Portable oxygen concentrators (POCs) are widely used to administer long-term oxygen therapy (LTOT) and employ pulsed delivery modes to conserve oxygen. Efficient pulsed delivery requires that POCs are triggered by patient inhalation. Triggering is known to fail for some patients during periods of quiet breathing, as occurs during sleep. This article describes a new nasal interface designed to improve triggering of pulsed oxygen delivery from POCs. In vitro experiments incorporating realistic nasal airway replicas and simulated breathing were conducted. The pressure monitored via oxygen supply tubing (the signal pressure) was measured over a range of constant inhalation flow rates with the nasal interface inserted into the nares of the nasal airway replicas, and then compared with signal pressures measured for standard and flared nasal cannulas. The triggering efficiency and fraction of inhaled oxygen (FiO2) were next evaluated for the nasal interface and cannulas used with a commercial POC during simulated tidal breathing through the replicas. Higher signal pressures were achieved for the nasal interface than for nasal cannulas at all flow rates studied. The nasal interface triggered pulsed delivery from the POC in cases where nasal cannulas had failed to do so. FiO2 was significantly higher for successful triggering cases than for failed triggering cases. The nasal interface improved triggering of pulsed oxygen delivery from a POC and presents a simple solution that could be used with commercially available POCs to reliably supply oxygen during periods of quiet breathing.
Long-term oxygen therapy (LTOT) is widespread in the treatment of chronic respiratory diseases. For patients with hypoxemic chronic obstructive pulmonary disease (COPD), LTOT administered for 15 h/day or more has been demonstrated to improve survival time [1,2]. However, the choice of delivery device for administering LTOT is driven by cost as much as by performance or clinical evidence . Technologies used to administer LTOT in the home and in daily life are evolving, with small, portable compressed gas cylinders and liquid dewars increasingly being replaced by portable oxygen concentrators (POCs). POCs range in size and weight; larger devices are typically transported in wheeled carts, whereas smaller (≤5 lb) devices may be carried in over-the-shoulder bags. Some POCs have been marketed as single-source, 24-hour-a-day solutions that meet reported patient preference by addressing oxygen needs both at home and during activity . However, an elevated frequency of nighttime hypoxemic events has been reported during use of POCs as compared with larger, stationary, continuous-flow oxygen sources . Broadly, a majority of LTOT patients report problems with their oxygen delivery systems, the most frequent being equipment malfunction and lack of physically manageable portable systems [5,6].
While POCs offer potential efficiencies for oxygen delivery, their performance is variable [7–9]. POCs rely on pressure swing adsorption technology, whereby an oxygen-rich gas stream (∼90–95% vol O2) is produced by passing compressed air through zeolite beds used to trap nitrogen. As POCs become smaller, the sizes and capacities of compressors, sieve beds, and batteries used in the pressure swing adsorption process are reduced. Accordingly, less oxygen can be produced in a given time interval. To compensate, POCs incorporate pulsed flow delivery to conserve oxygen. During pulsed flow delivery, an oxygen bolus, or “pulse” is delivered through nasal cannula only when patient inhalation is detected. This limits loss of oxygen to the surrounding atmosphere during exhalation. POCs commonly rely on pressure triggering to detect inhalation: as the patient begins to inhale, entrainment of room air creates a small drop in pressure monitored through the cannula supply tubing, referred to hereafter as the signal pressure. When the signal pressure exceeds the triggering threshold of the POC device, a pulse of oxygen is released. The signal pressure is a function of inhalation flow rate but is also highly sensitive to the variable fit of conventional nasal prongs within the nostrils of individual patients. This sensitivity leads to considerable intra- and intersubject variability in triggering efficiency [10,11]. In particular, patients may not be able to trigger oxygen delivery during sleep, mouth-breathing, talking, or other circumstances characterized by low nasal inhalation flow rates . Due to triggering inefficiencies, the conventional way to receive supplemental oxygen during sleep is from a stationary oxygen concentrator that uses continuous oxygen delivery.
