A snorkel mask with a 3D-printed adaptor to HEPA filter was used in a COVID-19 emergency ward to provide safe respiratory support.

Noninvasive positive pressure ventilation (NIPPV) devices can generate aerosols spreading the virus in hospital wards. Initial reports on the use of snorkel masks adapted for NIPPV1-3 offered hope. Landry et al.4 evaluated snorkel masks performance in the laboratory in Respirology (2020). However, a leak and lower fraction of inspired oxygen (FIO2) were registered, attributed to a one-way valve at the front of the snorkel mask.4 Connection to NIPPV should, therefore, incorporate high-efficiency particulate air (HEPA). Our team designed the ‘Molinari connector’ to adapt the snorkel mask to an HEPA filter, allowing connection with high flow systems and positive end expiratory pressure (PEEP) valve in continuous positive airway pressure (CPAP) or BI-Level. It was 3D-printed using thermoplastic ABS (acrylonitrile butadiene styrene) filament (1.75 mm, 3Dvale brand, ArtBox 3D; Figure 1). This was an extraordinary measure when no randomized clinical trials could be conducted (because the allocation was based on the availability of resources), and when no studies had been published in clinical scenarios. We evaluated whether the system would provide adequate respiratory support to patients with acute respiratory failure and suspicion of COVID-19. We prospectively evaluated a cohort of 60 consecutive patients admitted between August 2020 and September 2021 to a hospital providing services to the Brazilian health system. We monitored clinical variables, respiratory distress markers, arterial blood oxygen, intubation rate and survival. We compared patients using the snorkel (n = 30) versus the traditional mask (Respironics Performax Full Face Mask, Philips; n = 30). The choice of mask was based on stock availability and patient's consent (the intervention was not determined by this research). The Ethics Committee approved the protocol. Patients' mean age was 57.05 ± 13.82 years, 41 (68.9%) were men; 54 (90%) had COVID-19 confirmed. Even without randomization, the groups were comparable regarding illness severity, PaO2/FiO2 ratio and lung involvement (at scans). All 30 patients using the snorkel mask used CPAP; in the traditional mask, 20 used CPAP and 10 BI-Level. PaO2/FiO2 ratio increased significantly after 30 min of NIPPV for both groups, with at least 35% of improvement (143.20 ± 20.00 to 193.33 ± 23.80, p = 0.016 in snorkel mask; 176.90 ± 25.70 to 305.12 ± 36.64, p = 0.005 in traditional mask; between-groups p = 0.017). Patients using the snorkel mask had higher PEEP (9.77 ± 018 vs. 9.00 ± 0.30, p = 0.012) and higher HACOR score (p = 0.049). Snorkel mask users spent more days on NIPPV (p = 0.04), but not time of session or hospital stay. The intubation rates (45% vs. 47%, p = 1.000) and death (32% vs. 27%, p = 0.580) were similar between groups. All patients had significant improvements in all clinical variables comparing the moments before and 30 min after respiratory support except for mean arterial pressure and heart rate, which did not improve in NIPPV. In the first 30 minutes, there was a significant reduction of respiratory rate (29.33 ± 1.08 to 23.80 ± 0.96, p = 0.001 in snorkel mask; 28.13 ± 1.37 to 25.23 ± 0.95, p = 0.029 in traditional mask; snorkel vs. traditional p = 0.453) and increased SpO2/FiO2 (146.19 ± 13.61 to 166.92 ± 2.64, p = 0.014 in snorkel mask; 158.21 ± 15.49 to 167.93 ± 5.34, p = 0.017 in traditional mask; snorkel vs. traditional p = 0.525). The snorkel mask could be properly connected and safely attached to equipment providing flux over 80 L/min and PEEP ≤15 cm H2O. The possibility of an increase in dead space and rebreathing of exhaled air were not confirmed by the blood gas results (PaCO2 38.23 ± 1.99 to 40.40 ± 1.74, p = 0.131 in snorkel mask; 37.17 ± 1.35 to 40.34 ± 1.93 in full-face mask, p = 0.336, between-groups p = 0.051), intubation and death rates. Thus, the snorkel mask allowed ventilatory assistance without impairing gas exchange or causing death. The Molinari connector allowed 30 patients in this study, and almost 100 patients afterwards to receive respiratory support during the pandemic. During the outbreaks, it was impossible to purchase the traditional supplies in our region, not for financial reasons but mainly because they were not manufactured quickly enough to supply all hospitals—the snorkel mask was the alternative we found, proven at least as effective than the traditional and with no leaks and no major safety events. These patients would have had ventilatory support denied or converted to invasive modalities unnecessarily. The experience of designing the system created skills and knowledge in our team and preparedness in our hospital. None declared.