Randomized Controlled Trial of a Novel Communication Device Assessed During Noninvasive Ventilation Therapy

Abstract

Background

Noninvasive ventilation (NIV), a form of positive airway pressure (PAP) therapy, is the standard of care for various forms of acute respiratory failure (ARF). Communication impairment is a side effect of NIV, impedes patient care, contributes to distress and intolerance, and potentially increases intubation rates. This study aimed to evaluate communication impairment during CPAP therapy and demonstrate communication device improvement with a standardized protocol.

Research Question

How does an oronasal mask affect communication intelligibility? How does use of an NIV communication device change this communication intelligibility?

Study Design and Methods

A single-center randomized controlled trial (36 outpatients with OSA on CPAP therapy) assessed exposure to CPAP 10 cm H₂O and PAP communication devices (SPEAX, Ataia Medical). Communication impairment was evaluated by reading selected words and sentences for partners to record and were tabulated as %words correct. Each outpatient-partner pair performed three assessments: (1) baseline (conversing normally), (2) mask baseline (conversing with PAP), and (3) randomized to functioning device (conversing with PAP and device) or sham device. After each stage, both outpatients and partners completed Likert surveys regarding perceived intelligibility and comfort.

Results

While conversing with PAP, word and sentence intelligibility decreased relatively by 52% (87% vs 41%) and relatively by 57% (94% vs 40%), respectively, compared with normal conversation. Word and sentence intelligibility in the intervention arm increased relatively by 75% (35% vs 61%; P < .001) and by 126% (33% vs 76%; P < .001) higher than the control arm, respectively. The device improved outpatient-perceived PAP comfort relatively by 233% (15% vs 50%, P = .042) and partner-perceived comfort by relatively 245% (20% vs 69%, P = .0074).

Interpretation

Use of this PAP communication device significantly improves both intelligibility and comfort. This is one of the first studies quantifying communication impairment during PAP delivery.

Trial Registry

ClinicalTrials.gov; No.: NCT03795753; URL: www.clinicaltrials.gov

FOR EDITORIAL COMMENT, SEE PAGE 1324

With landmark trials demonstrating the efficacy of noninvasive ventilation (NIV) for treating cardiogenic pulmonary edema and acute exacerbations of COPD, NIV has changed the standard of care for acute respiratory failure (ARF) and reduced the amount of invasive mechanical ventilation (IMV).^{1, 2, 3, 4} The incidence of ARF increasing from 1 million (2001) to 1.9 million (2009) cases per year, and resultant costs increasing from $34 billion to $54 billion spent annually suggest that ARF requiring mechanical support is a progressively worsening problem.⁵ Finding ways to mitigate this cost becomes ever more important by making NIV more feasible. Even propensity-matched comparisons between NIV and IMV suggest that NIV hospitalizations are on average $35 thousand cheaper than hospitalizations using IMV.⁶

Communication impairment (CI) resulting from the use of a full oronasal mask—the default NIV interface in acute respiratory failure—is a widely known side effect of NIV.^7,8 In a study of 1,604 ICU staff, patients, and relatives from 32 French ICUs, Schmidt et al⁹ found that the inability of NIV patients to be correctly understood was significantly associated with high patient anxiety (P < .001) and a 16% greater chance of feeling dyspneic (P = .001). This perception holds true even for familiar observers; when relatives observed patients being misunderstood, patients were 25% more likely to be anxious. Increased anxiety was one reason why some physicians and nurses (P < .001) were less willing to provide NIV.⁹

Demoule et al¹⁰ demonstrated that, in 54 Belgian and French ICUs, moderate to severe patient anxiety during NIV treatment is associated with 4.9 times more NIV intolerance (P < .001) and 1.7 times more NIV failure (P = .027).¹⁰ NIV intolerance is associated with NIV failure requiring IMV. Carlucci et al¹¹ noted that up to 48% of patients using NIV for ARF needed early NIV discontinuation, 22% of which were specifically associated with patient refusal. Also, 77% of patients with early NIV discontinuation went on to require endotracheal intubation.¹¹ This delay in IMV treatment—initial NIV failure subsequently requiring IMV—is associated with worse outcomes, with 3-month mortality of up to 30% as opposed to 4% without initial NIV failure.¹²

