In recent years, the application of noninvasive positive pressure ventilation (NPPV) therapy in home mechanical ventilation (HMV) has increased  . In particular, NPPV is useful for patients with chronic obstructive pulmonary disease (COPD)   , which is treated with long-term home oxygen therapy (HOT)  . NPPV therapy may confer therapeutic effects, such as a reduction in respiratory work, sustained improvement of arterial blood gas status, reduction in hospitalization frequency, and improved survival and quality of life  . In addition, in NPPV therapy, ventilation is performed using a mask without tracheal intubation; thus, NPPV therapy facilitates the ability to have conversations and eating, and has benefits such as a lower risk of infection. However, as a condition for obtaining a therapeutic effect, it is imperative to fit the mask to the patient.
In NPPV, in contrast to intermittent positive-pressure ventilation (IPPV), FiO2 depends upon complex factors such as the amount of ventilation and leak that is associated with patients in addition to the devices  . A study reported that leaks in the mask may occur because of altered positioning of the mask resulting from a skin disorder or selection of an inappropriate mask on NPPV enforcement  . Leaks around the mask can compromise the effectiveness of NPPV, but NPPV ventilators can compensate for leaks  . However, when HOT device and NPPV therapy are used in combination, since the ventilator and HOT device are independent circuit configurations, it is difficult to compensate for the loss in oxygen concentration in the air taken in due to the leak (Figure 1). Therefore, in this study, we simulated lungs of COPD patients to determine the extent to which FiO2 supplied to the alveoli can be affected by the reduction in oxygen concentration due to a leak in the mask.
2.1. Experimental Materials and Methods
We had to provide spontaneous respiration because we simulated COPD patients in this study. We used the following method, as described in a previous study  . First, the left lung of the TTL model lung (Training and Test Lung®, Michigan Instruments, Kentwood, MI, USA) was used to produce mandatory ventilation, and the right lung provided an inspiratory effort. Spontaneous respiration can be produced by attaching a lift bar between the left and right lungs. Mandatory ventilation of the left lung was performed using LTV-1000 (Pacific Medico, Tokyo, Japan). The ventilator was set in pressure-control continuous
Figure 1. Circuit diagram. To simulate the NPPV mask, we developed an adapter with an O2 supply port and a leak port attached between the inspiration side circuit of VELA and Y piece. Through this adapter, O2 was supplied to the TTL model lungs and leaked. The left side of the figure of the TTL model lung represents the left lung, the right represents the right lung. Circuit diagrams show HOT only (a), using VELA without leakage (b), and spontaneous respiration with leakage (c).
mandatory ventilation (PCV-CMV) mode, spontaneous respiration was set at 12 times, inspiratory pressure was set at 15 cmH2O, and inspiratory time was set at 1 sec. In the right lung, spontaneous respiration was enabled using VELA (Care Fusion 207, Palm Springs, CA, USA) to simulate respiration in patients with COPD. VELA was set in the NPPV-pressure support ventilation (PS) mode. PS was set at 10 cmH2O, positive end-expiratory pressure (PEEP) was set at 5 cmH2O, and the flow trigger was set at 2 L/min. Airway resistance in the left lung of the TTL model lung was set at 20 cmH2O to produce spontaneous respiration, and compliance was set at 0.05 L/cmH2O. Airway resistance in the right lung of the TTL model lung was set at 50 cmH2O to simulate lungs of patients with COPD, and compliance was set at 0.04 L/cmH2O. To simulate the leaks from NPPV mask, we developed an adapter with an O2 supply port and a leak port, and attached it between the inspiration side circuit of VELA and the Y piece. Through this adapter, O2 was supplied to the right lung of the TTL model lungs and leaked (Figure 1).
2.2. Method of FiO2 Measurement Simulating No Air Leak Using HOT Device and VELA
For HOT, an oxygen concentrator (O2 Green SHIZUNE IT-5 L, TERUMO, Japan) was used. Six flow rates of O2 (0.5, 1, 2, 3, 4, 5 L/min) in the TTL model lung were used. The oxygen concentration meter was placed at two locations: the right lung of the TTL model lung, which simulated a COPD patient, and the connection port of the circuit (simulated patient mouth), used to measure FiO2 after 5 minutes. In order to obtain a stable O2 value, O2 was passed through the O2 supply port for 5 minutes. The O2 measurement was repeated ten times for each flow setting. For the FiO2 measurement of the simulated patient's mouth, the combined Y portion was measured using an oxygen concentration meter. Furthermore, after performing the measurements for 5 minutes, in order to eliminate the accumulated O2, FiO2 in the left lung of the TTL model lung was released into room air until the O2 concentration reached 21%, and then O2 supply was initiated.
