the paper).
6) Results, p14, para 1; NOW I'm lost as I looked at Table 1 expecting to find “green” 6/310 but I read 7/454; and for “red”, I read 20/466. WHERE do the numbers in line 3 come from – e.g., an earlier draft of paper before screening was completed?
We apologise to the reviewer for the confusion. In the text identified, we report outcomes from asymptomatic HCW screening in 21 specific clinical areas identified on the basis of risk or sickness patterns. Asymptomatic screening tests were also performed on an opportunistic basis on a small number of individuals working outside these areas, as well as in a subsequent intensified manner on ward F and ward Q after identification of clusters of positive cases on these wards.
We report the outcomes of this intensified screening programme on wards F and Q in Figure 4. In Figure 3, tests from HCWs working across >1 area are included in the statistics for each area where they worked, whereas for table 1 individual test results are aggregated according to the area of highest risk an individual HCW was exposed to. In table 1, we report all asymptomatic screen test results.
This explains the differences which the reviewer has noted. Since we recognise that this is confusing, we have added sentences of clarification the legends.
7) Results: I'd place the sentence "It was subsequently found... " at the end of the paragraph because the two persons therein confuse the reader who is trying to keep track of testable numbers, especially having already been disconcerted by inability to track in the preceding paragraph!
Many thanks, amended as suggested for improved readability.
8) Results: there were prior reasons to investigate wards F (high sickness rate and Q (PPE concerns). However, looking at Figure 3, I'd like to know more about locations O and T (both red) which had similar numbers tested as in F and Q but different % positives (~ 15% vs 5%).
Ward O was listed for further cluster investigation after the study period. We have now highlighted this in the Results section, and included a description of the ward type. Ward T was not prioritised for further cluster investigation due to the lower rate of positive tests.
9) Results: I make the count of positive tests in ward Q to be 7+1+1 = 9 but only four were sequenced. Why not the other FIVE? Please add explanation.
We have added detail that other isolates could not be sequenced due to a sample CT value >30.
10) 38.7% => 40% (12/30); data... suggest... probable symptoms (cf probability symptoms).
We thank the reviewer for noting these and have adjusted the manuscript accordingly (throughout).
11) Discussion, p17+18+19: I count 3 recommendations in main para on p18, fourth in last line of p17; two more in main para on p18; and final = 7th at top of page 18 and 8th in the next paragraph. I consider that any paper that makes 8 well-grounded and cogent recommendation from a single study has done a remarkable piece of work.
Very many thanks for this highly complementary assessment of our work.
12) Discussion might consider whether a corresponding swab-test study in patients would be warranted; or for care-home residents and staff.
Many thanks for this helpful suggestion. The sentence “Our approach has potential applications outside the hospital and may be of benefit in reducing transmission in other institutions, for example care-homes.” has been added to the closing paragraph to highlight this point.
13) Tables: Numbers don't match some cited in text.
We believe the reviewer is referring to some of the apparent discrepancies noted by reviewer 2 between the test results reported in Table 1, Figure 3, and the Results section. The reviewer may wish to note our response to reviewer 2 above in which we explain that the numbers are correctly reported, but in a manner that was initially confusing due to the different populations being studied. We have amended the text, the figure and the table legends to make this clearer. We have checked all other figures in the text and table and note that they are all in alignment.
14) Table 3. Should corresponding probability-ratings be shown for participants who tested negative? Seem to give only half the story. Answers cannot be got by subtraction between Tables 1 and 3.
We thank the reviewer for this suggestion. For those individuals who tested negative, we did not undertake the detailed symptom scoring required to assign a clinical risk group. We therefore regret that we are unable to provide this data.
15) Figure 1. The blue dots appear to be differently distributed and could almost be a mixture of two distributions. Does this notion have any subject-matter traction, eg use different blue plotting symbols for HCW (dot) and their symptomatic household contacts (square)?
We thank the reviewer for this interesting observation. We re-evaluated data in Figure 1 according to reviewer’s suggestion. A Mann-Whitney test of the differences in CT values between symptomatic household contacts and symptomatic HCWs was not significant (p=0.06), which may reflect the low number of symptomatic household contacts testing positive (n=4).
Nevertheless, further consideration of Ct values according to subgroups is clearly interesting, and we have compared values according to duration of symptoms as highlighted in our response to reviewer 2.
