INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

There is increasing interest in using respiratory input impedance measurements (Zin) as a pulmonary function test in infants. Zin measurements can be performed within seconds during tidal breathing and often do not disturb the infant during natural unsedated sleep. Few studies have reported Zin measurements in infants (1). Sly and colleagues (4) reported Zin measurements in healthy infants at low frequencies (0.2- 20 Hz) during reflex induced apnea. Desager and coworkers (3) and Marchal and colleagues (1, 2) reported Zin between 6 and 48 Hz during tidal breathing in wheezy and healthy infants, respectively. Finally, Jackson and colleagues (5) measured Zin between 20 and 256 Hz in healthy infants. However, in none of these studies were the Zin measurements made following induced airway constriction or dilation, whereby the change in airway mechanics was verified by a reference lung function technique.

As pointed out by Jackson and colleagues (5), it is important to measure Zin over a wide range of frequencies because the factors that influence Zin at low frequencies (< 2-10 Hz) are different from those that affect Zin at higher frequencies. For example, at high frequencies airway wall properties become important. The ability to measure a parameter sensitive to airway wall properties could be particularly important in infants. Much evidence covering the relationship between fetal lung growth and the subsequent risk of obstructive airway disease in infancy (6) has been established on partial flow-volume curves produced by the rapid chest-compression technique (RCT). A reduction of forced flow at FRC (maxFRC) in early infancy is a risk factor for subsequent wheezing disease (6). Although it has been stated that reduced maxFRC indicates "smaller airways," maximal flow is determined not only by cross-sectional area but also by airway wall properties (airway wall compliance) (9). It is known from animal work that airway wall compliance is greater in infancy than in later life (10), so that the relative contribution of airway wall properties to flow limitation is also likely to be greater in infancy.

As a first step, this study aimed to determine whether changes in Zin measured over a wide range of frequencies were sensitive to alterations in airway mechanics in infants. We used methacholine challenge to induce changes in airway mechanics and verified these by a reference lung function technique. Secondly, we wished to determine whether the induced changes in Zin were similar to those found in adults with airway obstruction. This would imply that similar analysis techniques could be applied to Zin data from infants in order to extract relevant physiological parameters.

Theoretical Aspects of Zin Data Interpretation

Studies using Zin (11, 12) to measure lung function in animals or adults have made measurements either at low frequencies (f < ~2 Hz) or at higher frequencies (f > ~2 Hz) (13). Zin at low frequencies is largely a function of the viscoelastic properties of the tissues and airway resistance (Raw). At higher frequencies, Zin is similarly influenced by Raw, influenced less by tissue properties, and influenced more by airway wall properties.

Zin data have been analyzed either using systems identification techniques (1, 11, 20, 21) or by considering changes in specific features of the Zin spectra (3, 17). For example, in adults the severity of airway obstruction is directly related to the degree of negative frequency dependence in the real part of Zin (Zin_re) and the resonant frequency (where Zin_re is a relative minimum and the imaginary part, Zin_im, crosses zero) and inversely related to the antiresonant frequency (where the Zin_re is a relative maximum and Zin_im crosses zero) (17). Unlike feature analysis, systems identification techniques provide estimates of physiological parameters by fitting a model to the Zin data. One such model, the DuBois six-element model (21) (Figure 1a) provides separate estimates of airway and tissue resistance (Raw, Rti), as well as thoracic gas volume (Vtg), which corresponds to the gas compression compliance (Cg). However, this model can be used only if the Zin data include an antiresonance that is related to the tissue inertance (Iti) and the alveolar gas compression compliance (Cg). There is such an antiresonance in dogs (13, 14, 16) and rabbits but not in adult humans (15). Instead, the antiresonances in adults are due to wave-propagation phenomena and are thus related to inertance of the gas within the airways and the compliance of the airway walls (15). Because the anti-resonances are related to wave propagation, estimates of Raw and Vtg are not possible in human adults; nevertheless, inferences about airway wall properties are possible (15, 17). It has only recently been shown that there is an antiresonance in healthy infants at approximately 120 Hz (5). Although it is not clearly understood what phenomena contribute to this antiresonance in infants, preliminary results indicate that it is related to the total respiratory system inertance (Irs) and at least partly due to the gas compression compliance in the face mask (5). A model was suggested (Figure 1b) for analyzing infant Zin data that, according to computer simulations, results in an estimate of total respiratory system resistance (Rrs) that may be related to airway caliber (5).

