Detection of severe acute respiratory coronavirus virus 2 (SARS-CoV-2) in the air near patients using noninvasive respiratory support devices

Noninvasive respiratory support (NIRS) devices have increasingly been used for the treatment of acute hypoxemic respiratory failure, including high-flow nasal oxygen (HFNO) and noninvasive positive pressure ventilation (NIPPV). During the coronavirus disease 2019 (COVID-19) pandemic, the use of these devices was subject to divergent recommendations¹ because of being framed as aerosol-generating procedures.^2,3 The initial reluctance to use NIRS partially contributed to an early intubation paradigm and uncertainty remains.⁴ We evaluated whether use of such devices increased severe acute respiratory coronavirus virus 2 (SARS-CoV-2) in air samples near COVID-19 patients. We detected SARS-CoV-2 positivity using real-time polymerase-chain reaction (RT-PCR).

Methods

We used environmental air samples obtained from intensive care units (ICUs) where patients had been placed in cohorts for COVID-19. From September 2, 2020, through February 12, 2021, we collected air samples in 3 ICU pods with 12 patient rooms (Supplementary Fig. 1 online). These units have central air conditioning, and each single room has air insufflation and exhaustion that recirculates with outside air, but upon recirculation the air passes through a high-efficiency particulate air (HEPA) filter.

NIPPV was delivered to patients with oronasal face masks with 2 levels of pressure through dedicated ventilators. Viral heated moisture exchangers (HMEs) were used between the single circuit and the face-mask piece. The positive end-expiratory pressure (PEEP) device was adjusted from 5 to 12 cmH₂O according to the patient’s fractionated oxygen (F_iO₂) need, and pressure support was adjusted for the patient’s comfort, synchrony, tidal volume, respiratory rate, and effort. HFNO was delivered adjusting F_iO₂ to the SpO₂ target (90%–95%) and flow adjusted to 40–60 L/min for maximum patient comfort.

Two authors, trained in air sample collection, always collected the air samples using Coriolis µ air sampler (Bertin Technologies, Montigny-le-Bretonneux, France) for 10 minutes at 300 L/min. Air samples were drawn at the bedsides of patients with recently (<72 hours) positive result for SARS-CoV-2 RT-PCR (cycle threshold [Ct] cutoff, <32), early during respiratory failure and under NIRS, either HFNO or NIPPV (Supplementary Fig. 1 online). Control air samples were obtained from the ICU aisle, ∼50 cm outside the patient’s room (Supplementary Fig. 1 online).

We tested the air samples for SARS-CoV-2 RT-PCR targeting genes E and S with RealStar SARS-CoV-2 RT-PCR Kit 1.0 (Altona Diagnostics, Hamburg, Germany). For RNA extraction, the QIAamp Viral RNA Mini Kit (Qiagen, Hilden, Germany) was used. RT-PCR positive samples underwent at least 3 blind passages for viral culture with Vero CCL81 cells (ATCC CCL-81) to assess for viable virus. Cultures were kept for at least 2 weeks and were observed daily for evidence of cytopathic effects (CPEs). Viral isolation would be confirmed only if a CPE was observed and confirmed by RT-PCR of culture supernatant with a lower Ct value on viral isolation compared to the Ct value of the sample.

Results

We obtained 49 air samples: 18 samples from 7 patients under NIPPV; 15 samples from 5 patients under HFNO; and 16 samples from the ICU aisle. The device utilization characteristics and days elapsed from symptom onset and air sample collection RT-PCR positivity are presented in Table 1. Overall RT-PCR positivity was 5 (10%) of 49 (Ct range, 33–38.8): 3 (17%) of 18 samples (for genes E and S) near NIPPV, with 2 near the same patient plus 1 (7%) of 15 samples near HFNO and 1 (6%) of 16 samples in the ICU aisle (Ct, 35.4) (both of these were positive only for the gene E). None of the RT-PCR–positive samples was positive in viral culture. The 6 retrieved patient Ct values varied from 19.75 to 36.

Table 1. Device Utilization Characteristics and Days Elapsed From Symptom Onset and RT-PCR Positivity To Air Sample Collection

Note. NIRS, noninvasive respiratory support; NIPPV, noninvasive positive pressure ventilation; TF, total face; FF, full face; HFNO, high-flow nasal oxygen; E(I)PAP, expiratory/inspiratory positive airway pressure; RR, respiratory rate; Ct, cycle threshold. Air samples from patients 1, 2, and 3 were collected after they were extubated. The other air samples were collected from patients who had not been intubated up to that moment.

^a Ct values exact number could not be retrospectively retrieved, but were lower than 32 (cycle threshold positivity value).

^b Gene E.

^c Gene S.

^d Second sample near the same patient.

Discussion

Air samples near patients under NIRS were infrequently positive for SARS-CoV-2 (10%), and none of these RT-PCR-positive samples was positive in viral culture. Our findings indicate that COVID-19 patients might use NIRS and that concerns regarding aerosolization should probably not be a reason to withhold NIRS strategies.

NIRS have been put under the umbrella of aerosol-generating procedures,⁵ though there is controversy. SARS-CoV-2 spreads when symptomatic or asymptomatic people talk or breathe as much as when they cough. A struggling patient with overt dyspnea and cough probably generates many particles to ambient air, whereas NIRS does not necessarily increase aerosol generation. If dyspnea is adequately controlled and the patient stays comfortable with either NIPPV or HFNO, then aerosol generation is probably not an important issue. Exhaled aerosols occur in multiple sizes that are associated with different generation sites and production mechanisms in the respiratory tract.⁶ In healthy individuals, neither HFNO nor NIPPV increase the risk of aerosolization when measured by an aerodynamic particle spectrometer.⁷

We compared PCR positivity for NIPPV, HFNO, and the ICU aisle. A recent study reported 30% positivity among 40 air samples, without detection of viable virus.⁸ They also reported that air contamination was no different from patients under invasive mechanical ventilation and HFNO, suggesting that the ventilatory strategy does not increase aerosol generation. Another study assessed whether HFNO or continuous positive airway pressure was associated with increased levels of viral RNA in air samples compared to standard oxygen therapy.⁹ Only 30% participants had at least 1 positive or suspected positive result and they showed that Ct values were much higher than a paired nasopharyngeal sample. Also, they did not observe viable culturable virus in their PCR positive samples.

Our study had several limitations. We did not prospectively collect Ct values from patients; however, positivity using our method was a Ct value <32, always lower than the corresponding positive air samples. Second, we did not sample surfaces as another surrogate of environmental spread, although fomites are likely not the main transmission route. Third, air samples were collected mainly from patients in the second week of disease and may have been less infective; nevertheless, this timing is representative of patients with the timing of respiratory failure onset and potential need for NIRS.

In our study, neither NIPPV nor HFNO appeared to have been associated with a higher risk of aerosol generation than being at the bedside of patients with COVID-19.