1) Thank you for this very important paper. We do really need data driven recommendations on screening of health care workers to reduce nosocomial infections of SARS-CoV-2. The major problem that needs to be address is that only of the 31 in your "asymptomatic" group are actually asymptomatic – the remainder are either postsymptomatic or pre-symptomatic. This needs to be much more clearly defined throughout the paper as it has major implications for infection control.
We thank the reviewer for their complimentary assessment of our work. The individuals in the HCW asymptomatic screening group were asymptomatic at the time of screening, however could be split into distinct sub-groups, with some experiencing symptoms prior to their test. We agree that this is unclear and have greatly clarified as requested, in particular in the section “Three subgroups of SARS-CoV-2 positive asymptomatic HCW”.
2) The title implies that you have built a screening program but you really have done a point prevalence survey and have not actually made recommendations as to what a screening program should look like – most importantly how often HCW should be tested. I would suggest you use your results to make recommendations on this topic or would suggest changing the title (and the objective) to reflect this as a point prevalence study to understand symptom screening and high risk wards for COVID-19 in HCW.
We respectfully disagree with the reviewer’s point. We have built a screening programme, and have described in detail how this operates in our supplementary information. The information presented is not a point prevalence survey. Rather, data were gathered over an initial 3-week period, and simply reflect the output of the programme so far. We have nevertheless changed the title as discussed above.
3) In the methods section, please define "asymptomatic" and "pauci-symptomatic" and I would recommend defining "pre-symptomatic" and "post-symptomatic" groups as well for clarity – and maintain the use of these group names throughout as your "asymptomatic" groups is not really asymptomatic – they are only without symptoms at the time of testing. Also, as these are terms that are used throughout and are important for understanding your results, please explain the schematics used for the high probability COVID symptoms and the color coding system for the wards in the text of the methods section – instead of making the reader go to the figures to understand these categorizations. Also please explain the rationale behind including the location-based designations. In the methods, you should also define Ct values and how you will use these to estimate the viral load.
Thank you for these comments. We have now defined “pauci-symptomatic” in the Methods section. We have mentioned schematics and colour coding in the methods. Location-based designations are ward locations which have been given a code so that the ward, staff and patients are not identifiable.(see Table S5). Details of CT values and relation to viral loads now in Methods.
4) In the Results section, I think it would be incredibly useful to know how many of the post-symptomatic HCW had low Ct values and culturable virus. This has incredibly important implications for return to work practices for HCW and is a point I was really hoping you would touch on. It appears that you have this data but it is not being presented in this way. Additionally, please provide data to the Results section on viral loads (either ranges of Ct values or median Ct values for each group). I think the subgroups of the asymptomatic HCW should be a major focus here as this incredibly important information for infection control programs. Additionally, please add some information in your Results section about the median amount of time that it took for HCW to perform the tests and for results to return and what your facility's specific recommendations were for those with positive results.
Many thanks for these interesting points. We were not able to collect data on culturable virus as at the time of the study we lacked the facilities to perform such analyses. We have referenced a previous study that examined this question (Wolfel et al., 2020). As described in our response to reviewer 2, we have now analysed Ct values according to subgroups. Information about time taken by HCW to perform the test was not collected, however sampling took place within a 10-minute appointment. The time from sample arrival time in the laboratory to a result being provided to the HCW was 12-36 hours, which was dependent on when each sample was taken during the working day.
5) I am a little lost in the discussion as to what your major conclusions and recommendations are based on your results. I would personally love to see Ct values for HCW by symptom groups – presymptomatic, symptomatic at the time of testing (maybe by high, medium, low clinical probability), and post-symptomatic. This would really highlight differences in infectivity and provide a great framework for the need for intermittent testing of all HCW.
Please see our response to reviewer 2, who also raised the same points.
In summary, I think this is really great work and could answer some really important questions about how all hospitals can prevent nosocomial SARS-CoV-2 infections in the future.
Many thanks again to this reviewer for their complimentary assessment of our work.
Congratulations on setting up a staff screening program in the midst of a pandemic.
1) I found it difficult to understand whether the primary aim of your manuscript is to describe the prevalence of carriage of SARS CoV2 amongst asymptomatic healthcare workers or to describe your infection control/occupational health program. Much of the manuscript is taken up with the latter, but I found it relatively uninteresting in the absence any data related to the success (or otherwise) of what you did. I suggest you omit it and focus on your finding that truly asymptomatic carriage amongst healthcare workers is rare, even during a major pandemic.