View larger version (14K): [in this window] [in a new window]   Figure 1.   (a) Six-element model (Reference 21), (b) four-element model (5). Zaw = airway impedance; Raw = airway resistance; Caw = airway compliance; Iaw = airway inertance; Zti = tissue impedance; Rti = tissue resistance; Cti = tissue compliance; Iti = tissue inertance; Cg = gas compression compliance; Cm = face mask shunt compliance; Rrs = respiratory system resistance; Crs = respiratory system compliance; Irs = respiratory system inertance.

Based on these theoretical considerations, our goals can be described in more detail. We were interested in determining whether induced airway obstruction resulted in increased frequency dependence in the real part of Zin and changes in the frequency of the resonance and/or antiresonance. Questions related to model analysis were whether the method suggested by Jackson and coworkers (5) provides an estimate of Rrs that is a sensitive indicator of changes in airway caliber.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. The study was performed in 13 infants and young children (6 girls, 7 boys), age 26-81 wk, with episodic or persistent wheeze who had been referred from the outpatient clinic for lung function tests (Table 1). Infants with other specific diseases such as tracheal or upper airway disease or an upper respiratory tract infection within the previous 3 wk were not included in the study. We did not study healthy infants for ethical reasons. The patients were sedated using a maximum dose of 150 mg/kg triclofos sodium, and lung function was measured during behaviorally defined quiet sleep. The protocol was approved by the Ethics Committee of the Royal Postgraduate Medical School, Hammersmith Hospital, London, where the study was carried out. Written consent was obtained from the patient's parents.

Forced-oscillation technique for infants. Zin was measured using loudspeaker-generated, low-amplitude pressure oscillations applied to the airway opening via a latex face mask (Rendell Baker Soucek, Size 1; Ambu International, Linthicum, MD). The setup and its calibration has been described previously (22). A pseudorandom noise signal containing frequencies between 2 and 256 Hz in 2-Hz increments was generated by computer and output via a digital-to-analog converter. Zin was measured using the wave-tube technique described in detail by Franken and colleagues (23). Briefly, in the forced-oscillation technique (FOT) oscillatory pressures are measured at two sites (6.7 cm apart) in a rigid tube (radius, 0.5 cm) between the loudspeaker mechanism and the mouthpiece. The first pressure transducer (P₁) was placed 3.3 cm and the second pressure tranducer (P₀) 10 cm from the airway opening (face mask).

Zin was computed as the load impedance of the tube by this equation:

(1)

where L is the distance between the two pressure transducers (6.7 cm), Zc is the characteristic impedance of the tube, and  is the propagation coefficient of the tube. Both Zc and  are computed from theoretical considerations using the diameter of the tube and L (24). P₁ and P₀ were measured with piezoelectric solid-state pressure transducers that were matched within ± 2% in magnitude and ± 2° of phase (EuroSensor, Model 33; London, UK). The electrical output of these transducers was band-pass filtered (8-2,000 Hz), amplified, and analog/digital (AD) converted at 2,048 Hz (MetraByte Model DAS 16/ 16F; Keithley Instruments, Reading, UK). The ratio of P₁/P₀ was estimated from the cross-power spectra of P₀P₁ and the auto-power spectra of P₁. The coherence function was computed using the method of Michaelson and colleagues (25). The wave-tube method of measuring Zin has the advantage that flow is not measured directly but is inferred by two pressure measurements made by transducers whose frequency response is linear from 4 to 1,000 Hz. When flow is measured using a pneumotachometer, care must be taken to compensate for the frequency response characteristics of the pneumotachometer; there is an upper frequency limit to where this can be done accurately (26). The wave-tube technique does, however, have the disadvantage that accurate and reliable measurements at low frequencies become difficult. With the particular tube used in this study, we were unable to measure Zin reliably below 16 Hz (23).

The rapid thoracic compression technique. Partial expiratory flows at FRC (maxFRC) were measured using the rapid chest-compression technique described by Clarke and colleagues (22). Infants wore an inflatable polyethylene thoracoabdominal jacket (Medical Engineering Department, Royal Postgraduate Medical School, Hammersmith Hospital, London) with arms in. Flow was measured using a face mask (Rendell Baker Soucek, Size 1; Ambu International) and a Fleisch No. 1 pneumotachograph (Validyne MP45, Northridge, CA). The linearity was estimated to be accurate within 2%. The maxFRC and the flow signal were AD-converted and assessed using Rasp software (Physiologic, Newbury, UK). From the 10 forced expiratory maneuvers, we calculated the mean and standard deviation of maxFRC and related the values to the reference values of Tepper and colleagues (27).

