INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lung function testing in infants is gaining increasing attention as investigators seek methods for the production of clinically useful data in this noncooperative age group. Much has been learned about the normal growth and development of the infant respiratory system by using measurements of forced expiration, produced by rapidly inflating a cuff around the infant's chest and abdomen at end-inspiration, and measuring the subsequent expiratory flows. This technique relies on achieving flow-limitation, a condition that may not be reached reliably in healthy infants. In addition, the resulting flows are referenced to the previous end-expiratory volume, a landmark notoriously variable in infants. Because of these difficulties, investigators have been searching for a technique that provides more dependable data.

Recent studies in both animals and humans have established that the pulmonary parenchyma plays an important role in determining the lung function and in determining the mechanical responses to various insults. Constrictor agonists have been shown to alter the lung function via actions on both airway and tissue compartments (1), the pattern of response depending on which agonist is used and on the route of delivery. Both under normal conditions and following challenge with various constrictor stimuli, the simplest model that explains the behavior of the respiratory system consists of an "airway" compartment (airway resistance and inertance) and a tissue compartment that describes the "viscoelastic" behavior of the respiratory tissues (1). This model also seems to explain the behavior of both the lungs and the respiratory system of human adults and children with various lung diseases (2, 12).

We recently developed a technique to collect low-frequency respiratory impedance (Zrs) data in normal infants, evoking the Hering-Breuer reflex at an elevated lung volume (12). We demonstrated that, by using an appropriate model, the airway and tissue parameters can be separated. This method therefore appears particularly promising for a clinical assessment of the respiratory mechanics in infants. A feature of this study (12) was that all measurements were made at a mean transrespiratory pressure (Ptr_mean) of 20 cm H₂O. Since the early work of Briscoe and DuBois (13), there has been a consensus that airway resistance (Raw) decreases with increasing lung volume (3, 5, 11, 13). Similarly, human studies on total respiratory elastance (Ers) confirmed that, in accordance with the shape of the quasi-static pressure-volume (P-V) curve, Ers increases with lung volume near the total lung capacity (17, 18). At lower lung volumes, but still above the elastic equilibrium volume of the respiratory system, where the quasi-static P-V curve would be expected to be linear, the dynamically determined Ers appears to decrease with increasing lung volume in normal adult humans (18, 19, 22) and dogs (23). In humans, the effect of the lung volume on the total respiratory tissue resistance (Rti) has not been systematically studied. Bachofen determined the parenchymal resistance (Rti,L) in adults and found a significant increase near the full inspiratory position (14). In that study, the respiratory rate and the tidal volume were not standardized, although these factors have significant effects on Rti,L (19, 22, 23). Rti,L, which appears to be a significant component of Rti (9, 11), has been reported to increase with increasing lung volume in dogs (5), cats (3), and rabbits (24). However, Suki and associates found no effect of variation of positive end-expiratory pressure (PEEP) between 2 and 7.5 cm H₂O on Rti,L in dogs, when it was determined with oscillatory amplitudes comparable to those of normal breathing (11).

The above examples clearly indicate that, while the effect of the lung volume on the mechanical properties of the airways and respiratory tissues has been well characterized in different animal species, a detailed investigation has not been carried out in normal adults, and no data at all are available on infants. In the present study, we demonstrate that reliable estimates of the low-frequency Zrs spectra can be obtained at FRC in normal infants. In addition, we report changes in airway and tissue parameters at Ptr_mean levels of 0, 10, and 20 cm H₂O.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The anthropometric data on the eight infants studied are shown in Table 1. None of the infants had respiratory symptoms at the time of the measurements. Four infants were healthy, with no history of respiratory symptoms (Nos. 1 through 4), while four had had recurrent episodes of wheeze (Nos. 5 through 8). All parents had given informed written consent to the study. The protocol was approved by the Human Ethics Committee of the Princess Margaret Hospital. The infants were studied in the supine position during quiet sleep after an oral dose of chloral hydrate (100 mg/kg).

