We performed an open label, cross-over, randomized, study in 2 tertiary care hospitals in France. Both hospitals had substantial experience in the management of NMD and treatment with NIV. The study was approved by our ethics committee (CPP Nord-ouest II) following French legislation, and written informed consent was obtained from all patients. The trial was registered in ClinicalTrials.Gov (number NCT03458507).

Patients with NMD treated with home NIV who were followed by the participating centers were screened for eligibility between May 2018 and May 2019. The inclusion criteria were: adults (>18 years old) treated with nocturnal non-invasive ventilation <15h/day (to exclude patient with severe respiratory insufficiency in which transition period should have been at risk) and diagnosed with a NMD (Becker muscular dystrophy, facio-scapulo-humeral dystrophy, limb-girdle dystrophy, myotonic dystrophy, Duchenne muscular dystrophy). Exclusion criteria were: occurrence of an acute cardiorespiratory or ear-nose-throat event during the month prior to inclusion, rapidly progressive NMD (such as ALS), severe nasal obstruction and maxillofacial deformities or previous upper airway surgery preventing the usage of one type of mask (nasal or oronasal).

Inclusion visit. Patients tested the interface they did not usually use (i.e. nasal mask for those treated with oronasal masks or oronasal mask for those treated with nasal masks) during a one-hour diurnal NIV session. The most appropriate alternative mask (in terms of brand and size) was chosen for each patient in order to minimize leakage and maximize comfort.

Follow up. At the end of the inclusion visit, patients were randomized to start either with their usual or alternative interface. During the week with the alternative interface, patients were asked to use this new interface with the aim to sleep overnight with it. During the week with the usual interface, patients were asked to pursue their treatment as usual. The number and nature of interface-related problems reported by patients and/or the need for a home care provider visit were collected during both weeks.

At the end of each week, an unattended type III nocturnal polygraphy under NIV was performed at home using a Somnolter® (Nomics, Liege, Belgium) device synchronized with transcutaneous partial pressure in CO2 (PtcCO2) monitoring by SenTec V-Sign™ System. This polygraph also records mandibular mouth opening.

Assessment of NIV side-effects was performed with the Modified SECI questionnaire14 (for each commonly reported side effect, the patient is asked to rate the frequency, magnitude and perceived impact on adherence on a five-point Likert-type scale).

Patients were evaluated on two occasions, at the end of each week with either the usual or the alternative interface (cross-over). The primary objective was to compare mean nocturnal oxygen saturation between oronasal masks and nasal masks (Mean nocturnal SpO2, measured by polygraphy, Somnolter®). We also assessed the following exploratory secondary outcomes: percentage of sleep with SpO2<90% (polygraphy, Somnolter®); oxygen desaturation index (polygraphy, Somnolter®); mean nocturnal transcutaneous partial pressure in CO2 (PtcCO2, SenTec V-Sign™); mean mouth opening during sleep (polygraphy, Somnolter®); level of non-intentional leaks (percentage of recording time spend with leaks assessed by SomnoHolter®); the side-effects for each type of mask (Modified SECI).

Participants were randomized (computer-generated random numbers list) to either begin the trial with their usual interface (for one week) or the alternative interface (for one week) in order to allow a balance between the numbers of patients who would begin with their usual interface versus with an alternative interface, the randomization was stratified for the type of interface usually used by the patient.

Very few studies have compared nasal and oronasal masks for noninvasive ventilation in chronic respiratory failure.10,12,13 Therefore, in this pilot study, the sample size estimation was based on our inclusion capability and was based on the following: 1) at least 100 eligible patients were followed in the two participating centers, 2) 25% to 30% of patients with NMD use oronasal masks and we wanted to recruit equal numbers of patients who usually used oronasal masks and nasal masks and 3) we estimated that about 50% of eligible patients would accept to participate. Therefore, we planned to 30 patients (15 with nasal mask and 15 patients with oronasal mask). This sample size was sufficiently powered to detect a mean difference of mean nocturnal SpO2 of 2±2% (considered as clinically significant) between oronasal versus nasal mask (alpha=0.05, power=80%).

