Clinical review: Helmet and non-invasive mechanical ventilation in critically ill patients

Non-invasive mechanical ventilation (NIV) has proved to be an excellent technique in selected critically ill patients with different forms of acute respiratory failure. However, NIV can fail on account of the severity of the disease and technical problems, particularly at the interface. The helmet could be an alternative interface compared to face mask to improve NIV success. We performed a clinical review to investigate the main physiological and clinical studies assessing the efficacy and related issues of NIV delivered with a helmet. A computerized search strategy of MEDLINE/PubMed (January 2000 to May 2012) and EMBASE (January 2000 to May 2012) was conducted limiting the search to retrospective, prospective, nonrandomized and randomized trials. We analyzed 152 studies from which 33 were selected, 12 physiological and 21 clinical (879 patients). The physiological studies showed that NIV with helmet could predispose to CO2 rebreathing and increase the patients' ventilator asynchrony. The main indications for NIV were acute cardiogenic pulmonary edema, hypoxemic acute respiratory failure (community-acquired pneumonia, postoperative and immunocompromised patients) and hypercapnic acute respiratory failure. In 9 of the 21 studies the helmet was compared to a face mask during either continous positive airway pressure or pressure support ventilation. In eight studies oxygenation was similar in the two groups, while the intubation rate was similar in four and lower in three studies for the helmet group compared to face mask group. The outcome was similar in six studies. The tolerance was better with the helmet in six of the studies. Although these data are limited, NIV delivered by helmet could be a safe alternative to the face mask in patients with acute respiratory failure.


Results
In the text the data are expressed as mean ± standard deviation. We analyzed 152 studies from which 33 were selected for this clinical review. Twelve of these were physiological studies, performed in healthy subjects, and 21 were clinical studies, performed in patients with acute respiratory failure ( Figure 2). Tables 1 and 2 summarize the main results.

Physiological studies Carbon dioxide rebreathing
Compared to the face mask the helmet, due to its larger internal volume, might facilitate carbon dioxide (CO 2 ) rebreathing. Patroniti and colleagues [7] found that, with continuous fl ow continuous positive airway pressure (CPAP) and a gas fl ow from 20 to 60 L/minute and positive end-expiratory pressure (PEEP) from 0 to 15 cmH 2 O, the inspiratory CO 2 concentration was always higher with the helmet than with the face mask (3.1 ± 0.15 versus 0.8 ± 0.3 mmHg, P < 0.01). Increasing the gas fl ow rate signifi cantly lowered the inspiratory CO 2 concentration.
Similarly, Taccone and colleagues [8] observed that the CO 2 concentration was similar between a continuous low fl ow CPAP (10 L/minute) and CPAP delivered by a mechanical ventilator (13.7 ± 6.6 versus 12.4 ± 3.2 mmHg). In addition, reducing the size of the helmet did not prevent CO 2 rebreathing, suggesting that the CO 2 rebreathing primarly depends on the fresh gas passing through the helmet and the amount of CO 2 produced by the patient.
Among the commercially available helmets, CO 2 rebreathing was lower than 5 mmHg and not diff erent during continuous high fl ow CPAP [9]. An antisuff ocation valve, which allows room air to enter the helmet during any interruption of gas fl ow, limited the CO 2 rebreathing but not the loss of external PEEP [9]. In a subsequent study, Milan and colleagues [10], testing three commercially available helmets supplied with antisuff ocation valves, found that the helmet with the largest valve had lower CO 2 rebreathing but a greater reduction in oxygena tion in case of interruption of the gas fl ow. Costa and colleagues [11] tested the helmet at diff erent PSV and PEEP combinations and did not fi nd any changes in CO 2 rebreathing, which ranged from 5.2 ± 3.1 to 6.7 ± 3.3 mmHg. However, during PSV of 5, 10 and 15 cmH 2 O with an increased respiratory muscle load, the helmet was always associated with more CO 2 rebreathing independent of the level of pressure support compared to the face mask (4.3 ± 0.5 versus 0.0 ± 0.0 mmHg, 3.5 ± 1.0 versus 0.4 ± 0.4 mmHg and 4.4 ± 1.3 mmHg versus 0.5 ± 0.6 mmHg; P < 0.0001) [12].
Racca and colleagues [13] evaluated if an intentional leak at the helmet expiratory port during PSV, by increasing the fl ow through the helmet, could ameliorate the CO 2 rebreathing. NIV and CPAP were delivered using closed and open circuit ventilators equipped with a plateau valve positioned at the helmet's expiratory port. CO 2 rebreathing was signifi cantly lower with the opencircuit ventilators. However, inspiratory pressure assistance signifi cantly dropped with these open-circuit ventilators, casting doubt on the choice of the optimal helmet ventilation setup.