Thank you for this suggestion. We have now described the rate of true asymptomatic infection in the Results sections of the revised manuscript and at the beginning of the Discussion. It is important to recognise, however, that the milder symptoms experienced by individuals testing positive for SARS-CoV-2 may be unrelated to COVID-19 in some instances, even if temporally related. The fact that only 15% of symptomatic HCWs tested positive for SARS-CoV-2 in our study serves as a reminder that there are many potential alternative causes of the symptoms associated with COVID-19, which may occur concurrently with SARS-CoV-2 infection. We have added a sentence to the discussion to this effect.
2) The title of the manuscript is misleading: A comprehensive healthcare worker (HCW) screening programme identifies asymptomatic HCWs as a source of nosocomial SARSCoV-2 transmission. You have identified asymptomatic carriers of SARS-CoV2 amongst healthcare workers but you have not demonstrated that they are a source of transmission.
Many thanks for this point. We have changed the title as suggested to: “A comprehensive healthcare worker SARS-CoV-2 screening programme to detect asymptomatic infection: a prospective cohort study.”
3) Please present all results in the Results section, rather than the Discussion section.
We have re-checked our manuscript, and found that the only data highlighted in the discussion that is not directly highlighted in the text of the Results section is: “57% of HCWs from asymptomatic screening group were pauci- or asymptomatic…”. However, this number is directly derivable from Figure 2.
4) Please describe the inclusion criteria for the asymptomatic screening program – what were the criteria for being defined as "staff" on those wards. Was it voluntary? What percentage of staff were actually tested?
We have added detail that testing was a voluntary process, and offered to all individuals working in a given ward during the time of sampling.
5) What were your criteria for "pauci-symptomatic"?
Please see our response to the reviewer’s first question above.
6) Please give the data for the prevalence of SARS-CoV2 positivity amongst staff who were both truly asymptomatic and had not been symptomatic previously (unless symptom onset was >24 days prior to testing). These data should also be given in the Abstract. If I have understood your results correctly it may be as low as 6/1032 (approximately 0.6%)
Thank you for this suggestion. We have added this to the Results section of the revised manuscript.
6) When reporting prevalence it is important for the reader to understand the context. For those who are not familiar with the situation in the UK (or even better in your region) at the time of your study (even those who are may not remember the situation when they read your paper in the future) please give some data on the situation in the community. I appreciate that low rate of testing in the UK makes this difficult but data such as COVID deaths per head of population will at least provide some context. You should also mention in your discussion that your prevalence (at least in green zones) may simply be the community prevalence or may be lower than the community rate. The incidence in asymptomatic pregnant women in New York City was 13.7%. (https://www.nejm.org/doi/full/10.1056/NEJMc2009316). Without knowing the prevalence of positivity in asymptomatic patients it is difficult to know whether testing all patients or all staff is likely to be a more effective measure.
Many thanks for highlighting this important point. We had already emphasised the issue with the lack of point-prevalence data from asymptomatic individuals in similar healthcare settings in our discussion, however we have now modified this point to read:
“This figure is difficult to contextualise due to paucity of point-prevalence data from asymptomatic individuals in similar healthcare settings or the local community.” For contrast, we discuss a publication examining prevalence of SARS-CoV2 infection in asymptomatic residents of a care home during an outbreak.
Furthermore, we have amended an additional sentence to read: “This may be partly explained by the relatively low rates of infection in the East of England in comparison to other areas of the UK (cumulative incidence 0.17%, thus far).”
With regards to prevalence in asymptomatic patients, this would be important information, however such data is not yet available.
7) Similarly more detail on infection control measures in your wards would be helpful: were FFP3 masks routinely fit tested, were they re-used, ventilation of wards (air changes per hour), use of high flow nasal oxygen and non-invasive ventilation, availability of single rooms and airborne infection isolation rooms, measures to limit staff-staff transmission? This is particularly important given your finding that the prevalence of positivity was higher in red zone wards.
Thank you for this suggestion. We have included a more detailed description of the hospital’s COVID-19 infection control and PPE practices as a caption to table S2 (formerly table s4) of the revised manuscript.