In this work, we report on the design and testing of a new nasal interface intended to improve triggering of pulsed oxygen delivery from POCs. Ideally, the elimination of triggering failures will allow POCs to be used at all times, even during sleep. Two hypotheses were evaluated using an established in vitro model of medical gas administration incorporating realistic upper airway replicas and a lung simulator [7,11,13,14]. First, it was hypothesized that the new nasal interface would produce higher signal pressures than standard or flared nasal cannulas at fixed inhalation flow rates. Second, it was hypothesized that using the new nasal interface with a commercial POC would improve triggering efficiency beyond that observed for standard or flared nasal cannulas, resulting in a greater in vitro fraction of inspired oxygen (FiO2) for the new nasal interface.
Three adult airway replicas formed from acrylic plastic and containing airways starting from the nares through the entrance of the trachea were used in this study. Replica geometries were based on magnetic resonance images previously reported . Geometric parameters for each of the three replicas were obtained using meshlab (Visual Computing Laboratory, Istituto di Scienza e Tecnologie dell'Informazione, Italy) and ParaView (Kitware, NY); these geometric parameters are listed in Table 1.
|Subject||Sex||Volume (cm3)||Wall surface area (cm2)||Nostril 1 inlet area (cm2)||Nostril 2 inlet area (cm2)|
|Subject||Sex||Volume (cm3)||Wall surface area (cm2)||Nostril 1 inlet area (cm2)||Nostril 2 inlet area (cm2)|
The three airway replicas were selected based on their propensity to trigger or not trigger pulsed oxygen delivery from a POC as reported in previous work by Chen et al. . The subject 2 replica was selected as a control, given relatively high signal pressures measured previously, whereas subjects 6 and 9 replicas were selected to evaluate improvement in pulse triggering for the new nasal interface concept, given relatively low signal pressures measured previously in each of these replicas .
The prototype shown in Fig. 1 was created to test the new nasal interface concept. The new nasal interface uses nasal pillows (Mirage Liberty Nasal Pillows, Small, 61,333; ResMed Ltd., Australia) to tightly fit the inner walls of the nares. The Mirage Liberty Nasal Pillows are made of skin-safe silicon and come in three standard sizes (small, medium, and large) to fit most adult nares. In addition, the new nasal interface includes four air entrainment ports on its bottom face. The number of holes left open was used in this work to create multiple “settings”:
Setting 1: one large hole open (providing 24 mm2 open area for air entrainment)
Setting 2: two large holes open (48 mm2 open area)
Setting 3: three large holes open (71 mm2 open area)
Setting 4: all four holes open (84 mm2 open area)
The hole sizes and settings for the presented prototype were chosen based on signal pressure measurements collected using an earlier edition prototype . VeroGray material (Stratasys, Eden Prairie, MN) was used to three-dimensional print the air entrainment body using a rapid prototyping machine (Eden 350 V; Stratasys). A standard (Hudson RCI RUS1103; Teleflex Medical, NC) and a flared nasal cannula (Hudson RCI RUS1104; Teleflex Medical) were used in this study for comparison with the new nasal interface.
Signal Pressure Tests.
Signal pressures were measured in a standard nasal cannula, a flared nasal cannula, and for each new nasal interface setting at constant inhalation flow rates of 10, 15, 20, 30 and 40 L/min drawn by vacuum through the airway replicas. Flow rates were measured using a mass flowmeter (TSI 4040; E & E Process Instrumentation, ON, Canada). Flow rates are reported as L/min at a standard temperature and pressure of 21.1 C and 101.3 kPa, respectively. The cannula or interface was inserted into the nares of each airway replica and a manometer (Digitron 2020 P-LIQ; ITM Instruments Inc., QC, Canada) was used to measure signal pressures (Fig. 2).
Simulated Breathing Tests.
Using a setup similar to that used by Chen et al. , oxygen concentration and flow rate were measured across a series of simulated breathing tests (Fig. 3). The resistance imposed by the nasal cannula or interface was also measured during each trial. A lung simulator (ASL 5000 Breathing Simulator; IngMar Medical, PA) was used to induce breathing patterns through the nasal replicas, representative of a COPD patient during sleep (inspiratory time = 1.79 s, expiratory time = 2.93, breathing frequency = 13 breaths/min, tidal volume = 520 mL) . This breathing pattern assumed 100% nasal flow, with a peak inspiratory flow rate of 27 L/min. Additionally, a breathing pattern was tested with the same inspiratory time, expiratory time, and breathing frequency, but with substantially reduced nasal ventilation . For these tests only 45% of the sleeping COPD patient tidal volume (234 mL) was used, consistent with the lowest nasal proportion in oronasal breath partitioning data reported by Gleeson et al. . This additional breathing pattern is representative of cases where COPD patients breathe partially through their mouths during sleep. The peak nasal inspiratory flow rate for this pattern was 12.5 L/min. During sleep, a range of breathing patterns may be present depending on the phase of sleep cycle the person is in . The lowest nasal proportion reported by Gleeson et al. was tested, as this was anticipated to represent a challenging scenario for oxygen pulse triggering.
