Introduction

Noninvasive positive pressure ventilation (NPPV) may provide effective ventilatory support in selected patients with acute respiratory failure [1]. NPPV can be delivered as continuous positive airway pressure (CPAP), pressure support ventilation (PSV), or volume- and pressure-cycled systems by means of a nasal or face mask (FM) [2]. Both types of masks can reduce the work of breathing (WOB) in patients with respiratory failure [3, 4, 5]. However, dyspnoic patients often breath through their mouths, causing air leakage and lowering the efficacy of NPPV when the nasal mask is used [6]. Therefore the FM is preferable in this kind of patient because it greatly improves alveolar ventilation [2]. NPPV may fail in a significant number of cases on account of technical problems, such as gas leaks around the mask [6, 7], skin lesions [8, 9], and mask discomfort [10, 11].

The "helmet" is a transparent plastic hood originally used to deliver the desired oxygen fraction during hyperbaric oxygen therapy and occasionally to deliver NPPV [12]. In contrast to the FM, the helmet does not make contact with the patient's face, and it therefore causes no skin lesions and improves the patient's comfort, thus permitting longer periods of NPPV delivery. The helmet can also be used in difficult anatomical situations such as in edentulous patients and in patients with facial trauma. However, because it is larger than the FM, the pressure within the system during NPPV may be dissipated against the high compliance of the helmet, thus interfering with correct pressurization and ventilator function. Nevertheless it does appear that the helmet can be equally as effective as the FM for administering NPPV to patients with acute respiratory failure with better tolerance and fewer complications [13].

The present study evaluated in a lung model, and in healthy subjects compared two helmets of different size vs. a FM for delivering CPAP, with continuous high- and low- flow systems and with a modern mechanical ventilation, and PSV.

Materials and methods

Interfaces

The helmet (Castar, Starmed, Modena, Italy) is a transparent, latex free, polyvinylchloride hood, joined by a metal ring to a soft polyvinylchloride collar (Fig. 1). Two underarm straps attached to the ring keep it from flying upwards when the gas flow pressurizes it.

Fig. 1.
figure 1

The helmet and the face mask

Two differently sized helmets were tested, a large one (L) and a small one (S), with 18 and 15 l internal gas volume, respectively. When the head is inserted into the helmet, the internal gas volume is reduced to 15 and 12 l. The two helmets have equal compliance (65 ml/cmH2O between 10 and 30 cmH2O of pressure; our own data). The large one was originally designed to deliver CPAP and the small one to deliver PSV.

The FM (Medivalv, Vital Signs, Totowa, N.Y., USA) is a transparent rigid FM (0.3 l internal gas volume, large size) with a soft inflatable edge and kept in place by a dedicated rubber headstrap (Fig. 1). When the mask is put on the face, the internal gas volume is reduced to 0.2 l. The edge was inflated with 20–30 ml air to adhere to the face, and the appropriate size (small, medium, or large) was chosen to provide the best fit and comfort.

The internal volume of the helmet and FM with and without the head or the face were measured by filling the devices with water.

Experimental protocol

Lung model study

The lung model consisted of a pneumatic lung simulator in which a piston inside a rigid plastic box was driven by an electrical power with a preset frequency and excursion (Starmed, Modena, Italy; Fig. 2). The lung simulator generated a sinusoidal wave flow pattern and a negative pressure in the respiratory breathing circuit during the inspiratory phase. The lung simulator was set to mimic a "high" and "low effort." The "high effort" was obtained with a tidal volume (VT) of 800 ml and a respiratory rate (RR) of 20 bpm (i.e., inspiratory time of 1.5 s, peak inspiratory flow rate of 1.2 l/s and a mean inspiratory flow rate of 0.5 l/s) while the "low effort" (VT of 800 ml) with a RR of 10 bpm (i.e., inspiratory time of 3.0 s, peak inspiratory flow rate of 0.6 l/s and a mean inspiratory flow rate of 0.25 l/s).