8) The discussion needs to be tightened up significantly so that it focuses on the findings of the study and does not overstate implications of those findings. Examples include: First two paragraphs of the discussion are not based on your findings or any data and are possibly specific to the UK. My hospital runs on-site PCR assays five times daily, without the need for a "substantial collaboration". It would be better to present your major findings first.
Thank you for this feedback. We have restructured the Discussion section, and removed UK-centric content. We have chosen to highlight the utility of collaborative efforts in enabling rapid development and upscale of a HCW screening programme in the context of limited pre-existing clinical laboratory capacity, as this is a challenge that we believe will be shared by many other institutions internationally.
9) "Our data clearly demonstrate that a relentless focus on the testing of individuals fitting a strict clinical case definition for COVID-19 will inevitably miss a substantial burden of asymptomatic and pauci-symptomatic disease". I would suggest that a prevalence of <1% (after you exclude previously symptomatic staff) is better described as small but possibly important, rather than substantial.
We have removed the words “a substantial burden of” a suggested.
10) "The value of this approach is illustrated by our detection of a cluster of cases in ward F, where the consequences of uncontrolled staff-to-patient transmission had the potential to cause substantial morbidity and possible mortality". I could not find any data demonstrating staff-to-patient transmission.
Thank you for questioning this. We have now re-worded this section of the discussion to improve clarity, as follows “The value of this approach is illustrated by our detection of a cluster of cases in ward F, where the potential for uncontrolled staff-to-staff or staff-to-patient transmission could have led to substantial morbidity and mortality in a particularly vulnerable patient group”.
11) “Coronaviruses generally have very low mutation rates (~10-6 per site per cycle)28, with the first reported sequence of the current pandemic only published on 12th January 2020. Whilst viral genomics is a powerful tool in the investigation of potential transmission events when combined with epidemiological data, given that SARS CoV-2 was only introduced into the human population in late 2019, there is at present a lack of diversity in circulating strains. As the pandemic unfolds and more detailed epidemiological and genome sequence data from patient and HCW clusters are generated, real-time study of transmission dynamics will become an increasingly important means of informing disease control responses. Prospective epidemiological data will also inform whether hospital staff are more likely infected in the community or at work, and may identify risk factors for the acquisition of infection, such as congregation in communal staff areas or inadequate access to PPE".
This seems a little speculative and I can't really see the relevance to your findings. You argue that in the face of falling community transmission screening of HCWs becomes more important. If community infection is uncommon then the most likely source of transmission to a HCW is a patient. If so, then doesn't screening of all patients make more sense? This is particularly so as, even when community transmission is relatively high (based on the high deaths per capita in UK), the prevalance is approximately 0.6% after you exclude HCWs who were previously symptomatic.
The reviewer is correct in that as cases drop, the ability to identify hospital acquired infections rapidly will be critical to ensuring infection control measure are effective and that to achieve this testing of both HCW and patients will be key. Our intention was to highlight that HCW screening should be implemented as part of a package, which would of course include patient screening. We have updated the text in the discussion as follows:
“However, as the pandemic unfolds and detailed epidemiological and genome sequence data from patient and HCW clusters are generated, real-time study of transmission dynamics will become an increasingly important means of informing disease control responses and rapidly confirming (or refuting) hospital acquired infection. Importantly, implementation of such a programme would require active screening and rapid sequencing of positive cases in both the HCW and patient populations. Prospective epidemiological data will also inform whether hospital staff are more likely to be infected in the community or at work, and may identify risk factors for the acquisition of infection, such as congregation in communal staff areas or inadequate access to PPE.”
12) A lack of data on staff behaviour outside of work should be given as a weakness of the study.
We have added to the limitations section of our discussion: “Our study is limited by the relatively short time-frame, and small number of positive tests and a lack of behavioural data.”
13) Minor comments:
Discussion: Parameterise not parametrise
Abstract, findings. Use of the word "conversely" implies to me that being symptoms >7 days prior to testing and being truly asymptomatic or pauci-symptomatic were mutually exclusive categories. Presumably they are not.
I didn't find the vignettes to be at all useful.
As requested by reviewer 2, we have updated the vignettes to add more detail. We have removed the word “conversely”. (See Table 2.)