Breathing pattern, inspiratory eff ort and comfort
Besides the larger internal volume that aff ects CO 2 rebreathing, the higher compliance of the helmet might delay ventilator assistance and may promote patientventilator asynchrony.  Chiumello and colleagues [14] evaluated the breathing pattern and work of breathing (WOB) with helmet and face masks during continuous fl ow CPAP, mechanical ventilator CPAP and PSV. During continuous fl ow CPAP, mechanical ventilator CPAP and PSV there was no diff erence in breathing pattern and WOB; on the contrary, during PSV the face mask signifi cantly reduced the WOB compared to the helmet. Th e helmet requires a signifi cant portion of the ventilator pressure in the initial phase of inspiration to pressurize its inner volume and not the patient, resulting in less assistance (that is, it takes a longer time to reach the required level of pressure support).
In a subsequent study Costa and colleagues [11], raising the level of pressure support from 5 to 15 cmH 2 O, found that the respiratory rate and inspiratory eff ort with the helmet progressively decreased and tidal volume increased compared to spontaneous breathing. Th e highest level of pressure support (15 cmH 2 O) signifi cantly increased the discomfort.
Racca and colleagues [12] compared the helmet and face mask during PSV with normal and high respiratory muscle load to mimic dyspneic patients. With normal muscle load the breathing pattern and inspiratory eff ort was not diff erent with helmet and face mask, but with high respiratory muscle load the inspiratory eff ort was signifi cantly higher with the helmet than with the face mask.
Helmet devices may predispose to auto-cycled phenomena. Its elastic properties thus predispose to fl ow variations not tracked by eff ective inspiratory or expiratory eff orts. Autocycled breathing was more common with helmet ventilation -on average double that with face mask ventilation [12]. Th e dyspnea score was signifi cantly higher with high respiratory muscle load with helmet compared to face mask ventilation [12].

Patient ventilator synchrony
PSV, the most commonly used ventilatory support during NIV, is regulated by pneumatic triggering based on fl ow criteria. To improve patient ventilator synchrony, it is possible in most current mechanical ventilators to adjust the pressurization time and the expiratory cycling off criteria to better match the neural time with the ventilator time [15][16][17][18]. Costa and colleagues [19] examined the eff ects of diff erent pressurization times and diff erent To ameliorate the asynchrony that can be present with conventional pneumatic ventilator triggering, neural triggering using diaphragm electrical activity has been developed [20]. Moerer and colleagues [21] compared neurally and pneumatically triggered PSV delivered with a helmet with regard to synchrony and patient comfort. Th e pneumatic trigger was delayed compared to the neural trigger and directly increased with the level of PSV; during pneumatic triggering the number of wasted eff orts increased with high PSV, while wasted inspiratory eff orts did not occur during neural triggering. Th e expiratory delay was always lower with pneumatic compared to neural triggering. Comfort of breathing was also lower during pneumatic triggering compared to neural triggering.

Humidity and noise
Although the optimal level of humidifi cation of inspired gases during NIV is unknown, inadequate humidifi cation can cause patient distress and favour intolerance [1,22,23]. Similar to CO 2 rebreathing related to the helmet's high internal volume, the humidity of expired gases can  mix with the fresh inspired gases, which are more dry and cold, thus increasing the level of heat and humidity, avoiding the necessity of active humidifi cation. Chiumello and colleagues [24] reported that during continuous fl ow CPAP without an active humidifi er, the temperature and humidity levels of the inspired gases were signifi cantly higher compared to non-humidifi ed medical gases and they were directly dependent on the gas fl ow passing throught the helmet. Compared to the face mask the helmet can expose the entire head to positive pressure, which may injure the tympanic membranes. Cavaliere and colleagues [25] evaluated the performance of the middle ear by recording the tympanometry and the acoustic refl ex after one hour of PSV with both the helmet and the face mask. During PSV with the helmet, the tympanometry showed a slight increase in acoustic compliance but returned to basal values after one hour, while it did not show a change with the face mask. In both groups the acoustic refl ex did not change. Th ese data may suggest the use of ear plugs in selective cases, such as during long-term use and when high airway pressures are used.

