Aerosol emission and exposure in non-invasive ventilation

In this study, we investigated aerosol generation in noninvasive respiratory support with CPAP or HFNC to assess the risk of airborne pathogen transmission. We did not observe any statistically significant differences in aerosol emissions between CPAP or HFNC compared to NBA during activities such as breathing, speaking or coughing. The finding reinforces the notion that, contrary to previous understanding, respiratory support treatments are not inherently significant aerosol-generating procedures.

Our results are in line with previous studies for both HFNC and CPAP, but also for positive pressure non-invasive ventilation (NIV)^22,24. Similar results for HFNC have also been reported for patients with respiratory infection²³. However, increased particle production when using breathing aid has also been reported, but the effect remained smaller than the difference between respiratory activities²⁴.

In accordance with existing literature, the aerodynamic diameter of the particles observed in our study were mainly < 1 μm²⁶. During coughing maneuvers, CPAP demonstrated a non-significant decrease in aerosol emissions compared to NBA, while HFNC showed a slight increase. Gaeckle et al. reported similar non-significant trend when using NIV, indicating that positive pressure and the tightly fitted face mask may help to limit the number of emitted particles²². Similarly, Pearce et al.. (2016)²⁷ and Hamilton et al.. (2022)¹⁸ reported a reduction in aerosol emissions with CPAP compared to no intervention. Pearce et al. found a 15% reduction in smaller particles with CPAP use, and Hamilton et al. reported reduced aerosol generation even in the presence of large air leaks. A similar result was observed by Bem et al. (2021) in their study of high-flow nasal cannula (HFNC), where they found that the variation between subjects was more prominent than the effect of the intervention. In all the discussed studies, the individual variation was much larger than the differences between breathing aids. The clinical implication is that both HFNC and CPAP can be freely used, as the possible increase in infection risk is likely to be an order of magnitude smaller than the natural variation between patients.

The differences between these experimental setups were variations in measurement environments, ventilation conditions, and participant characteristics (healthy versus critically ill). Sample sizes across studies ranged from 10 to 25 participants, and experimental settings varied from controlled hospital settings to emergency departments.

There is ongoing debate and conflicting results about the concept of “super-spreaders” in the transmission of infectious diseases. It has been proposed that a small percentage (20%) of individuals are responsible for over 80% of the transmission of pathogens^28,29. Likewise, in the present study, about 37.5% of participants contribute to 80% of the emissions. This suggests that the likelihood of aerosol-based transmission of respiratory infections is influenced by unique characteristics of the individual in addition to medical intervention. We also observed that some participants exhibited notably higher particle emissions than their counterparts during all studied activities. This aligns with the concept of “super-spreaders,” who disproportionately contribute to the spread of infectious diseases due to their higher particle emissions. Given this significant individual variation, it is important to treat all patients as potential super-spreaders, as it is not possible to identify these individuals in advance.

Our study was conducted in an operating room (OR) with a high air change rate, which presented both strengths and limitations. The high ventilation rate allowed us to perform the measurements in a virtually particle-free environment, minimising contamination from external sources. However, this environment also posed a significant limitation: it accelerates the removal of particles compared to a standard clinical setting. As a result, the particle concentrations reported in our study are unlikely to reflect the absolute values observed in less ventilated areas. Despite this limitation, the high ventilation setting enabled meaningful comparisons and interpretations of the relative effects of different breathing aids, which remains valuable.

We followed a standardised protocol, with specific instructions for each phase of the study (normal breathing, deep breathing, speaking, and coughing), ensuring consistency and reproducibility of results. The use of healthy volunteers further helped to standardise the baseline emission levels, making it easier to detect differences attributed to the devices tested. However, this choice also limits the generalisability of our results, as patients with respiratory conditions may exhibit different emission patterns²⁸. Additionally, the sample size was relatively small, which reduced the statistical power to detect significant differences. Small sample sizes are a common challenge in aerosol research, given the complexity and resource-intensive nature of the measurements. Moreover, the aerosol distribution in our data exhibited a wide range, from zero emissions to very high values. This distribution, with a few “super spreaders” and many zero values, is difficult to model, presenting a challenge also encountered by other studies with similar datasets.

Finally, our measurements were taken from a distance representative of real-life exposure scenarios, without the use of a funnel. Some studies have employed funnels or close-proximity measurements to capture aerosol emissions more accurately²⁴, but such methods may not reflect actual exposure conditions. By measuring at realistic distances, our study prioritised ecological validity, though this approach may have slightly reduced the precision of particle capture.