Clinical review: Humidifiers during non-invasive ventilation - key topics and practical implications

Inadequate gas conditioning during non-invasive ventilation (NIV) can impair the anatomy and function of nasal mucosa. The resulting symptoms may have a negative effect on patients' adherence to ventilatory treatment, especially for chronic use. Several parameters, mostly technical aspects of NIV, contribute to inefficient gas conditioning. Factors affecting airway humidity during NIV include inspiratory flow, inspiratory oxygen fraction, leaks, type of ventilator, interface used to deliver NIV, temperature and pressure of inhaled gas, and type of humidifier. The correct application of a humidification system may avoid the effects of NIV-induced drying of the airway. This brief review analyses the consequences of airway dryness in patients receiving NIV and the technical tools necessary to guarantee adequate gas conditioning during ventilatory treatment. Open questions remain about the timing of gas conditioning for acute or chronic settings, the choice and type of humidification device, the interaction between the humidifier and the underlying disease, and the effects of individual humidification systems on delivered humidity.

Th e human airway has an important role in heating and humidifying inspired gas, and recovering heat and moisture from expired gas. Th e amount of water vapor in a gas mixture can be measured as absolute humidity (AH) or relative humidity (RH) in relation to the tempera ture. AH is the total water present in the gas (mg H 2 O/L) and RH is the amount of water present expressed as the percentage of maximum carrying capacity at a given temperature [1]. Th e human airway must provide gas at core temperature and 100% RH at the alveolar surface in order to optimize gas exchange and protect lung tissue [2].
Non-invasive ventilation (NIV) is a mechanical venti lation modality that does not utilize an invasive artifi cial airway (endotracheal tube or tracheostomy tube) [3]. NIV is usually delivered through a nasal or oro-nasal mask so the inspired gas passes through the upper airway where it is conditioned. Like during spontaneous breathing, patients under NIV require adequate humidifi cation and heating of the inspired air (that is, gas conditioning) [3]. NIV delivers inspired air at high fl ow rates, which may overwhelm the usual airway humidifi cation mechanisms. Inadequate gas conditioning has been associated with anatomical and functional deterioration of nasal mucosa (ciliary activity, mucus secretion, local blood fl ow, nasal resistance). In addition, there are also negative eff ects on tolerance to NIV when a patient breathes inadequately humidifi ed air [1,3-5] (Table 1).
Metaplastic changes and keratinization of the nasal epithelium and submucosa have been reported in patients on home-NIV when the level of humidifi cation was inadequate for long periods [5]. Th ese histopathological fi ndings were confi rmed by our recent survey, which found similar structural changes of the nasal mucosa in four patients with acute respiratory failure treated for 7 days with NIV without a humidifi cation system added (unpublished data; Figure 1). Th is suggests that changes in the nasal mucosa occur relatively early after starting NIV in an acute setting and that humidifi cation should be considered even when only short-term use of NIV is expected.
Inadequate airway gas conditioning may have serious consequences in critically ill patients using NIV [6,7]. Diffi culties were recently reported with intubation in patients failing a trial of NIV delivered at high inspiratory oxygen fraction with a low level of humidifi cation [6]. One case study showed ispissated secretions, causing life-threatening airway obstruction in a patient using NIV for hypoxemic respiratory failure [7]. A high fraction Abstract Inadequate gas conditioning during non-invasive ventilation (NIV) can impair the anatomy and function of nasal mucosa. The resulting symptoms may have a negative eff ect on patients' adherence to ventilatory treatment, especially for chronic use. Several parameters, mostly technical aspects of NIV, contribute to ineffi cient gas conditioning. Factors aff ecting airway humidity during NIV include inspiratory fl ow, inspiratory oxygen fraction, leaks, type of ventilator, interface used to deliver NIV, temperature and pressure of inhaled gas, and type of humidifi er. The correct application of a humidifi cation system may avoid the eff ects of NIV-induced drying of the airway. This brief review analyses the consequences of airway dryness in patients receiving NIV and the technical tools necessary to guarantee adequate gas conditioning during ventilatory treatment. Open questions remain about the timing of gas conditioning for acute or chronic settings, the choice and type of humidifi cation device, the interaction between the humidifi er and the underlying disease, and the eff ects of individual humidifi cation systems on delivered humidity. of inspired oxygen was also entrained into the NIV circuit and this extra anhydrous gas may have contributed to the airway obstruction.
Inadequate humidifi cation can also cause signifi cant discomfort for chronic NIV users. In an experimental setting, Wiest and colleagues [8] showed that drynessrelated symptoms started to appear when AH was lower than 15 mgH 2 O/L. Normal subjects undergoing NIV without humidifi cation scored discomfort signifi cantly higher on a visual analogue scale than when heated humidifi cation (HH) was used, and this was associated with increased nasal airway resistance (NAWR). HH improved comfort and compliance to NIV [6,9].
Similarly, Lellouche and colleagues [10], in healthy subjects under continuous positive airway pressure (CPAP) delivered by a bucco-nasal mask, found the level of comfort was signifi cantly lower when no humidifi cation was used. When a humidifi cation device was applied and AH was higher than 10 to 12 mgH 2 O/L comfort was signifi cantly better [10].
In a very comprehensive compliance analysis in patients in home-NIV, Nava and colleagues [11] compared two humidifi cation systems, HH and a heat and moisture exchange fi lter (HME); compliance was much better (75% of patients) with the former. However, other symptoms, such as dry throat, the number of hospital admissions and the rate of complications caused by infection (mainly pneumonia), were similar with the two systems [11]. Similar results were published by Massie and colleagues [12], who stressed the importance of 'early' humidifi cation from the very start of ventilatory treatment, so as to ensure the best possible compliance in home-NIV patients. Some other authors, however, maintain that humidifi cation has no really signifi cant eff ect on adherence to chronic NIV [13].