2.3. FiO2 Measurement Method Simulating Air Leak Using HOT Device and VELA
In order to simulate NPPV by VELA, oxygen was supplied from the oxygen concentrator to the lungs of the TTL model lung and at the same time a ventilator was attached. In addition, oxygen at concentrations ranging from 0.5 to 5 L/min was released from the adapter. From the summed-up value of the ventilation volume per minute from the respirator and the oxygen volume per minute in the oxygen concentrator, leaked at 5 L/min, the leak amount was precisely measured using the flow analyzer (PF-300, Ver. 3.1, imt Medical, Switzerland). From there, FiO2 was measured for 5 minutes in the TTL model lung part and Y piece part. Measurements were repeated 10 times for each setting.
2.4. Statistical Analysis
The FiO2 measurements using HOT device and VELA simulating each state were compared between three groups of leaks: 0 L/min, 5 L/min, and 10 L/min. For comparison between groups and all variables, multiple comparisons were performed by the Tukey HSD method. Statistical analyses were performed with statistical software “R” (Version 3.5.0, The R Foundation for Statistical Computing, Vienna, Austria), and P < 0.01 was considered statistically significant.
When supplying O2 to the lungs using the oxygen concentrator alone, FiO2 > 90% was obtained at an O2 flow rate of 4 - 5 L/min. At an O2 flow rate of less than 4 L/min, FiO2 was 88.9% at 3 L/min, 82.3% at 2 L/min, 60.8% at 1 L/min, and 35.9% at 0.5 L/min. When the O2 flow rate was set at less than 3 L/min, FiO2 was markedly decreased. Furthermore, when NPPV therapy was used in combination, FiO2 decreased to 28% to 41% at an O2 flow rate of 2 to 5 L/min. The largest reduction in FiO2 using NPPV therapy combined with HOT device was 55.5% at 3 L/min O2 flow rate (Figure 2).
When leakage was 5 or 10 L/min and O2 flow rate was 3 to 5 L/min, FiO2
Figure 2. Variation in FiO2 with O2 flow. Using the oxygen concentrator alone, more than 90% of FiO2 was obtained at an O2 flow rate of 4 - 5 L/min. At O2 flow rates of less than 4 L/min, FiO2 was 88.9% at 3 L/min O2, 82.3% at 2 L/min O2, 60.8% at 1 L/min O2, and 35.9% at 0.5 L/min O2. When the O2 flow was less than 3 L/min, FiO2 markedly decreased. When using a ventilator, the highest reduction in FiO2 was 55.5% at an O2 flow rate of 3 L/min.
However, FiO2 decreased by 2.8% at 1 L/min O2 flow and 1.4% at 0.5 L/min O2 flow when the leakage was increased from 0 to 10 L/min. At low rates of O2 flow, FiO2 was minimally affected by the leakage (Table 1, Figure 3).
These results revealed a marked difference between each leak amount, and
Table 1. Estimation of FiO2 in lungs imitating an NPPV mask leak.
Figure 3. Comparison of FiO2 with changes in leakage and O2 flow. When comparing 5 and 10 L/min rates of leakage with 3 and 5 L/min O2 flows, we observed that FiO2 decreased by approximately 3% in both leakage cases. Thus, when the leakage increased from 0 to 10 L/min, a maximum reduction of approximately 6% of FiO2 occurred. However, FiO2 decreased by 2.8% at 1 L/min O2 flow and 1.4% at 0.5 L/min O2 flow when the leakage increased from 0 to 10 L/min. At low rates of O2 flow, FiO2 was not markedly affected by leakage.
This study differs from previous study by Goutorbe et al. (2013) in two respects. First, HOT device has not been used in the previous study, but in this study, HOT device was used to measure lung oxygen concentration. Second, in the previous study, an oxygen cylinder was connected to the ventilator, but in this study, oxygen concentration was measured by connecting the HOT device to the respiratory circuit instead of the ventilator. Here, it is worthwhile to note that in Japan, oxygen therapy at home is performed using NPPV therapy and HOT device in combination, and the HOT device is connected to a ventilator circuit.
Three noteworthy points were confirmed in this study. First, the FiO2 value when using NPPV therapy and HOT device was clearly lower than the FiO2 value using only HOT device. The difference was the maximum when the O2 flow rate was 3 L/min. Second, upon setting the O2 flow rate at less than 1 L/min, the FiO2 value markedly decreased. Third, the FiO2 value decreased as the leak amount increased. This phenomenon has not been observed in the previous study.
