INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: forced oscillation; monitoring; noninvasive ventilation; respiratory resistance

Proportional assist ventilation (PAV) is a new modality of artificial respiratory support designed to improve the interaction between the ventilator and the patient (1). PAV has been shown to be effective in reducing muscle effort and in improving ventilation (2). Moreover, preliminary data suggest that PAV may reduce the need for intubation during acute respiratory failure (8), facilitate exercise rehabilitation (9, 10), and improve gas exchange in chronic hypercapnia (11, 12). The essence of PAV is that the ventilator applies a pressure at the airway opening (Pao) that is proportional to the driving pressure generated by the patient's inspiratory muscles (1). To this end, Pao has one component to help the breathing muscles to generate flow () through respiratory system resistance, and another component to increase lung volume (V) by overcoming respiratory elastance. Accordingly, Pao = FA ·  + VA · V, where FA is a flow assist setting proportional to patient resistance and VA is a volume assist setting proportional to patient elastance. Therefore, tailoring FA and VA to each patient condition, which is the key issue of PAV, demands the measurement of the patient's respiratory resistance and elastance (1).

In intubated patients the assessment of respiratory mechanics for setting PAV can be carried out by means of occlusion techniques after patient sedation/paralysis (2) or by employing an esophageal balloon (6). However, these procedures are not applicable in noninvasive PAV, which is one of the most promising applications of this mode of ventilatory support (7). In the absence of a conventional noninvasive method for assessing respiratory mechanics in spontaneously breathing patients, setting of PAV has been carried out by indirectly estimating respiratory resistance and elastance by means of the so-called runaway method (1). Such procedure is based on the hypothesis that the application of a pressure support exceeding the value required by the patient's mechanics results in detectable ventilator overassist (runaway). According to published data about noninvasive PAV (7), this titration procedure is more suitable for setting VA than FA. As a consequence, in this application FA is also set in terms of patient comfort (7, 8, 12) by assuming that the patient feels uncomfortable when FA exceeds respiratory resistance. As these titration procedures are based on the trial of different FA and VA values, setting PAV is a relatively complex process. In particular, these titration procedures are not practical for routinely modifying FA and VA in accordance with the changes experienced by the patient's mechanics. In this regard, several authors have pointed out that the lack of a noninvasive simple procedure for setting and updating PAV limits the benefits of PAV and also hinders its clinical application (4, 6, 12, 13).

The forced oscillation technique (FOT) is potentially useful in determining the FA setting of noninvasive PAV. Indeed, the FOT allows the assessment of patient mechanics by applying a low-amplitude high-frequency pressure oscillation at the airway opening without interfering with the ventilator pattern (14). Total respiratory system resistance is continuously and automatically computed from the pressure and flow signals recorded at the mask (15). Accordingly, the FOT could be used for setting and updating FA while VA could be set by the runaway method. Moreover, earlier data indicate that the instrumentation required by the FOT could be substantially simplified by generating the oscillatory pressure with the same ventilator (19), thereby facilitating its routine application in clinical practice. Consequently, the aim of this study was to ascertain whether the FOT applied by the same PAV ventilator is suitable for assessing respiratory resistance during noninvasive application of PAV in patients with chronic obstructive pulmonary disease (COPD).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The study was carried out with 14 male patients (67 ± 6 yr, 71 ± 23 kg, 165 ± 6 cm; mean ± SD) with severe COPD (FVC = 56 ± 18% pred, FEV₁ = 33 ± 9% pred, FEV₁/FVC = 0.44 ± 0.12). At the time of the measurements the patients were in the recovering phase (Pa_O2 = 57 ± 8 mm Hg, Pa_CO2 = 53 ± 7 mm Hg, pH = 7.41 ± 0.04) from the exacerbation that led to their hospital admission owing to hypercapnic respiratory failure. The study was approved by the ethics committee of the Hospital Clinic Provincial (Barcelona, Spain) and informed consent was obtained from the patient.

