Vol. 88, Issue 3, 997-1005, March 2000
Lung resistance and elastance in spontaneously breathing
preterm infants: effects of breathing pattern and demographics
Paresh B.
Pandit1,
Kee H.
Pyon1,
Sherry E.
Courtney1,
Sandra E.
England2, and
Robert
H.
Habib3
1 Department of Pediatrics, The Children's
Regional Hospital at Cooper Hospital and Robert Wood Johnson Medical
School, Camden 08103; 2 Department of Pediatrics,
Robert Wood Johnson Medical School, New Brunswick, New Jersey 08901;
and 3 Department of Pediatrics, Mercy Children's
Hospital at St. Vincent Mercy Medical Center, Medical College of Ohio,
Toledo, Ohio 43608
 |
ABSTRACT |
Reported values of lung resistance
(RL) and elastance (EL) in spontaneously
breathing preterm neonates vary widely. We hypothesized that this
variability in lung properties can be largely explained by both inter-
and intrasubject variability in breathing pattern and demographics.
Thirty-three neonates receiving nasal continuous positive airway
pressure [weight 606-1,792 g, gestational age (GA) of
25-33 wk, 2-49 days old] were studied. Transpulmonary pressure was measured by esophageal manometry and airway flow by face
mask pneumotachography. Breath-to-breath changes in RL and
EL in each infant were estimated by Fourier analysis of
impedance (Z) and by multiple linear regression (MLR).
RLMLR (RLMLR = 0.85 × RLZ 
0.43; r2 = 0.95) and ELMLR
(ELMLR = 0.97 × ELZ + 8.4; r2 = 0.98) were
highly correlated to RLZ and
ELZ, respectively. Both RL
(mean ± SD; RLZ = 70 ± 38, RLMLR = 59 ± 36 cmH2O · s · l
1)
and EL (ELZ = 434 ± 212, ELMLR = 436 ± 210 cmH2O/l)
exhibited wide intra- and intersubject variability.
Regardless of computation method, RL was found to decrease
as a function of weight, age, respiratory rate (RR), and tidal volume
(VT) whereas it increased as a function of
RR · VT and inspiratory-to-expiratory
time ratio (TI/TE). EL decreased
with increasing weight, age, VT and female gender and
increased as RR and TI/TE increased. We
conclude that accounting for the effects of breathing pattern
variability and demographic parameters on estimates of RL
and EL is essential if they are to be of clinical value.
Multivariate statistical models of RL and EL
may facilitate the interpretation of lung mechanics measurements in
spontaneously breathing infants.
impedance; multiple linear regression; frequency dependence; amplitude dependence
| |
INTRODUCTION |
BABIES WITH VERY LOW BIRTH WEIGHT (VLBW) due to
prematurity almost invariably require respiratory support because of
surfactant deficiency, underdeveloped lungs, and/or immature
respiratory control (6, 7, 16, 18-21, 27, 33). However, VLBW
infants are increasingly being supported by administration of exogenous surfactant followed by less invasive respiratory support such as nasal
continuous positive airway pressure (NCPAP; Refs. 7, 12). Allowing
spontaneous ventilation while preventing alveolar collapse by NCPAP, as
opposed to positive pressure mechanical ventilation, is believed to
decrease lung barotrauma and allows easier access to the infant (8,
12).
Two important limitations of NCPAP-spontaneous ventilation support are
1) the possibility of failure, leading to intubation, and
2) the lack of a well-defined method to guide weaning neonates off this support. Instead, weaning is often a trial-and-error process
that is highly variable among care providers. In theory, if reliable
and reproducible, serial lung mechanics measurements probing the
progression or regression of the underlying lung disease may provide a
quantitative basis for weaning infants from NCPAP support. This
approach, however, is hindered by the difficulty of interpreting these
measurements given the large inter- and intrasubject variability of
lung resistance (RL) and elastance (EL)
estimates (1, 2, 6, 7, 10, 13, 15, 17-21, 27, 33).
We hypothesized that breathing pattern and demographic factors are
responsible for a substantial portion of this variability. The main
objective of this study was to elucidate how breathing pattern
parameters and patient demographics influence RL and
EL in spontaneously breathing preterm infants and to
examine whether these RL and EL dependencies
are altered by the method used to compute these properties.
| |
METHODS |
Subjects
Lung mechanics measurements were performed on a total of 37 infants on
NCPAP support in the neonatal intensive care nursery. Persistent leaks
around the face mask did not allow for data analysis in four infants.