Fig. 2.
figure 2

The lung simulator

Different modes of ventilation were evaluated: (a) Continuous flow CPAP delivered by a valveless high-flow (100 l/m, CPAPHF) and a low-flow system (50 l/m, CPAPLF) ("Down's Flow Generator", Vital Signs) plus a latex reservoir bag (Harold, Milan, Italy) with a volume of 10 l at atmospheric pressure and 20 l at 20 cmH2O of pressure [14]. We used a spring loaded mechanical positive end-expiratory pressure valve (Medivalv) in both systems [15]. (b) Ventilator CPAP (CPAPVENT), delivered by a Servo 300 C ventilator (Siemens, Elema Solna, Sweden) set in pressure support at 0 cmH2O assistance with a flow trigger at the highest sensitivity, with the shortest (SIRT, 0%) and longest (LIRT, 10%) inspiratory rise time. The bias flow rate during flow trigger was 2 l/m. (c) PSV, delivered by the same mechanical ventilator set at 10 and 20 cmH2O of pressure support on flow trigger with the shortest and longest inspiratory rise time (as above). The PEEP levels during CPAP and PSV were 5 cmH2O.

Human study

The study included six healthy subjects (mean age 29±4 years, weight 73±8 kg). CPAPHF, CPAPVENT, and PSV were applied to each subject in random order. Unlike in the lung model study, only one level of PSV (i.e., 5 cmH2O) and the SIRT were used. The FIO2 was kept constant at 21% and the PEEP at 5 cmH2O. In all subjects recordings were made during unassisted breathing (i.e., spontaneous breathing without any modes of ventilation and interfaces). The subjects were studied in supine position. After a sufficient adaptation period (15–20 m) respiratory mechanics were recorded under each of the study conditions.

Measurement

Data were collected through an analog to digital converter at a sample rate of 200 Hz and stored for subsequent analysis by means of dedicated software (Colligo, Elekton, Milan, Italy).

Lung model study

The gas flow rate was measured with a pneumotachograph (Fleish no. 2, Fleish, Switzerland) positioned between the helmet or the FM and the lung simulator. Airway pressure (Paw) was measured by a pressure transducer (MPX 2010 DP, Motorola, Phoenix, Ariz., USA) at the side port of the helmet or FM. Three breaths at each experimental setting were evaluated. The following inspiratory variables were calculated. During CPAP: the negative airway pressure time product during CPAP (areaCPAP) measured as the area defined by the Paw curve under the PEEP level from the onset to the end of inspiratory flow. This is an index of the capability of the interface to maintain constant the airway pressure at the PEEP level during the inspiration [15], and ideally it should be 0. During PSV: (a) The airway pressure time product during PSV (areaPSV) measured as the area defined by the Paw curve from the onset to the end of inspiratory flow. It is an index of the capability of the interface to pressurize the lung simulator [16]. (b) The time to reach the set level of pressure support (TPS), measured from the onset of inspiratory flow to the point when Paw reaches the set PSV level [17]. TPS is an index of the delay of the interface to pressurize the lung simulator, and ideally TPS should be 0.

Human study

Paw was measured with the same set up as used in the lung model study. Changes in thoracic pressure were estimated from changes in esophageal pressure (Pes) measured by an esophageal balloon (Smartcath, Bicore Monitoring Systems, Irvine, Calif., USA) connected to a pressure transducer (MPX2010 DP, Motorola). Changes in abdominal pressure were estimated from changes in gastric pressure (Pga) measured by a gastric balloon. The appropriate position of the esophageal and gastric balloons was confirmed by visual analysis of the pressure curves [18].

Measurement of VT by pneumotachography may be unreliable with the helmet because part of the gas delivered by the ventilator or the continuous flow system is compressed inside the helmet and does not reach the airway. When a pneumotachograph is connected to a mouthpiece inside the helmet, the connecting wires and the nose clip constitute are cumbersome and might affect the breathing pattern. To overcome this problem we indirectly measured the VT by a chest wall movements analysis using a noninvasive optoelectronic plethysmography (OEP) [19]. OEP showed a highly accuracy of measurements of the VT during spontaneous and assisted ventilation [20]. However, our OEP system is calibrated only for the supine position [19]. A detailed description of this method has been published elsewhere [19].

In the first 2 min of each measurement period we recorded consecutive undisturbed breaths to measure VT, the relative contribution of rib cage movement to VT (RC/VT), RR, TPS (see above), duty cycle (TI/TTOT), the inspiratory work of breathing per liter (WOB), and the dynamic intrinsic PEEP (PEEPI). WOB was computed from Pes/VT loops using a modified Campbell's diagram by calculating the area under the inspiratory Pes/VT on the one hand and under the static Pes/VT curve of the chest wall on the other [21]. The theoretical value for chest wall compliance, which in normal subjects in supine position is 5% of the predicted vital capacity per cmH2O, was used to trace the static Pes/VT curve for the chest wall [21]. We also calculated the WOB performed by the rib cage compartment, applying the same method as above, plotting the Pes tracing not with the VT but considering the rib cage displacement. PEEPI was considered equal to the difference between the Pes value at the onset of negative deflection and its value corresponding to the first point at zero flow [21].