Contributors to gas conditioning during non-invasive ventilation
Th e optimal hygrometric values of AH or RH in the diff erent NIV applications have not yet been established, since most data come from experimental studies that took direct (nasal cavities) or indirect (thermometers) measurements [14]. However, analysis of the need for humidifi cation during NIV must clearly take account of the following parameters: 1) air leaks; 2) interface for NIV delivery; 3) type of ventilator; 4) room temperature; 5) temperatures of inhaled gas and the vaporization chamber; 6) airfl ow and pressure at the entrance of the humidifi cation system; and 7) type of humidifi cation system.

Air leaks
Mouth and/or peripheral mask air leaks, especially in tachypneic patients under NIV, cause unidirectional nasal air fl ow, so the mucosa recovers less heat and moisture during expiration. Th is may cause a continuous drop in AH. An increase in NAWR is the typical consequence of large mouth air leaks during NIV from a nasal mask. Th is refl ects the nasal vasoconstrictive response to prolonged inhalation of dry air [4,5]. Clinically, increased NAWR is likely to lead to unsuccessful acclimatization to NIV in the chronic setting, and to the failure of NIV to improve gas exchange and dyspnea in the acute setting. Air leaks will also aff ect the performance of HMEs because the HME recovers less moisture when the expired air is drier.
Th e site of the leak, either unintentional (around the mask) or intentional (mask or circuit) also aff ects humidity. While the unintentional leaks are likely to have a substantial impact on AH because of the compensatory increase in inspiratory fl ow, the intentional leaks are not known to contribute to insuffi cient humidifi cation of air during NIV. We can only speculate that ineffi cient washout of the exhaled air throughout an unintentional leak system (that is, plateau valve, anti-rebreathing valve) is likely to involve a substantial degree of humidity, especially in interfaces with high dead space, such as a total face mask.
In a study in adult volunteers, nasal CPAP with a mouth leak resulted in a three-fold increase in NAWR that was substantially attenuated by eff ective humidi fi cation [15].  Th ese fi ndings were confi rmed in adult volunteers undergoing NIV [16]. Th is high NAWR would result in substantially lower eff ective pressure being transmitted to the nasopharynx and subsequently to the distal airway ( Figure 2).

