Helmet noninvasive ventilation has been used in patients with acute hypoxemic respiratory failure for over two decades, and its use has increased during the coronavirus disease 19 (COVID-19) pandemic [1]. Compared with mask interface, the helmet interface has the advantage of delivering prolonged and uninterrupted treatment with higher levels of positive airway pressure [2]. Additionally, helmet use, compared with mask interface, is associated with less air leak, lower risk for skin injury, and lower aerosol generation [3, 4]. However, the use of helmet interface has not gained wide popularity because of the limited availability, increased cost, limited experience with its use, and the perceived intolerance of some patients to helmet interface due to claustrophobia. In this paper, we provide some insights into the differences in the design of three selected helmet trials to inform the interpretation of the findings [5,6,7].

In the trial by Patel et al., helmet noninvasive ventilation was compared with mask noninvasive ventilation. This single-center randomised controlled trial (RCT) was planned to include 206 patients but was terminated at 83 patients after observing significantly reduced risk of intubation and lower 90-day mortality with helmet noninvasive ventilation [5]. A follow-up study at 1 year of hospital discharge confirmed the reduction in mortality and demonstrated a positive effect of helmet noninvasive ventilation on functional independence [8]. The authors hypothesized that a greater tolerance of the helmet interface and higher positive end-expiratory pressure (PEEP) levels achieved could explain these findings. The HENIVOT trial (n = 109) compared helmet noninvasive ventilation with high-flow nasal oxygen (HFNO) in patients with acute respiratory failure due to COVID-19. Helmet noninvasive ventilation compared with HFNO did not improve the primary outcome of respiratory support-free days, but resulted in a lower intubation rate [6]. In a secondary analysis aimed at identifying subgroups that preferentially benefited from helmet, physiologic markers of excessive inspiratory effort while receiving standard oxygen identified a subgroup that had a greater benefit from helmet compared to HFNO [9]. This finding suggests that helmet noninvasive ventilation may have a protective effect for patients at risk for self-inflicted lung injury through effective lung recruitment. The Helmet-COVID trial (n = 320) evaluated helmet noninvasive ventilation compared to usual respiratory support in patients with COVID-19, including mask noninvasive ventilation, HFNO, and conventional oxygen. There were no differences between the two groups in 28-day mortality or intubation rate [7]. A follow-up study at 180 days of randomization confirmed the lack of mortality difference and demonstrated no difference in health-related quality of life measures [10]. This could be attributable to the notion that the physiology of acute hypoxemic respiratory failure is continually evolving, and therefore a strategy that may involve a series of modalities may be effective.

How can one reconcile the findings of these trials (Table 1)? First, the two studies that showed benefits with helmet noninvasive ventilation (Patel et al. and HENIVOT trials) were relatively small. It has been demonstrated that small RCTs and those stopped early for benefit may have implausibly large treatment effects [11]. Second, the three trials differed in the selection of the control group. One interpretation of the findings is that respiratory support with helmet may be better than respiratory support with mask noninvasive ventilation or HFNO if used alone but not better than respiratory support that allows using different or alternating modalities based on patient response. Third, the three trials differed in the level of PEEP used in the helmet and control group. This is an important issue. One study found that helmet used with higher levels of pressure support and PEEP compared with the mask settings, together with the shortest possible pressurization time resulted in better patient–ventilator synchrony [12, 13]. In the trial by Patel et al., helmet was used with a median PEEP of 8 cmH2O and mask noninvasive with a median PEEP of 5 cmH2O. In the HENIVOT trial, the helmet was used with a median PEEP of 12 cmH2O and was compared with HFNO. In Helmet-COVID, a median PEEP of 10 cmH2O was applied in patients who received the helmet noninvasive ventilation and patients in the control group who received mask noninvasive ventilation. The Patel et al. and HENIVOT trials in which there were differences in PEEP levels between the helmet and control group were associated with a reduction in intubation, but not the Helmet-COVID trial in which the same levels of PEEP were applied in the two groups. Fourth, the trials were conducted in different populations (COVID-19 versus non-COVID-19). Further studies are needed to evaluate whether helmet noninvasive ventilation effects vary across patients' populations. Fifth, the three trials enrolled patients with different severity of hypoxemia. A post hoc analysis of the HENIVOT trial demonstrated that helmet noninvasive ventilation compared to HFNO was associated with lower intubation rate among patients with severe hypoxemia expressed as arterial oxygen partial pressure (PaO2)/[fraction of inspired oxygen (FiO2) × dyspnea visual analog scale] < 30 and/or arterial carbon dioxide partial pressure (PaCO2) of less than 35 mm Hg [9]. However, subgroup analyses in the Helmet-COVID did not show any differential effect by PaO2/FiO2 ratio or PaCO2 level. Sixth, timing and duration of helmet noninvasive ventilation may be important. Although reporting the duration before enrollment may have not been standardized across the three trials, patients in the HENIVOT and Patel et al. appear to have studied interventions applied within a shorter time compared to the HELMET-COVID. In addition, duration of helmet non-invasive ventilation (NIV) application after randomization differed across studies as did respiratory support provided when off helmet NIV. Seventh, helmet noninvasive ventilation has a learning curve, and its effects may depend on the experience with its use. Eighth, the use of dexmedetomidine was common in the Helmet-COVID trial; its role in improving synchrony requires further study. Ninth, awake proning was applied in 25–30% of patients in the Helmet-COVID trial. In the HENIVOT trial, 32 patients (60%) in the HFNO group vs. 0 in the helmet group underwent prone positioning. Feasibility of prone positioning with helmet noninvaisve ventilation has been demonstrated [14]. A recent systematic review showed that awake prone positioning in patients with acute hypoxemic respiratory failure reduces intubation, although has no impact on mortality [15].

Table 1 Key differences between the three trials on helmet noninvasive ventilation
Full size table

Until further evidence becomes available, the three trials collectively suggest that helmet noninvasive ventilation is at least as effective as mask noninvasive ventilation and HFNO, especially when used in centers with experience with this approach. Better outcomes with helmet noninvasive ventilation are probably more likely with early use and with allowing the alternate use of helmet noninvasive ventilation, mask noninvasive ventilation, and HFNO to increase tolerance and with applying stringent intubation criteria. Future trials should consider early timing of the intervention to prevent undue influence of pre-randomization respiratory support practices on clinical outcomes. Randomization may need to be stratified based on physiological parameters of severity of hypoxemia or inspiratory effort to identify important sub-phenotypes that may benefit from helmet NIV. Post-intervention respiratory support may need standardization to minimize variation in post-randomization usual respiratory care clinical practices that could affect patient outcomes.