key: cord-0881157-kv6o4u4k
authors: Tomisa, Gábor; Horváth, Alpár; Farkas, Árpád; Nagy, Attila; Kis, Erika; Tamási, Lilla
title: Real-life measurement of size-fractionated aerosol concentration in a plethysmography box during the COVID-19 pandemic and estimation of the associated viral load
date: 2021-09-03
journal: J Hosp Infect
DOI: 10.1016/j.jhin.2021.08.025
sha: e0d4d25235c5fb293bf19210570340a217f7efc0
doc_id: 881157
cord_uid: kv6o4u4k

INTRODUCTION: There are concerns about pulmonary function tests (PFT) being associated with aerosol generation and enhanced virus transmission. As a consequence, the number of PFTs was significantly reduced during the COVID-19 pandemic. However, there are no robust data supporting this fear. OBJECTIVES: The aim of this work was to perform real-life measurement of aerosol concentration in a PFT laboratory to monitor the concentration of particles near the patient and to model the associated potential viral load. METHODS: Two optical particle counters (OPC) were used to sample the background concentration and the concentration of particles in the vicinity of the mouth of the patients in the whole-body plethysmography box. A statistical evaluation of the measured particle concentration time series was completed. The particle exhalation rate was assessed based on the measured particle concentration data by applying the near-field/far-field theory. The number of exhaled viruses by an infected patient during the test was assessed in comparison with the emission of viruses during quiet breathing and speaking. RESULTS: Twenty-five patients were involved in the study. Eighteen patients produced a significant increase of the aerosol concentration, which was 1910±593 particles/L. Submicron particles dominated the number size distribution of the generated particles, but large particles represented a higher volume fraction in the generated particles compared to the background. An average gene exhalation rate of 0.2/min was estimated from this data. This is one order of magnitude higher than the release rate of the same infected person during quiet breathing and of the same order of magnitude with the release rate during normal speaking. CONCLUSIONS: Present results demonstrated that PFT is an aerosol generating procedure. Based on current results, the moderate increase of viral load does not underpin the stopping of such examinations.

Pulmonary function tests (PFT) are important and necessary diagnostic tools in respiratory medicine; however, there are concerns that a forced breathing manoeuvre may result in enhanced particle (droplet and aerosol) amount exhaled by the patient. In addition atypical breathing can provoke coughing, which may also be associated with increased particle exhalation. As the patient does not wear a facemask during the PFT measurements, the chances of pathogen transmission may be potentially increased. There is increasing evidence that important routes of SARS-CoV-2 transmission are via inhalation of virus containing droplets and aerosols [1] - [2] , which deposit in different regions of the airways [3] . Therefore, shortly after the onset of the COVID-19 pandemic the number of PFTs was drastically reduced or even stopped in some places. National and international medical and respiratory societies (e.g. ERS-European Respiratory Society, ARTP-Association for Respiratory Technology and Physiotherapy) released statements and position papers with warnings on the possible increased risks linked to PFTs and recommended different measures of risk mitigation. Some of these recommendations were updated as experience and knowledge accumulated [4] . However, the precaution is partly based on theoretical considerations rather than a solid experimental basis.

In the months following the first pandemic peak in 2020 efforts were spent to reveal whether PFTs are significant sources of aerosol generation and whether the increase in transmission of risk associated with PFT is real or not [5] - [8] . The values of the increase in particle concentration attributable to PFT documented in the above studies span between a few hundreds of particles up to more thousands of particles per litre, but there were notable differences among the studies regarding the circumstances. On the other hand, speaking without a facemask may also produce an extra load of a few hundreds of particles per litre, and it was also demonstrated [9] that the intersubject variability is more than an order of magnitude in the concentration of aerosol generated. As it can be seen, to date there are only a few results regarding the number of particles attributable to PFT. Some of the related works concluded that aerosol generation during PFT is significant, but the excess risk of infection due to these particles remained unknown. Moreover, in all but one of the previous studies the subjects were not real patients, and the conditions were more or less far from 'real-life' (e.g. waiting artificially long times to reach the background particle level, use of ultraclean, laminar theatre for the measurements).