Measurement Apparatus

The Zrs data were collected by using apparatus described in detail previously (12). Briefly, the lungs were inflated through a balloon valve by a computer-controlled pump. Transrespiratory pressure (Ptr) was measured at the airway opening (SPM-02PG; Sugikura, Tokyo, Japan) and was used to control lung inflations by means of BRATLAB software (RHT-INFODAT, Montreal, Canada). The nose and mouth were covered with a face mask (Vital Signs, Totowa, NJ). The mask size and the pressure in the air-cushion rim were adjusted to form a leak-free seal around the nose and mouth. Additionally, a thin layer of petroleum gel was smeared onto the air-cushion rim to make the sealing perfect. A loudspeaker-in-box system, which generated the forcing signal for the Zrs measurements, was attached between the face mask and the pump outlet via a lateral port. The forcing signal contained 16 components in the frequency range of 0.5 to 21 Hz and produced < 2 cm H₂O peak-to-peak pressure excursions in the face mask. The components were integer multiples of the fundamental frequency (0.5 Hz). The energies of the low-frequency components were enhanced to attain a sufficient signal-to-noise ratio over the entire frequency range. The phases of the components were optimized to produce minimal peak-to-peak pressure at the airway opening.

Oscillatory flow () was measured with a screen pneumotachograph connected to a differential pressure transducer (ICS 33NA002D). An identical transducer was used to measure oscillatory pressure (Prs). The frequency responses of the transducers sensing and Prs were sufficiently similar for no correction to be necessary. The and Prs signals were low-pass-filtered at 25 Hz, sampled at 128 /s by the analog-digital board of an IBM PC-compatible computer (Dell 486 / 66), and stored for further analysis.

Measurement Protocol

The uncorrected input impedance of each infant's respiratory system (Zrs*) was measured at Ptr_mean levels of 20, 10, and 0 cm H₂O in the following manner. Prior to the oscillatory measurements at 20 cm H₂O, three deep inspirations to 20 cm H₂O were produced by the pump. Between the deep inspirations, the pump was automatically switched off for 0.5 s to let the infant exhale passively. At the end of the third inspiration, the balloon valve was closed, holding the infant at the raised lung volume and inducing a 6- to 8-s apneic period. The oscillations were recorded for 4-s intervals. Following each measurement, the breathing resumed spontaneously after removal of the face mask.

Zrs* data at lower Ptr levels were collected by allowing the infant to exhale passively to a target Ptr_mean of either 10 or 0 cm H₂O. The volume history was similar, but the number of deep inspirations had to be increased up to six prior to the measurements. During oscillations, the Prs versus traces were monitored on the computer screen. The measurements were rejected if there was a leak around the face mask (as indicated by a slow decrease in mean Prs) or any respiratory effort (irregularities in the Prs versus loops) during data collection. Four to six consecutive measurements were made at each pressure level.

After completing the measurement protocol in each infant, we determined the impedance of the apparatus dead space (Zm) by closing the balloon valve and pressing the face mask against a smooth, flat surface.

Computations

Zrs* and Zm were computed from the Prs and signals by fast Fourier transformation using 2-s time windows and 95% overlapping (2). Zm was regarded as a lumped shunt impedance in parallel with Zrs*, and the corrected input impedance of the respiratory system (Zrs) was calculated accordingly (12). The Zrs spectra from the repeated measurements were ensemble-averaged, and the mean data were evaluated by model fitting. As a consequence of the small-amplitude oscillations, poor signal-to-noise ratios, biased estimates, and low reproducibility characterized the impedance at oscillation frequencies that coincided with the heart rate and its harmonics. Therefore, the Zrs values at these frequencies were excluded from the model fit.

The model contained an airway resistance (Raw) and inertance (Iaw) and a constant-phase tissue compartment including tissue damping (G) and elastance (H) (8). The fitting was accomplished in the interval of 0.5 to 15 Hz, by minimizing the root-mean-square error (F) between the modeled and measured impedance data (12). Tissue hysteresivity () (4) was calculated as G/H.