Measurements

Noninvasive PAV was applied by means of a ventilator prototype (BiPAP Vision system; Respironics, Murrysville, PA). The software of the experimental ventilator used in this study was especially modified by the manufacturer to allow application of a forced oscillation pressure (5 Hz, ± 1 cm H₂O) superimposed onto the PAV pressure waveform (Figure 1). Nasal pressure (Pn) was measured by a transducer (143PC; Honeywell, Freeport, IL) directly connected to the mask. Breathing flow () was measured by a mesh-wire screen pneumotachograph placed between the exhalation port of the ventilation system and the nasal mask. The pressure drop across the pneumotachograph was sensed by a differential pressure transducer (143PC; Honeywell). Esophageal (Pes) and gastric (Pga) pressures were measured in seven of the patients by means of two balloons (4-cm perimeter) filled with 1 ml of air and connected to pressure transducers (143PC; Honeywell) through 90-cm-long catheters (0.12-cm i.d.). , Pn, Pes, and Pga were sampled at 100 Hz and stored for subsequent analysis.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Example of the nasal pressure signal applied by the ventilator in one patient with COPD. The low-amplitude 5-Hz forced oscillation was superimposed onto the pressure waveform corresponding to proportional assist ventilation.

Procedure

The study was carried out with the patient in a sitting or semirecumbent position according to his preference. The esophageal and gastric balloons were positioned after application of nasal topic anesthesia. The correct position of the esophageal balloon was verified by the occlusion test (20). The nasal mask was carefully fitted to minimize leaks. To this end, a continuous positive airway pressure (CPAP) of 4 cm H₂O was applied and then the patient was required to stop breathing for a brief period. If necessary, adjustments of the mask were made until the leak was less than 40 ml/s (i.e., leak resistance greater than 100 cm H₂O·s/L). Special attention was paid to ensure that the mouth of the patient was closed during measurements with the nasal mask. The patient was then allowed to become familiar with the equipment under the baseline condition (spontaneous breathing with CPAP = 4 cm H₂O) for a period of 10-15 min. Then, flow (FA) and volume (VA) assistance were set independently. VA was determined by the method usually employed in noninvasive PAV (11, 12). To this end, CPAP was kept to 4 cm H₂O, FA was fixed to 1 cm H₂O · s/L, and VA was progressively increased from 2 cm H₂O/L by steps of 1 cm H₂O/L until the patient felt uncomfortable or the runaway condition was reached. VA_max was the higher value of VA before the patient reported discomfort or the runaway condition was reached. The flow-assistance setting FA was determined by keeping CPAP at 4 cm H₂O, setting VA to 2 cm H₂O/L, and progressively increasing FA from 1 cm H₂O · s/L by steps of 1 cm H₂O · s/L until the runaway condition appeared or until the patient did not tolerate the pressure applied. FAmax was the higher value of FA before the patient reported discomfort or the runaway condition was reached. Once VA_max and FA_max were determined, PAV (with the addition of CPAP = 4 cm H₂O) was applied for 15-20 min with FA and VA set to 80% of FA_max and VA_max, respectively. To determine the effects induced by changing from nose breathing to mouth breathing, the whole protocol was repeated in five of the patients after replacing the nasal mask by a full face mask and by asking each patient to breathe through the mouth.