The volume of the lung simulator chamber and the pressure at the entrance of the lung simulator were recorded at a sampling rate of 512 Hz using the ASL 5000 operating software. Oxygen concentrations were recorded using a gas analyzer (GA-200; iWorx, NH) and were corrected for sampling delay and time constant as described previously .
A SimplyGo Mini (Philips Respironics, PA) POC was used in this study. The SimplyGo Mini includes multiple nominal pulse delivery settings, each transmitting a different oxygen pulse volume. POC settings 2 and 4 were tested, which are reported by the manufacturer to deliver pulse oxygen volumes of 22 mL and 44 mL, respectively .
Experiments were conducted in triplicate for each combination of airway replica, nasal cannula or interface, breathing pattern, and POC setting. To ensure the oxygen concentration profile reached steady-state, each trial lasted approximately 3 min (∼40 breaths). All experiments took place in a controlled laboratory where temperature and humidity did not vary significantly over the course of the study.
Signal pressure measurements were repeated in triplicate for each replica and nasal cannula/interface combination at 10, 15, 20, 30, and 40 L/min. The average signal pressure and standard deviation across each set of repeated measurements were calculated.
Oxygen concentration and flow rate waveforms were used to calculate the FiO2 for each simulated breath . First, the flow rate of oxygen passing through the trachea was determined by multiplying inhalation flow rate by oxygen concentration on each time point. Next, the oxygen flow rate was numerically integrated between start and end of inhalation using the trapezoidal rule to calculate the volume of oxygen inhaled. The volume of inhaled oxygen was then divided by the tidal volume to calculate FiO2.
FiO2 values for either 15 or 18 breaths were included in the average FiO2 calculation for each POC setting, breathing pattern, airway replica, and nasal cannula/interface combination. In scenarios where the POC successfully triggered on all breaths for all three repeated trials, five breaths from each trial were used to calculate the average FiO2, for a total of 15 breaths per scenario. Conversely, when the POC did not trigger, it defaulted to a timed pulse setting. The SimplyGo Mini timed pulse setting operates with a 12 breath/min frequency , whereas the simulated breathing frequency was 13 breaths/min for all tests. As a result, the timed pulses cycled between being completely in-phase with inhalation and being completely in-phase with exhalation; to capture this entire cycle in the average FiO2 calculation, 18 breaths were required. After the data were split into groups by subject number, a one-way analysis of variance (ANOVA) for each POC setting/breathing pattern combination was executed, with FiO2 and the nasal cannula/interface setting being the dependent and independent variables, respectively. For example, using the FiO2 data collected during the 100% nasal flow POC setting-2 trials, three one-way ANOVAs were conducted: one for each subject number. Since four POC setting/breathing pattern combinations were tested, 12 one-way ANOVAs were conducted in total. Tukey-Kramer tests further evaluating FiO2 differences were conducted after each one-way ANOVA. A significance level of 5% was used in all cases.
The pressure drop (below ambient) measured at peak inspiratory flow was used to evaluate imposed inspiratory resistance for each nasal cannula/interface setting. To calculate the imposed inspiratory resistance, a baseline pressure drop for each airway replica, with no cannula or interface in place, was measured at peak inspiratory flow rate in triplicate. After averaging the three measurements, the appropriate baseline pressure drop was subtracted to calculate the imposed pressure drop at peak inspiratory flow due to the nasal cannula or interface. Then, each imposed pressure drop at peak inspiratory flow was divided by the flow rate to obtain the imposed resistance at peak inspiratory flow, measured in units of [cm H2O*s/L].