Statistical Analysis

The scatter in the parameters was expressed by SD values. One-way analysis of variance on repeated measurements with the Student-Newman-Keuls multiple comparison method was used to assess the effect of the lung volume on the respiratory parameters. Statistical significance was accepted at the 5% level.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The real (Rrs) and imaginary parts (Xrs) of the measured Zrs spectra, together with the model fits from the three different Ptr_mean levels from one subject, are shown in Figure 1. The measurements were highly reproducible: large standard deviations can be seen only at those points that coincided with the heart frequency or its harmonics. The impedance curves at each Ptr_mean exhibit the same characteristic frequency dependence. The change in the high-frequency level of Rrs indicates that increase of the lung volume caused a monotonous decrease in airway resistance. Total respiratory tissue resistance, reflected by the quasi-hyperbolically decreasing components of the low-frequency real part, was highest at FRC, was markedly decreased at 10 cm H₂O, and was increased again at 20 cm H₂O. The lowest-frequency values of the imaginary part show that elastance was minimal at 10 cm H₂O and was approximately the same at FRC and 20 cm H₂O. The model fitted the impedance spectra without any noticeable systematic error up to ~ 10 Hz; the slight increases in the high-frequency Rrs data are probably attributable to distortion of the parabolic velocity profile. Because this phenomenon was not included in our model, Zrs data above 15 Hz were discarded. The average value of F was 1.1 ± 0.5 cm H₂O · s/L and showed no significant difference between the volume levels.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Real (upper panel ) and imaginary (lower panel ) parts of the total respiratory impedance in a representative infant (No. 2) at Ptrmean of 0 (triangles), 10 (squares), and 20 (circles) cm H2O. Mean ± SD values were obtained by averaging five successive measurements at each Ptrmean level. Lines denote the corresponding model fits. Impedance data corrupted by cardiac noise and those > 15 Hz were omitted from the model fitting (open symbols).

The results of the model fitting are summarized in Figure 2. The parameters displayed a similar volume dependence in each infant over the Ptr_mean range studied. There were no apparent differences in the pattern of volume dependence of any parameter between the healthy infants and those with a history of recurrent wheeze. The mean Raw at FRC was 20.6 ± 4.9 cm H₂O · s/L and decreased highly significantly on increase of Ptr_mean from 0 to 10 (12.5 ± 2.7 cm H₂O · s/L, 62 ± 7% of the value at FRC) and 20 cm H₂O (8.7 ± 2.4 cm H₂O · s/L, 42 ± 7%). At FRC, the average value of Iaw was 0.037 ± 0.014 cm H₂O · s²/L. Although some infants showed marked changes (increase or decrease) in Iaw with Ptr_mean, the volume dependence of this parameter was on average inconsequential. The value of G was highest at FRC, with a mean value of 39.6 ± 10.3 cm H₂O/L and was significantly lower at a Ptr_mean of 10 cm H₂O (26.4 ± 6.0 cm H₂O/L, 68 ± 10% of the value at FRC). Further lung volume increase caused a significant increase in G, but the difference between G at FRC and that at a Ptr_mean of 20 cm H₂O (32.1 ± 10.2 cm H₂O/L, 81 ± 18%) remained statistically significant. The group mean value for H at FRC was 146.8 ± 35.2 cm H₂O/L. H also exhibited a minimum at 10 cm H₂O (112.0 ± 24.1 cm H₂O/L, 77.6 ± 6%) and reached 143.0 ± 44.0 cm H₂O/L at 20 cm H₂O (98 ± 20%). Relative to the average value of 0.27 ± 0.05 at FRC,  was slightly but statistically significantly decreased at higher lung volumes, decreasing to 89 ± 14% (0.24 ± 0.06) at 10 cm H₂O and to 84 ± 14% (0.22 ± 0.02) at 20 cm H₂O.

View larger version (25K): [in this window] [in a new window]   Figure 2.   Airway and tissue parameters of the respiratory system in eight infants measured at three Ptrmean levels. Closed symbols: infants with no respiratory symptoms. Open symbols: infants with previous recurrent wheezy episodes. *p < 0.05 compared with the parameter value at Ptrmean of 0 cm H2O. #p < 0.05 compared with the parameter value at Ptrmean of 10 cm H2O.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In a previous study (12), we demonstrated that the apneic period produced by invoking the Hering-Breuer reflex allows good-quality measurements of the low-frequency respiratory impedance in sedated infants and that this simply implemented technique offers the potential of partitioning the lung mechanics into airway and tissue compartments noninvasively in infants. Nevertheless, because this involved inflating the respiratory system to a Ptr of 20 cm H₂O, the relationship of the derived airway and tissue parameters to those obtained with conventional methods was not straightforward. The results of the present study demonstrate that Zrs measurements in sleeping infants can be extended to lung volumes as low as FRC. Although we have had a high success rate in data collection at an inflation pressure of 20 cm H₂O in every infant successfully sedated (n = 45 from 47 in this and other protocols), five of the 13 infants we attempted to measure at lower lung volumes could not be included in the present study because their apneic interval at low Ptr_mean (either 0 or 10 cm H₂O) was shorter than the minimum required for data collection. Consequently, a higher inflation pressure (20 cm H₂O) clearly provides more favorable conditions for the routine clinical measurement of low-frequency Zrs. The respiratory system impedances at each inflation pressure are in good qualitative agreement with those obtained previously in adults (2) and in various animal species (1, 5, 6, 8) and are very similar to those reported in our earlier study (12). Apart from the positive frequency dependence of Rrs at high frequencies (> 10 Hz), not accounted for by the four-parameter model, the latter was consistent with the Zrs data measured at all Ptr_mean levels. Therefore, the volume dependence of Zrs will be discussed below in terms of the model parameters.