Clinical studies Acute cardiogenic pulmonary edema
During acute pulmonary edema the main benefi cial eff ects of NIV besides the improvement of gas exhange are the reduction of preload and afterload, which improves the cardiac performace [26,27].
In an observational study of 121 patients with presumed acute pulmonary cardiogenic edema, Foti and colleagues [28], in a prehospital setting applying a continous fl ow CPAP with a helmet, found a signifi cant improve ment in arterial oxygen saturation (79 ± 12% versus 97 ± 3%, P < 0.01) and in hemodynamics (systolic blood pressure 175 ± 49 mmHg versus 145 ± 28 mmHg, P < 0.01). Th e helmet CPAP was well tolerated in all enrolled patients.
In a prospective pilot study with a matched control group of patients with hypoxemic acute respiratory failure related to cardiogenic pulmonary edema, Tonnelier and colleagues [29] reported that helmet CPAP signifi cantly reduced respiratory rate and heart rate and improved oxygenation (158 ± 94 mmHg versus 145 ± 28 mmHg), similar to the mask. Control patients were selected from a group of patients with acute respiratory failure due to cardiogenic pulmonary edema treated with a facial mask. Th e helmet allowed a longer period of CPAP without any adverse event and good tolerance.

Hypoxemic acute respiratory failure
In a large multicenter survey of patients with hypoxemic acute respiratory failure, NIV was successful in avoiding intubation in 70% of the patients [30]. Patients with a high severity score (simplifi ed acute physiology score (SAPS) II >34), older age, acute respiratory distress syndrome or pneumonia, severe metabolic acidosis, severe hypoxemia (partial pressure of oxygen in arterial blood (PaO 2 )/fraction of inspired oxygen (FiO 2 ) <170) or failure in improvement in PaO 2 /FiO 2 after one hour of treatment were at higher risk of failure [31].
In a matched control study, Antonelli and colleagues [32] evaluated the effi ciency of PSV delivered by helmet or face mask in patients with hypoxemic acute respiratory failure. Both groups had a similar improvement in oxygenation within the fi rst hour; however, at support discontinuation the increase in oxygenation was higher for patients who received PSV by helmet (267 ± 104 mmHg versus 224 ± 81 mmHg, P < 0.05). Th e total duration of PSV was similar (40 ± 30 hours versus 42 ± 31 hours), as well as the intubation rate and hospital mortality. No patients in the helmet group compared to 38% of the mask group had intolerance to NIV.
Th e application of periodic deep insuffl ation (sighs) during invasive mechanical ventilation may improve gas exchange [33][34][35]. In a prospective cross-over study, Cammarota and colleagues [36] found that during CPAP with either the helmet or face mask, the sigh (that is, an increase of airway pressure from 10 to 20 cmH 2 O for 8 seconds every minute) signifi cantly improved the arterial oxygenation and reduced the respiratory rate. Independent of sigh, the helmet CPAP group had higher oxygenation while the tolerance was similar in the two groups.
Similarly to the previous study [36], in hypoxemic acute respiratory failure Isgrò and colleagues [37], applying a periodic sigh or two CPAP levels (similar to bi-level positive airway pressure (BIPAP) ventilation) during continuous fl ow CPAP with a helmet, found a signifi cant improvement in oxygenation compared to basal CPAP (109.2 ± 33.9 mmHg versus 124.5 ± 45.2 mmHg and 128 ± 52 mmHg). Th ere was no signifi cant diff erence between sigh and BIPAP regarding oxygenation levels.
Compared to PSV, neurally adjusted ventilatory assist (NAVA) improved ventilator synchrony in healthy subjects [21]. Cammarota and colleagues [38] evaluated the short-term physiologic eff ects of NAVA compared to PSV delivered with helmet in postextubation acute respiratory failure. NAVA signifi cantly increased the ventilator inspiratory and reduced expiratory time. No asynchrony was present with NAVA, while there were no diff erences in gas exchange and respiratory rate.

Immunocompromised patients
Th e respiratory complications in immuncompromised patients remain the main cause of morbidity and mortality; thus, respiratory support that avoids or reduces pulmonary complications could be useful [39,40].
Principi and colleagues [41], in a prospective clinical study of hematological malignancy patients with hypoxemic acute respiratory failure compared to historical matched controls, observed that no patient failed helmet CPAP due to intolerance of the technique compared to 11.4% of patients in the mask group, and that helmet CPAP could be applied continuously over a much longer period of time than mask CPAP (28.4 ± 0.2 hours versus 7.5 ± 0.4 hours, P < 0.001). Th e oxygenation improvement was equal in the two groups but the intubation and mortality rates were lower with the helmet (0 versus 41% and 23 versus 47%, P < 0.001).
Rocco and colleagues [42], in a matched controlled study of immunocompromised patients of diff erent etiologies of acute respiratory failure, showed that patients receiving PSV with helmet had signifi cantly lower NIV discontinuations in the fi rst 24 hours than patients treated with mask; also, fewer complications related to device were reported (that is, skin necrosis, P = 0.01). Oxygenation, intubation and mortality rates were similar (202 ± 61 mmHg versus 224 ± 111 mmHg, 37% versus 47%, and 47% versus 31%, respectively).
In an observational study of immunocompromised patients with acute respiratory failure, Rabitsch and colleagues [43] reported that helmet PSV signifi cantly improved arterial oxygenation and respiratory rate, and only two patients (20%) were intubated.