The aim of the present study was to contribute to the evidence base related to PFT measurements during COVID-19 and other pandemics by performing real-life aerosol concentration monitoring without any artificial intervention or change during several working J o u r n a l P r e -p r o o f hours in a PFT laboratory. The measurements were carried out during the standard tests on real patients (with proven disease or under the establishment of the diagnosis) performed by the medical professional in order to capture the possible changes in the realistic aerosol environment due to the examinations. Another aim was to use the results of the measurements and data from the literature to estimate the potential risk associated with PFT during the pandemic.

A single-centre observational study was conducted in the whole-body plethysmography box in the PFT laboratory of the Department of Pulmonology, Semmelweis University (Budapest, Hungary), on a summer day (28 July 2020). The plethysmography box (PDT-111/pd, Piston Medical) was 0.7×0.9×1.7 m large and it was located in a room of size 5.6×3.5×3.0 m. Figure   1 demonstrates the layout of the room. The position of the doors, windows, body box and other major objects is also indicated in the figure. The room air characteristics were monitored by a mobile temperature and humidity data logger (Testo 174H, Testo SE & Co. KGaA). The same quantities were automatically monitored by built-in sensors inside the body box for the correction of spirometric data. There was no air conditioner in operation during the measurements. The windows of the laboratory were halfopened. The doors of the laboratory were usually closed, but the patient entrance door was opened every time a patient entered or left the room. The air exchange rate of the laboratory was 6.5 times per hour. The door of the plethysmography box was opened between two tests J o u r n a l P r e -p r o o f and closed or opened during the measurements depending on the type of the measurement.

Normal spirometry, lung volume measurement (plethysmography) and diffusion testing (DLCO) was conducted by a PFT technologist standing outside the box while patients were sitting in the plethysmography box. Two observers sat in the laboratory (outside the transparent cabin at 2 m distance) and documented the timeline of the relevant events and conditions. All the medical personnel and the observers wore FFP2 facemasks during the whole period of the measurements. The patients took off their facemask exclusively during their stay in the plethysmography box. Disposable bacterial and viral filters (PBF-100) were used.

The participating patients were volunteers and provided written consent. The study was completed based on ethical approval (no. SE RKEB 212/2020).

Since the study aimed to observe real-life situation, it was not always enough time to make sure that the particle concentration reaches a relatively constant background level (as it was in some previous studies) in between the tests. Therefore, it was necessary to use two similar sampling devices in parallel, with one device sampling the changing of background particle concentration, the second one measuring the aerosols near the patient. For this purpose, two identical optical particle counters (OPC -Grimm Aerosoltechnik, Portable Aerosol Spectrometer, model 1.109) were used with 6 s time resolution. The size distributions were recorded in 31 size bins between 0.25 and 32 µm. The upper concentration limit (< 5% coincidence error) of the instrument is 2 million particles/L, which was not expected to occur in a PFT laboratory in the given size range. The small size of the OPC devices (24 × 13 × 7 cm 3 ) and their quiet operation made them suitable for the measurements. OPC-A had a fixed position inside the cabin in the furthest possible position from the source, that is, at about 140±15 cm distance from the mouth of the patient, depending on the patient's height. Based on previous studies [6] , it is not expected that particle concentration increases at 1.4 m from the patient, so OPC-A provided the background concentration in the cabin. OPC-B sampled the aerosols inside the body box while the patient was inside the cabin (at 30 cm from the patient's mouth) and inside the laboratory but outside the cabin when the cabin was empty (between two measurements).

Statistical evaluation of the measured particle concentration time series was performed by OriginPro 2021 (version 9.8.0.200) software. Background concentration and near-patient concentration time series were compared by using two-sample t-tests. Concentration time series measured outside and inside the box were compared by correlation analysis (Pearson coefficient). The agreement between the two devices (OPC-A and OPC-B) sampling the same environment at the same time was verified by the Bland-Altman test [10] .