Experimental Protocol

Measurements were made with the infant in the supine position. Transcutaneous Po₂ (Radiometer, Copenhagen, Denmark) and transcutaneous saturation Sa_O2 (BioxIII; Omeda, Boulder, CO) were monitored. The head position was standardized based on the experience of Desager and colleagues (3), apart from the fact that the oral airway opening was not taped because of safety considerations. The head position was not changed between measurements using the different techniques or subsequent methacholine challenge. In order to achieve a complete seal and to reduce gas volume, the face mask was filled with putty (Therapeutic Putty; Carters, Bridgend, UK). During quiet regular tidal breathing, two sets of eight Zin() measurements were performed using the forced-oscillation technique; each set required 5 s. The second set was taken to estimate the reproductiblity of the Zin measurements. Thereafter, the inflatable jacket was wrapped around the infant's chest and 10 forced expirations were performed at baseline.

Methacholine challenge test. In nine of the 13 infants, a complete methacholine challenge test was performed as described previously (22) using cumulative doses of methacholine (saline, then 0.5, 1, 2, 4, 8, 16, 32, and 64 g/L methacholine). Four infants awoke before completion of the test. At each provaction dose, five maxFRC measurements were taken and averaged. Once a provocation dose resulted in a decrease of maxFRC of 30% (i.e., the PC₃₀), no higher concentrations of methacholine were given and a final set of Zin measurements was taken. Zin was not measured at intermediate steps. As already stated, head and face mask position were not changed during or between the tests. We kept the sequence of measurements constant (FOT, FOT, RTC, challenge, RTC, FOT), because putting on and removing the squeeze jacket can disturb the child. Using this sequence, we only put the jacket on once and removed it once for each child.

Data analysis of spectral features. Four parameters from the Zin spectra related to airway obstruction (1) were analyzed: the resonant (fr,1), and the anti-resonant frequencies (far,1), the relative maximum in the real part at far,1 [Zin_re(far,1)] and the frequency dependence of Zin_re at f < fr,1. The latter parameter was quantified by the slope  of a linear regression between 16 Hz and 30 Hz, as well as between 16 Hz and fr,1 (₃₀ and _fr,1, respectively). Each of these four parameters was assessed from the mean of eight single runs. The short-term repeatability of these four parameters was estimated by the standard error of the mean of the differences between the first and second parameters derived from the two sets of baseline Zin measurements. The reproducible parameters were displayed in Bland Altman plots. To estimate the change in the parameters derived from Zin during the bronchial challenged tests, we compared the first measurement set at baseline to the postchallenge set using paired t tests.

Data analysis using systems identification technique. The Zin() data were analyzed using a lumped four-element model (Figure 1b) as proposed by Jackson and colleagues (5). The parameters in this model were estimated by minimizing the following performance index, PI, given by

(2)

where n = the number of data points, Z_rs,d = Zin data at a given frequency f, and Z_rs,m = model-predicted Zin data at a given frequency using the model parameters. The model parameter of particular interest as a possible index of airway caliber was Rrs (Figure 1b). Similarly, we calculated short-term repeatability of Rrs using the standard error of the mean differences, and we estimated the change in Rrs after bronchial challenge using a paired t test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Reference lung function technique. At baseline, the mean (± SD) maxFRC was found to be 196 ± 93 ml · s¹, which corresponds to 64 ± 27% of predicted values (27). In the subgroup of nine infants who had a challenge test, maxFRC decreased significantly from 234 ± 97 ml · s¹ to 125 ± 69 ml · s¹ (p < 0.0005), confirming that a significant change in airway mechanics was induced by the methacholine challenge test.

Description of Zin data. In all subjects, the coherence was greater than 0.95 within the frequency range 16 to 256 Hz. At baseline, Zin() spectra in all 13 infants showed a first resonant frequency (fr,1) at 37 ± 9 Hz and a first antiresonant frequency (far,1) at 155 ± 28 Hz (Figure 2). The relative maximum of Zin_re at far,1 [Zin_re(far,1)] was found to be 51 ± 10 cm H₂O · L¹ · s. Most of the infants showed a certain degree of frequency dependence of resistance at frequencies below fr,1. The frequency dependence of resistance was 0.25 ± 0.35 cm H₂O · L¹ · s Hz¹ between 16 and 30 Hz and 0.21 ± 0.27 cm H₂O · L ¹ · s Hz¹ and between 16 Hz and fr,1.