Community-acquired pneumonia
Severe community-aquired pneumonia with intensive care admission and associated hypoxemic acute respiratory failure can require respiratory support in up of 60% of patients [44]. Carron and colleagues [45], in a prospective observational study including 64 consecutive patients with acute respiratory failure due to communityaquired pneumonia, investigated the failure of NIV. NIV was delivered as PSV with helmet. It was found that NIV succeeded in 43% of patients and failed in 56%. Th e only two independent factors associated with failure were changes in arterial oxygenation and oxygenation index between admission and after 1 hour of NIV.
In a large multicenter randomized controlled trial in patients with hypoxemic acute respiratory failure due to community-aquired pneumonia, Cosentini and colleagues [46] compared continuous fl ow CPAP delivered by helmet and oxygen therapy for improving oxygenation. Th e primary end point was the time required to reach a PaO 2 /FiO 2 ratio above 315 mmHg. Th is was reached in a signifi cantly shorter time in the helmet group compared to the control group (1.5 hours versus 48 hours, P < 0.001). In the helmet group 95% of patients reached this end point compared to 30% of the controls (P < 0.001). No patients required intubation or died during the study.
In the last recent pandemic due to infl uenza A (H1N1) it was reported that patients admitted to intensive care for acute respiratory failure with severe hypoxemia required respiratory support in up of 80% of cases [47,48]. A retrospective observational study evaluated the use of NIV in all patients admitted to intensive care with presumed or confi rmed infl uenza A (H1N1) infection and hypoxemic acute respiratory failure [49]. NIV was delivered as CPAP and PSV with a helmet or face mask. Th ere was a signifi cant improvement in gas exchange and respiratory and heart rates decreased. None of the patients required orotracheal intubation (100% success) and all the patients survived.
In a matched-control study, Conti and colleagues [54] also found that in patients with acute respiratory failure after major abdominal surgery, PSV delivered by a helmet signifi cantly improved oxygenation and reduced intubation rate (20% versus 48%, P = 0.036) compared to PSV with facial mask. Th e complications (mask intolerance, major leaks and ventilator-associated pneumonia) were signifi cantly lower in the helmet group compared to the face mask group (12% versus 32%, P = 0.06).
In a prospective observational study evaluating helmet CPAP in postoperative patients who developed acute hypoxemic respiratory failure, Redondo-Calvo [55] and colleagues found that 74.7% of the patients did not require intubation. Th e intubated patients presented higher levels of illness and lower improvement in oxygenation and CPAP duration. Th e intubated patients had a longer hospital stay and higher rate of hospital deaths compared to unintubated (30.2 ± 20.1 days versus 12.7 ± 8.2 days, P < 0.001, and 44% versus 15%, P = 0.004).
Although the NIV is commonly used to treat acute respiratory failure in postoperative patients, it has also been used to prevent acute respiratory complications after surgery [50]. Barbagallo and colleagues [56] randomized 50 patients after elective lung resection to a prophylactic continuous fl ow CPAP with helmet for two hours or to oxygen therapy. Th e helmet group had significantly higher oxygenation without any diff erence in postoperative complications and mortality.

