Investigating Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Surface and Air Contamination in an Acute Healthcare Setting During the Peak of the Coronavirus Disease 2019 (COVID-19) Pandemic in London

Abstract

Background

We evaluated severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) surface and air contamination during the coronavirus disease 2019 (COVID-19) pandemic in London.

Methods

Prospective, cross-sectional, observational study in a multisite London hospital. Air and surface samples were collected from 7 clinical areas occupied by patients with COVID-19 and a public area of the hospital. Three or four 1.0-m3 air samples were collected in each area using an active air sampler. Surface samples were collected by swabbing items in the immediate vicinity of each air sample. SARS-CoV-2 was detected using reverse-transcription quantitative polymerase chain reaction (PCR) and viral culture; the limit of detection for culturing SARS-CoV-2 from surfaces was determined.

Results

Viral RNA was detected on 114 of 218 (52.3%) surfaces and in 14 of 31 (38.7%) air samples, but no virus was cultured. Viral RNA was more likely to be found in areas immediately occupied by COVID-19 patients than in other areas (67 of 105 [63.8%] vs 29 of 64 [45.3%]; odds ratio, 0.5; 95% confidence interval, 0.2–0.9; P = .025, χ2 test). The high PCR cycle threshold value for all samples (>30) indicated that the virus would not be culturable.

Conclusions

Our findings of extensive viral RNA contamination of surfaces and air across a range of acute healthcare settings in the absence of cultured virus underlines the potential risk from environmental contamination in managing COVID-19 and the need for effective use of personal protective equipment, physical distancing, and hand/surface hygiene.

SARS-CoV-2, COVID-19, transmission, air contamination, surface contamination

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has rapidly spread around the world since it emerged in late 2019, resulting in a coronavirus disease 2019 (COVID-19) pandemic [1]. Evidence from SARS, influenza, and SARS-CoV-2 suggests droplet and contact spread as primary transmission routes, with evidence of airborne spread during aerosol-generating procedures (AGPs) [1, 2].

Hospital-onset COVID-19 infection (HOCI) has been reported and is probably linked to ineffective implementation of infection prevention and control measures [1, 3–5]. The transmission dynamics in healthcare environments are unclear and likely to be multifactorial. Contaminated surfaces and air have been a key part of the transmission dynamics of SARS, MERS, influenza, and other organisms in hospitals [1, 2, 6]. Laboratory evidence suggests that SARS-CoV-2 can survive on dry surfaces and in aerosols for days to weeks, particularly on nonporous surfaces [7, 8]. Furthermore, SARS-CoV-2 RNA has been detected on surfaces and in the air in hospitals where COVID-19 patients are being treated [9–17].

However, our understanding of the role of surface and air contamination in the transmission of SARS-CoV-2 is limited. To date, most studies have relied on polymerase chain reaction (PCR) and have not attempted to culture virus, thereby limiting the ability to interpret PCR-based detection; have focused on 1 geographical region (Asia); and have included a limited selection of clinical and nonclinical areas with no evidence from operating theatre environments [9, 10, 12, 13, 15, 16]. In mid-April 2020, the United Kingdom experienced the first wave of the COVID-19 pandemic. During this period, there was evidence for HOCI [5]. Therefore, to inform and optimize infection prevention and control interventions, we evaluated SARS-CoV-2 surface and air contamination across a range of clinically relevant locations (including operating theatres) and public areas during the peak of the COVID-19 pandemic in London using both reverse-transcription (RT) PCR and viral culture. We also performed supporting laboratory experiments to assess SARS-CoV-2 viability on surfaces, with associated limits of detection, to qualify our findings.

METHODS

Setting

Sample collection for this prospective, cross-sectional study was performed between 2 April 2020 and 20 April 2020 on selected wards within a large North West London teaching hospital group comprising 5 hospitals across 4 sites with 1200 acute beds. Most sampling was conducted at 1 hospital site during the peak of the COVID-19 pandemic (Supplementary Figure 1) when most patients were managed in cohort wards.

Clinical Areas Selected for Air and Surface Sampling

Seven clinical areas (emergency department, an admissions ward, 2 COVID-19 cohort wards, theatres during tracheostomy procedures, an admissions ward, an intensive care unit (ICU), and a 6-bed bay converted into a negative-pressure area for management of continuous positive airway pressure [CPAP] on patients with COVID-19) and a public area of the hospital were selected to represent the diversity of clinical environments (Supplementary Table 1).