The patient characteristics of the remaining 33 are shown in Table
1. The ethnic distribution was Caucasian: 15, African-American: 11, Hispanic: 6, and Asian: 1. At the time of
measurements, the degree of respiratory distress in these infants was
mild (median NCPAP = 5 cmH2O, median inspired
O2 fraction = 0.25). This study was approved by the
Institutional Human Investigation Committee and was performed with
parental consent.
Measurements and Protocol
Airway flow (Flow) was measured in preterm infants during quiet sleep
with a calibrated neonatal fixed-orifice pneumotachograph (Novametrix,
Wallingford, CT; dead space = 0.8 ml) via a neonatal face
mask. Tidal volume (VT) was calculated by
numerical integration of Flow. Leaks around the face mask were removed
by repositioning the face mask until none was detected. Detection of
airway leaks was possible from monitoring of VT obtained
via real-time integration of Flow. In a few cases, petroleum jelly was
used around the mask to ensure that no leaks were present. Only
stretches of breaths with no detectable leaks were considered for analysis.
Airway opening (Pao) and esophageal (Pes) pressures were measured with
pressure transducers (Microswitch 743PC). For Pes measurement, each
infant received a neonatal esophageal balloon catheter (Ackrad Laboratories, Cranford, NJ) inserted so that the balloon was, in the
esophagus, at the level of the lower third of the trachea. Proper
positioning of the esophageal balloon was checked by continuous on-line
monitoring of Pes and adjusted until a high correlation (r2 > 0.90) was obtained between Pao and Pes
tracings with the airway occluded (5). To facilitate quiet sleep,
infants were generally fed via a nasogastric tube after instrumentation
was complete and before initiation of measurements. At completion of
feeding, the nasogastric tube was withdrawn to avoid possible
measurement artifacts.
Flow, Pao, and Pes were zeroed and calibrated at the beginning of each
experiment. Their time signals during spontaneous breathing were then
sampled at 100 Hz, monitored on-line, and stored on a computer
(Ventrak, Novametrix) for later analysis. Data collection took
place in stretches of 30-60 s, which were repeated when
necessary to allow for variability in breathing patterns
of individual subjects. Measurements were done with NCPAP and
O2 support discontinued.
Data Analysis
First, time domain data from each infant were examined, and all breaths
with evidence of airway leaks around the face mask were excluded. Then,
breath-to-breath estimates of RL and EL were derived in all infants from transpulmonary pressure (Ptp; Ptp = Pao Pes), Flow, and VT data using two methods of calculation.
Lung impedance.
Before calculation of the breath-to-breath lung impedance (Z), the
sampled time signals [Ptpn and
Flown] for each breath were processed as
follows: 1) the mean Ptp and Flow over the full breath were
subtracted from Ptpn and Flown
so that both signals were of zero mean; and 2) both
Ptpn and Flown were then padded
with zeros so that the number of points (N) is increased to the
next higher power of two (m) so that N = 2m. For example, the zero mean breath
data corresponding to a respiratory rate (RR) of 60 min1
(or 100 samples given the 100 Hz sampling rate) would be padded with 28 zeros at the end of the breath to allow the use of 128-point fast
Fourier transform (FFT). Lung resistance (RLZ)
and lung elastance (ELZ) as determined by
Fourier analysis (Matlab, The Math Works, Natick, MA) were then
calculated as follows
![R<SC>l</SC><SUB>Z</SUB> = Real[Z(f<SUB><IT>k</IT> = 1</SUB>)]; <IT>k</IT> = 1 . . <IT>N</IT>](/content/vol88/issue3/fulltext/997/img002.gif)
|
(1)
|
Where
RLZ and ELZ are the
lung mechanical properties at fk = 1 or the
breathing frequency (3, 24).
Multiple linear regression.
Lung resistance (RLMLR) and elastance
(ELMLR), as determined by multiple linear
regression (MLR), were estimated by least squares fitting (Matlab, The
Math Works) of the time domain Ptpt data over the
entire breath in terms of Flowt, VT,
and the elastic recoil pressure at end expiration (P0)
using a series resistance-elastance model of the lungs (16, 33)
|
(2)
|
Where
Ptpmod,t represents the MLR model
(mod) estimate of Ptpt.