Interfaces
Th e most common interfaces used to deliver NIV are nasal and facial masks, the former being tolerated better in the chronic than the acute setting. However, nasal masks tend to have more leaks than face masks, and this can result in inadequate gas conditioning of inspired air. Th e use of an oro-nasal mask avoids the changes in RH related to mouth leaks [17]. Th is choice is likely to be crucial to the success of NIV in selected categories of acute patients (for example, need for prolonged venti latory support, diffi culty in spontaneously clearing secretions, mouth-breathers) [18,19].
Physicians using the helmet to deliver NIV must be particularly careful. Compared to the other more popular interfaces (nasal, oro-nasal, total face masks), the helmet has a much larger inner space (more than 10 liters), which may act as a 'reservoir' of humidity because of the amount of exhaled gas that remains there [19]. Th e clinician must therefore carefully adjust the humidification of the inhaled gas depending on the type of interface and the resulting leak pattern.

Ventilator types
Intensive care ventilators, home-care mechanical ventilators, and high-fl ow CPAP systems operate by providing a very high inspiratory fl ow to compensate for the inspira tory demand of a patient with acute respiratory failure and the air leaks when applied in a NIV mode. Th is issue was originally studied by Poulton and Downs [20] with high airfl ow CPAP systems delivered with a non-invasive interface.
Th e technical aspects are more sophisticated for the diff erent NIV equipment introduced in the last few decades. Th e impact on humidifi cation depends on the type of ventilator used in acute and chronic settings. Single-circuit home ventilators and/or dedicated NIV ventilators equipped with a turbine or piston diff er from double-circuit ICU ventilators, which are pneumatic and supplied with high-pressure sources of gas. Home models -that is, blower-based devices -compress room air and have higher humidity than ICU ventilators, which obtain dry gas directly from a mains hospital outlet. Th is was clearly depicted by Wiest and colleagues [21], who demonstrated that ICU ventilators provided a lower level of AH during NIV than turbine mechanical ventilators (5 versus 13 mgH 2 O/L) [21]. Here we have to bear in mind that AH lower than 5 mgH 2 O/L is 'critical' for the likelihood of complications related to inadequate gas condition ing of inspired air [21].
Given the decrease in humidity resulting from the anhydrous nature of medical oxygen, physicians have also to remember that the higher the oxygen fraction delivered during NIV, the greater the risk of inadequate gas conditioning without the addition of HH or a HME.
Chiumello and colleagues [22] compared the hygrometric values in a helmet system during CPAP with or without HH, delivered by either a mechanical ventilator or a continuous low (40 L/minute) or high (80 L/minute) fl ow CPAP, in nine patients with acute respiratory failure and ten healthy subjects. Th e HH system raised the AH during ventilator CPAP (from 18.4 ± 5.5 mgH 2 O/L to 34.1 ± 2.8 mgH 2 O/L) and with CPAP at low fl ow (from 11.4 ± 4.8 mgH 2 O/L to 33.9 ± 1.9 mgH 2 O/L) and with CPAP at high fl ow (from 6.4 ± 1.8 mgH 2 O/L to 24.2 ± 5.4 mgH 2 O/L). Without HH, the AH was signifi cantly higher with ventilator CPAP than with the continuous low fl ow and high fl ow. Th e level of comfort was similar for all three modalities, with or without the HH. Th e fi ndings in healthy individuals were similar to those in patients with acute respiratory failure.

Ambient temperature
Even though insuffi cient humidifi cation during NIV can be caused by specifi c climatic or environmental conditions in the place where the treatment is implemented (for example, an excessively cold sleeping area), the majority of cases suff ering from excessive airway dryness are due to technical factors related to the ventilation process itself and its interaction with the patient (interface, leaks, inspired oxygen fraction, respiratory rate, use of humidifi ers). Th e eff ects of ambient temperature probably only need be considered for patients who sleep in very cold premises.