A treatment effect test, a period effect test, and a test for the interaction between treatment and period were used to assess respective effect of treatment (nasal versus oronasal), treatment sequence, and the first-order carryover risk.

Paired student t-tests were conducted to assess differences between the two interfaces for normally distributed outcomes. Wilcoxon signed-rank tests for paired values were used to compare non-normally distributed outcomes. Significance was set at p<0.05. Missing data were not imputated.

All randomized patients were analyzed in intention to treat (ITT): if a patient did not tolerate the alternative interface, polygraphy was performed with their usual interface and data collected during this evaluation was used in the analysis. A per-protocol (PP) was secondarily performed using data from patients who completed both polygraphies with the appropriate masks without any protocol deviation. Finally, a post hoc comparison using the same statistical models was performed to compare the usual and alternative interfaces to assess the impact of mask change among those patients.

The study flow is depicted in Fig. 1, 30 patients with NMD were included. Among them, 13 used an oronasal mask at home. During the week with the alternative interface, 5 patients had major intolerances (2 with the nasal mask and 3 with the oro-nasal) despite home care provider interventions and thus resumed the use of their usual interface (Fig. 1).

Fig. 1.

Consort flow diagram.

(0.53MB).

Table 1 presents the anthropometric characteristics of patients and ventilator settings. Patients were middle aged and equally distributed in terms of sex. Myotonic dystrophy was the most represented neuromuscular disorder (36.7%). Pressure support mode was used by 86.7% of patients with no differences in IPAP and EPAP settings between those who usually used nasal and those who used oronasal masks. There was no difference in mean NIV adherence, residual AHI or tidal volume estimated by NIV built-in software between patients who used nasal and those who used oronasal interfaces.

There were no significant differences in mean nocturnal SpO2 between the interfaces (ITT, p=0.73; PP, p=0.59) (Fig. 2).

Fig. 2.

Effect of mask type on SpO2 and PtcCO2 parameters in patients with neuromuscular diseases. Legend: ODI, oxygen desaturation index; tcCO2, transcutaneous carbon dioxide; SECI, side effect to CPAP inventory.

(0.15MB).

There were no differences in secondary outcomes between the interfaces (Fig. 2): percentage of sleep recording spent with SpO2<90% (ITT, p=0.33; PP, p=0.57); oxygen desaturation index (ITT, p=0.74; PP, p=0.59); mean nocturnal transcutaneous partial pressure in CO2 (ITT, p=0.13; PP, p=0.11); mean mouth opening during sleep (ITT, p=0.1931; PP, p=0.054); level of non-intentional leaks (ITT, p=0.58; PP, p=0.80) or side-effects according to each type of mask (ITT, p=0.46; PP, p=0.98). Additionally, no differences were found between masks in terms of mean NIV adherence, residual AHI, nadir SpO2, time spent with leaks or time spent with tcCO2>55mmHg.

There were no differences in the ratio of patients who reported interface problems or the number of home care provider interventions between interfaces (Table 2).

Post hoc analysis of interface transitions (Fig. 3)

Post hoc comparisons showed that changing from one interface to another reduced NIV efficacy: oxygen desaturation index (ITT, p=0.01; PP, p=0.01); mean nocturnal transcutaneous partial pressure in CO2 (ITT, p=0.048; PP, p=0.09); side-effects (ITT, p=0.008; PP, p=0.02).

Fig. 3.

Effect of usual vs alternative mask on SpO2 and PtcCO2 parameters in patients with neuromuscular diseases. Legend: ODI, oxygen desaturation index; tcCO2, transcutaneous carbon dioxide; SECI, side effect to CPAP inventory.

(0.15MB).

During the week with the alternative interface, mask problems ratio and home care provider intervention numbers were higher (p<0.01) and mean adherence was lower (p=0.02) (Table 2). Accordingly, patients perceived an impact on adherence during the week with the alternative interface (Table 3).

Table 3.

Side effects according to the type of mask (details of the SECI score).