Hypercapnic acute respiratory failure
In COPD patients, NIV is recommended to improve gas exchange, and to decrease respiratory workload and the need for tracheal intubation [1,4].
In a cohort study, Antonelli and colleagues [57] evaluated the eff ect of PSV delivered by helmet or by face mask on gas exchange and intubation rate in COPD patients with acute exacerbation. After one hour both groups presented a signifi cant reduction of partial pressure of carbon dioxide in arterial blood (PaCO 2 ) and improvement in pH. However, the decrease in PaCO 2 was lower in the helmet group compared to the face mask group (75 ± 15 mmHg versus 66 ± 15 mmHg, P = 0.01). Also, on discontinuing support, PaCO 2 was higher (P = 0.002) and pH lower (P = 0.02) in the helmet group. Th e improvements in oxygenation and respiratory rate were similar as well as the intubation rate (30% versus 42%). Intensive care and hospital mortality were not diff erent between the groups.
Antonaglia and colleagues [58] ventilated a series of patients with severe exacerbation of COPD using PSV delivered by face mask for two hours; subsequently, only those in whom gas exchange improved were randomized to helmet or face mask. After four hours of NIV, the face mask group had a signifi cantly lower PaCO 2 compared to the helmet group (63 ± 14 mmHg versus 70 ± 4 mmHg, P = 0.01) with no diff erence in oxygenation or respiratory rate. However, 9 of the 20 patients (45%) in the mask group compared to 2 of 20 (5%) in the helmet group required intubation (P < 0.01).
In a small group of hypercapnic patients with severe COPD recovering from acute exacerbation, Navalesi and colleagues [59] evaluated PSV delivered by a helmet or face mask in random order. Compared to spontaneous breathing, NIV reduced PaCO 2 with both devices (from 55.9 ± 7.3 mmHg to 52.0 ± 7.1 mmHg with helmet, P < 0.05; and from 55.5 ± 7.7 mmHg to 51.7 ± 8.5 mmHg with face mask, P < 0.05). Ineff ective inspiratory eff orts were signifi cantly more common with the helmet and although the WOB decreased to a similar extent as for spontaneous breathing, with the helmet the delay between inspiratory eff ort and ventilator support was signifi cantly longer.

Interfaces
In a prospective cross-over study Vargas and colleagues [60], in a group of patients at high risk for respiratory distress, compared three diff erent NIV settings: PSV delivered by face mask; PSV at the same pressure support and PEEP with helmet; and PSV with 50% increases in pressure support and PEEP with helmet. At the same level of pressure support the helmet had a low inspiratory eff ort compared to face mask. Th e increase of PSV reduced the inspiratory eff ort to a similar extent as with the face mask. Patient ventilator asynchrony was more frequent with the helmet, while respiratory rate and patient comfort were similar among the three conditions.

Novel indications
Arterial oxygenation lower than 75 mmHg with an oxygen fraction higher than 50% is considered a contraindication to fi beroptic bronchoscopy [61]. Antonelli and colleagues [62] investigated the feasibility and safety of fi beroptic bronchoscopy with bronchoalveolar lavage during NIV delivered with a helmet in patients with acute respiratory failure. Oxygenation did not change throughout the procedure and dropped only 2% at the end of the fi beroptic bronchoscopy. No patients required sedatives or analgesics.

Sedation
Similar to invasive mechanical ventilation, sedation has been advocated to improve NIV tolerance and reduce the rate of failure [63,64].
In a prospectively uncontrolled study, Rocco and colleagues [65] evaluated the continuous infusion of remifentanil in patients with hypoxemic acute respiratory failure during NIV with helmet or mask. Th e mean remifentanil dose administered was 0.07 ± 0.03 μg/kg/ minute and infusion lasted 52 ± 10 hours in the success group. Th irty-six patients were enrolled and 22 (61%) continued NIV treatment; after one hour respiratory rate decreased and oxygenation increased with both helmet and face mask. Fourteen patients failed (39%) and required endotracheal intubation because of persistence of discomfort.
Th e main application of NIV with helmet was for acute cardiogenic pulmonary edema, hypoxemic acute respiratory failure, community-acquired pneumonia, hypercapnic acute respiratory failure, and in post-operative and immunocompromised patients. Th e main favourable characteristics of the helmet (Table 3), such as low distensibility, absence of any contact with the face, minimum presence of air leaks, the possibility to deliver continous fl ow CPAP as well as non-invasive positive pressure ventilation, can extend the application of NIV in patients with acute respiratory failure. However, the high internal volume can promote higher CO 2 rebreathing, patient ventilator asynchrony and lower reductions in WOB compared to the face mask. Higher levels of pressure support and faster pressurization rates, however, could improve the effi ciency of the helmet to be comparable to the face mask. Table 4 summarizes general recommendations to optimize NIV with the helmet.

Conclusion
Th e helmet has been shown to be an eff ective interface for the application of NIV, but compared to the face mask it may increase patient ventilator asynchrony and CO 2 rebreathing. However, the helmet is better tolerated, allowing longer use. Further studies are required to defi ne the ideal patient populations and open up new clinical indications for NIV with the helmet.

Competing interests
The authors declare that they have no competing interests.

Pressure-support ventilation
Apply higher PEEP and pressure support level (50% higher than those applied with face mask) Use the maximum pressurization rate