All inpatient wards were fully occupied by adult patients with COVID-19 at the time of sampling, apart from the emergency department. In the part of the emergency department dedicated for patients with confirmed or suspected COVID-19, 2 of the cubicles were occupied and 1 patient was in the ambulatory wait area at the time of sampling.

All areas were disinfected daily with an additional twice-daily disinfection of high-touch surfaces using a combined chlorine-based detergent/disinfectant (Actichlor Plus, Ecolab).

In each clinical area, between 3 and 5 air samples were collected. High-touch surfaces in the immediate vicinity of each air sample were sampled, including bed rails, clinical monitoring devices (blood pressure monitors), ward telephones, computer keyboards, clinical equipment (syringe pumps, urinary catheters), and hand-cleaning facilities (hand-washing basins, alcohol gel dispensers). In each clinical area, sampling was performed in both patient (ie, bays and single rooms) and nonpatient (ie, nursing stations and staff rooms) care areas. Environmental sampling was conducted during 3 tracheostomy procedures. During the first procedure, air sampling was performed before and during the procedure; for the other procedures, air sampling was performed during the procedure only.

Sampling Methods

All 1-m³ air samples were collected into a conical vial containing 5 mL Dulbecco’s minimal essential medium (DMEM) using a Coriolis μ air sampler (Bertin Technologies). Surface samples were collected by swabbing approximately 25-cm² areas of each item using flocked swabs (Copan) moistened in DMEM and then deposited into 1.0 mL of DMEM. Temperature, humidity, and time of day were recorded at the time of sampling.

Detection and Quantification of SARS-CoV-2 Viral RNA Genome and Viral Culture

Viral RNA detection and absolute quantification were performed using quantitative RT-PCR. Samples were extracted from 140 µL of the DMEM medium using the QIAamp viral RNA mini kit (Qiagen, Germany). Negative controls (water) were extracted and included in the PCR assays. SARS-CoV-2 viral RNA was detected using AgPath-ID One-Step RT-PCR reagents (Life Technologies) with specific primers and probes that target the envelope (E) gene [18]. A standard curve with 6 serial dilutions of 1 × 10⁵ to 1 × 10⁰ copies/µL E gene was included in each RT-qPCR run. E gene copies per cubic meter of air and copies per swab were calculated. Duplicate PCRs were run from each sample. Samples were defined as positive if both duplicates had a cycle threshold (Ct) <40.4 and defined as suspected if 1 of the 2 had Ct <40.4, equivalent to 1 genome copy.

For the viral culture, Vero E6 (African green monkey kidney) and Caco2 (human colon carcinoma) cells were used to culture virus from air and environmental samples using a method adapted from one previously used to culture influenza virus [19]. The cells were maintained in DMEM supplemented with heat-inactivated fetal bovine serum (10%) and penicillin/streptomycin (10 000 IU/mL and 10 000 µg/mL). For virus isolation, 200 µL of samples was added to 24 well plates. On day 0 and after 5–7 days, cell supernatants were collected, and RT-qPCR to detect SARS-CoV-2 was performed as described above. Samples with at least 1 log increase in copy numbers for the E gene (reduced Ct values relative to the original samples) after 5–7 days propagation in cells compared with the starting value were considered positive by viral culture.

We performed a laboratory experiment to determine the limit of detection for culturing SARS-CoV-2 dried on surfaces. A log₁₀ dilution series from solution that contained 8.25 × 10⁶ plaque-forming units/mL SARS-CoV-2 (titered by plaque assay in Vero cells) from 10^–3 to 10^–6 (covering Ct values from 26 to 36 and E gene copies from 10⁶ to 10³) was produced in DMEM, and 50 µL of each dilution was inoculated in triplicate onto the surface of plastic (standard keyboard key) or stainless steel (2 cm × 1 cm × 0.2 cm) pieces. The inoculated surfaces were dried in a safety cabinet for 2 hours after which they were visibly dry. They were then sampled using flocked swabs. Swabs were deposited into 1.5 mL of DMEM for 1 hour, and then 100 µL was used to inoculate wells of Vero E6 cells culture in 24-well plates. RT-qPCR was used to determine viability following 7 days of culture as follows: 140 µL was used for RNA extraction and qPCR immediately (0 days post-inoculation [dpi]) and after incubation for 7 days in a 24-well plate with Vero E6 cells (7 dpi). Samples with an increase in copy numbers for the E gene (reduced Ct values relative to the original samples) after propagation in Vero E6 cells were considered positive by viral culture.

