Introduction

Non-invasive continuous positive airway pressure (nCPAP) and non-invasive pressure support ventilation (nPSV) can both be delivered with different interfaces [13]. In patients with acute respiratory failure, facial masks are effective in unloading the inspiratory muscles [4, 5], but problems such as air-leaks [6], skin breakdown [7], mask placement instability [6], and patients discomfort [1, 6] may lead to an interruption of non-invasive ventilation (NIV) and increase the risk of NIV failure. These problems can be partially overcome by the head helmet. This interface has been found to be better tolerated by patients over time, leading to longer continuous use of NIV and a lower rate of NIV interruption [811]. The helmet, however, is less efficient in pressurizing the airway [5]; moreover, the armpit braces that maintain the helmet in place may cause patient discomfort and axillary skin lesions, leading to NIV intolerance and failure [12].

Efforts to overcome, at least partially, these side-effects has resulted in the recent introduction of a new helmet (NH) into clinical practice in Europe. This NH is characterized by an anular openable ring placed underneath an inflatable cushion that secures the helmet without the need of armpit braces. In addition to eliminating side-effects associated with the armpit braces, this interface should improve, compared to the standard helmet (SH), the rate of pressurization by avoiding or at least reducing to a large extent the downward displacement of the soft collar during ventilator insufflation.

We designed this bench study to evaluate and compare the performance of the SH and NH in delivering NIV in either nCPAP or nPSV mode.

Materials and methods

Non-invasive continuous positive airway pressure and nPSV were applied to a mannequin connected to an active test lung system (ASL 5000; Ingmar Medical, Pittsburgh, PA) set using a single-compartment model, an active inspiration simulated by a semi-sinusoidal pressure waveform, and near-normal mechanical properties of the respiratory system (resistance 4 cmH2O/l/s and compliance 60 ml/cmH2O). Both modes were applied via the SH (CPAP: Castar; NIV: Castar-R; Starmed, Mirandola, Italy) (Fig. 1a) and NH (CPAP: Castar Next; NIV: Castar-R Next; Starmed) (Fig. 1b). A detailed description of the two interfaces is provided in the Electronic Supplementary Material (ESM).

Fig. 1
figure 1

a Standard helmet (SH), composed of rigid plastic ring (1), soft collar (2), two padded armpit braces (3), transparent hood (4), and an inflatable cushion (5) with its inflating line (6). The helmet is secured to the axillas by the armpit braces. b New helmet (NH), composed of a rigid plastic ring (1), inflatable cushion (2) with its inflating line (3), anular openable ring secured to the rigid ring underneath the cushion (4), and transparent hood (5). The helmet is secured to the head by the inflated cushion placed around the neck below the jaw

nCPAP (10 cmH2O) was applied at a simulated respiratory rate (RRsim) of 20 breaths per min (b/min) and two simulated inspiratory efforts (Pmus) of 10 and 15 cmH2O, respectively, using a standard flow-meter delivering a flow rate of 60 l/min. nPSV was delivered at 2 RRsim (20 and 30 b/min) and a Pmus of 6 cmH2O with the mechanical ventilator (Puritan Bennet 840; Covidien Health-Care, Mansfield, MA) set with an inspiratory pressure support (iPS) of 15 cmH2O, positive end-expiratory pressure (PEEP) of 8 cmH2O, the fastest rate of pressurization, and a cycling-off flow threshold of 25 and 50 % of the peak inspiratory flow. This settings were chosen to enhance the performance of the SH [3, 1315]. Airflow (V′) and airway pressure (Paw) at the helmet inlet during the inspiratory phase were measured using a pneumotachograph and a pressure transducer (see ESM).

The measured variables assessed during nCPAP were the inspiratory trough (ΔPawi) and the expiratory peak (ΔPawe), calculated as differences from the preset CPAP level. The measured variables in nPSV were (1) inspiratory trigger delay (Delaytrinsp); (2) expiratory trigger delay (Delaytrexp); (3) time of synchrony (Timesync), expressed as absolute value and percent of inspiratory effort duration; (4) trigger pressure drop (ΔP trigger); (5) pressure–time product during the triggering phase (PTPt); (6) pressure–time product at 200 ms from the onset of the ventilator pressurization (PTP200), as the index of pure pressurization performance [16]; (7) pressure–time product at 300 and 500 ms from the onset of the simulated effort, expressed as percentage of the area of ideal pressurization (PTP300-index and PTP500-index), as indexes of overall performance [17].

Data are expressed as the mean ± standard deviation. The SH and NH were compared for each variable using unpaired Student’s t tests. P values of ≤0.05 were considered to be statistically significant.

Results

During nCPAP, the differences in ΔPawi and ΔPawe between the two interfaces at both Pmus, although significant, were minor and clinically irrelevant (Fig. 1E in ESM). At Pmus of 10 cmH2O, the PawI was 1.2 ± 0.0 and 1.3 ± 0.0 with the SH and NH (p < 0.01), respectively, while the ΔPawe was 1.3 ± 0.0 with the SH and 1.1 ± 0.0 with the NH (p < 0.01). At Pmus of 15 cmH2O, the ΔPawi was 1.8 ± 0.0 with the SH and 1.9 ± 0.0 with the NH (p < 0.01), while the ΔPawe with the SH and NH was 1.7 ± 0.0 and 1.3 ± 0.0 (p < 0.01), respectively.