View larger version (35K): [in this window] [in a new window]   Figure 2.   Impedance spectra (mean and standard deviation for all 13 infants) from 16 to 256 Hz. The mean resonant frequency fr,1 was 37 ± 9 Hz, and the first antiresonance occurred at 155 ± 28 Hz.

The standard error of the mean differences in the spectral features for fr,1, far,1, Zin_re(far,1), _fr,1, and _30Hz as a measure of short-term repeatability were 3.03 Hz, 2.64 Hz, 1.78 cm H₂O · L¹ · s, 0.07, and 0.09, respectively. Bland Altman plots are only presented (Figure 3) for the parameters far,1 and Zin_re(far,1), which we found to be sensitive to changes in airway mechanics. These findings indicate a relatively good short-term repeatability for all parameters. The repeatability of Zin_re(far,1) in a single subject (Figure 3) was poor. This subject was an infant with recurrent wheeze since birth, who suffered from oligohydramnios antenatally, and who was suspected of having lung hypoplasia. There were no distinctive clinical features among the other three subjects whose values exceeded the 95% confidence interval.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Bland Altman plot representing short-term repeatability of the first antiresonance ar,1. The difference in frequency far,1 and amplitude Zinre( far,1) derived from the first and second measurement sets are presented as a function of their mean values. The solid lines represent the 95% confidence intervals of the mean differences. Asterisk indicates repeatability of Zinre(far,1) in a single subject.

Lumped four-element model. An example of an impedance spectrum from 16 to 256 Hz in one infant and the fit of the four-element model are shown in Figure 4. In fitting the Zin data with this model, data at frequencies where the real part of Zin had a negative frequency dependence were eliminated (f < 27 ± 9 Hz). In one infant, Zin data at f > 220 Hz could not be followed by the model, so those data were not used in the system identification process. In all other infants, the model fitted the data reasonably well up to 256 Hz. From the first set of Zin measurements for the whole group, the mean Rrs was found to be 20.6 ± 4.7 cm H₂O · L¹ · s; Crs, 1.87 × 10³ ± 2.42 × 10³ L · cm H₂O¹; Irs, 0.025 ± 4.5 × 10³ H₂O · L¹ · s²; and the face mask shunt compliance Cm, 2.2 × 10⁵ ± 1.1 × 10⁵ L · cm H₂O¹ (equivalent to a face-mask dead-space volume of 21 ml). The standard error of the mean of the differences between Rrs from the first and the second Zin measurement was found to be 2.32 cm H₂O · L ¹ · s.

View larger version (21K): [in this window] [in a new window]   Figure 4.   Mean impedance spectrum of a set of eight measurements. In a single subject, the first resonance occurs at 34 Hz, and the first antiresonance occurs at 200 Hz. The four-element model fit is represented by a solid line (Cm = 2.20 × 105 L · cm H2O1, Rrs = 17.13 cm H2O · L1 · s, Crs = 1.19 × 103 L · cm H2O1, and Irs = 2.2 × 105 cm H2O · L1 · s2).

Changes of Zin-and RCT-derived parameters during methacholine challenge. At baseline, there was a tendency for fr,1 to be higher and far,1 to be lower if maxFRC was low; however, the correlations were not significant. There was no correlation between , Rrs, and maxFRC under baseline conditions for the group of 13. Because of the protocol design, we cannot exclude an effect of repeated rapid thoracic compression on impedance parameters.

A representative example of Zin at baseline and after methacholine challenge is shown in Figure 5 for one subject. Both frequency (far,1) and the relative maximum in the real part [Zin_re(far,1)] of Zin increased following methacholine challenge. On average, far,1 increased significantly (t = 5.34, p = 0.007) from 148 ± 26 to 203 ± 15 Hz (Figure 6). Zin_re(far,1) also changed significantly for the whole group (t = 3.0791, p = 0.015), from 51 ± 10 to 70 ± 15 cm H₂O · L¹ · s, but as seen in Figure 6, Zin_re(far,1) did not change in three of the infants.

View larger version (23K): [in this window] [in a new window]   Figure 5.   Representative measurement sets (mean ± SD of eight Zin runs) in a single subject assessed at baseline (closed circles) and after methacholine challenge (open circles). The frequency far,1 and amplitude Zinre(far,1) of the first antiresonance increased during the challenge.