Multivariate Statistical Models of RL and
EL
In all multivariate modeling, all breath-to-breath estimates of
RLZ, ELZ,
RLMLR, and ELMLR from
all infants were considered in conjunction with their corresponding
breathing pattern and demographic variables. The following breathing
pattern variables were considered: RR, breath period (T), inspiratory
time (TI), expiratory time (TE),
TI/TE, VT (ml), 1/VT,
effective minute ventilation (
Eeff = RR · VT; ml/min), mean inspiratory flow
(MIF = VT/TI; l/s), maximum inspiratory flow
(l/s), and mean expiratory flow (l/s). Demographic parameters included
in the analysis were birth weight (BW, kg), test weight (wt, kg), 1/wt,
gestational age (GA, wk), age (wk), corrected gestational age (CGA,
wk), and gender (male = 0, female = 1).
Multivariate linear models (SigmaStat, Jandel Scientific, San Rafael,
CA) of the form RL/EL = a × wt + b/wt + c × age + . . . . . +
g × RR + ... + constant (where a-i
represent the constants defined in Table 3) were used to determine the
independent breathing pattern and demographic predictors of
RLZ, ELZ,
RLMLR, and ELMLR. P < 0.05 was used for covariate retention in the final model.
| |
RESULTS |
Data from a representative example infant are illustrated in Fig.
1A. The breath-to-breath changes in
pattern that are shown resulted in corresponding changes in the
pressure-to-volume
[Ptpt-to-VT] relationship,
reflecting the alterations in the underlying mechanical properties
(Fig. 1B).
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 1.
A: transpulmonary pressure (Ptp), volume, and flow shown for 9 breaths of varying pattern in an example infant (male, 1.15 kg, 12 days
old). B: volume-Ptp curves corresponding to breaths 2, 5, and 8 spanning observed variability in respiratory rate
(RR; 103-136 min1) and tidal volume
(VT; 3.2-6.9 ml). Here, a substantial change in
pressure-to-volume slope (elastance, E) and hysteresis area
(resistance, R) indicates how changes in pattern lead to changes in
between-breath estimates of lung R and E (RL and
EL).
|
|
A total of 450 breaths from all 33 infants were considered in the
analysis. The number of breaths analyzed in each infant differed
(5-18) mainly according to the degree of observed breathing pattern variability in each infant. Figure
2 illustrates the inter- and intrasubject
variability in RR and weight-corrected VT (ml/kg), respectively. The range of measured VT and RR values shown
for each infant indicated significant spontaneous breathing pattern variability in 28 of 33 infants. No trends were seen for
RR and VT as a function of infant size. The average
breathing pattern and lung mechanics data from all infants are
summarized in Table 2.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 2.
Preterm infants' breathing patterns show large variability in RR
(A) and VT (B) both within
as well as among subjects. Each weight value or line on these graphs
represents an individual subject. There were no distinct tendencies for
RR and VT variability with infant size.
|
|
Lung mechanical properties estimated by the impedance
(RLZ, ELZ) and MLR
(RLMLR and ELMLR)
methods as a function of infant weight are graphically illustrated in
Fig. 3. Both RL (Fig.
3A) and EL (Fig. 3B) showed a clear
tendency to decrease (near hyperbolically) with increasing weight,
independent of the method of calculation. Significant within-subject
variability in both RL and EL, presumably due
to changes in breathing pattern, is depicted by the range of values at
each of the weight values. These variabilities were similar for both
the Z and MLR methods of calculation.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 3.
Both RL (A) and EL (B)
estimated by either impedance (Z) or multiple linear regression (MLR)
method (RLZ and ELZ vs.
RLMLR and ELMLR,
respectively) decreased with increasing weight or intersubject
variability. Both also showed significant within-subject variability.
Near-hyperbolic drop in RL and EL as a function
of weight is depicted by regression lines through MLR- (solid line) and
Z-derived (dotted line) values.
|
|
The correlation and agreement between the mechanical properties
(RLZ vs. RLMLR and
ELZ vs. ELMLR) obtained
with the two methods of calculation are presented and discussed in the
APPENDIX.
