key: cord-0289357-46cvdgz2
authors: Gall, E.; Laguerre, A.; Noelck, M.; Van Meurs, A.; Austin, J. A.; Foster, B. A.
title: Aerosol generation in children undergoing high flow nasal cannula therapy
date: 2020-12-14
journal: nan
DOI: 10.1101/2020.12.10.20245662
sha: 7d934e5ad6c9a410960f1eadac3934bdd45f3d41
doc_id: 289357
cord_uid: 46cvdgz2

Objective High flow nasal cannula therapy (HFNC) may increase aerosol generation, putting health care workers at increased risk of infection, including from SARS-CoV-2. This study examined whether use of HFNC increases aerosols and if there is a relationship between flow rate and near-field aerosol concentrations. Patients and Methods Subjects between 30 days and 2 years of age were enrolled. Each child received HFNC therapy at different flow rates over time. Three sampling stations with particle counters were deployed to measure aerosol generation and dispersion in the room: station one within 0.5 m of the subject, station two at 2 m, and station three on the other side of the room. We measured carbon dioxide (CO2) and relative humidity. Station three (far-field) measurements were used to adjust the station one (near-field) measurements for room conditions. Results We enrolled ten children ranging from 6-24 months (median 9 months), two with respiratory illness. Elevated CO2 indicated the near-field measurements were in the breathing plane of the subjects. Near-field breathing plane concentrations of aerosols 0.3-10 microns are elevated by the presence of the patient with no HFNC flow, relative to the room far-field, by 0.45 #/cm3. While we observed variability between subjects in their emission and dispersion of particles, we did not find an association between HFNC and near-field elevations of particle counts. Conclusion Near-patient levels of particles in the 0.3-10 micron range was not affected by the use of HFNC in healthy patients. Further study on older children and children with increased mucus production may be warranted.

High flow nasal cannula therapy (HFNC) provides respiratory support for hospitalized 74 children across a range of ages and diagnoses including asthma, pneumonia and bronchiolitis. 75 World Health Organization (WHO) guidance suggests that HFNC does not cause wide-spread 76 dispersion of droplets from patients. 1 However, empirical data in clinical settings is lacking on 77 whether HFNC contributes to aerosol generation. While children typically have more mild and 78 even asymptomatic infections with SARS-CoV-2, respiratory disease and co-infection with other 79 viruses have been reported. 2 During the COVID-19 pandemic, HFNC has been treated as an 80 aerosol generating procedure (AGP) in the United States given concern around particle 81 generation, typically characterized in the health care field as transmission by droplet (≥5µm) and CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint model to evaluate enhancement of exhaled air, measured by the extent of light-scattering as a 96 function of distance from the patient. They showed an increase in "exhaled air dispersion" from 97 65 mm with HFNC flow of 10 L/min to 172 mm with HFNC flow of 60 L/min. Using smoke 98 particles as tracers and an adult human head with a lung model attached, Elshof et al. 7 examined 99 the dispersion of 100 µm droplets using HFNC with a lung simulator. They described an 100 estimated dispersion range of 100 µm droplets of between 18.8 and 33.4 cm from the individual 101 using flow rates between 30-60 L/min. They also noted that HFNC increased the distance of 102 exhaled smoke to nearly one meter under several conditions whereas a non-rebreather or Venturi 103 mask did not influence the distance beyond normal breathing. 7

This study sought to examine whether HFNC therapy use in children generates elevated 105 particle levels in the near-field (0.5 m) of the patient's breathing plane. We measured 106 concentrations of particles and carbon dioxide (as an exhaled breath tracer) in rooms in a clinical 107 care facility with varying HFNC flow rates for each patient. Our study addresses several 108 knowledge gaps concerning HFNC and particle generation and dispersion as it: i) addresses an 109 unstudied population, children, ii) was conducted in a clinical care facility with human subjects, 110 and iii) directly measured aerosols with diameter 0.3 -10 µm and carbon dioxide in the near-111 field breathing plane and room far-field. The goal was to generate data to inform the safe use of 112 this therapy and inform resource management and infection control measures.