Since virus detection was not completed within the present work, only a theoretical estimation on the number of the emitted gene copies was performed in this study. Since the number of viable viruses is two or three orders of magnitude lower than the total number of detectable viruses [11] , hereon we systematically use 'gene copies' instead of 'virus copies'.

As the number of exhaled gene copies depends on many factors, a comparative estimation was performed by comparing the number of copies due to PFT to those that would be emitted by the same person in the same environment, but breathing normally or speaking. For the evaluation of the number of gene copies that an infected patient may emit during a pulmonary function test it was necessary to evaluate the emitted particulate mass (at the patient's mouth) starting from the data obtained at 30 cm from the subject's mouth. For this purpose, the equations of the "near-field/far-field" well-mixed room model were used [12] - [13] . The timedependent mass concentration of the emitted particles was expressed as the sum of near-field and far-field concentrations obtained by solving mass balance equations

and − =

In equation (2) g denotes the rate of particle generation by the patient (mass/time),  is the interzonal flow rate (volume/time) between near-field and far-field zones, t is the time from the exhalation of particles and VN-F the volume of the spherical near-field zone (with 0.6 m radius).

The interzonal flow rate (volume/time) was expressed as

where S is the surface area of the above-defined sphere and uavg the average wind speed in the room (= 0.1 m/s). In equation (3) VF-F denotes the volume of the PFT laboratory (52 m 3 ), q is the room ventilation rate (5.6 m 3 /min) and t1/2 stands for virus half-life on aerosols. In this study the value of t1/2 = 66 min was adopted from the publication of Doremalen et al. [14] . For the near-field component the loss due to virus inactivation (see equation 2) was neglected because it is much less than the loss due to interzonal air exchange. In addition to the exhaled particle losses due to near-field and far-field air exchange, the detected mass of particle can be lower also because of gravitational settling. A 32 m diameter particle (which was the upper size limit J o u r n a l P r e -p r o o f of the sampled particles) may fall outside of the near-field zone within 20 seconds, a 10 m particle within 200 seconds [15] . Since the examinations took in average 2.5 minutes, it is plausible to consider only the PM10 fraction of the particles and compare the number of estimated viruses in PM10 emitted during PFT to the available data on virus load of PM10 due to other types of activities [16] . Another relevant phenomenon is droplet evaporation. As data on viral load is available for the emitted mass, it is important to convert the sampled mass into emitted mass by also considering evaporation. The time needed for a droplet containing water and non-volatile solutes to reach its equilibrium size can be estimated by the formula

where D0 is the initial diameter of the droplet,  is 4.2×10 -10 m 2 /s [15] , RH is the relative humidity and 0 is the initial volume fraction of solutes in the droplet. The value of 0 may vary depending on the NaCl, surfactant and protein content of the droplet. In this work the value of 0.025 was considered, which is the average of the volume fraction values characteristic of saliva droplets with low (3 mg/mL) and high (76 mg/mL) protein content [17] . The final (after evaporation) diameter of the droplets can be estimated by the expression

A conservative approach has been applied for the estimation of the particulate mass that an infected patient may emit during a pulmonary function test assuming that all the generated particles originate from the exhalation of the patient. By the same token, ev defined in equation 5 was low in comparison with the duration of the PFT implying that all the droplets have already reached their equilibrium size when detected. All these approximations lead to the highest possible particle mass at the mouth of the patient, that is, the upper bound of the number of viruses the patient may have emitted. The number of viruses was estimated by assuming that each mL of emitted particle contains 10 6 gene copies, which is characteristic of a normal emitter [18] . The number of viruses generated during PFT was compared to the number of the viruses the same person would emit in the same environment during normal breathing or speaking.