Airway Resistance

We estimated Raw from the total respiratory impedance. Thus, our Raw parameter includes, in addition to the "true" flow resistance of the intrathoracic and extrathoracic airways, the Newtonian resistance of the chest wall (Rw) and also that of the lung parenchyma. The latter has been found negligible in the dog lung (10), and Rw has been shown to comprise a small proportion of the high-frequency Rrs (9, 17). Nevertheless, the fact that Rw appears to be constant (3, 17, 23), while the flow resistance of the airways decreases with increasing lung volume, may result in an increase in the chest wall contribution to our estimated Raw as the lung volume increases. Although we cannot fully exclude the possibility that the inclusion of Rw into our parameter Raw may result in an overestimation of the true airway resistance, particularly at high lung volumes, we argue that this overestimation is not significant. Indeed, Nagels and coworkers found no significant difference in the chest wall contribution to the overall Newtonian respiratory resistance at 20 Hz, when these parameters were determined at 25% (18%), 40% (13%), and 70% (21%) of VC (16).

Our Raw values at FRC (in the range of 14.8 to 29.3 cm H₂O · s/L) are slightly lower than the length-corrected plethysmographic value (~ 30 cm H₂O · s/L, corresponding to our mean length of 78 cm) reported in a consensus statement of infant respiratory mechanics (25). The differences in the values can be attributed to the differences in techniques. During conventional measurements, nonlinearities may be introduced by the relatively high respiratory flows occurring in the nasal pathways during spontaneous breathing (as encountered in the plethysmographic measurements of Raw). In addition, the plethysmographic estimates of Raw obtained during spontaneous breathing reflect the cyclic changes of the glottic aperture throughout the respiratory cycle. In our oscillatory measurements, made during apneic periods, not only were these cyclic changes absent, but we may also assume that the glottic aperture was maximal after the hyperventilatory period. The fact that our Raw values at FRC are slightly lower than the previously reported plethysmographic values implies that our resistance estimates are not dominated by the resistance of the upper airways to a greater degree than those obtained by using body plethysmography.

We are unaware of previous reports on the volume dependence of Raw in infants. However, our observation that Raw decreases significantly with increasing Ptr_mean is consistent with most previous observations on normal adults (13, 22) and various animal species (3, 5, 11, 23, 24). Furthermore, in the present study, we observed a sharp decrease in Raw as the lung volume increased, falling to 63% and 42% of the value at FRC when Ptr_mean increased to 10 and 20 cm H₂O, respectively. The degree of volume dependence reported here is in good agreement with that reported by others. Oostveen and associates measured normal humans and found a 55% decrease in Raw as the lung volume increased from 30% to 70% of the vital capacity (17). Hantos and colleagues reported a 27% decrease in Raw when the transpulmonary pressure was elevated from 4 to 8 cm H₂O in dogs (5). Nagels and coworkers found a ~ 50% decrease in Raw between 25% and 70% of the vital capacity in adults (16). All of these studies involved the use of forced oscillations and Raw was determined from high-frequency Rrs or RL data. However, when Raw was calculated by using end-inspiratory occlusion techniques, the decrease in Raw with increasing lung volume was either considerably smaller (20% decrease [18]) or absent (20, 21, 26). This apparent discrepancy is most probably to be explained by the differences in the techniques applied. End-inspiratory airway occlusion techniques allow estimations of airway and tissue parameters exclusively at higher Ptr levels. For example, during a baseline measurement (PEEP = 0), the effective Ptr levels at end-inspiration, where the measurements are made, are at least 7 to 10 cm H₂O (3, 18). Elevation of PEEP to 20 cm H₂O can increase the end-inspiratory pressure up to 65 cm H₂O (26), depending on the ventilation pattern. Thus, the results of these reports may be expected to differ from those in which oscillatory techniques were used, where the Ptr levels are generally lower.