Temperature of the inhaled gas
Conditioning inspired gas involves both heating and humidifi cation. To facilitate gas exchange and protect lung tissue, inspired gas must reach body temperature by the time it arrives at the alveolar surface [23]. How easily this can be achieved depends on the temperature of the inhaled gas. Th e ventilator power source aff ects gas temperature, as turbine-driven systems create more heat than piston-driven ventilators. Increasing levels of inspiratory positive airway pressure (IPAP) have also been reported to raise the gas temperature with a turbine-driven NIV device [24]. Th is tended to preserve AH at high IPAP levels, but it was not suffi cient to maintain adequate humidity when ambient RH was low. In this bench study, HH set at its highest temperature was most eff ective in countering the deleterious eff ects of NIV on delivered humidity. In clinical practice the temperature setting for HH may be based on the patient's tolerance.

Airfl ow at the entrance to the humidifi er
Th e impact of the airfl ow entering the HH humidifi cation chamber is one of the most important physical phenomena aff ecting airway humidity. Th is was largely studied by Wenzel and colleagues [25], who analyzed the factors underlying humidifi cation capacity over a variable range of airfl ows at the entrance to the humidifi cation chamber (20,55, and 90 L/minute). At high fl ow rates, many commer cial humidifi ers were unable to generate adequate RH. Clinically, this suggests that a humidifi er alone may not be enough to ensure adequate humidifi cation of inspired gases, particularly at high fl ow rates [26].

Types of humidifi cation system
In general, HH and HME technically produce similar AH levels (25 to 30 mgH 2 O/L), which are adequate for the physiological functioning of the upper airway. However, the gas conditioning performance of each system may vary over a range of respiratory rates, especially at the ventilator airfl ow that enters the system through the humidifi cation chamber [18,27,28].
Th e choice of active HH or HME may have repercussions on respiratory mechanics -such as tidal volume, minute volume, and work of breathing -and gas exchange [29,30]. For NIV, signifi cant disadvantages have been observed with the HME compared to the HH (Table 2). HMEs have been associated with greater dead space and possibly also CO 2 retention. Th is was shown by Jaber and colleagues [29], who reported higher partial pressure of arterial carbon dioxide (PaCO 2 ) during NIV with HME than with HH. Similarly, Lellouche and colleagues [30] found inspiratory eff ort was greater in patients with hyper capnia during HME than HH. Th is has also been associated with increased work of breathing. On the basis of these studies, therefore, it would seem that HH is superior to HME during acute NIV but this advantage is probably limited to acute rather than chronic settings.
Th e choice of the humidifi er for NIV must also take into consideration the interface, and the amount of unintentional leaks. In the absence of substantial leaks, the AH was no diff erent with HH or HME when using a face mask during NIV [10]. However, when there were excessive leaks the AH signifi cantly dropped when using HME (around 40%). Th e leak aff ects the HME's performance by changing the diff erence between inspired volume (cool air) and expired volume (warm, moist air).
In their bench study, Holland and colleagues [31] examined the eff ects of mechanical ventilation parameters on RH and AH, and the eff ect of a HH system on RH, AH, and ventilator performance during NIV. Without humidifi cation, RH in the NIV circuit (range 16.3 to 26.5%) was substantially lower than the ambient RH (27.6 to 31.5%) at all ventilatory settings. Increasing the IPAP led to a signifi cant decrease in RH (Spearman's rho = 0.67, P < 0.001), which returned to normal when HH was applied. Changing the respiratory rate or inspiratoryexpiratory ratio had no signifi cant eff ect. RH and AH both rose with the addition of humidifi cation, and the air was fully saturated at the maximum heater setting. A key conclusion of this study was therefore that the incorporation of HH systems in NIV ventilators increased the RH.
Esquinas and colleagues [32] analyzed AH values in a series of patients with hypoxemic acute respiratory failure, with NIV administered by a turbine-driven ventilator and a face mask, over a wide range of inspired oxygen fractions in four diff erent NIV environments: 1) without humidifi cation; 2) with a HH-MR850; 3) with a HH730; 4) with a HME booster. Th e main fi ndings were that the increase in inspired oxygen fraction led to a proportional decrease in AH and this eff ect was greater in an environment without humidifi cation than during NIV delivered with HH and HME-booster systems, and that AH levels were 'critical' when the inspired oxygen fraction was higher than 60% [32].
In agreement with Holland and colleagues [31], Esquinas and colleagues [32] found AH was higher with a humidifi cation system. When HH and HME booster systems were compared, AH was higher with the latter; however, the HME booster caused more patient-ventilator asynchrony and hypercapnia.
Th ere is as yet no uniform recommendation for any one device and there is also only limited epidemiological information on the best hospital practices and protocols for humidifi cation and device selection. Recently, our international group for the study of humidifi cation (the Humivenis Working Group) carried out a survey in 15 units with NIV expertise and found that usual practice was to use HH more often than HME for acute NIV applications (53% versus 6.6%). Surprisingly, despite the importance of gas conditioning during NIV, there were relatively few hospital protocols referring to humidifi cation practice in the participating centers (55%) [6].