View larger version (16K): [in this window] [in a new window]   Figure 6.   The frequency far,1 (t = 5.34, p = 0.007) from 148 ± 26 to 203 ± 15 Hz and Zinre(far,1) increased significantly for the whole group of nine infants after methacholine challenge (t = 3.0791, p = 0.015) from 51 ± 10 to 70 ± 15 cm H2O · L1 · s.

The frequency dependence of resistance represented by _30Hz (postchallenge, 0.27 ± 0.34, p = 0.28) and _fr,1 (postchallenge, 0.22 ± 0.23, p = 0.34), as well as the resonant frequency fr,1 (postchallenge, 38 ± 8 Hz, p = 0.77), did not change significantly during methacholine challenge. None of the respiratory system parameters (Rrs, Crs, Irs) derived from the system identification technique using the four-element parameter model changed significantly following methacholine (t = 0.05, p = 0.95), as shown in Figure 7. Although the numbers were small, there were no systematic differences between boys and girls.

View larger version (19K): [in this window] [in a new window]   Figure 7.   Rrs derived from system identification of Zin() using a four-element model in comparison to the maxFRC following methacholine challenge. The Rrs did not change during the methacholine challenge test.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Zin measurements are typically reported over one of two frequency ranges: a low frequency range (f < about 2-10 Hz), where tissue properties dominate the impedance spectrum (4, 11, 12), and a higher frequency range where tissue properties become less important and airway caliber and wall properties become more influential (13). Zin measurements at the low frequency range are potentially useful for measuring the mechanical properties of the respiratory tissues, lung, and chest wall. Furthermore, as argued by Sly and colleagues (4), parameters derived from Zin measurements at these lower frequencies may be more pertinent to conditions of natural breathing. However, the goal of the current study was not to investigate the behavior of the infant respiratory system under conditions of natural breathing but to test the hypothesis that Zin measurements well above 2 Hz are sensitive to changes in airway caliber and airway wall properties in infants. Zin measurements have been reported in infants (1), but not to high enough frequencies to reveal an antiresonance where Zin is thought to be sensitive to airway wall properties. It has also not been shown that Zin is sensitive to changes in airway mechanics by comparing values before and after induced bronchial obstruction, which was verified with an independent lung function measurement. We report here for the first time Zin in infants to frequencies that include an antiresonance during baseline conditions and following bronchconstriction induced by methacholine.

Our Zin measurements are qualitatively similar to those reported in the literature. In most of our subjects, the real part showed negative frequency dependence at frequencies below fr,1. Frequency dependence has been reported in subjects with wheezing disorders (1, 3). Marchal and colleagues (1) also found frequency dependence of the real part below fr,1 in healthy infants, but Jackson and colleagues did not (5). These findings are consistent with findings in adult patients who had airway obstruction (28) compared to healthy adults. The frequency dependence of resistance in adults is thought to be due to parallel inhomogeneity or to nonrigid behavior of the upper or central airways (14). The results by Marchal and coworkers (1) might be explained by recent computer predictions (5) suggesting that because infant airways are so compliant the real part could be frequency-dependent even in healthy infants.

Our mean fr,1 (37 ± 9 Hz) was slightly lower that the fr,1 reported by Jackson (5) and Desager (3) and their colleagues. The frequency of the first antiresonance (155 ± 28 Hz) in wheezy infants was significantly higher than the one reported (113 ± 10 Hz) in healthy infants (5). This would be consistent with our observation that far,1 increased after airway obstruction, as well as Chalker and colleagues' observation (17) that far,1 was higher in adults with chronic airway obstruction than in healthy individuals, possibly as an expression of altered airway wall properties.