Stepwise multivariate linear analyses indicated that RL and
EL may be predicted from combinations of breathing pattern
and demographic variables. Specifically, RL
(cmH2O · s · l1)
is a function of weight, 1/wt, age, RR, TI/TE,
VT, Eeff, and a
constant k. Alternatively, EL (cmH2O/l)
is predicted from weight, 1/wt, age, gender, RR,
TI/TE, VT and a constant k.
The dependence of RL and EL on both weight and
1/wt indicated a power law dependence of these properties on weight.
Thus, by using nonlinear regression analysis, the term a × wt + b/wt + k in both multivariate models was
simplified to a power law relationship of the form a × wtb. With this change, the final multivariate
models for RL and EL obtained with the two
methods of computation were as follows (Table 3)
|
(3a)
|
|
(3b)
|
and
|
(4a)
|
|
(4b)
|
Through use of the above equations, simulations depicting the
separate RR and VT dependence of RL and
EL are illustrated in Fig. 4.
Alternatively, because RR and VT are likely to change simultaneously, simulations showing the model-predicted effects of
shallow-rapid, slow-deep, and intermediate breathing on RL and EL while minute ventilation was maintained are shown in
Fig. 5.
View this table:
[in this window]
[in a new window]
|
Table 3.
Multivariate models of RL and EL in terms of
breathing pattern and demographic parameters: R/E = a · wtb + c · gender + d · age + e · RR + f · VT + g · VEeff + i · TI/TE
|
|
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 4.
Simulated RL and EL based on multivariate
models (Eqs. 3a and 4a). A: VT
dependence of RL. B: RR dependence of
RL. C: VT dependence of EL.
D: RR dependence of EL. All simulations were done
assuming age = 1 wk, male, TI/TE = 1. Weight
was varied between 0.6 and 1.8 kg, and effective minute ventilation
(Eeff) was computed as
RR · VT.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 5.
Simulated RL (A) and EL (B)
based on multivariate models (Eqs. 3a and 4a) with
fixed Eeff
(RR · VT) and assuming varying breathing
strategies: rapid shallow breathing (solid lines), slow deep breathing
(dotted lines), and an intermediate pattern (dashed lines). All
simulations were done assuming age = 1 wk, male,
TI/TE = 1. Weight was varied between 0.6 and
1.8 kg.
|
|
| |
DISCUSSION |
Important pharmacological advances (e.g., exogenous surfactant) have
improved survival and decreased mechanical ventilatory dependency of
VLBW infants (7, 15, 25, 29). Indeed, a large percentage of these
infants are allowed to ventilate spontaneously while their lung volumes
are maintained by NCPAP. A well-defined quantitative method to guide
the weaning of infants from NCPAP remains elusive.
Use of serial lung mechanics measurements to guide weaning is a
possible method. However, the wide variance in the reported RL and EL values in spontaneously breathing
preterm infants (1, 2, 6, 7, 10, 13, 15, 17-21, 27, 34) casts
doubt on its applicability. We postulated that a substantial component of this variability in RL and EL derives from
differences in patient demographics and the significant breathing
pattern variability characteristic of spontaneously breathing VLBW
infants. Hence, quantifying these effects in multivariate models
predicting RL and EL may facilitate their
interpretation and enhance their clinical value. In this study, we
confirmed that both RL and EL are altered significantly by breathing and demographic parameters, and we quantified these effects in the form of multivariate models.
These models were essentially identical whether RL and
EL were derived by the MLR or impedance method. Thus, from
this point forward, the Z and MLR subscripts will be dropped for simplicity.
Multivariate Models of RL and EL
The multivariate statistical models of RL and
EL (Eqs. 3 and 4) indicate that both
properties varied as a function of breathing pattern and demographic
factors. These factors, however, did not account for all RL
and EL variability in preterm infants as other factors
affecting their estimation could not be controlled for. These include
errors due to cardiogenic oscillations, extraneous noise (given that no
averaging is possible), changes in absolute lung volumes and gas shunt
effects within the face mask, as well as breath-to-breath changes in
glottal geometry. Also, between-subject differences in 1) lung
disease (albeit mild) and 2) sleep-wake state may have also
added to the variability in RL and EL. The sleep state is a major contributor to breathing pattern variability in
both term and preterm infants (16). Hence, changes in the sleep state
during measurements can consequently alter the underlying mechanical properties.