Humidifi cation and CPAP therapy in patients with obstructive sleep apnea
Obstructive sleep apnea (OSA) is one of the most important indications for chronic use of CPAP at home. CPAP therapy is the gold standard for OSA patients and compliance with therapy reduces morbidity and mortality [33]. Nevertheless, compliance and adherence to CPAP for OSA patients remains a major problem. Th e most frequently reported factors associated with low acceptance and adherence rates include side eff ects such as nasal congestion, dry nose or throat, discomfort related to cold air, and/or allergy to the interface material; these are reported by as many as 65% of patients using nasal CPAP [34]. Chronic nasal congestion in particular can compromise a patient's ability to use CPAP successfully. Th e nasal mucosa has a considerable capacity for heating and humidifying inspired air but even so it can be overwhelmed at high fl ow rates with CPAP and when there is unidirectional fl ow caused, for instance, by mouth leaks. Th e fl ow of cold air through the nose dries the mucosa, resulting in the release of vasoactive and proinfl ammatory mediators. Th ese boost superfi cial mucosal blood fl ow and cause engorgement of deeper capacitance vessels, leading to increased nasal resistance. Th is in turn promotes mouth breathing, setting up a vicious circle.
Th ese complications can be largely prevented by humidifying the inspired air. Most current CPAP devices come with an integrated HH system. As these deliver more moisture than cold pass-over humidifi ers they may be more eff ective in patients with mouth leak and nasal congestion [12].
HH reduces nasal symptoms and nasal resistance, consequently attenuating infl ammatory cell infi ltration and fi brosis of the nasal mucosa [35,36]. Th erefore, the American Academy of Sleep Medicine has recommended the use of HH to improve CPAP compliance and adherence as a standard of practice [37]. According to the scientifi c literature, however, HH should only be added when patients complain of bothersome upper airway symptoms that are unresponsive to simpler measures (for example, intranasal steroids, temporary use of local vasoconstrictors) as there is no reliable information on the improvements in CPAP acceptance, adherence, or quality of life resulting from 'prophylactic' humidifi cation [36,38].

Conclusion
During NIV, adequate gas conditioning is essential because of the deleterious eff ects of inhalation of dry air, which may then negatively infl uence adherence to and the success of the ventilatory treatment. Several parameters, mostly involving the technical aspects of NIV, are determinants of ineffi cient humidifi cation. Th e correct application of an appropriate humidifi cation system may help prevent NIV-induced airway dryness. However, there are still open questions about when exactly to apply a humidifi er in acute or chronic settings, the best type of humidifi cation device in each situation, the interaction between the humidifi er and the underlying disease and the eff ects of individual ventilators on delivered humidity.