Data Analysis

The last 2 min of stable signals recorded during baseline and during PAV were processed to assess the ventilation pattern, the breathing effort, and lung resistance (Anadat; RHT-Infodat, Montreal, Canada). Tidal volume (VT), breathing frequency (f), the ratio between inspiratory time and breathing period (TI/TT), and minute ventilation (E) were determined from the volume signal (V) obtained by digital integration of the recorded flow (). Breathing effort was assessed by means of the esophageal pressure swing (Pes) and by computing the pressure-time product of transdiaphragmatic (Pdi = Pes  Pga) pressure multiplied by breathing frequency (PTPdi · f). Lung resistance (RL) was computed for each breathing cycle by multilinear regression of the transpulmonary pressure, flow and volume signals as described in detail in Farré and coworkers (18). Specifically, only breathing cycles resulting in positive resistance and elastance values and with a fitting error with the resistance-elastance model of less than 10% were accepted. On average, for each patient and measurement (baseline and PAV), data were obtained from 30 accepted breathing cycles. Total respiratory resistance measured by the FOT (Rrs) was computed by Fourier analysis of the 5-Hz oscillatory pressure and flow components of the recorded mask pressure and flow signals (15). Results are shown as means ± SD. The changes in the parameters of ventilation pattern and breathing effort from baseline to PAV and the differences between resistance values were analyzed by paired t tests. The relationships between Rrs with RL and with FA_max were assessed by linear regression. Statistical significance was considered at the p = 0.05 level.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nasal PAV (VA = 9.2 ± 2.6 cm H₂O/L, FA = 7.6 ± 2.3 cm H₂O · s/L) significantly improved patient ventilation: VT and E increased by 55 and 38%, respectively, from baseline to PAV (Table 1). Moreover, PAV markedly unloaded the respiratory muscles: Pes and PTPdi · f decreased by 38 and 41%, respectively, from baseline to PAV (Table 1).

The relationship between Rrs noninvasively measured by the FOT and RL determined with the use of an esophageal balloon is depicted in Figure 2. Rrs and RL were close both during nasal (Rrs = 8.9 ± 3.1 cm H₂O · s/L, RL = 9.0 ± 2.6 cm H₂O · s/L; n = 7) and mouth (Rrs = 5.6 ± 2.1 cm H₂O · s/L, RL = 5.8 ± 1.4; cm H₂O · s/L; n = 5) breathing. Patient resistance was significantly lower when breathing through the mouth than during nasal breathing.

View larger version (20K): [in this window] [in a new window]   Figure 2.   (Top) Scatter plot of total respiratory resistance (Rrs) measured by FOT and lung resistance (RL) measured with an esophageal balloon when PAV was applied during nasal (solid circles) and mouth (open circles) breathing. Dashed line: identity line. (Bottom) Corresponding difference between Rrs and RL as a function of RL. Solid line, mean of the differences; dashed lines, mean ± 2 SD of differences.

Respiratory resistance measured by the FOT during nasal PAV in the whole population (14 patients) was Rrs = 11.1 ± 5.4 cm H₂O · s/L, which is a figure slightly, but nonsignificantly, greater than the maximum FA value applied during the setting of PAV (FA_max = 9.5 ± 2.9 cm H₂O · s/L). Moreover, during measurements in which PAV was applied when breathing through the mouth, the differences observed between Rrs and FA_max were not significant (5.6 ± 2.1 and 4.7 ± 1.9 cm H₂O · s/L, respectively). Figure 3 is a scatter plot of FA_max versus Rrs for both nasal and mouth breathing, indicating that these two variables exhibited a significant linear regression (r = 0.72). However, their agreement was better for the lowest values of resistance.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Maximum value of flow assistance achieved during the procedure of PAV setting (FAmax) versus respiratory resistance noninvasively measured by forced oscillation (Rrs) during nasal (solid circles) and mouth (solid triangles) breathing. Solid line, linear regression; dashed line, identity.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main result obtained in this study was that forced oscillation superimposed onto airway pressure by means of the PAV ventilator allowed the noninvasive and automatic assessment of respiratory resistance. Rrs measured by the FOT at 5 Hz was close to lung resistance measured by means of an esophageal balloon.