Statistical Analyses

A χ² test was used to compare the proportion of environmental samples (surfaces or air) that were positive or suspected for SARS-CoV-2 RNA in areas immediately occupied by patients with COVID-19 with other areas. The mean concentration of air and surface contamination in each of the areas was log transformed and then compared using 1-way analysis of variance followed by Tukey’s multiple comparisons test.

Ethics Approval

The work was registered locally as an NHS service evaluation (#434).

RESULTS

There were 114 of 218 (52.3%) surface samples suspected (91 of 218 [41.7%]) or positive (23 of 218 [10.6%]) for SARS-CoV-2 RNA, but no virus was cultured (Table 1). The proportion of surface samples with suspected or positive RNA varied by item, including >80% of computer keyboards/mice, alcohol gel dispensers, and chairs and >50% of toilet seats, sink taps, and patient bed rails (Figure 1).

Table 1.

Polymerase Chain Reaction Results from Surface and Air Samples

		Surface Samples							Air Samples
Area Sampled		Total	Positive	% Positive	Suspect	% Suspect	Positive or Suspect	% Positive or Suspect	Result	Concentration (copies/m3)	Notes
Cohort ward A	Staff room	6	0	0.0	2	33.3	2	33.3	Negative
	Nurse station	6	1	16.7	3	50.0	4	66.7	Negative
	Toilet B (outside the patients’ bay)	6	0	0.0	2	33.3	2	33.3	Negative
	Cohort bay B	6	3	50.0	2	33.3	5	83.3	Positive	7048
Cohort ward B	Staff room	4	0	0.0	0	0.0	0	0.0	Negative
	Patients’ toilet (in the ward)	7	0	0.0	1	14.3	1	14.3	Suspect	464
	Male bay	12	1	8.3	4	33.3	5	41.7	Suspect	1335
	Male bay (side room)	8	2	25.0	5	62.5	7	87.5	Suspect	163
Adult acute admission unit	Ward manager’s office	5	1	20.0	2	40.0	3	60.0	Negative
	Nurse station	7	0	0.0	5	71.4	5	71.4	Positive	404
	Patient bay 2	8	0	0.0	2	25.0	2	25.0	Negative
	Patient bay 1	10	0	0.0	8	80.0	8	80.0	Negative
Adult emergency department	“Green” majors	10	1	10.0	5	50.0	6	60.0	Negative
	Nurse station	4	2	50.0	0	0.0	2	50.0	Negative
	Ambulatory waiting	3	2	66.7	1	33.3	3	100.0	Negative
	Patient assessment cubicles	3	0	0.0	1	33.3	1	33.3
	Male toilet (next to the nurse station)	2	0	0.0	1	50.0	1	50.0
	Resuscitation bay (last patient > 2 hours)	10	0	0.0	4	40.0	4	40.0	Suspect	35
Hospital public areas	QEQM main entrance	7	1	14.3	4	57.1	5	71.4	Suspect	1574
	Male toilet at QEQM main entrance	7	1	14.3	3	42.9	4	57.1	Suspect	1545
	Lift area QEQM ground floor	10	0	0.0	4	40.0	4	40.0	Negative
Temporary CPAP ward	Nurse station	5	1	20.0	2	40.0	3	60.0	Suspect	1922
	CPAP unit	19	2	10.5	12	63.2	14	73.7	Suspect	31	<1 m from 2 patients
									Negative		>2 m from patients
	Personal protective equipment doffing area	5	0	0.0	2	40.0	2	40.0	Negative
Adult ICU	Staff room	10	0	0.0	6	60.0	6	60.0	Suspect	249
	Nurse station inside ICU	6	1	16.7	0	0.0	1	16.7	Negative
	Bay area	11	0	0.0	5	45.5	5	45.5	Suspect	164
	Side room bay area	8	2	25.0	4	50.0	6	75.0	Suspect	307
Theatres	Theatres	13	2	15.4	1	7.7	3	23.1	Negative		Before tracheostomy
									Negative		During tracheostomy
									Suspect	1163	During tracheostomy
									Negative		During tracheostomy
	Total	218	23	10.6	91	41.7	114	52.3	2 of 31 (6.4%) positive; 12 of 31 (38.7%) suspect

Abbreviation: CPAP, continuous positive airway pressure; ICU, intensive care unit; QEQM, Queen Elizabeth Queen Mary.