Modified SECI	Nasal mask	Oronasal mask	p-Value	Usual mask	Alternative mask	p-Value
Total score (/225)	71 [58; 92]	82 [63; 92]	0.493	67 [58; 87]	84 [62; 104]	0.063
Side effects
Frequency (/75)	26 [22; 40]	29 [25; 35]	0.437	27 [21; 35]	29.5 [25; 37]	0.224
Magnitude (/75)	24 [20; 35]	30 [22; 32]	0.582	23 [20; 32]	30 [22; 35]	0.161
Perceived impact on adherence (/75)	16.5 [15; 25]	17 [15; 26]	0.877	15 [15; 19]	21.5 [16; 27]	<.01
Incomfort related to
Blocked up nose (/15)	6 [3; 7]	4 [3; 7]	0.159	5 [3; 7]	6 [3; 8]	0.515
Runny nose (/15)	5 [3; 7]	3 [3; 6]	0.105	5 [3; 7]	3 [3; 6]	0.390
Nose bleed (/15)	3 [3; 3]	3 [3; 3]	0.661	3 [3; 3]	3 [3; 3]	0.460
Dry mouth (/15)	5 [3; 8]	5 [3; 8]	0.403	5 [3; 8]	5 [3; 8]	0.950
Irritated eyes (/15)	4.5 [3; 8]	3 [3; 7]	0.670	3 [3; 7]	4.5 [3; 8]	0.699
Upset bowel (/15)	3 [3; 7]	3 [3; 7]	0.980	3 [3; 7]	3 [3; 9]	0.769
Transient deafness (/15)	3 [3; 3]	3 [3; 3]	0.788	3 [3; 3]	3 [3; 3]	0.534
Using NIV in front of others (/15)	3 [3; 5]	3 [3; 7]	0.226	3 [3; 4]	3 [3; 5]	0.353
Increased number of awakenings (/15)	6 [3; 9]	7 [4; 10]	0.604	7 [3; 8]	7.5 [4; 11]	0.255
Mask pressure (/15)	5.5 [3; 8]	7 [4; 9]	0.140	6 [3; 7]	8 [4; 10]	0.083
Mask leakage (/15)	6 [4; 9]	7 [4; 9]	0.676	6 [4; 8]	7.5 [4; 12]	0.375
Cold air (/15)	3 [3; 6]	3 [3; 6]	0.958	3 [3; 4]	3 [3; 6]	0.549
Noise (/15)	3 [3; 8]	4 [3; 8]	0.559	3 [3; 7]	3.5 [3; 8]	0.495
Problems exhaling (/15)	3.5 [3; 7]	3 [3; 5]	0.278	3 [3; 5]	3.5 [3; 7]	0.231
Anxiety during treatment (/15)	3 [3; 7]	5 [3; 8]	0.095	3 [3; 7]	4 [3; 9]	0.157

Complaints reported by participants about the alternative interface were predominantly mask intolerance, leaks, mouth dryness and cutaneous lesions. Problems relating to the underlying condition and impaired autonomy were reported especially with the use of an oronasal interface (impaired communication leading to the incapacity to call for help if needed, compromised vocal command device use, compromised hydration autonomy at night, etc.)

Although our study was powered to detect a mean difference of mean nocturnal SpO2 of 2±2% between oronasal and nasal masks, it is possible that the study was under powered due to the fact 5 patients did not tolerate the alternative interface.

The patients were not naive to NIV and the heterogeneity of the underlying diseases and level of autonomy could also have impacted the results. Yet, this recruitment reflects daily practice in tertiary care hospitals who deals with respiratory care in NMD which remain rare diseases (therefore it would have not been possible to complete this study in a reasonable time-frame with naive patients).

In our study, oronasal mask patients appeared to not have significant enough issues with upper airway resistance. We should have focused on more carefully selected patients with high upper airway resistance, residual events or low efficacy of oronasal NIV and assess the benefits to switch interface in those patients.

Finally, the period of adaptation to the alternative interface might have been too short.

Because of the high specificity of the sample included, the results of this study should not be extrapolated to other populations such as people with obesity hypoventilation syndrome and chronic obstructive pulmonary disease.