Signal pressure test results are summarized in Fig. 4. Subjects 6 and 9 recorded lower signal pressures using the standard and flared cannula than when using any new nasal interface setting at any flow rate. However, subject 2 recorded signal pressures using a flared cannula that were higher than those of interface setting 4 at all flow rates, and higher than those of interface setting 3 at all flow rates except for 40 L/min. At least one new nasal interface setting resulted in higher signal pressures than those of the standard and flared nasal cannulas at all flow rates, for all subjects.
Three main results were recorded during every simulated breathing trial: triggering type, FiO2, and imposed inspiratory resistance. Sample oxygen concentration and flow rate profiles for successful triggering, inconsistent triggering and failed triggering types are shown in Fig. 5. In Figs. 6 and 7, the average FiO2 for each scenario is provided.
There were three cases of inconsistent triggering. First, while using the flared cannula subject 6 successfully triggered the POC during 2 of 3 100% nasal flow POC setting-4 trials. On the third trial however, Subject 6 triggered the POC for only half of the breaths. As a result, the overall triggering success rate for this scenario was approximately 83%, so 15 out of 18 FiO2 values included in the scenario average FiO2 calculation were taken from breaths where successful triggering occurred. For the 45% nasal flow POC setting-2 trials, both subjects 6 and 9 inconsistently triggered during all trials while using nasal interface setting 3. The overall triggering success rate was approximately 50% for subject 6, and approximately 55% for Subject 9. Overall triggering success rates were reflected in the average FiO2 calculations for these scenarios as well.
Nasal interface setting 1 was not tested during any trial where 100% nasal flow was assumed. Since interface settings 2, 3, and 4 successfully triggered for all 100% nasal flow trials, imposing higher breathing resistance to increase signal pressure further was not necessary.
While using nasal interface setting 3, all subjects either failed to trigger or inconsistently triggered the POC under the 45% nasal flow pattern. Nasal interface setting 4 was not tested during any 45% nasal flow pattern because doing so would have led to lower signal pressures and failed triggering cases.
For all one-way ANOVA tests conducted, the selection of nasal cannula/interface setting was found to significantly influence FiO2 (p < 0.01). Tukey–Kramer posthoc tests indicated that, for a fixed subject number, breathing pattern, and POC setting, average FiO2 values from successful triggering cases were significantly higher than those from failed triggering cases, regardless of nasal cannula/interface group. Additionally, the inconsistent triggering cases with success rates of 50% and 55% led to mean FiO2 values significantly lower than those for all successful triggering cases within the same subject number, breathing pattern, and POC setting scenario. However, during the 100% nasal flow POC setting-4 trials, the successful triggering cases observed while subject 6 used nasal interface settings 2 and 4 led to mean FiO2 values not significantly different than the mean FiO2 value corresponding to the inconsistent triggering case observed while subject 6 used the flared cannula (with 83% success rate). The inconsistent triggering scenario with an 83% success rate was the only case in which an inconsistent or failed triggering scenario was not significantly lower than those for all successful triggering cases within the same subject number, POC setting, and breathing pattern group.
The imposed resistances at peak inspiratory flow for each combination of nasal cannula/interface setting, breathing pattern, and airway replica are depicted in Fig. 8. Error bars in Fig. 8 represent the propagated error from baseline pressure drop measurements and the pressure drop measurements at peak inspiratory flow for each trial.
This article describes a new nasal interface designed to improve triggering of pulsed oxygen delivery from POCs. The nasal interface incorporates pillow-type nasal prongs to snugly fit the nostrils, and draws ambient air through a series of entrainment ports. By adjusting the area of air entrainment ports (the “setting”) on the nasal interface, higher signal pressures were achieved than for standard and flared nasal cannulas at all flow rates, and for all three nasal airway replicas, studied. As a result, the nasal interface was able to trigger a POC in cases where flared and standard nasal cannulas failed to trigger. For a fixed subject group, breathing pattern, and POC setting, average FiO2 values for successful triggering cases were significantly higher than values for failed triggering cases, regardless of nasal cannula/interface group.