To exclude any possible expiratory muscles recruitment that could cause errors in calculating WOB and PEEPI the gastric pressure was measured [22]. Expiratory work was measured as the area enclosed by the expiratory portion of the Pes-VT loop to the right of the chest wall compliance line [14]. Missing efforts were defined as when the patient's respiratory rate determined from the esophageal pressure recording differed from the rate determined from the flow/airway pressure recordings.

Subjective and air leakage evaluation

At the end of each study condition we assessed the levels of comfort, noise, claustrophobia using a continuous scale. Subjects were asked to place a mark on a continuous line (length of 10 cm) from "worse "(cm 0), "poor" (cm 2.5), "sufficient" (cm 5), "good" (cm 7.5), to "best feeling"(cm 10) in response to the question: "How do you feel for each of the above subjective variables." The subjects were carefully instructed on the appropriate use of the scale before the beginning of protocol. The air leakage was evaluated by the physician who passed a hand around the mask or the collar of the helmet and used the same scoring system as above from: "high air leakage" (cm 0) to "no air leakage" (cm 10).

Statistical analysis

Results are expressed as mean ±SD. For each mode of ventilation the differences between the three systems were evaluated by three-ways repeated-measures analysis of variance for the lung model and one-way repeated-measures analysis of variance for the human study, followed, when appropriate, by Bonferroni's t test. A p value less than 0.05 was considered significant.

Results

Lung model

During CPAPLF the areaCPAP was significantly higher using the FM at high and low effort than with either helmet (Fig. 3, upper part). During CPAPHF the areaCPAP was similar for all the three interfaces at high and low effort. The areaCPAP at high effort was higher than at low effort regardless of the interfaces or CPAP settings.

Fig. 3.
figure 3

The areaCPAP of three interfaces in the lung model during continuous flow CPAP (above) and ventilator CPAP (below). Data are expressed as mean ±SD. White bar Large helmet; black bar small helmet; gray bar face mask. Above: CPAP LF continuous low flow CPAP; CPAP HF continuous high flow CPAP. $ p<0.01 vs. large and small helmet, *p<0.01 vs. "low effort," ^p<0.01 vs. CPAPLF. Below: L IRT longest inspiratory rise time; S IRT slowest inspiratory rise time; $ p<0.01 vs. large and small helmet, *p<0.01 vs. "low effort," ^p<0.01 vs. longest inspiratory rise time

During CPAPVENT the areaCPAP was significantly lower with the FM than with either helmet at both high and low effort (Fig. 3, lower part). The SIRT significantly reduced the areaCPAP at high and low effort with all three interfaces. However, this reduction was greatest with the FM.

During PSV the FM significantly increased the areaPSV compared to helmet L and helmet S regardless of effort, inspiratory rise time, and level of PSV (Fig. 4, upper part). The SIRT significantly increased the areaPSV both at high and low effort with all three interfaces. At 20 cmH2O of PSV both helmets had a similar value of areaPSV as to these obtained at 10 cmH2O of PSV using the FM for the same inspiratory effort and inspiratory rise time.

Fig. 4.
figure 4

The areaPSV (above) and time to reach pressure support (T PS , below)of the three interfaces in the lung model during pressure support ventilation. Data are expressed as mean ±SD. White bar Large helmet; black bar small helmet; gray bar face mask; L IRT longest inspiratory rise time; S IRT slowest inspiratory rise time; $ p<0.01 vs. large and small helmet, *p<0.01 vs. "low effort," ^p<0.01 vs. longest inspiratory rise time

Independently of effort, inspiratory rise time, and PSV level the TPS was significantly lower with the FM than with either helmet (Fig. 4, lower part).