The gas volume within the face mask represents a gas flow shunt pathway
that may cause some variability in the data, particularly among
different subjects. This gas volume will vary on the basis of mask size
and facial anatomy, which may have varied considerably among the
infants we studied. If we assume similar gas volumes within the
mask, the shunt compartment is more competitive to inspiratory flow in
smaller compared with larger infants given their high vs. low
downstream impedance. In this study, we minimized this effect by always
using the smallest neonatal face masks with which leaks could be avoided.
Changes in absolute lung volumes [e.g., end-expiratory lung
volumes (EELV)] may have occurred during measurements. Also,
measurements from different infants may have been done at comparatively
different lung volumes. Variations in EELV can alter RL and
EL considerably, particularly if it is sufficiently low
such that airway closure may have occurred (30, 32). Lung volumes in
each of the infants we studied were maintained by NCPAP (4-6
cmH2O) up to the point when we did our measurements. Also,
the measurements were typically brief (30-60 s) and in no case did
we observe substantial changes in pulse oximetry and
electrocardiography that might indicate large changes in oxygenation
during data acquisition. Also, inasmuch as Pes at end expiration
reflects changes in lung volume, our measurements did not include gross
changes in Pes at end expiration (see example in Fig. 1), and hence we
do not expect that our measurements included large within-subject
changes in EELV.
Neonates largely rely on their glottal aperture to maintain their
functional residual capacity (8, 12). Thus it is possible that infants
altered their glottal geometry during measurements and consequently
changed the lung mechanical properties estimated at the airway opening.
Changes in glottal aperture are greatest between inspiration (wide) and
expiration (narrow) and are a main reason for the higher effective
resistance during expiration. Therefore, to illustrate these effects,
we compared the changes in resistance estimated from expiration
(Rexp) to that estimated over the full breath (R).
Breath-to-breath changes in Rexp can be quite large, and
the corresponding changes in R computed over the entire breath are also
evident but are relatively smaller (Fig.
6A). These smaller changes probably
reflect the fact that R is a weighted average of inspiratory and
expiratory resistance, as well as reflecting the effects of other
breathing pattern changes such as
TE/TI, RR, and VT (Fig. 6B)
that contribute to variability of the estimated mechanical properties.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 6.
A: breath-to-breath changes in RL calculated from
the full breath vs. expiration only in an example infant. Note
1) the larger expiratory resistance (Rexp) for any
given breath at least partly reflecting the narrowed glottal aperture
during expiration compared with inspiration and 2) the larger
interbreath variability in Rexp compared with a weighted
average of both inspiratory and expiratory resistance (R). R and
Rexp were computed by the MLR method. B:
corresponding changes in VT, RR, and
TE/TI (relative to breath 1) that are
perhaps related to changes due to glottal aperture and that also
contribute to variability of R and E.
|
|
Demographic variables.
In preterm infants, RL and EL decreased with
lung growth and maturity as indicated by their negative dependence on
weight and age, respectively. These findings conform to previous
studies of postnatal lung mechanics (3, 10, 15, 24). In infants and
young children, Gerhardt et al. (18), Lanteri and Sly (26), and Galal
et al. (17) independently demonstrated a negative size dependence of
RL and EL as a function of lung volume, height, and weight, respectively.
The separate dependence of RL and EL on weight
vs. age may also indicate that the effects of development during late
gestation on lung mechanics are not simply a function of changes in
lung size. Other maturational changes during the third trimester,
reflected in the age parameter, such as increases in surfactant pool
sizes (23) and alterations in alveolar architecture (11), may have also contributed.
EL also showed a gender dependence, that is, girls had a
~10% lower EL compared with boys (Eq. 4). Over
all infants, RL estimates were ~15% less in girls vs.
boys, but this difference did not prove significant in the multivariate
model. These results are consistent with the studies by Stocks et al.
(35) in preterm neonates and by Hanrahan et al. (22) in full-term
babies that considered gender differences in lung mechanics.
Breathing pattern variables.