Unlike IMV, where patients are completely unable to phonate without use of their larynx, NIV permits the synthesis of natural audible speech, but the mask interferes with its acoustic transmission. Because NIV interfaces can be more easily removed and replaced, it is well known that intermittent NIV mask removal during NIV treatment provides more effective communication than IMV but risks derecruitment, which can be detrimental in ARF and can lead to further lung injury.^{13, 14, 15, 16, 17}

Characterizing the degree of CI is fundamental both to understanding how communication is altered and for devising methods to improve this communication problem. In terms of CI, NIV mask-associated CI is similar to speech dysarthria-associated CI in stroke patients, where speech is audibly generated but deviated from baseline. Consequently, speech dysarthria metrics may be useful in evaluating the CI inflicted by oronasal mask NIV.¹⁸ One example includes the Assessment of Intelligibility of Dysarthric Speech tool, which is currently used in stroke-associated speech dysarthria and has been used to characterize intelligibility of electrolarynx-generated speech in IMV patients¹⁹ and in tracheostomy patients.^{18, 19, 20}

When augmentative and alternative communication (AAC) methods have been used to ameliorate CI in IMV,²¹ improved patient satisfaction,²² and reduced symptoms of fear, anxiety,²³ and distress^21,24,25 occurred. The improvement of CI in NIV may analogously decrease fear, anxiety, and distress, and could possibly reduce NIV intolerance while improving quality of care. Rose et al¹⁹ associated electrolarynx-improved communication in IMV patients with significantly reduced anxiety (P = .007, n = 24). Finally, similar to the investigation of AAC methods to improve communication in IMV patients, there is value in the investigation of AAC methods to enhance communication and consequently reduce symptoms of fear, anxiety, and distress in NIV patients.²⁶ Analogous problems exist with fighter pilots, astronauts, and scuba divers—roles in which full face masks or helmets are used to facilitate breathing via positive pressure ventilation.^27,28

This study sought to characterize the level of CI that patients may experience while using CPAP and study the CI from a new AAC NIV communication device (SPEAX, Ataia Medical, Fig 1), similar to analogous solutions for fighter pilots. NIV as used for acute and chronic forms of respiratory insufficiency can be administered by both bilevel positive airway pressure (BPAP) and CPAP. We chose to place our study subjects on CPAP with a pressure of 10 cm H₂O to simplify and standardize the protocol and mitigate the risk of hyperventilating these patients with OSA while still characterizing communication impairment from the oronasal NIV mask.

Figure 1.

PAP communication device setup on orofacial mask for this trial. (PAP communication device: SPEAX, Ataia Medical. Mask: Phillips Respironics AF531.) PAP = positive airway pressure.

Methods

Study Design and Oversight

This study was a randomized, controlled, patient-partner blinded (device on or off) Phase IIB trial evaluating intelligibility and outpatient comfort outcomes for adult patients prescribed positive airway pressure (PAP) therapy for OSA within 2 years and seen at the Emory Sleep Center from March to June 2019. Each participant also had to bring a partner (friend or relative) to assist with evaluating the participant’s communication. The Emory institutional review board approved this study (IRB00097529). Written informed consent was obtained from all outpatients and partners. This study was submitted to ClinicalTrials.org (NCT03795753).²⁹ The first author, although involved in study design, was not permitted to conduct experiments or intervene in study enrollment. All patient enrollment, experiments, and studies were conducted by nonconflicted study staff.

Study Setting

This study took place in a standard sleep study room at the Emory Sleep Center, using PAP communication devices (SPEAX, Ataia Medical, Dallas, TX). Outpatients and partners were seated 2 m apart and asked to sit with their backs against the chair. A speaker was placed adjacent to the patient and calibrated to play standardized ICU background noises on continuous loop to a decibel meter placed on the partner’s chair at a volume of 55 ± 3 dB.^30,31

Outpatient Selection and Randomization

Inclusion Criteria

Adult outpatients (≥18 years) prescribed PAP therapy for OSA within the past 2 years and seen at the Emory Sleep Center from March to June 2019 were considered. Outpatients and partners were compensated with $25 gift cards for their participation.