Human study

Ventilatory variables during CPAP breathing are summarized in Table 1. During CPAPHF and CPAPVENT there was no difference in breathing pattern or inspiratory WOB with all three interfaces. PEEPI and expiratory WOB were essentially zero with all the three interfaces. Ventilatory variables during PSV are summarized in Table 2. Figure 5 shows the experimental recording tracings of tidal volume, esophageal and airway pressure during PSV and spontaneous breathing. Compared to spontaneous breathing all three interfaces reduced the inspiratory WOB and increased VT (p<0.05). There were no significant differences between the two helmets. The FM significantly reduced TPS and inspiratory WOB compared to both helmets while there were no differences in the RC/VT ratio. There were no missing efforts with the helmets or FM. Subjective evaluations are summarized in Table 3. The levels of comfort and claustrophobia were similar for the three interfaces, but noise was significantly lower with the FM than with the helmets. The air leakage was comparable for all three interfaces in all conditions.

Table 1. Breathing pattern and work of breathing during CPAP (V T tidal volume, RC/V T relative contribution of rib cage movement to tidal volume, RR respiratory rate, T I /T TOT duty cycle, WOB-L inspiratory work of breathing per liter, RC-WOB-L rib cage inspiratory work of breathing per liter)
Table 2. Breathing pattern and work of breathing during pressure support ventilation and spontaneous breathing (V T tidal volume, RC/V T relative contribution of rib cage movement to tidal volume, RR respiratory rate, T PS time to reach pressure support, T I /T TOT duty cycle, WOB-L inspiratory work of breathing per liter, RC-WOB-L rib cage inspiratory work of breathing per liter)
Fig. 5.
figure 5

An example of tidal volume (V T ), esophageal pressure (Pes), and airway pressure (Paw) tracings during pressure support ventilation using the three interfaces and during spontaneous breathing

Table 3. Subjective evaluations

Discussion

The lung model showed that: (a) during CPAPLF both helmets significantly reduced the negative airway pressure time product compared to FM; (b) during CPAPHF the three interfaces were comparable; (d) during CPAPVENT the FM significantly reduced the negative airway pressure time product compared to both helmets; and (d) during PSV the FM significantly increased the airway pressure time product compared to both helmets.

The observations in healthy subjects showed that: (a) during CPAPHF and CPAPVENT the three interfaces were comparable; (b) during PSV all three interfaces significantly reduced the inspiratory WOB compared to spontaneous breathing, but this reduction was greatest with the FM; (c) the levels of comfort, claustrophobia, and air leakage were comparable for all the three interfaces, with least noise using the FM.

Evaluation of performance: continuous flow and ventilator CPAP

The most effective CPAP (i.e., which reduced at the minimum level the inspiratory and expiratory effort) is performed when the airway pressure remains constant at the PEEP level during all the respiratory cycle [23]. Airway inspiratory fluctuations are due mainly to an insufficient gas delivery related to the individual need of flow during inspiration [14]. To minimize these fluctuations during both CPAPHF and CPAPLF, a high compliance reservoir bag is often used [24]. In the lung model when CPAP was delivered by a continuous low flow system plus a reservoir bag the areaCPAP of the large and small helmets was lower than with the FM. Thus the helmet probably serves as an additional reservoir of volume in addition to the reservoir bag.

In contrast to the lung model using CPAPHF, the three interfaces were comparable. A high inspiratory flow of gas thus achieves a better match with the inspiratory flow demand of the lung simulator regardless of the interface.

In the lung model during CPAPVENT the FM reduced the areaCPAP compared to the helmets. The mask has a lower compressible volume than a helmet, and therefore it may maintain the PEEP level better during flow delivery by the ventilator during the inspiration.

Finally, we found that independently of the interface the shortest inspiratory rise time reduced the areaCPAP more with FM than with the helmets. This was likely due to the fact that the shortest inspiratory rise time was associated with a higher initial peak flow that could match the initial inspiratory flow demand better [17].

The breathing patterns and inspiratory WOB in the healthy subjects were similar using all three interfaces with both CPAP settings. These results are slightly different from those obtained during CPAPVENT in the lung model (see above). This may be because of physiological variability in inspiratory effort, tidal volume, and inspiratory-expiratory ratio, which renders the experimental conditions less controllable than in the lung model. This suggests that "in vitro" studies are important to identify possible differences in technical behavior of different systems, but that the results must be complemented by clinical studies before they are widely considered.

Evaluation of performance: pressure support ventilation

In the lung model during PSV the FM increased the areaPSV compared to the helmets. Part of the gas delivered by the ventilator is used to pressurize the helmet, causing a longer time to reach pressure support ventilation, with a delayed rise in the inspiratory airway pressure. This may explain the lower efficiency of both helmets. Helmet S was designed by the manufacturer to reduce the compressible gas volume and to increase the efficiency during PSV. However, the 3 l decrease in the internal gas volume did not give the desired effect.