Regarding the first observation, at an O2 flow rate of 5 L/min and using only HOT device, the FiO2 was 93%, and a difference of approximately 51% was observed compared to the measurement obtained using a mechanical ventilator with HOT device. A maximum difference of about 55% was observed at a flow rate of 3 L/min. When O2 was delivered in combination with a ventilator in a previous study, FiO2 was approximately 39% in the absence of leakage and at an O2 flow rate of 4 L/min. In the present study, the FiO2 was approximately 38%, which is not a substantial difference. However, it is important to note that a maximum reduction in FiO2 of 55.5% is observed when using HOT device with a mechanical ventilator, with an O2 flow rate of 3 L/min or more, as compared to the FiO2 obtained when HOT device is used alone.
Regarding the second observation, in addition to the performance of the adsorptive oxygen concentrator, the maximum ventilation volume of the TTL model lung is thought to be related to the performance. Because the maximum ventilation volume of the TTL model lung used in this study was 2 L, when the O2 flow was set to less than 1 L/min, the lungs were not sufficiently filled with oxygen, as measured at 5 minutes. Therefore, it is thought that FiO2 was markedly reduced at O2 flow of less than 1 L/min. When we set the O2 flow at 1 L/min and measured FiO2 for 30 minutes twice, then similar values were obtained compared with measurements recorded for 5 minutes. Thus, we found that the FiO2 concentration was difficult to increase in the simulated lung at an O2 flow rate of less than 1 L/min. In the guidelines for respiratory therapy, it remains unclear how much FiO2 decreases when a mechanical ventilator is added to an oxygen concentrator.
As regards the third observation, this difference is thought to be due to the difference in the mode of oxygen supply. In the previous study, an oxygen cylinder was used, and O2 was mixed directly from a ventilator and sent to the simulated lung. Therefore, FiO2 increased despite the rate of leakage increasing from 5 to 10 L/min, at O2 flow rates of 1, 2, and 4 L/min. This is considered to be a problem when the leak amount is adjusted. Although an effect of the leak on FiO2 has been reported previously by Miyoshi et al. (2005), the reduction in FiO2 when using HOT device has not yet been clarified. Furthermore, in the study by Goutorbe et al. (2013), the increase in FiO2 with an increase in the leak amount could not be confirmed. By contrast, in this study, we mainly considered home care in Japan, so we used HOT device to supply O2 to the middle of the inspiratory side circuit of VELA. In this research method, it is considered that the cause of O2 decrease that cannot be compensated by the leakage from the HOT device occurs due to the difference in the method of mixing air and O2. Therefore, it was found that when the O2 flow rate was 3 L/min or more, FiO2 decreased by about 6% as the leakage increased at this flow rate (Table 1).
O2 is supplied through a piping in the clinical setting. In our study, O2 was supplied from HOT device because there is no oxygen piping facility at home. In a previous study, O2 was supplied on the same principle as piping from a ventilator  . This is an indicator when using respiratory equipment in the hospital. We believe that this research provides very useful and important knowledge for home medical applications.
In our study, important findings on the use of HOT device in combination with a ventilator have been reported, but there are some limitations to this study. First, we simulated COPD patients using TTL model lungs, but we have not studied this phenomenon in humans. Therefore, inter-individual differences such as gender and physical differences were not reflected in the reduction of FiO2. We expect to obtain more robust results in future by conducting the same experiments on humans and increasing the sample size.
Second, errors in FiO2 measurement may occur due to the performance of the ventilator. Currently, there are many types of ventilators used at home care, and the performance of the equipment varies from manufacturer to manufacturer. Future research is warranted to identify the correct FiO2 value by performing a number of validation experiments using various types of ventilators. Third, even for the same ventilator, it is necessary to compare the difference between FiO2 supplied from oxygen piping and FiO2 supplied from HOT device. In the previous study, the measurement was based on the former method, so the result showed that FiO2 increased at a leakage rate of 10 L/min compared to a leakage rate of 5 L/min. In this study, different findings were obtained using the latter method. This may have been caused by different oxygen supply methods. We believe that this discrepancy can be resolved by performing more comparative experiments. Our current research provides scientific evidence to obtain multiple measurements of FiO2 reduction caused by leaks when using HOT device in the future.
In this study, we observed that FiO2 decreases because of a leak in the NPPV mask. Moreover, we measured the FiO2 for each oxygen flow rate supplied from the oxygen concentrator and the reduction in FiO2 with the use of a mechanical ventilator. However, our results differ from those of a previous study; therefore, we believe that further studies should be conducted using the same method of oxygen supply as that used in the previous study. To elucidate these differences, it will be important to reconsider the selection of respirators and oxygen concentrators.
The authors thank Kohta Kuradomi (Oita University Graduate School of Medicine, Japan) for providing experimental support. We would like to thank Editage (www.editage.jp) for English language editing.