While only three nasal airway replicas were tested in this study, these replicas were carefully selected from a larger set of fifteen replicas studied in our previous work . As reported by Chen et al. , relatively high signal pressures were measured for the subject 2 replica, whereas relatively low signal pressures were measured for subjects 6 and 9 . Therefore, these three replicas were included in this work to evaluate the new nasal interface concept against conventional nasal cannulas in cases where triggering both was, and was not, expected. The promising in vitro results presented herein provide rationale to proceed to in-human studies, where variability between individuals can be further assessed.
Although the nasal interface clearly improved POC triggering in this in vitro study, the acceptability of such an interface to patients is unproven. Given that nasal pillows contact the nostrils, the nasal interface may be less comfortable than conventional nasal cannula when worn for extended periods of time. Additionally, esthetic concerns may make some patients hesitant to wear the nasal interface outside the home. However, as triggering issues are known to occur most often during sleep, some patients might use a pillow-style nasal interface during sleep only. In the sleep setting, there is a strong precedent for the use of nasal pillow interfaces in the delivery of continuous positive airway pressure for treatment of obstructive sleep apnea.
A further consideration is the resistance imposed by the nasal interface. As shown in Fig. 8, the imposed resistance at maximum inspiratory flow rate caused by the nasal interface is higher than that caused by the standard or flared cannula. In selecting an appropriate setting for the nasal interface, a tradeoff exists between increasing signal pressure and increasing imposed resistance. For the 100% and 45% nasal flow breathing patterns, successful triggering was, respectively, observed at settings 4 and below, and at settings 2 and below (Figs. 6 and 7). At these settings, the imposed resistance at maximum inspiratory flow was on average ∼1.5 and ∼2 cm H2O*s/L for 100% and 45% nasal flow breathing patterns, respectively (Fig. 8). These values are similar to the resistances of bacterial/viral breathing circuit filters  and those permitted in certification testing of N95 respirators used as personal protective equipment , and are below the resistances of gas masks, which have been shown to have only a slight impact on the respiratory effort of stable COPD patients . Whether or not these imposed resistances are acceptable to patients requiring LTOT remains to be evaluated.
Collectively, the results presented in Figs. 6–8 indicate that the optimal device setting (that is, the optimal area of air entrainment ports) differed between the breathing patterns studied. During early development, a mockup version of the device that allowed for continuous adjustment of the open port area was created. Signal pressures at constant inhalation flow rates were measured at several different open port areas. To estimate the optimal open area at each inhalation flow rate, the signal pressure versus open port area data were compared to the threshold signal pressures of existing POCs, leading to the concept of a device with adjustable “settings” . Experiments to test if port location makes a significant difference in the relationship between signal pressures and open port area have not been conducted to date. However, for the device design presented here, which incorporates a central plenum between the air entrainment ports and sensing ports, any influence of port location is expected to be minor.
Breathing simulations that assumed 45% nasal flow were conducted to study the effects of reduced nasal ventilation on POC triggering. However, the replicas used in this study do not include oral cavities or mouth openings. The new nasal interface, standard nasal cannula, and flared nasal cannula all impose resistance to nasal breathing, which may cause some fraction of inhaled air to be shunted through the oral passage when the mouth is open. In addition, because the replicas used in this study have a closed oral opening, the FiO2 measured for all 45% nasal flow trials represents the FiO2 prior to dilution with the orally inhaled fraction of air. While these limitations corresponding to the present nasal airway replicas could potentially be addressed by fabricating replicas that include both nasal and oral cavities, such an improvement would be complicated by the changing position of the velum as breathing transitions from predominantly nasal to predominantly oral. Alternatively, the potential influence of imposed nasal resistance on nasal-oral flow partitioning may be better assessed by conducting in-human trials.
One of the problems most frequently reported by LTOT patients is the lack of physically manageable portable systems [5,6]. Based on reported patient preference, the ideal oxygen supply is a single source unit that can be used during sleep, rest, and activity . Pairing POCs with the nasal interface presented here would provide control over, and increase the attainable range of, the signal pressures sent to POCs, and could thus minimize events where triggering fails. The interface therefore presents a simple solution that could be used with commercially available POCs to provide patients with a physically manageable oxygen source that can reliably supply oxygen in all common daily use circumstances.
I2I Grant from the Canadian Natural Sciences and Engineering Research Council (NSERC) (Grant No. I2IPJ 538442-19; Funder ID: 10.13039/501100000038).