This study was focused on the assessment of respiratory resistance during noninvasive PAV. Accordingly, the measurements were carried out on patients with COPD, given that this disease is characterized by increased values of respiratory system resistance during the stable phase and, especially, during acute exacerbations. In the absence of a conventional method for noninvasively assessing respiratory mechanics in spontaneously breathing patients, the settings of PAV (FA and VA) were determined on the basis of the runaway phenomenon and/or patient comfort, which are the most commonly used criteria for setting noninvasive PAV (7, 8, 11, 12). In agreement with previously reported data, the PAV settings in this study were close to those employed in patients with similar characteristics (7, 11, 12). A CPAP level of 4 cm H₂O was applied during both the spontaneous breathing baseline and during PAV in all the patients. A certain low CPAP level was required by the ventilator to ensure adequate air renewal in the circuit. CPAP was not tailored to each patient because this would require the use of an esophageal balloon, which is a procedure not routinely applicable in noninvasive ventilation. Instead, a value of 4 cm H₂O was applied to all the patients because this figure was close to the CPAP value that proved to be beneficial in counterbalancing intrinsic positive end-expiratory pressure in COPD patients subjected to nasal PAV (6). The PAV applied in this study improved the breathing pattern and unloaded the breathing muscles (Table 1) consistently with values previously reported for COPD patients during nasal PAV (4, 12).

The FOT was previously employed to measure Rrs in patients subjected to artificial airway pressure such as conventional invasive mechanical ventilation (21, 22) or nasal CPAP (15, 18). Nevertheless, the FOT has not been used to date for assessing respiratory resistance during noninvasive PAV. However, it has been reported that the FOT at a frequency of 5 Hz is suitable for assessing respiratory resistance in spontaneously breathing COPD patients subjected to nasal CPAP (18). The results obtained in the present study when applying the FOT during mask PAV confirm the agreement between Rrs and RL. Although these two resistance indices do not yield the same physiological information (Rrs is total respiratory system resistance at 5 Hz and RL is lung resistance at the spontaneous breathing frequency) they are expected to provide similar figures in the patients, in whom both Rrs and RL were probably determined mainly by airway resistance. On the one hand, Rrs should be higher than RL owing to chest wall resistance. However, the difference should be small because chest wall resistance at 5 Hz makes a negligible contribution to total respiratory resistance in severely obstructed patients. On the other hand, Rrs at 5 Hz should be lower than resistance at the spontaneous breathing frequency (RL) on account of the negative frequency dependence of resistance caused by viscoelasticity and inhomogeneity and by the extrathoracic upper airway shunt. However, the potential effect of the frequency dependence of resistance on the difference between Rrs and RL was minimized by using a low FOT frequency (5 Hz) and because applying the FOT through a nasal mask reduces the effect of the extrathoracic upper airway (18). In contrast to the measurement of resistance, the assessment of respiratory elastance by the FOT during spontaneous breathing is not easy because it would require the use of low forced oscillation frequencies, which could potentially interfere with breathing. From a practical point of view, it should be pointed out that FOT measurements during noninvasive ventilation may be affected by air leaks at the mask level. Specifically, actual respiratory resistance could be underestimated by measured Rrs in case of nonnegligible leaks. In routine application, special attention should be paid to detecting and quantifying air leaks; for example, as in the present work, by estimating leak resistance from the nasal pressure and flow signals (15, 18).

In the earlier studies in which the FOT was employed in patients subjected to mechanical ventilation or CPAP, an oscillation generator capable of withstanding the ventilator pressure was required (21, 23). The inclusion of such an FOT generator is not a practical limitation in the already instrumented environment of an ICU or sleep laboratory. However, additional FOT instrumentation may reduce the routine applicability of the technique during noninvasive ventilation in a hospital ward or at home. A novel feature of the approach employed in this study is that the FOT was applied by the same ventilator and, consequently, no additional device was required to measure Rrs. The procedure was implemented in a specific blower-based ventilator because of the previous experience gained in modifying the control system of this type of device (19, 24, 25). However, ventilators based on other technologies could be modified to incorporate a forced oscillation for measuring Rrs during PAV or a more conventional mode of ventilation.