This study was a prospective study looking at aerosol generation and dispersion from 115 pediatric patients on high-flow nasal cannula (HFNC) in a typical pediatric hospital room.

Subject eligibility and recruitment: 117 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint hereafter, and shown in Figure 1 ) and one procedure room at a tertiary care hospital in the Pacific is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint Ambient air in the hospital room was sampled with the door closed and no patient present 138 (background condition) for 15 minutes. The child was then connected to the HFNC, flow was 139 then increased from 0 to 0.5 L/kg/min, to 1 L/kg/min then finally to 2 L/kg/min, then back to 0 140 L/kg/min and repeated the cycle one more time for a total of two measurements per subject at We recruited ten children ranging from 6-23 months (median nine months) and their 148 parents to participate in the study between September and November 2020. The median weight 149 of participants was 9.8 kg (range 7.3-14.0 kg). The flow rates were calculated for each child at Particle and carbon dioxide measurement: 158 Three sampling stations were deployed in the room of each study participant prior to their 159 arrival, excepting P08 and P10 who were present prior to sampling. The main sampling location 160 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint (station 1, Figure 1 ) consisted of a common 0.9 m sampling line with inlet installed in the 161 patient's breathing plane at distance ~0.5 m from the breathing zone, similar to O'Neil et. al 8 .

The tubing was 0.95 cm outer diameter conductive tubing (Bev-a-line) with the inlet directed 163 towards the patient. The sampling line was connected to a stainless-steel manifold with ports for 164 four instruments. An optical particle sizer (TSI/OPS 3330) and scanning mobility particle sizer 165 (TSI/NanoScan SMPS 3910) counted particles ranging from 0.01 to 10 μm at a time resolution 166 of one-minute. A condensation particle counter (TSI, P-Trak 8525) measured particles ranging In this study, we normalize the data reported by station 1 (near-field) to that of station 3 180 (far-field), which we take as the ambient room particle and CO2 level. Note that we lacked 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint Field co-location of instruments: 184 The Purple Air (PA) sensors and Onset CO2 sensors were co-located with the OPS and 185 LICOR CO2 analyzer during a background period, where the room was unoccupied for 15 min. 186 We used these periods to develop correction factors that were applied to the far-field (station 3) respectively. In this study, the complex fluid mechanics occurring in the patient's breathing 290 plane due to exhaled breath, HFNC airflow, and the room airflows complicate further theoretical 291 calculations of particle concentrations or emission rate originating from the patient. Humans also 292 generate particles from activity, 10 particles originating from the respiratory system versus, e.g., 293 patient movement, cannot be differentiated here.

Median values of ΔPM0.3-10 decreased slightly, though not statistically significantly, with 295 increasing HFNC flow rate. We speculate this may be the result of enhanced mixing between 296 forced air from subject and room air with higher velocities at higher HFNC flow conditions. We 297 evaluate statistical significance of differences in medians of ΔPM0.3-10 and ΔCO2 across HFNC 298 flow rates using a Wilcoxon rank sum test for ΔPM0.3-10 and a student t-test for ΔCO2, based on a 299 Kolmogorv-Smirnov test for normality. Tests for normality and statistical testing employed the 300 average of each HFNC condition conducted in duplicate across the six subjects (i.e., 12 301 independent samples of ΔPM0.3-10 and ΔCO2 for each condition). There are no statistically 302 significant differences across ΔPM0.3-10 or ΔCO2 for any comparison of flow conditions. We set 303 the threshold of significance as p < 0.0083 for 95% confidence with Bonferonni correction for 304 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint Results shown in Figure 4 reveal high variability in near-patient concentrations of PM 307 and CO2. The explanation for the mechanism behind these observations is beyond the scope of 308 this paper, though we speculate it is possible that patients with negative ΔPM0.3-10 may be low 309 emitters of particles or positioned in the space such that enhanced particle deposition is occurring 310 in the turbulence generated from airflows interacting with the patient and associated equipment 311 (bedding, instruments, etc.). Particles also deposit in the respiratory system. 11 Patient 06 and 312 Patient 04 measurements were conducted during relatively high room background PM levels, 313 perhaps contributing to the negative ΔPM0.3-10 observed. We note that prior studies have 314 observed large variability in particle emission rate and concentrations in exhalations of humans 315 during breathing and speaking. 12-16 There is debate on the size of particles that are considered 316 infectious, with droplet nuclei playing a larger role than previously considered 17our study 317 measured a broad range of potentially infectious particles including droplet nuclei.