RL and EL in preterm infants exhibited
significant dependence on breathing pattern variables. RL
decreased with increasing RR (min1) and
VT (ml/kg), whereas it increased with increasing
Eeff (ml · kg1 · min1)
and TI/TE (Eq. 3). EL
decreased as VT increased, but it increased for higher RR
and TI/TE ratio (Eq. 4). It is
important to note that
Eeff is equal to RR × VT and that TI/TE is
mathematically equivalent to the term (RR × VT)/(MIF RR × VT), where MIF (ml/min) is the mean
inspiratory flow rate. The breathing pattern dependence of both
RL and EL can thus be narrowed to the same
three variables (RR, VT, and MIF), but these effects are
complex and differ for RL compared with EL.
As in previous studies in infants and young children (14, 17, 31), both
RL and EL exhibited a negative VT
dependence over the entire range of infant weights (Fig. 4). The
degree of amplitude dependence, given a fixed RR and
TI/TE, seems to be more important in the larger
babies. Such negative VT (or amplitude) dependence of
EL is similar to what Fletcher et al. (14) and Galal et al.
(17) reported in infants and young children. This has also been
reported in healthy and diseased lungs and is hypothesized to reflect
either static hysteresis of lung tissues (28, 32) or nonlinear
viscoelasticity (36).
There was generally (weight < 1,200 g) a negative RR, or frequency
(f), dependence of RL concomitant with a positive f
dependence of EL (Figs. 4B and 4D). Such f
dependence of lung mechanical properties is consistent with
stress-adaptive, or viscoelastic, properties of lung tissues (2, 17,
31, 33). A similar decrease of RL with increasing RR has
been described in older infants and children (17, 31). Also, a small
but significant increase in EL as a function of RR was
reported in mechanically ventilated infants and young children (17). In
the largest babies (>1.2 kg), the simulations in Fig. 4B show
a reversal from negative to positive f dependence of RL.
Given the fixed TI/TE = 1 for these simulated
data, this reversal may indicate the effects of the necessary increase
in MIF. Higher flows and larger airways can lead to higher airway
resistance due to turbulence (17). This effect is manifested in Eq.
3 by the parameters TI/TE and Eeff.
Control of breathing is geared toward maintaining gas exchange and
hence alveolar ventilation. We simulated (Fig. 5) how RL and EL are altered given a fixed
Eeff (500 ml · min1 · kg1)
that is achieved either by 1) rapid shallow breathing
(VT = 5 ml/kg, RR = 100 min1), 2) slow deep breathing
(VT = 10 ml/kg, RR = 50 min1), or 3) an intermediate
pattern (VT = 7 ml/kg, RR = 71 min1). These indicated that
EL is systematically lower as VT increases and
RR decreases, indicating that deeper and slower breathing is
mechanically advantageous (Fig. 5B). This advantage is
relatively more important in the larger infants. RL changes
were not systematic (Fig. 5A). A rapid shallow breathing
strategy resulted in lower RL (~15%) only in the
smallest infants and was reversed for the larger neonates.
Clinical implications.
Theoretically, RL and EL measurements can
provide valuable insight to clinicians regarding the patient's
underlying lung function and perhaps about the changes in mechanics
after treatments. The main contribution of this study is
that we illustrate and quantify the effects of variability in breathing
pattern and demographics on RL and EL estimates
in spontaneously breathing preterm infants. These effects are a major
obstacle hindering the routine use of such measurements clinically in
infants whether intubated or not (1, 2, 6, 7, 10, 13, 15, 17-21).
Indeed, given the magnitude of the variability in RL and
EL, our results indicate that these properties must be
adjusted (corrected) for changes in breathing pattern for proper
interpretation to be possible.
In the presence of breathing pattern variability and changes in other
aforementioned factors, it is difficult to know whether decreases in
RL and EL really signal readiness for weaning
off the respiratory support. However, when such factors are unchanged, one may argue that a reduction in RL and EL,
combined with other available clinical data (e.g., blood gases,
oximetry, etc.), indicates an improvement in lung mechanics. In such
instances, the likelihood of the weaning being successful is probably greater.
Controlling all factors contributing to the variability of
RL and EL estimates in clinical settings is
difficult (or impossible in spontaneously breathing subjects). Thus
models quantifying the effects of breathing pattern and lung volume
changes might provide a tool for the interpretation of lung mechanics
measurements. It is important to note that the models derived in this
study for RL and EL are by no means exhaustive
and are based on data from 33 mildly diseased neonates. These results
should be reproduced and validated in larger patient groups. Moreover,
other factors (not addressed in this study) such as changes in lung
volume, sleep state, and the effects of the severity and type of lung disease need to be incorporated into future models.