Exclusion Criteria

The minimum self-reported education level for both outpatients and partners was a 6th grade reading and writing level. All outpatients and partners were required to be native English speakers. To reduce the impact of variable baseline intelligibility, participants and partners could not have a self-reported history of speaking (including dysarthria and aphasia) or hearing impairment. A formal hearing screening was not conducted on patients or partners.

Study Procedure

Outpatients were randomized to a 2:1 intervention (functional device):control (sham device). Variable block-sizes of three or six outpatients were arranged randomly by the statistics team before the event allocated by providing sequentially numbered envelopes. Both outpatients and partners were blinded to the assignment. Study staff were not blinded to the assignment. Outpatients and partners were seated a fixed distance (2 m) apart, facing each other. Chairs were standardized, and both outpatients and partners were instructed to not lean forward.

This study was conducted in three stages: a baseline stage (with outpatients conversing normally), a mask baseline stage (with outpatients conversing while using PAP), and the device stage (with outpatients randomized to [1] conversing while using PAP and the device or [2] conversing while using PAP). Supplemental Video 1 and 2 demonstrate study protocol. To mitigate the risk of hyperventilating stable OSA patients while characterizing communication impairment from the oronasal NIV mask, regardless of their home setting, all outpatients in the mask baseline and device stage were put on CPAP with a pressure of 10 cm H₂O. To standardize mask interfaces, all patients were given this CPAP by a standard oronasal mask (Phillips Respironics AF531) at the commencement of the mask phase. Masks were fitted per AF531standard guide. The mask-based component (communicator patch) was attached to the oronasal mask. All outpatients, regardless of intervention and control, were given actual patches; no sham devices were used. Communicator device output was disabled for the control arm. Patients, partners, and staff were not allowed to change the volume on the device, to protect blinding. The chosen oronasal mask (Phillips AF531) and the mechanical ventilator (Phillips V60) are the predominant NIV methods in ARF in the United States and was therefore the PAP equipment used in this study. Once in the device stage, communicator microphones were attached to the flat surface of the PAP mask and the communicator set to 70% volume, a volume loud enough to be functional but soft enough to not be clearly heard by the patient, to preserve study blinding.

Each stage had a word and sentence evaluation section, each of which was derived from the Assessment of Intelligibility of Dysarthric Speech tool.¹⁸ For the word section, the outpatient read a list of 25 words, which the partner would have to identify from 12 similar-sounding words for each word read. For the sentence section, the outpatient would read 5- to 15-word sentences, which the partner would have to transcribe verbatim. Words and sentences were randomized by section and outpatient pair to reduce bias.

Before the baseline and mask stages, outpatient-partner pairs were given the opportunity to practice the tasks. After each stage, both outpatients and partners completed a Likert survey about perceived intelligibility and comfort, which took 3 minutes to complete.

Statistical Power/Sample Size

The study was designed for statistical power of 0.80 and a type I error rate of 0.05, with a 2:1 intervention (functional communicator): control (sham communicator) split. Presuming at least 1 SD difference between the average of the intervention and the control arm would require 24 intervention and 12 control patients, for a total of 36 patients. Outpatients were randomly assigned to intervention or control arms by variable block-size (three or six) randomization assignment to functional or sham communicator. All outpatients were analyzed by intention-to-treat principle.

Word intelligibility was assessed by the number of words correctly selected divided by the total number of words. Sentence intelligibility was assessed by the number of words correctly transcribed divided by the total number of words.

Comparisons between groups for percent word intelligibility, percent sentence intelligibility, and Likert scale results were performed using analysis of covariance (ANCOVA) and paired t tests. Data were arcsine transformed to stabilize variances of group proportions. Paired t tests were used to compare the difference between baseline conversion and mask baseline conversion for word and sentence intelligibility. Independent group t tests were used to compare the change between intervention and control arm intelligibility. ANCOVA was used to compare the intervention and control arm intelligibility adjusting for baseline mask. All calculations were performed in Microsoft Excel (Version 16.34).

Statistical Analysis Results

From March to June 2019, 3,835 unique outpatients were seen at the Emory Sleep Center. Chart review identified 1,139 unique outpatients who were prescribed PAP within the past 2 years and were called to assess for appropriateness of inclusion. One thousand one hundred twelve of these were invited to participate in the study. We recruited 36 outpatients that agreed to be involved in this trial. The study recruitment was stopped on completion of the study recruitment goal of 36 outpatients. Patient demographics are described in Table 1. The CONSORT diagram is listed in Figure 2.