The performances of the two helmets with respect to triggering performance, synchrony, and pressurization during nPSV are shown in Table 1. The differences in ΔP trigger and PTPt, although statistically significant, were overall small and of limited clinical relevance. In contrast, Delaytrinsp, regardless of the setting, was significantly (p < 0.01) and noticeably shorter with the NH than with the SH. Delaytrexp was significantly shorter with the NH than with the SH, irrespective of the setting; however, the differences observed were relevant at RRsim 30 b/min, while they were less important at RRsim 20 b/min. It is noteworthy that with the NH at an RRsim of 20 b/min and a 50 % expiratory threshold, the ventilator slightly anticipated the cycling-off with respect to Pmus, as indicated by the negative value. Timesync, both in absolute value and percentage of inspiratory effort duration, was significantly longer with the NH than with the SH, in particular at a RRsim of 30 b/min. PTP200 significantly improved with NH, compared to SH (p < 0.01), irrespective of the setting.

Table 1 Synchrony and triggering and pressurization performance during non-invasive pressure support ventilation

Both PTP300-index and PTP500-index were significantly higher with the NH, compared to SH, irrespective of the RRsim and cycling-off threshold (p < 0.01 for all comparisons). This increase associated with NH use, compared to use of the SH, in terms of the PTP300-index (dark-gray bar) and PTP500-index (light-gray bar) is shown in Fig. 2 for all four experimental settings and ranged from 17 to 48 %. The differences in PTP500-index were remarkably smaller than those observed for PTP300-index at RRsim 20/min, but not at RRsim 30/min. Moreover, while the difference in PTP300-index was larger at RRsim 20/min than at RRsim 30/min, the setting did not influence the extent of increase in PTP500-index.

Fig. 2
figure 2

Increase in PTP300-index (dark-gray bar) and PTP500-index (light-gray bar) determined by the NH with respect to the SH. PTP 300-index airway pressure–time product during the initial 300 ms from the onset of the simulated effort, expressed as the percentage of the area of ideal pressurization, PTP 500-index airway pressure–time product during the initial 500 ms from the onset of the simulated effort, expressed as the percentage of the area of ideal pressurization, RR 20, RR 30 respiratory rate at 20 and 30 b/min, respectively, ET expiratory trigger (cycling-off flow threshold expressed as percentage of the peak inspiratory flow)

Discussion

In our study, the two helmets, despite minimal differences, performed equally well during nCPAP. During nPSV, however, the NH outperformed the SH for most of the variables considered and in most of the simulated settings.

Our study is characterized by the intrinsic limitations of all bench evaluations that attempt to mimic situations in a strictly controlled environment that are definitely more complex and variegated in the clinical setting. In particular, compared to natural breathing, the differences between respiratory cycles are minimal, as indicated by the very small standard deviation, which makes clinically irrelevant scanty differences statistically significant. In addition, our experimental setting does not allow evaluation of crucial determinants of the choice of the interface, such as comfort and tolerability [9, 11].

We limited our comparison only to one interface, the helmet. In nPSV, at equally applied inspiratory and expiratory pressures the facial mask provides greater inspiratory muscle unloading than the helmet [3, 5]. However, analogous unloading can be achieved by the helmet when a specific ventilator setting is applied, such as by increasing both PEEP and iPS by 50 % with respect to the values used with the face mask and by setting the fastest pressurization rate [3]. Also, although several models of SH using armpit braces are available for clinical use, we tested only one SH model. The reason for this choice was twofold: (1) the results of a previous study [13] showed that this SH model is one of the best performing models; (2) the SH tested is of a similar design and is made of the same materials as the NH tested, thereby allowing a head-to-head comparison related only to the novelty introduced by the new interface, i.e., the reduced upward displacement of the interface during the course of ventilator insufflation.

We applied nCPAP at 10 cmH2O because this is the value most commonly applied in the clinical setting regardless of the underlying pathology [18, 19]. The values of ΔPawi and ΔPawe during nCPAP observed in our study are slightly higher than those reported by Patroniti et al. [20] in healthy volunteers. These differences are likely due to the smaller inspiratory effort exerted by normal subjects, in contrast to the simulated Pmus values in our investigation. In fact, we found lower values of ΔPawi and ΔPawe at Pmus 10 cmH2O than at Pmus 15 cmH2O. Because the two helmets performed equally, the only advantage of the NH, relative to the SH, is represented by the absence of the armpit braces: whether this difference really improves comfort and the potential clinical benefit of this improvement will require further evaluation.

For nPSV, the values of the simulated effort, RR, resistance, and compliance, were those already utilized in previous investigations [13, 14]. The ventilator settings, i.e., fastest rate of pressurization, PEEP (8 cmH2O), and iPS (15 cmH2O), were set to mimic the specific helmet settings proved to be effective by Vargas et al. [3] in the clinical setting. With respect to SH, at both simulated RRsim and cycling-off thresholds, triggering performance, patient–ventilator synchrony, rate of pressurization, and overall performance were improved by the NH.

In conclusion, this bench study shows that, compared to the SH, the NH is equally effective in delivering nCPAP and more effective in delivering nPSV, while avoiding the need for armpit braces. Further evaluations are necessary to assess whether these advantages translate into physiological and clinical improvements.