Figure 1.

Proportion of environmental samples suspected or positive by item sampled. The number on the x-axis represents the number of each item sampled.

There were 14 of 31 (38.7%) air samples suspected (12 of 31 [38.7%]) or positive (92 of 31 [6.4%]) for SARS-CoV-2 RNA, but no virus was cultured (Table 1). A total of 10¹ to 10³ genome copies/m³ of SARS-CoV-2 RNA were detected in air from all 8 areas tested (Table 1). There was no significant difference in mean viral RNA concentration across those 8 areas (P = .826). Similarly, 10¹ to 10⁴ genome copies/swab SARS-CoV-2 RNA were detected in surface samples from all 8 areas tested (Figure 2). There was a significant difference in the mean SARS-CoV-2 surface viral load across those 8 areas (P = .004), with both cohort ward A (mean = 1.76 log₁₀ copies/swab, P = .015) and the temporary CPAP ward (mean = 1.69 log₁₀ copies/swab, P = .016) showing higher levels of viral RNA than the adult ICU (mean = 0.0018 log₁₀ copies/swab).

Figure 2.

Severe acute respiratory syndrome coronavirus 2 E gene copy number from surface swabs. The quantity of E gene copy number per swab is shown. Suspect samples = blue dots; positive samples = red dots; negative samples = black dots. Abbreviations: CPAP, continuous positive airway pressure; ICU, intensive care unit; PPE, personal protective equipment.

SARS-CoV-2 RNA was detected in several clinical areas where AGPs are commonly performed, including a resuscitation bay in the emergency department and a bay temporarily converted for CPAP where SARS-CoV-2 RNA was detected from air both within and outside the bay. No patient was undergoing CPAP at the time of sampling, but 1 patient was undergoing high-flow nasal cannula (HFNC) oxygen therapy. In operating theatres, 1 of 3 air samples collected during 3 tracheostomy procedures were positive.

SARS-CoV-2 RNA was detected in surface and air samples in parts of the hospital that hosted staff but were not being used for direct patient care, including the ICU staff room, the nursing station outside of the CPAP unit, and the hospital main entrance and public toilets. However, SARS-CoV-2 RNA detection in air and surface samples was significantly more likely in areas immediately occupied by COVID-19 patients than in other areas (67 of 105 [63.8%] in areas immediately occupied by COVID-19 patients vs 29 of 64 [45.3%] in other areas; odds ratio, 0.5; 95% confidence interval, .2–.9; P = .025).

Since viable virus was not cultured from any of the air or surface samples, we performed laboratory experiments to determine the limit of detection of SARS-CoV-2 dried onto surfaces. Four dilutions of virus deposited onto 2 nonporous surfaces determined that dried inocula with a Ct value <30 (corresponding to an E gene copy number of ≥10⁵ per mL) yielded SARS-CoV-2 that could be cultured (Table 2). In our study, all surface and air samples from the hospital environment had a Ct value >30.

Table 2.

Viability of Severe Acute Respiratory Syndrome Coronavirus 2 Dried Onto Steel or Plastic Surfaces from a Dilution Series

Inoculum (Plaque-Forming Unit)	Steel Surface Swab (Ct)	E Gene Copies/mL	After Culture (Ct)	Interpretation	Plastic Surface Swab (Ct)	E Gene Copies/mL	After Culture (Ct)	Interpretation
41.25	26.23 ± 0.30	1.86 × 106 ± 3.66 × 105	12.65 ± 0.51 Pos	Culturable	25.95 ± 0.06	2.23 × 106 ± 9.74 × 104	11.16 ± 0.19 Pos	Culturable
4.125	29.27 ± 0.04	2.30 × 105 ± 5.10 × 103	12.86 ± 0.01 Pos	Culturable	29.51 ± 0.29	1.97 × 105 ± 3.71 × 104	12.58 ± 1.47 Pos	Culturable
0.4125	32.54 ± 0.06	2.47 × 104 ± 9.23 × 102	36.48 ± 1.80 Neg	Nonculturable	32.67 ± 0.07	2.23 × 104 ± 1.04 × 103	37.39 ± 0.21 Neg	Nonculturable
0.041 25	39.22 ± 5.13	1.68 × 103 ± 1.92 × 103	41.33 ± 3.45 Neg	Nonculturable	36.55 ± 0.23	1.63 × 104 ± 2.85 × 102	39.76 ± 4.61 Neg	Nonculturable

Viability determined through reverse-transcription polymerase chain reaction from cultures immediately after drying and 0 days post-inoculation (dpi) with Vero E6 cells compared with after culture (7 dpi). Means and standard deviations of Ct values are shown.