This manuscript report the results of COVID-19 screening study among healthcare workers. The study was conducted between 6th and 24th of April 2020 and 10,32 asymptomatic HCWs were screened in a 1300-bed teaching hospital, targeting wards considered to be at higher risk of transmission. There were 31 asymptomatic cases during the study period of which 12 previously experienced symptoms compatible with COVID-19 and 55% of the cases were truly asymptomatic. The authors may wish to consider the following comments:
1) Abstract: I would state in the Methods section whether each HCW was tested once or more during the three week period. In the Findings section I would report n(%) to avoid confusion. I assume the 75% refer to 75% of the 12 HCWs (38.7%) who experiences symptoms.
Many thanks for this suggestion. We have added details as suggested to the Findings section. 21 individuals underwent repeat testing for a variety of reasons, discussed in the first paragraph of the Results section. We have not added these details to the Methods section of the Abstract due to space constraints.
2) The Methods section should provide more details about the study design and procedures with the full description in the appendix. For example, how many tests were done by HCW? Which departments/clinical areas were included in the study? How the data about symptoms were collected? There is a mention of a telephone consultation, but was this done on the day of the test, was there a follow-up? A paragraph providing a summary of these issues in the Methods section would be helpful so that the reader would not need to read the appendix to understand the study.
Thank you for this suggestion. We have added content to the first paragraph of the methods section to further explain these aspects of our study design. The number of HCWs or household contacts that underwent repeat testing is included in the first paragraph of the Results section (See Methods and Results sections)
3) What are the "defined groups" that are mentioned in the data analysis section in relation to Fisher's exact test? Similarly, what are the "categories" that are mentioned in relation to Man-Whitney test? Also, what was/were the variable(s) tested using Man-Whitney test? This section should be more specific to the present study and currently it is very generic and the same text could be used in any study.
We apologise for the lack of clarity regarding the statistical tests applied. In line with the suggestion made by the reviewer, we have adjusted the methods section accordingly and added clarity as to which tests have been applied and where.
4) Figure 1: I would report the median and interquartile range in each group, in the figure legend. It is suggested that the 95% CIs of the median were calculated, how were they calculated? It is stated "…viral loads were significantly lower for those in the HCW asymptomatic screening group than in those tested due to the presence of symptoms (Figure 1)". Looking at Figure 1, the data suggest that the median in the asymptomatic group is slightly above 30 while the median in the symptomatic group is approximately below 25. I am not sure if I missed something here.
We apologise to the reviewer for this error in the figure legend. The figure does indeed show the median and interquartile range, as the reviewer suggests it should. We have corrected the legend accordingly.
In terms of the relationship between CT value and symptom group, it should be noted that there is an inverse relationship between CT value and viral load (i.e. individuals with higher viral loads will have the presence of virus detected at an earlier cycle in the RT-PCR process and hence have lower CT values). We note that further clarity on this point was also requested by reviewer 4. We have amended the figure legend and the methods to make this clearer (see Figure 1).
5) Figure 3: Were "% of total positives" calculated for total population or within green, amber and red separately?
The % of total positives refers to all positive test results in total population, and not treating green/amber/red separately. We have amended the figure legend for clarity (Figure 1—figure supplement 1).
6) Figure 1—figure supplement 1: This legend of this graph mentions linear regression, and this should be stated in the statistical analysis section. Also, if linear regression were used, why are the data points being reported as median rather than mean? Were there concerns about whether the variable is approximately normally distributed? If yes, how was this handled in the linear regression considering that the assumptions of the linear regression may have been violated?
We thank the reviewer for this observation which is in line with those made by reviewers 2 and 4, and already addressed above. In brief, we agree that linear regression was not the appropriate statistical method given that normality cannot be assumed for these data. We have replaced Figure 1—figure supplement 1 with a non-parametric analysis. This analysis does not change our findings (of a lack of relationship between Ct value and symptom duration). (See Figure 1—figure supplement 1.)
7) I am not sure how informative the p-values from Fisher's exact tests are? The number of positive cases are small and using descriptive statistics may be more appropriate. I suggest that all the comparisons done between cases and non-cases could be presented in on table. This could be either n(%) for binary/categorical variables (such as symptomatic vs asymptomatic and red vs green vs amber etc.) and median (IQR) for continuous non-parametric variables (such as viral load). Such a table would provide a clear summary of the main results and conclusions of the study in one place.