Differences in near-field to far-field CO2 were larger and more pronounced than for PM.

CO2 levels in human breath are ~100x higher than ambient levels (~38,000 vs. 400 ppm). 18 In 320 contrast, particle concentrations in human breath in the size range 0.3 -10 um are expected to be 321 similar or lower than background levels measured in the patient rooms, e.g., Fairchild and 322 Stampfer 12 report particles in exhaled breath of <0.1 to 4 #/cm 3 . Notably, there is wide variation 323 in particle concentrations in human breath and particle generation rates during coughing, with the 324 presence of a respiratory infection and causing increased particle generation rate. 19

In contrast to the variability in ΔPM0.3-10 shown, ΔCO2 is variable but more consistently 326 positive (Figure 4b) , implying that measurements were generally made in the breathing planes of 327 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint near the median value reported. Again, we speculate that this is a result of differences in particle 331 generation across subjects that are not related to metabolism (e.g., unknown physiological factors 332 that have been previously suggested as explaining "superemission" of aerosol during speech 13 ). the patient alone, though we observe variability across patients. However, we caution that most 349 of our empirical data were collected for healthy patientsour limited data for HFNC in children 350 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint with respiratory illness showed one patient with substantially elevated near-field ΔPM0.3-10 while 351 for another patient we observe a small decrease in this metric. Further study of the impacts of 352 HFNC on particle generation and dispersion in patients with respiratory illness is warranted.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) preprint

The copyright holder for this this version posted December 14, 2020. ; https://doi.org/10.1101/2020.12.10.20245662 doi: medRxiv preprint

Clinical Management of Severe Acute Respiratory Infection When Novel 356

2019-NCoV) Infection Is Suspected. World Healthy Organization

Coinfection and Other Clinical Characteristics of COVID-19 in 360 Children

High-flow nasal cannula for acute hypoxemic 362 respiratory failure in patients with COVID-19: systematic reviews of effectiveness and its 363 risks of aerosolization, dispersion, and infection transmission

Assessment of the potential for pathogen dispersal 366 during high-flow nasal therapy

The impact of high-flow nasal cannula (HFNC) on 369 coughing distance: implications on its use during the novel coronavirus disease outbreak

Exhaled air dispersion during high-flow nasal cannula 372 therapy versus CPAP via different masks

High-flow nasal cannula for COVID-375 19 patients: risk of bio-aerosol dispersion

Characterization of Aerosols Generated During Patient 378 Care Activities

Modality of human expired aerosol size 380 distributions

Source strengths for indoor human activities that 383 resuspend particulate matter

Aerosol Technology: Properties, Behavior, and Measurement of Airborne 386 Particles. 2 edition. Wiley-Interscience

Particle Concentration in Exhaled Breath

Aerosol emission 391 and superemission during human speech increase with voice loudness

Inhaling to mitigate exhaled bioaerosols

The Size Distribution of Droplets in the Exhaled Breath of 396 Healthy Human Subjects

Fennelly KP. Particle sizes of infectious aerosols: implications for infection control. The 401 Lancet Respiratory Medicine

Risk of indoor airborne infection transmission estimated from 403 carbon dioxide concentration

Quantity and Size Distribution of Cough-405

Generated Aerosol Particles Produced by Influenza Patients During and After Illness

Size Distribution, and Characteristics of Cough-408 generated Aerosol Produced by Patients with an Upper Respiratory Tract Infection

We thank the families who volunteered to participate in the study. Paula