In conclusion, RL and EL in spontaneously
breathing VLBW infants were similar whether computed by the MLR or Z
methods but varied significantly with breathing pattern and patient
characteristics. Specifically, in infants 27-34 wk postconception
(0.6-1.8 kg), RL was found to decrease as a function
of weight, age, RR, and VT and to increase as
Eeff
(RR · VT) and
TI/TE increased. EL decreased with
increasing weight, age, VT, and female gender and increased
as RR and TI/TE increased. On the basis of
these findings, we conclude that these dependencies must be accounted
for (when assessing pulmonary function in preterm infants) if
RL and EL are to be of clinical value. The
derived multivariate equations of RL and EL
represent an initial attempt for deriving models to assist
neonatologists in the interpretation of serial RL and EL measurements in VLBW infants with mild lung disease
during quiet sleep. These multivariate models may differ significantly 1) in the presence of moderate and/or severe lung disease, as opposed to the mild lung disease in this group, due to associated changes in lung mechanics and breathing pattern, and 2) in
awake infants or during different sleep states.
| |
APPENDIX |
Comparison of Lung Resistance and Elastance Estimated by MLR vs. Z
Method
The RL and EL obtained by using either
computational method were comparable to those previously reported in
preterm infants of similar size and GA (5, 6, 23). Also, the inter- and intrasubject breathing pattern variability resulted in similar variability in RLZ and
RLMLR (Fig.
A1A) as well as for
ELZ and ELMLR (Fig.
A1B).
View larger version (45K):
[in this window]
[in a new window]
|
Fig. A1.
RL (A) and EL (B)
estimated by MLR method agreed closely with those obtained from Z
method as illustrated by linear regression comparisons.
RLMLR was generally less (~15%) than
RLZ (A). EL estimates were
essentially identical with both methods (B). Moreover, this
agreement did not depend on breathing pattern. C and D:
bias and limits of agreement (±2 SD) for RL and
EL estimates, respectively, by the two methods.
|
|
RLMLR (RLMLR = 0.85 × RLZ 0.43; r2 = 0.95) and ELMLR
(ELMLR = 0.97 × ELZ + 8.4; r2 = 0.99) were
highly correlated to RLZ and
ELZ, respectively (Fig. A1). These high
correlations probably indicate that RL and EL
estimates from both methods change in similar fashion with both
breathing pattern and demographic variables. The 0.85 slope relating
the two independent estimates of RL, however, indicates
that RLMLR generally underestimates
RLZ by ~15% (Fig. A1A). In contrast,
the 0.97 slope relating ELMLR and
ELZ indicates that the lung elastance estimate
is not significantly influenced by calculation method (Fig.
A1B).
Consistent with the slopes relating the MLR- and Z-derived
mechanical properties, analyzing the difference
RLMLR RLZ
(
RL) relative to the mean RL reveals a 20%
bias compared with minimal bias (3%) for EL (Fig. A1,
C and D). Moreover, after accounting for the bias in
each, the wider limits of agreement for RL (Fig. A1C) compared with EL (Fig. A1D)
indicate a smaller variability due to method of calculation in
EL vs. RL estimates.
ELZ and ELMLR were
essentially identical (Fig. A1B) with little bias (3%) and a
very high correlation (r2 = 0.99) between them.
These estimates were in close agreement over the entire range of
measured ELZ and ELMLR
values as illustrated by the Bland-Altman (9) plot (Fig. A1D).
Alternatively, RLMLR generally underestimated
RLZ by ~15% (Fig. A1A), but the two
independent estimates were again highly correlated
(r2 = 0.95).
Even after the larger bias was accounted for, differences between
RLZ and RLMLR were
generally larger (Fig. A1C) than those found between the
corresponding EL estimates. These larger differences and
underestimation of RLZ by
RLMLR may partly reflect the fact that
RLZ is the effective resistance at the
breathing frequency (fb) only. In contrast,
the RLMLR estimate is derived from the time
data, which probably carry information from a wider range of
frequencies (e.g., harmonics). RLZ at the
harmonics of the fb is typically lesser in magnitude
(4, 24, 31), and RLMLR is perhaps
closer to a weighted average of RLZ at the
fb and all harmonics contributing significantly to the
pressure and flow time domain data.