Model fitting using system identification techniques. The lumped four-element model, as well as the six-element model of DuBois (21), is not capable of fitting the negative frequency dependence of the real part of Zin for frequencies below fr,1. The four-element model is, however, able to follow the Zin data between about 16 and 256 Hz, which includes the resonance and antiresonance, providing estimates for Rrs, Crs, and Irs. We therefore determined whether this was true in infants (5). As expected, the lumped four-element was not capable of fitting the negative frequency dependence of the real part of Zin for frequencies below fr,1, but it was able to follow the Zin data between about 16 and 256 Hz, which includes the resonance and antiresonance (Figure 4). The mean baseline value for Rrs (20.6 ± 4.7 cm H₂O · L¹ · s) was similar to that in a patient population similar to ours (3), but slightly higher than the values reported in healthy infants by Marchal and colleagues (1) (17.4 ± 5.3 cm H₂O · L¹ · s) and by Jackson and colleagues (5) (15.6 ± 3.6 cm H₂O · L¹ · s), consistent with studies in adults (28). Our baseline values for Crs (1.87 ± 2.42 × 10³ L · cm H₂O¹) were between the values reported by Marchal and colleagues (1) (4.95 ± 8.26 × 10³ L · cm H₂O¹) and by Jackson and colleagues (5) (1.03 ± 0.58 × 10³ L · cm H₂O¹) in healthy infants. This difference is consistent with the age distribution of the patients in the different studies. However, one would expect much higher values for Crs (1- 2 × 10³ L · cm H₂O¹ per kg body weight) under quasistatic conditions or at tidal breathing frequencies. One contribution to this difference could be the frequency-dependent decrease of Crs. However, the imaginary part of Zin even for frequencies surrounding fr,1 is significantly influenced by Caw and Cg (13). As a consequence, the Crs extracted from the Zin data using the four-element model would be a complex function of all three compliances (Cti, Caw, and Cg). Although we found values of Rrs and Crs under baseline conditions similar to those of other research groups, the more important question is whether parameters derived from the four-element model are sensitive to changes in airway mechanics during induced airway obstruction.

Behavior of the maximal expiratory flows (RCT) and Zin during induced airway obstruction. In comparison to the reference values of Tepper and coworker (27), our baseline values for maxFRC showed mild airway obstruction at baseline and significant airway obstruction following methacholine challenge. Despite this, there was no significant change in fr,1 or the frequency dependence of the real part of Zin (). It appears there is no simple parameter in the Zin spectrum between 16 and approximately 100 Hz (below the first antiresonance) that is sensitive to methacholine-induced changes in airway mechanics.

However, at higher frequencies (> 100 Hz) we found that the antiresonance frequency (far,1) and the relative maximum in the real part at far,1 [Zin_re(far,1)] of the first antiresonance were highly sensitive to changes in airway mechanics. As mentioned above, at these higher frequencies, Zin is influenced by Raw and by airway wall properties but less by tissue properties. We conclude that in infants Caw significantly influences Zin at frequencies surrounding the far,1. Because the four- element model assumes that all of the shunt compliance is assigned to the face mask, it is not able to explain the changes in Zin during methacholine challenge; as a consequence, the resulting respiratory resistance (Rrs) does not reflect the changes in airway resistance during induced airway obstruction. We suggest that more complex models that include the separate face-mask gas compression as well as the airway wall compliance might be needed to interpret Zin over this frequency range in infants. However, the change in far,1 during the methacholine challenge is evidence that far,1 is, at least in part, due to wave-propagation phenomena and is thus related to inertance of the gas within the airways and the compliance of the airway walls, as in adults (15). If the far,1 and its higher harmonics are related to wave propagation, then Zin at these frequencies is a function of airway geometry (length and diameter), gas density, and wall mechanical properties (15, 17- 20, 29). This concept has been experimentally verified in dogs (16) and in human adults (15), where it has been shown that far,1 increased as a function of gas density.

Although the high-frequency Zin data in infants seem to behave similarly to those in adults, extraction of physiologically relevant parameters using distributed parameter models is not possible, as it is in adults and in dogs (18). In order to apply models to what is at the moment empirical data, more anatomic information of infants' upper and lower airways in health and disease is needed. Secondly, measurements of Zin to frequencies higher than 256 Hz must be made in order to include the second, and hopefully the third, antiresonance.

Summary

We have shown that high-frequency Zin in infants contain an antiresonance whose frequency is repeatable and which changes significantly during induced airway obstruction. The frequency of the first antiresonance is higher following airway obstruction. This behavior is consistent with the findings in human adults (17). High-frequency Zin data might therefore be useful for assessing changes in airway mechanics noninvasively in infants. Since measurements only take a few seconds, they are often possible in unsedated infants during natural sleep (unpublished observations). This makes the technique very useful for the rapid assessment of changes in airway mechanics during bronchial challenge. We have shown that the four-element model cannot explain high-frequency Zin data, probably because airway wall compliance becomes increasingly important at higher frequencies. We suggest that, as in human adults, changes in frequency far,1 during airway obstruction might be explained by changes in wave-propagation velocity in airways with changing wall compliance. We further suggest that the combined measurements of maxFRC and far,1 from Zin could provide additional information about changes in airway caliber versus changes in airway wall compliance. For example, if maxFRC were to decrease while far,1 remained constant, it would imply a change in airway caliber but not in airway wall compliance.