Abbreviation: Ct, cycle threshold.

Discussion

SARS-CoV-2 RNA was detected frequently from surface and air samples but was not cultured. SARS-CoV-2 RNA was identified across the 8 areas that we tested and was detected more frequently in areas occupied by COVID-19 patients than in other areas.

A direct comparison between our findings and those from other studies in which SARS-CoV-2 surface and air contamination was evaluated is not possible due to differences in environmental sampling strategy, experimental methods (including sampling methods), local SARS-CoV-2 epidemiology, the physical layout of buildings and clinical spaces, patient characteristics and shedding [4, 20], and patient and staff testing approaches and cleaning/disinfection protocols. Nonetheless, our finding of widespread detection of viral RNA on surfaces (114 of 218, 52.3%) and to a lesser extent air (14 of 31, 38.7%) is broadly consistent with the findings of most others, although the proportion of surface and air samples positive for viral RNA is higher in our study [9–14]. For example, Ye et al found that 14% of 626 surface samples were positive for viral RNA, with a higher proportion of surface samples positive in the ICU (32% of 69 samples) when sampling a range of clinical settings in a hospital where patients with COVID-19 were being cared for in Wuhan, China [10]. However, other studies have identified very little or no surface or air contamination [9, 11]. Other studies have observed higher frequencies of contamination in patient-care vs non–patient-care areas [9, 10, 12] and variation in the frequency of contamination across different clinical areas, which is in line with our findings [10, 12]. One surprising finding in our study was that the level of contamination on ICU surfaces was lower than in a cohort general ward or in the temporary CPAP ward, which is in contrast to other findings [10]. This may be because patients sampled in the ICU were on closed-circuit ventilation systems through cuffed endotracheal tubes, which may have a lower risk of producing surface and air contamination than other ventilation systems such as CPAP.

We did not identify viable virus in any surface or air sample. Few studies have attempted to culture SAR-CoV-2 from healthcare environments, and no viable virus was detected [11, 15]. Our laboratory study of the viability of virus dried on surfaces helps to qualify our findings and the findings of others, suggesting that Ct values >30 that correspond to an E gene copy number of <10⁵/mL are unlikely to be culturable (Table 2). This finding parallels studies of viral infectivity from clinical specimens [21, 22]. Bearing in mind that the viral RNA detected in the hospital setting might have been deposited more than 2 hours previously, we cannot determine whether our inability to culture virus from the samples is explained by the low RNA levels or the length of time since deposition, which may reflect nonviable viral RNA. It is also possible that virus was infectious but not culturable in the laboratory.

Surface contamination was detected on a range of items, especially computer keyboards, chairs, and alcohol dispensers. Other studies have also identified computer keyboards and/or mice as a risk for SARS-CoV-2 RNA contamination [9, 10, 12]. Many of the computers that we sampled were in shared staff clinical areas (such as nursing stations), so this argues for frequent disinfection of these items. The contamination of alcohol gel dispensers is unsurprising since staff activate these before hand hygiene is performed. However, alcohol gel dispensers should be included in routine cleaning and disinfection protocols or designed such that they can be activated without touching.

We sampled several areas where AGPs are commonly performed There was no difference in the viral load of the air across the 8 areas sampled, suggesting that AGPs do not produce persistently high levels of air contamination. However, we did not sample the air over time, and our air sampling method did not differentiate particle size and so we were unable to distinguish droplets from aerosols (<5 µM). One recent study identified SARS-CoV-2 RNA at low levels (in the range of 10¹–10² copies/m³) in patient care areas in a permanent and field hospital in Wuhan, China [14]. Positive samples were identified in a range of particle sizes, including those <5 µM, which would typically be considered as aerosols [2]. It seems likely, therefore, that the positive and suspected air samples identified in our study included a range of particle sizes spanning 5 µM, particularly in areas where AGPs are common.

While we sampled in a temporary CPAP ward, no patient was undergoing CPAP at the time of sampling. However, 1 patient was undergoing HFNC during sampling, and air contamination was identified <1 m from this patient. A recent summary of evidence concludes that HFNC is a lower-risk procedure in terms of aerosol generation than CPAP, which should be a topic for future studies [23].