Twenty-five patients were involved in the study, suffering from different diseases: two asthmatics, eight transplant patients, four COPD patients, one patient with CF (cystic fibrosis), two patients with idiopathic pulmonary fibrosis (IPF), two patients with interstitial lung diseases (ILD), one smoker suspected of asthma, two subjects suspected of TB, one lung cancer patient, one sarcoidosis patient and one patient with pulmonary hypertension. The total number J o u r n a l P r e -p r o o f of measurements was 27 as there were two measurements on two of the patients (reversibility test). One participant (suffering from sarcoidosis) was excluded because in his case the increase in aerosol concentration (much higher than the average increase) was generated by the patient's cloth handling. The results of the measurements for the remaining 24 patients are summarized in Table 1 . The mean total concentration increase was 1910±1018 particles/L. No statistically relevant correlation could be demonstrated between the type of disease and particle concentration enhancement due to the measurement. The correlation between FEV1 and FVC and particle concentration increase was also weak. The concentration increase was usually higher for longer PFT durations, but the correlation was weak (r=0.34). In addition, at this sample size the three types of examinations did not result in significantly different particle concentration increases.

The left panel of Figure 3 depicts the average number of particles of different sizes generated by 16 patients (18 measurements). As the figure demonstrates, submicron particles dominate in the newly generated particles. However, the number of gene copies in infected cases correlate with droplet volume rather than with droplet number [18] - [19] . Therefore, it is plausible to consider the volume size distribution as well. The right panel of Figure 3 for the droplets to reach their final size by evaporation (less than one second, see eq. 5) it can be considered that droplets had enough time to evaporate to their minimum size. In such circumstances, eq. 6 can be applied to get the initial volume of the droplets, which yields a value of 2×10 -7 mL/min for the release rate of fresh droplets, equivalent to the emission of 0.20 gene copies/min in the case of a normal emitter (1 mL of droplet corresponds to 10 6 gene copies).

A number of mechanisms (e.g. airflow driven fragmentation of viscoelastic filaments in the upper airways, elasto-capillary bursting of fluid films in the bronchioli) by which droplets are created and emitted from different anatomical regions of the respiratory tract are still under scientific debate [21] - [22] . The amount and size of the emitted droplets depend on the region of origin in the airways and also on the type of activity (breathing, speaking, singing, coughing, sneezing). It is scientifically plausible to hypothesise that within the same type of activity the intensity of the activity is also a key factor. For instance, it has been demonstrated that the amplitude of vocalisation (normal speaking versus loud speaking or quiet singing versus loud singing) affects the number of generated aerosol particles [1] , [9] . In a similar way, the mode of breathing (normal, deep, fast) may influence the size and amount of the emitted droplets [20] , J o u r n a l P r e -p r o o f [23] . Forced breathing leads to higher air velocity and modified airway calibre, which do have an effect on the formation and emission of droplets by the above mentioned mechanisms. Based on this rationale, PFT must be an aerosol generation activity and it is also likely that the number of emitted particles exceeds the number of particles emitted during quiet breathing. Present results seem to underpin this hypothesis, as in the case of 69% of the measurements we could detect a statistically significant increase of particle concentration. The increase was slightly higher than that obtained by [8] but lower than those reported by [7] . It is worth noting that four probably not the result of the forced expiration alone but a combination of forced breathing, quiet breathing before the test, speaking in some cases and even coughing for some patients.

Nevertheless, these events are all at play during a PFT, thus our data reflect the real-life situation.

It is an important question to what extent the surplus of particles due to PFT can increase the risk of infection. When analysing this issue, it is worth considering that the particle concentration increase was significant for only a part of the patients and only at short distance from the mouth of the patients. In addition, the increase compares small to the background concentration and it is less than its variation due to some other patientcare-related activities.

For instance, our measurements demonstrated that in a hospital room the increase in particle concentration could be of the order of 10 4 particles/L due to some activities, such as a medical visit or bedding (data under publication). Similarly, our measurements during bronchoscopy indicated particle generation of the order of 10 3 -10 4 particles/L (unpublished data). The flush of the toilet may generate up to 10 5 droplets/litre of air [24] and a high SARS-CoV-2 load was detected in a toilet of a hospital [25] . Thus, the risk of transmission of the virus by aerosols depends on several factors, especially on the number of viable and inhalable viruses which are more likely to be present in the exhaled air than on the clots. 