Table 1.

Patient Demographics

Demographic	Control Subjects (n = 13)	Intervention (n = 23)
Outpatient % male	53	45
Outpatient age, y (SD)	59.9 (17.1)	55.4 (13.3)
Outpatient race	9 (69.2%) White	10 (43.5%) White
	4 (30.8%) Black	11 (47.8%) Black
		1 (4%) Hispanic
		(4%) multiracial
Mean no. of years on NIV (SD)	6.0 (6.7)	4.8 (5.1)
Median (IQR)	2.0 (1.0-11.5)	4.0 (1.0-9.0)
Partner % male	31	26
Partner age (SD)	54.3 (15.2)	49.3 (19.6)
Partner race	5 (38.5%) White	12 (52.2%) White
	5 (38.5%) Black	9 (39.1%) Black
	1 (7.7%) Hispanic	2 (8.7%) multi-racial
	1 (7.7%) other
	1 (7.7%) multi-racial
PAP modality	8 (62%) auto CPAP	13 (57%) auto CPAP
	4 (31%) fixed CPAP	9 (39%) fixed CPAP
	1 (8%) ASV	1 (4%) auto BPAP
Usage	86.3% (25.0%)	86.6% (23.3%)
Usage ≥ 4 h	72.5% (34.6%)	78.0% (32.5%)
Usage < 4 h	27.9% (36.2%)	16.8% (25.8%)
Average, h	7.0 (1.9)	7.8 (6.6)
Residual AHI	2.6 (1.7)	3.0 (4.0)
Auto-CPAP characteristics	n = 8	n = 13
Min CPAP (cm H2O)	5.5 (1.9)	6.0 (1.8)
Max CPAP (cm H2O)	17.1 (3.2)	17.5 (2.8)
Mean (cm H2O)	8.4 (3.4)	9.9 (2.5)
95% (cm H2O)	11.3 (3.5)	12.5 (2.3)
Fixed CPAP characteristics	n = 4	n = 9
Set pressure (cm H2O)	9.7 (0.6)	10.3 (2.3)

AHI = apnea-hypopnea index; ASV = adaptive servo-ventilation; BPAP = bilevel positive airway pressure; NIV = noninvasive ventilation; PAP = positive airway pressure.

Figure 2.

CONSORT diagram.

Based on randomization, the intervention-control split was 23:13. This variation from the planned 24:12 split occurred because of the first randomization draw being used as a prototype for workflow testing purposes and was deleted from statistical analysis.

Word and Sentence Accuracy

When conversing normally, outpatient word intelligibility was 87%, and sentence intelligibility was 94%. From baseline conversation to mask baseline conversation, word intelligibility significantly decreased relatively by 53% (87% vs 41%; P < .001), and sentence intelligibility significantly decreased relatively by 58% (94% vs 40%; P < .001).

Compared with the control arm, the intervention arm word intelligibility significantly increased relatively by 75% (35% vs 61%, ANCOVA, P < .001; two-group t test, P = .006), and sentence intelligibility significantly increased relatively by 126% (33% vs 76%, ANCOVA, P < .001; two-group t test, P = .011) over the control arm. Intelligibility results are shown in Figure 3.

Figure 3.

Proportion of words and sentences correct by section.

Likert Scale Results

Fifty percent of intervention outpatients found that the device significantly improved NIV comfort, as opposed to 15% of control patients (35% absolute improvement, 233% relative improvement, P = .04).

Sixty-nine percent of intervention partners found that the device significantly improved PAP comfort, as opposed to 20% of control partners (49% absolute improvement, 245% relative improvement, P = .007).

As shown in Figure 4A and 4B, outpatients noted a significant improvement of 35% (4.87 vs 3.62, P = .04) in absolute mask comfort. Partners noted a similar, but nonsignificant, improvement of 33% (4.30 vs 3.23, P = .09) in absolute mask comfort.

Figure 4.

A, Outpatient and partner responses to whether the device was believed to improve mask comfort. B, Absolute average comfort ratings by outpatients and their partners.

Complications

No adverse events were recorded during or after this study.