We found that the shortest inspiratory rise time markedly increased the areaPSV with the FM and the helmets. The shortest inspiratory rise time causing a lower time to reach pressure support ventilation, due to a higher initial peak flow can sooner pressurize the helmet or the FM compared to a the longest inspiratory rise time. Similar findings have been reported in vitro [16, 25] and in vivo studies [17].

In healthy subjects the FM reduced the WOB more efficiently than either helmet. Because the RC/VT ratio was the same for all three systems we can assume that all subjects activated the same thoracic muscles during inspiration [26]. The pressure applied to the respiratory system is generated against two elastic structures placed in series, the compliance of the respiratory system and of the interfaces. Because the helmet has a higher compliance than the FM it needs a higher inspired volume to reach the same airway pressure causing an increase in time to reach the selected airway pressure and a lower respiratory assistance in the initial phase of inspiration.

The expiratory WOB and PEEPI in our subjects were almost zero, indicating the absence of expiratory resistances and a good interface between the helmet or the FM and the ventilator.

Subjective and air leakage evaluation

PSV delivered by a FM reduces the inspiratory WOB and avoids respiratory muscle failure [1]. However, FM is often poorly tolerated by distress patients. The tight fitting of the mask necessary to provide adequate ventilation often requires an uncomfortable pressure on the face and can cause discomfort and skin damage [6, 7, 10, 11]. Up to 18% of NPPV failures requiring endotracheal intubation are due to the patient's intolerance [11]. The tolerance is essential to improve the efficacy and reduce the rate of failures during NPPV [8]. Although for short term the subjective tolerance of the three interfaces was tested by an analogical scale [7]. We found that the levels of comfort, claustrophobia, and air leakage were comparable across the interfaces. The noise due to gas flow was higher in both helmets than with the FM, probably because the side port of gas entry into the helmet is near the ear. In the clinical use, however, the noise never caused premature interruptions of NIPPV [13]. Air leakage is a frequent problem with a FM, particularly in edentulous patients and in patients with a full beard. The helmets by avoiding face contact can be used in any difficult anatomical situations.

Limitations

In the lung model study the expiratory effort was not measured because expiration was actively performed by the lung simulator. In healthy subjects we chose to apply a moderate level of PSV to avoid totally abolishing the work of breathing and to hyperventilate the subjects, which could have masked any difference in the interfaces.

Potential limitations of the helmets must be clarified. First, the presence of a high volume (12–15 l) inside the helmet may increase the dead space and cause carbon dioxide rebreathing. We did not measure gas exchange. However, our healthy volunteers showed no increase in ventilatory effort or changes in tidal volume or respiratory rate with either the helmets or the FM, suggesting that under these experimental conditions the bias gas flow rate of the ventilator and the high gas flow rate during CPAP help to wash out carbon dioxide. This confirms previous reports of successful clinical application of the helmet [13]. However, periodical measurement of arterial blood gas tension is advisable in the clinical setting. Second, with the helmet there would be a possible risk of asphyxia when power failure occurs; hence the helmet should always used under strictly medical vision.

Third, with the helmet the measurement of the tidal volume with traditional spirometry is not possible. The method used in this study is not clinically applicable in its current form. However, other pletismographyc techniques such as the use of "respitrace" are available, but they are all somewhat cumbersome and less reliable than classical flow spirometers. Fourth, we used only one type of large FM with a higher internal volume than others. If we had selected a smaller FM, however, we would have expected even worse efficiency during continuous flow CPAP but better efficiency during ventilator CPAP and PSV.

Conclusions

Our results, suggest that using CPAPHF the three interfaces performed in a similar way. Using CPAPVENT and PSV the FM was more efficient than the helmets at increasing the airway pressure-time product. However, the human study was performed in healthy subjects and not in acutely ill patients in whom pulmonary mechanics might interfere with the mechanical characteristics of the interfaces.

The possibility of reducing the WOB by the helmet with performances comparable to those obtained with FM simply by raising the pressure support level expands the applicability of noninvasive ventilation to those patients who do not tolerate the mask and need long-term ventilatory assistance. However, the possible advantages of the helmet on the patient tolerability and comfort must be balanced with its interference with the ventilation, greater dead space, and compliance compared to face mask.