This study has a number of limitations. First of all, due to the operational complexity of the study, the number of patients was limited. As the study was observational, it was not possible to select as many as patients with the same disease, so there was a large span of diseases and heterogeneous underlying comorbidities. Therefore, it was not possible to have patient groups with large populations which implies that the statistical power of the analysis is not very strong.

Although the patients were not tested before the PFT, probably most of them were not infected by the new coronavirus (they were asymptomatic). The viral load due to PFT was calculated assuming normal emitter infected patients, which is an approximation. It was also assumed, that all the generated aerosol particles originate from the airways of the patients, which may lead to an overestimation of the viral load. In addition, the particle concentration was measured at some J o u r n a l P r e -p r o o f distance from the patient's mouth, the concentration of exhaled aerosol particles was determined theoretically based on these measurements, which may increase the uncertainty of the assessment.

In this observational study conducted under real-life conditions the concentration of particles around the patients during pulmonary function tests was successfully sampled. The excess of particles due to lung function testing demonstrated that PFT is indeed an aerosol-generating procedure. The estimated number of the released viruses of a hypothetical normal emitter infected person revealed that the viral load of such a measurement is comparable to the load associated with speaking of the same person without facemask. Current results revealed that the excess risk due to lung function examinations is not negligible, especially taking into account that new variants (e.g. Delta variant) are more contagious. However, taking into account that PFT measurements provide essential information on the status of the patient, stopping of PFTs is not recommended. Evidently, all the activities related to PFTs must be carried out with prevention, considering all the possible safety measures.

The coronavirus pandemic and aerosols: Does COVID-19 transmit via expiratory particles?

Breathing Is Enough: For the Spread of Influenza Virus and SARS-CoV-2 by Breathing only

Deposition distribution of the new coronavirus (SARS-CoV-2) in the human airways upon exposure to cough-generated droplets and aerosol particles

Italian Respiratory Society. Suggestions for lung function testingin the context of COVID-19

Aerosol Generation during Spirometry

Small droplet emission in exhaled breath during different breathing manoeuvres: Implications for clinical lung function testing during COVID-19

Airborne Particulate Concentrations During and After Pulmonary Function Testing

Risk mitigation of aerosolisation from lung function testing: results from the AERATOR study

Comparing aerosol concentrations and particle size distributions generated by singing, speaking and breathing

Statistical methods for assessing agreement between teo methods of clinical measurement

SARS-CoV-2, SARS-CoV, and MERS-CoV viral load dynamics, duration of viral shedding, and infectiousness: a systematic review and meta-analysis

Interzonal airflow rates for use in near-field far-field workplace concentration modeling

The near/far field model with constant application of chemical mass and exponentially decreasing emission of the mass applied

Aerosol and surface stability of SARS-CoV-2 as compared to SARS-CoV-1

Mechanisms of Airborne Infection via Evaporating and Sedimenting Droplets Produced by Speaking

Simulation of SARS-Cov-2 aerosol emissions in the infected population and resulting airborne exposues in different indoor scenarios

Physico-chemical characteristics of evaporating respiratory fluid droplets

Virological assessment of hospitalised patients with COVID-2019

Estimation of viral aerosol emissions from simulated individuals with asymptomatic to moderate coronavirus disease 2019

Th effect of exhalation flow on endogeneous particle emissopn and phospholipid composition

Strching and break-up of saliva filaments during speech: A route for pathogen aerosolisation and its potential mitigation

Exhaled particles and small airways

The mechanisms of breath aerosol formation

Aerosol generation by modern flush toilets

Aerodynamic analysis of Sars-CoV-2 in two Wuhan hospitals

The authors would like to thank Sándor Nyáguj, Éva Paksiné Veres and Erika Bölkény for their help in organising the aerosol sampling in the PFT laboratory. The authors would like to acknowledge Dr Balázs Madas for the useful piece of advice on the statistical analysis of the measured data.

None.J o u r n a l P r e -p r o o f