A nice feature of Fisher's exact test is that it can be used with small sample sizes because it does not rely on asymptotic approximations. See, for instance, the “Handbook of Biological Statistics” by John H. McDonald, available at http://www.biostathandbook.com/fishers.html. From the Summary: “Use the Fisher's exact test of independence when you have two nominal variables and you want to see whether the proportions of one variable are different depending on the value of the other variable. Use it when the sample size is small.” For this reason, we have elected to maintain the use of Fisher’s exact test in Results section, but thank the reviewer for their suggestion.
I have to congratulate Rivett et al. for their great work in providing much need justification for testing of healthcare workers during this COVID-19. I just have a few minor comments.
Very many thanks for the complimentary assessment of our work.
1) Is there any available data from the study to address the following questions, as they may provide useful insights to the transmission dynamics:
a) Were there any cases that were tested negative initially, but developed COVID-19 symptoms and were tested positive later?
No HCWs in the symptomatic arm had previously tested negative, likely because prior to the inception of the screening programme none had been screened. The period of screening examined reflected the first three weeks of the programme. We anticipate seeing such individuals as the programme develops.
2) Are HCWs from symptomatic household contacts tested? If so, how many positive HCWs cases are link to household contacts?
HCWs with symptomatic household contacts could be tested simultaneously if requested. In all four of the cases where household contacts tested positive, the HCW who lived with them also tested positive.
3) Did any family members of HCWs with positive tests from the asymptomatic group subsequently develop COVID-19?
Thank you for this question, this information was not collected.
4) It is interesting to note that in your study, appropriately 32% of HCWs with positive tests from the asymptomatic group in this study, had experience symptoms compatible with COVID-19 > 7 days prior to testing. 75% of those had returned to work after 7 days self-isolation according to UK recommendations. Does this strengthen the need to test HCWs before they return to work?
This may strengthen the need to test HCWs prior to their return to work, however given the evidence from Wolfel et al., 2020, it seems that viable virus is not culturable beyond 8 days of symptom onset. Admittedly this was a small study involving healthy participants with mild disease. This would suggest that the current 7 days of isolation after onset of symptoms is relatively safe from an infection control standpoint. On the basis of this evidence, we believe that we are likely to be detecting prolonged shedding of non-viable viral RNA. More studies are clearly required to elucidate how long viable virus can actually be shed to better inform screening/return to work policies. This is now fully considered in the Discussion.https://doi.org/10.7554/eLife.58728.sa2
- Michael P Weekes
- Stephen Baker
- Ian G Goodfellow
- Ian G Goodfellow
- Paul J Lehner
- Kenneth G C Smith
- Paul J Lehner
- Ian G Goodfellow
- Stephen Baker
- Michael P Weekes
- Nicholas J Matheson
- Nicholas J Matheson
- John R Bradley
- M Estee Torok
- Afzal Chaudhry
- Gordon Dougan
- M Estee Torok
- Richard J Samworth
- Richard J Samworth
- Jamie Young
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
This work was supported by the Wellcome Trust Senior Research Fellowships 108070/Z/15/Z to MPW, 215515/Z/19/Z to SGB and 207498/Z/17/Z to IGG; Collaborative award 206298/B/17/Z to IGG; Principal Research Fellowship 210688/Z/18/Z to PJL; Investigator Award 200871/Z/16/Z to KGCS; Addenbrooke’s Charitable Trust (to MPW, SGB, IGG and PJL); the Medical Research Council (CSF MR/P008801/1 to NJM); NHS Blood and Transfusion (WPA15-02 to NJM); National Institute for Health Research (Cambridge Biomedical Research Centre at CUHNFT), to JRB, MET, AC and GD, Academy of Medical Sciences and the Health Foundation (Clinician Scientist Fellowship to MET), Engineering and Physical Sciences Research Council (EP/P031447/1 and EP/N031938/1 to RS),Cancer Research UK (PRECISION Grand Challenge C38317/A24043 award to JY). Components of this work were supported by the COVID-19 Genomics UK Consortium, (COG-UK), which is supported by funding from the Medical Research Council (MRC) part of UK Research and Innovation (UKRI), the National Institute of Health Research (NIHR) and Genome Research Limited, operating as the Wellcome Sanger Institute