We identified surface and air contamination during 3 tracheostomy procedures. Several studies and commentaries have evaluated the potential for surgical procedures to produce aerosols for patients with COVID-19 [24–26]. One study evaluated the spread of droplets during tracheostomies, although it did not include sampling for SARS-CoV-2 [24]. Our findings highlight a potential theoretical risk of COVID-19 transmission during these procedures. However, a larger sample size is required to understand this risk.

Strengths of our study include our sampling strategy, which encompassed contemporaneous surface and air samples from a range of patient and non–patient-care areas with diversity in physical layout and ventilation and including operating theatres and areas dedicated to known and potential AGPs. Each sample was tested using PCR and viral culture, and we performed laboratory experiments to quality our findings. The sampling was conducted during the peak of the pandemic (and so likely represents a worst-case scenario). Limitations include not collecting patient samples to better understand how our findings link to patient samples, particularly during tracheostomies and AGPs. There was no asymptomatic patient or staff testing at the time of sampling, which means asymptomatic patients and staff could have shed SARS-CoV-2. There were challenges in interpreting the significance of samples with low viral loads, and there was a lack of resolution of particle sizes for contamination of the air. Because no longitudinal sampling was performed, these findings represent a “snapshot.”

Our findings may have implications for future policy and guidelines. Most international guidelines recommend enhanced surface disinfection during the management of COVID-19. For example, Public Health England recommends enhanced disinfection using a chlorine-based disinfectant (or a disinfectant with effectiveness against coronaviruses) [27]. Our finding of widespread RNA contamination of clinical areas used to care for patients with COVID-19 supports this. Physical distancing is recommended by most governments, and personal protective equipment (PPE) is recommended during contact with patients with COVID-19 plus higher levels of PPE when performing AGPs. While we did not measure particle sizes during our air sampling, our findings highlight a potential role for contaminated air in the spread of COVID-19. Our finding of air contamination outside of clinical areas should be considered when making respiratory PPE recommendations in healthcare settings [28].

While SAR-CoV-2 RNA was detected within healthcare environments, further research linking patient, staff, and environmental samples is required to better understand transmission routes. Longitudinal environmental and clinical sampling across healthcare settings is required to understand the risk factors associated with viral shedding and transmission. Our findings can be used to parameterize mathematical models of COVID-19 transmission. Our methods can be used to assess the risks associated with various procedures including surgery, AGPs, and nebulization of medications. Findings from these studies may influence PPE recommendations for specific procedures [29–31].

Our findings of extensive viral RNA contamination of surfaces and air across a range of acute healthcare settings in the absence of cultured virus underlines the potential risk from surface and air contamination in managing COVID-19 and the need for effective use of PPE, physical distancing, and hand and surface hygiene.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. J. Z. and J. A. O. conceived the study, collected and analyzed data, and wrote the manuscript. J. R. P. conceived the study, collected data, and contributed to the manuscript. C. C., D. M. G., P. R. B., and S. M. collected data and contributed to the manuscript. F. B., A. H. H., and W. S. B. conceived the study, analyzed data, and contributed to the manuscript. J. A. O. is the study guarantor.

Acknowledgments. We acknowledge the staff teams and patients who supported the sampling during the peak of the challenges posed by this pandemic.

Disclaimer. The views expressed in this publication are those of the authors and not necessarily those of the National Health Service (NHS), the National Institute for Health Research, the Department of Health and Social Care, or Public Health England. A. H. H. is a National Institute for Health Research (NIHR) senior investigator. International Severe Acute Respiratory and Emerging Infection Consortium provided funding for J. Z. and laboratory materials used for this study.

Financial support. The authors acknowledge the support of the NIHR Imperial Biomedical Research Centre, the NIHR Health Research Health Protection Research Unit (HPRU) in HCAI and Antimicrobial resistance (AMR), and the HPRU in Respiratory Infections at Imperial College.

Potential conflicts of interest. J. A. O., reports personal fees from Gama Healthcare Ltd and Pfizer outside the submitted work. J. K. reports grants from H2020 Innovative Training Networks (ITN grant), NIHR (i4i grant), CRUK fellowship, J+J Educational grant, personal fees from Verb robotics/Ethicon and Medtronic, and other relationships with Cerulean Health, One Welbeck day surgery, and LNC therapeutics. All remaining authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

Author notes

J. Z. and J. A. O. are joint first authors.