Three intervention outpatients had device failures during the intervention arm of the trial. In accordance with intention-to-treat analysis principle, these outpatients’ data were categorized with their original arm assignment.

Discussion

We conducted the first known study to quantify communication impairment in PAP by measuring communication improvement by a mask-based NIV/PAP communication device. We further demonstrated significant improvement in intelligibility and outpatient comfort while wearing the PAP mask. As stated, compared with the control arm, the intervention arm word intelligibility was significantly higher relatively by 75% (61% vs 35%, P < .001), and sentence intelligibility was significantly higher relatively by 126% (76% vs 33%, P < .001) over the control arm.

There are several limitations to this study.

This study was blinded against information transfer from study staff to participants. Some of this blinding may have been limited by the volume of the ICU sound placed right behind the outpatient (which was scaled to the volume measured at the partner’s chair, and therefore, louder at the outpatient’s chair) and that the communicator was directed away from the outpatient toward the partner. Possibly the outpatients and partners discerned whether the device was a sham. If this occurred, this may have subsequent subconscious influence on the Likert surveys and evaluations of comfort and intelligibility.

The study staff were not blinded to the assignment. However, we attempted to reduce as much bias as possible, because intelligibility and Likert assessments were only performed by the blinded patient and partner. Study staff had no role in rating comfort or intelligibility.

The intervention arm had 3 of 23 (13%) device failures, still classified as intervention arm patients given intention-to-treat , which may lead to an underestimate of device benefit. These device failures were due to a manufacturing defect that had been fixed. To maintain the intention-to-treat analysis, the study was completed with the original design. Despite this, there was still a significant improvement in intelligibility and comfort. Given the size of the treatment effect (word intelligibility +75%, sentence intelligibility +126%) already observed, likely the true difference between control and intervention arms is greater than currently observed and is great enough to overcome the failed devices.

There was a higher proportion of patients of White race in the control arm and a higher proportion of patients of Black race in the intervention arm. The significance of this difference is unclear (9 [69.2%, control] vs 10 [43.5%, intervention] White, 4 [30.8%, control] vs 11 [47.8%, intervention]).

Study generalizability to the inpatient setting is limited by the single-center trial in stable OSA outpatients. Furthermore, we acknowledge that CPAP, although sharing full-face masks with NIV, is not completely equivalent to BPAP. BPAP is more commonly used in the in-hospital setting for ARF. In this study, CPAP was selected to avoid hyperventilating stable OSA outpatients. However, because many of these patients may be started on prehospital CPAP, many patients still may experience therapeutic CPAP during their course of treatment.³² Williams et al³³ conducted a systematic review of 14 studies on prehospital CPAP in 2013, many in the setting of acute pulmonary edema.³³ Goodacre et al³⁴ conducted a meta-analysis that found that prehospital CPAP improved both intubation and mortality rates in ARF due to acute pulmonary edema or COPD exacerbation.³⁴ Furthermore, the outpatients in this study had all been prescribed PAP; as such, these are outpatients who have significant experience in PAP usage and may experience less psychological distress and claustrophobia than NIV patients in ARF experience. We theorize that the findings from this study would serve as the minimum effect size that patients naive to NIV may experience. Conversely, ARF-related dyspnea may limit underlying sound production—thus limiting underlying signal to be amplified—and may limit overall effect. Inpatient use may be further complicated by use of wall power. Because NIV machines require wall power, we do not foresee this to prevent use. This study has tested only Phillips AF531 masks and has not been explicitly studied in other mask types, but performance should not vary significantly in rigid plastic masks. These advantages and limitations must be explored in a subsequent inpatient study.

Although this is a single-center, outpatient trial, the results should be widely applicable in NIV usage. Because the study population included stable OSA outpatients on PAP, direct assessment of clinical outcomes in the acute care setting with patients on NIV (BPAP) would quantify the impact of this device on patients with ARF, including its effects on several pertinent clinical endpoints. Furthermore, this may be useful in the outpatient setting.

In conclusion, use of this novel PAP communication device significantly improved both intelligibility and comfort of NIV use. This is one of the first studies quantifying PAP communication impairment. Future studies assessing the use of this device in the acute care setting could clarify the clinical outcome benefit of this intervention.