The CITIID-NIHR COVID-19 BioResource Collaboration
Principal Investigators: Stephen Baker, John Bradley, Gordon Dougan, Ian Goodfellow, Ravi Gupta, Paul J Lehner, Paul A Lyons, Nicholas J Matheson, Kenneth GC Smith, M Estee Torok, Mark Toshner, Michael P Weekes
Infectious Diseases Department: Nicholas K Jones, Lucy Rivett, Matthew Routledge, Dominic Sparkes, Ben Warne
SARS-CoV-2 testing team: Josefin Bartholdson Scott, Claire Cormie, Sally Forrest, Harmeet Gill, Iain Kean, Mailis Maes, Joana Pereira-Dias, Nicola Reynolds, Sushmita Sridhar, Michelle Wantoch, Jamie Young
COG-UK Cambridge Sequencing Team: Sarah Caddy, Laura Caller, Theresa Feltwell, Grant Hall, William Hamilton, Myra Hosmillo, Charlotte Houldcroft, Aminu Jahun, Fahad Khokhar, Luke Meredith, Anna Yakovleva
NIHR BioResource: Helen Butcher, Daniela Caputo, Debra Clapham-Riley, Helen Dolling, Anita Furlong, Barbara Graves, Emma Le Gresley, Nathalie Kingston, Sofia Papadia, Hannah Stark, Kathleen E. Stirrups, Jennifer Webster
Research nurses: Joanna Calder, Julie Harris, Sarah Hewitt, Jane Kennet, Anne Meadows, Rebecca Rastall, Criona O,Brien, Jo Price, Cherry Publico, Jane Rowlands, Valentina Ruffolo, Hugo Tordesillas
NIHR Cambridge Clinical Research Facility: Karen Brookes, Laura Canna, Isabel Cruz, Katie Dempsey, Anne Elmer, Naidine Escoffery, Stewart Fuller, Heather Jones, Carla Ribeiro, Caroline Saunders, Angela Wright
Cambridge Cancer Trial Centre: Rutendo Nyagumbo, Anne Roberts
Clinical Research Network Eastern: Ashlea Bucke, Simone Hargreaves, Danielle Johnson, Aileen Narcorda, Debbie Read, Christian Sparke, Lucy Warboys
Administrative staff, CUH: Kirsty Lagadu, Lenette Mactavous
CUH NHS Foundation Trust: Tim Gould, Tim Raine, Ashley Shaw
Cambridge Cancer Trials Centre: Claire Mather, Nicola Ramenatte, Anne-Laure Vallier
Legal/Ethics: Mary Kasanicki
CUH Improvement and Transformation Team: Penelope-Jane Eames, Chris McNicholas, Lisa Thake
Clinical Microbiology & Public Health Laboratory (PHE): Neil Bartholomew, Nick Brown, Martin Curran, Surendra Parmar, Hongyi Zhang
Occupational Health: Ailsa Bowring, Mark Ferris, Geraldine Martell, Natalie Quinnell, Giles Wright, Jo Wright
Health and Safety: Helen Murphy
Department of Medicine Sample Logistics: Benjamin J. Dunmore, Ekaterina Legchenko, Stefan Gräf, Christopher Huang, Josh Hodgson, Kelvin Hunter, Jennifer Martin, Federica Mescia, Ciara O’Donnell, Linda Pointon, Joy Shih, Rachel Sutcliffe, Tobias Tilly, Zhen Tong, Carmen Treacy, Jennifer Wood
Department of Medicine Sample Processing and Acquisition: Laura Bergamaschi, Ariana Betancourt, Georgie Bowyer, Aloka De Sa, Maddie Epping, Andrew Hinch, Oisin Huhn, Isobel Jarvis, Daniel Lewis, Joe Marsden, Simon McCallum, Francescsa Nice, Ommar Omarjee, Marianne Perera, Nika Romashova, Mateusz Strezlecki, Natalia Savoinykh Yarkoni, Lori Turner
Epic team/other computing support: Barrie Bailey, Afzal Chaudhry, Rachel Doughton, Chris Workman
Statistics/modelling: Richard J Samworth, Caroline Trotter
Human subjects: As a study of healthcare-associated infections, this investigation is exempt from requiring ethical approval under Section 251 of the NHS Act 2006 (see also the NHS Health Research Authority algorithm, available at http://www.hra-decisiontools.org.uk/research/, which concludes that no formal ethical approval is required). Written consent was obtained from each HCW described in the anonymised case vignettes.
- Jos WM van der Meer, Radboud University Medical Centre, Netherlands
© 2020, Rivett et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.