The maximum flow assistance applied during the PAV setting procedure (FA_max) was, on average, similar to the values of patient resistance Rrs measured by the FOT (Figure 3). Such an agreement between these two variables was expected because, according to the rationale for PAV, FA_max is an indirect estimate of patient resistance. However, as shown in Figure 3, FA_max tended to be lower than Rrs in most patients with high resistance. This observation is in keeping with previous nasal PAV studies of COPD patients showing values of FA_max lower than patient resistance (7) or lower than the value expected according to the clinical status of the patient (11). In this regard, it should be mentioned that, as reported (7), the flow runaway condition was difficult to recognize. Moreover, the criterion of setting FA on the basis of patient comfort is difficult to implement in practice. Indeed, it is not easy to establish the progressively increasing FA value that determines the transition from patient comfort to discomfort. The value of FA_max obtained may differ depending on the patient threshold for discomfort tolerance and on the instructions given to the patient by the technician/physician performing the setting. This could result in increased variability when conventionally setting FA during noninvasive PAV.

The fact that FA_max was lower than Rrs in patients with higher resistance could be explained in terms of patient comfort. For instance, taking into account the pressure components corresponding to CPAP (4 cm H₂O) and to the small volume assist (VA = 2 cm H₂O/L), the nasal PAV pressure that should be applied by setting FA = Rrs in a patient with a high, but in no way exceptional, degree of obstruction (Rrs = 20 cm H₂O · s/L) would be as high as 20 cm H₂O. By contrast, if the level of flow assistance was reduced by 50% (FA = 10 cm H₂O · s/L) the maximum nasal pressure would be considerably decreased to 12 cm H₂O. It is feasible that such a patient could prefer underassistance with a relatively low nasal pressure to full assistance at the cost of a potentially uncomfortable nasal pressure, with the result that FA_max < Rrs. The application of a PAV level of assistance lower than that which would be theoretically ideal to unload the patient would not substantially reduce the effectiveness of this ventilation mode. Indeed, the synchrony between the ventilator and the patient, which is a major feature of PAV, is kept regardless of the magnitude of pressure assist (26, 27).

The influence of the route of breathing (nose versus mouth) on the FA settings of PAV was investigated. To this end, and in contrast to previous works in which noninvasive PAV was exclusively applied through a nasal mask (7, 11, 12), we also applied PAV through a full face mask and asked the patients to breathe through the mouth. As expected we found that, on average, patient resistance was ~ 3 cm H₂O · s/L higher when breathing through the nose than through the mouth (Figure 2). This value is consistent with the reported magnitude of nasal pathway resistance (28, 29). As nasal resistance may represent a nonnegligible fraction of total respiratory resistance (~ 30%), changing the breathing route from nose to mouth or vice versa could have an impact on the adequacy of the settings of FA during PAV. In fact, in some cases we observed patient intolerance due to overassistance when PAV with settings determined during nose breathing was applied during mouth breathing. Conversely, when breathing through the nose the patient could be considerably underassisted if PAV was applied according to the settings determined during mouth breathing. This is an issue to be considered when devising long-term or unattended applications with a full face mask, that is, when the patient may spontaneously change the route of breathing while subjected to noninvasive PAV.

In conclusion, this study demonstrated that forced oscillation can be routinely superimposed onto airway pressure by means of the PAV ventilator and that the Rrs noninvasively assessed by the FOT allowed the measurement of the patient's respiratory resistance during PAV. As this procedure facilitates the setting and updating of the level of flow assistance in accordance with the actual value of patient resistance, its routine implementation would enhance the benefits of PAV (4, 6, 12, 13). Moreover, as suggested (30) from more recent data (31), continuous monitoring of patient resistance during artificial ventilation could also help in optimizing the suctioning of secretions. From a practical viewpoint it is worth noting that the concept that a modern microprocessor-controlled ventilator incorporates the FOT to compute Rrs can easily be extended to any of the conventional invasive and noninvasive ventilation modes. This could contribute to the recommended aim of developing systems (25, 31) that automatically adapt ventilatory assist to changes in the patient's condition (32).