Inertance

As long as the contribution of the chest wall to the total respiratory inertance can be neglected (9, 16), the parameter Iaw determined from the Zrs spectra corresponds to the inertive properties of the intrapulmonary gas. The results of the few previous studies reported on estimations of Iaw in infants are difficult to interpret. Dorkin and associates (27) used a forced oscillatory technique at higher frequencies to measure Iaw in intubated infants in a similar age range to that reported on in the present study. Values close to zero or occasionally negative were obtained after subtraction of the inertance of the tracheal tube. In that study, the tracheal tube bypassed the upper airways and the trachea, the anatomical loci in which most of Iaw is expected to be located. The Iaw values obtained in the present study at 20 cm H₂O are well within the range of those reported in our previous study at this pressure (12).

Iaw is proportional to the length and inversely related to the cross-sectional area of the airways (28). Assuming that all dimensions of the airways vary with the cube root of the lung volume, we would expect Iaw to decrease with increasing Ptr_mean. We obtained various patterns for the Iaw versus Ptr_mean relationship, and the changes were not statistically significant for the whole population. Similar results were reported by Oostveen and associates in adults (17). We argue that the absence of the assumed volume-dependent changes in Iaw was due to the subordinate role played by Iaw in determining Xrs at low frequencies and to the biasing effects due to processes not incorporated in our model, e.g., the compliance of the cheeks and upper airways. Even if the face mask supported the cheeks firmly during the measurements, the potential remains for the upper airway shunt compliance (Caw) to influence the high-frequency Xrs at low lung volumes. In addition, the increase in the absolute value of the overall impedance (mainly due to the increase in Raw) around FRC tends to magnify the effect of Caw on our Iaw estimates. This interpretation is supported by the results of Nagels and coworkers (16), who measured Zrs in adult volunteers and found a significant increase in the high-frequency Xrs (the major determinant of Iaw) after correction for the shunt impedance of the upper airways.

Tissue Resistance (Damping)

The parameter G determined from the low-frequency Zrs spectra reflects the viscous resistance of both the lung tissues (Rti,L) and the chest wall (Rti,W). Studies on mammalian species with techniques similar to those applied in the present measurements suggest that the two compartments contribute about equally to Rti (9, 11). Whereas the results of the few earlier investigations indicate that the chest wall contribution to the respiratory elastance may be less significant in infancy (25), no such data are available for Rti in infants.

Previous studies have demonstrated that ventilation inhomogeneities, likely to develop during severe bronchoconstriction, may add an artifactual component to Rti,L (1). This phenomenon might influence the pulmonary component of our G estimates if ventilation inhomogeneities develop at low lung volumes. However, because the healthy lung behaves as a reasonably homogeneous system even at FRC (1), our estimates of G probably reflect the properties of the respiratory tissues accurately at each inflation pressure. In support of this concept, we found no obvious differences in the pattern of the volume dependence of G between healthy infants and those with recurrent wheeze.

Because of the difficulties in measuring Rti or Rti,L accurately, there have been only a few studies of these parameters in humans. Polgar and String measured Rti,L in newborn infants by subtracting the plethysmographic Raw from RL (29). The Rti,L value they reported (8.7 cm H₂O · s/L) appears to be consistent with our Rti value of 15.3 cm H₂O · s/L, calculated from the Rrs data at 0.5 Hz (Rti = G/), if methodologic differences (spontaneous breathing versus small-amplitude oscillations) and the contribution of the chest wall impedance in our study are taken into account. The significantly lower values obtained by Bachofen and Duc (30) for Rti,L in 7- to 11-yr-old children (1.08 cm H₂O · s/L) and by Bachofen (14) in adults (0.12 to 1.2 cm H₂O · s/L) may be ascribed to the large differences in lung volume between their groups and our own.

Bachofen measured Rti,L in adults (14) and found a significant increase in it at a lung volume near full inspiration. However, the fact that neither the respiratory rate nor the tidal volume was standardized in that study may bias the results. Human data on the volume dependence of the viscoelastic pressure drop across the respiratory tissues were obtained by analyzing the stress relaxation, manifested in a slow secondary change (Rs) in the tracheal pressure, after end-inspiratory airway occlusion. This technique revealed no change (26) or a slight increase (18, 21) in Rs with increasing lung volume. In view of the fact that the minimum volumes at which measurements were made were those obtained with quasi-static Ptr levels of 7 to 10 cm H₂O, the significant increase we observed in G on increase of Ptr_mean from 10 to 20 cm H₂O is consistent with those results. The findings of recent animal studies in which different techniques were used suggest that Rti,L is constant (11) or increases as the lung volume increases above FRC (3, 5, 6, 24). In an interpretation of these data, it must be taken into account that these studies were performed with alveolar capsules in open-chest animals.

Elastance

At FRC, the values of the respiratory compliance (Crs) calculated from the 0.5-Hz Xrs data in the present study (group mean, 5.9 ml/cm H₂O; range, 4.1 to 8.5 ml/cm H₂O) are smaller than those reported in the consensus statement on infant respiratory mechanics (8 to 16 ml/cm H₂O) after correction for body weight (25). The differences in technique used, and particularly the small-amplitude pressure excursions applied in the present study (which would be expected to yield lower values for tissue compliance, especially in the chest wall [19]), most probably explain this discrepancy. The H values obtained at 20 cm H₂O in the present study display excellent agreement with those obtained previously (12) under similar conditions.

The increase in H above 10 cm H₂O is consistent with previous reports on either static or dynamic estimates of Ers (16, 23, 26). Whether the elastance increases progressively with lung volume, or only following an initial decrease, is likely to be determined by the initial starting volume. If this is below the volume at which airway closure occurs, an initial decrease in H or Ers may be expected with increases in lung volume. Nagels and coworkers demonstrated a significant decrease in the dynamically determined Ers when the lung volume increased from 25% to 40% of VC (16). Barnas and colleagues recently confirmed this pattern by using various amplitudes of sinusoidal forcing in normal humans (19) and in dogs (23). Our finding that H exhibits decrease above FRC is fully consistent with the result of those studies.

Tissue Hysteresivity

Whereas numerous studies have investigated the changes in  under a wide variety of conditions in different species, our previous report on infants was the first to report values of this parameter in humans (12). In both the previous and the present study, similar values for  were obtained to those reported for the respiratory system in various mammalian species (1, 4, 8, 24). Our current observation that  is independent of the lung volume is also in complete agreement with the results of previous animal studies, where the changes in G were proportional to those in H (5, 24). Parallel changes in G and H support the notion that the increases in G and H seen at a Ptr of zero accurately reflect the mechanical properties of the respiratory tissues. The small but statistically significant increase in  at the lowest lung volume may reflect either intrinsic changes in the tissue properties or an artifactual increase in G secondary to ventilation inhomogeneity (1). In the latter case, we would expect the increase to be more marked in infants with recurrent wheeze. The fact that this did not occur raises the possibility that either the degree of ventilation inhomogeneity (producing airway closure) was similar in the normal infants and those with recurrent wheeze or that the relatively greater increase in G reflects the tissue mechanics.

Conclusions

The respiratory impedance data measured in the present study exhibit the same qualitative frequency dependence, independently of whether they were obtained at a Ptr of 0, 10, or 20 cm H₂O. Although the parameters estimated from the Zrs data and representing the mechanical properties of the airways and the respiratory tissues are characteristic of the measurement technique, i.e., small-amplitude oscillations in a low-frequency range, the changes in the parameters in the response to an altered Ptr_mean are in accordance with previous findings relating to the volume dependence of the respiratory mechanics.

The mechanical parameters estimated from oscillations at the level of FRC or at moderate hyperinflation (Ptr_mean of 10 cm H₂O) are more relevant to the conditions characterizing spontaneous respiration and theoretically are preferable to the higher-pressure estimates. From a practical point of view, however, the measurements at the lung volume achieved with a Ptr of 20 cm H₂O are advocated for general use, as they can be carried out at a low incidence of insufficiently short apneic periods.