Another possible reason for the differences between
RLZ and RLMLR may be
the edge effects on Fourier analysis. This is particularly possible
because averaging could not be done when considering breath-to-breath
variability. To minimize these effects in our analysis, pressure and
flow data were always centered (so that both were of zero mean) before
zero padding and application of the FFTs. Although we cannot discount
the possibility of edge effects on RLZ and
ELZ, we speculate that these effects should influence both the real and imaginary components of impedance. Hence,
the small bias and excellent agreement found for
ELZ and ELMLR suggest
that edge effects are of secondary importance. This is also
corroborated by the similarity of the effects of breathing pattern and
demographic parameters on MLR- and impedance-derived properties (Fig.
A1, A and B, Eqs. 3 and 4).
| |
ACKNOWLEDGEMENTS |
This study was supported by a biomedical engineering grant from
the Whitaker Foundation and an equipment grant from Novametrix.
| |
FOOTNOTES |
The costs of publication of this
article were defrayed in part by the
payment of page charges. The article
must therefore be hereby marked
"advertisement"
in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Address for reprint requests and other correspondence: R. H. Habib,
Mercy Children's Hospital, SVMMC-MCO, 2213 Cherry St., ACC 309, Toledo, OH, 43608 (E-mail: robert_habib{at}mhsnr.org).
Received 27 August 1999; accepted in final form 8 November 1999.
| |
REFERENCES |
1.
American Thoracic Society-European Respiratory Society.
Respiratory mechanics in infants: physiological evaluation in health, and disease.
Am. Rev. Respir. Dis.
147:
474-496,
1993[Web of Science][Medline].
2.
Bachofen, H.,
and
G. Duc.
Lung tissue resistance in healthy children.
Pediatr. Res.
2:
119-124,
1968.
3.
Barnas, G. M.,
P. Harinath,
M. D. Green,
B. Suki,
D. W. Kaczka,
and
K. R. Lutchen.
Influence of waveform and analysis technique on lung and chest wall properties.
Respir. Physiol.
96:
331-344,
1994[Web of Science][Medline].
4.
Barnas, G. M.,
D. Stamenovic,
and
K. R. Lutchen.
Lung and chest wall impedances in the dog in normal range of breathing: effects of pulmonary edema.
J. Appl. Physiol.
73:
1040-1046,
1992[Abstract/Free Full Text].
5.
Baydur, A.,
P. K. Behrakis,
W. A. Zin,
M. Jaeger,
and
J. Milic-Emili.
A simple method for assessing the validity of the esophageal balloon technique.
Am. Rev. Respir. Dis.
126:
788-791,
1982[Web of Science][Medline].
6.
Bhutani, V. K.,
and
S. Abassi.
Relative likelihood of bronchopulmonary dysplasia based on pulmonary mechanics measured in preterm neonates during the first week of life.
J. Pediatr.
120:
605-613,
1992[Web of Science][Medline].
7.
Bhutani, V. K.,
S. Abassi,
A. L. Walker,
and
J. S. Gerdes.
Pulmonary mechanics and energetics in preterm infants who had respiratory distress syndrome treated with synthetic surfactant.
J. Pediatr.
120:
S18-S24,
1992[Web of Science][Medline].
8.
Bhutani, V. K.,
E. M. Sivieri,
and
S. Abbasi.
Evaluation of pulmonary function in the neonate.
In: Fetal and Neonatal Physiology (2nd ed.), edited by R. A. Polin,
and W. W. Fox. Philadelphia, PA: Saunders, 1998, p. 1143-1164.
9.
Bland, J. M., and D. G. Altman.
Statistical methods for assessing agreement between two methods of
clinical measurement. Lancet: 307-310, 1986.
10.
Davis, G. M.,
and
L. C. Lands.
Measurement of infant pulmonary mechanics: comparative analysis of techniques.
Pediatr. Pulmonol.
23:
105-113,
1997[Web of Science][Medline].
11.
Dunhill, M. S.
Postnatal growth of the lung.
Thorax
17:
329-333,
1962[Free Full Text].
12.
Ehrenkranz, R. A.,
and
M. R. Mercurio.
Bronchopulmonary dysplasia.
In: Effective Care of the Newborn Infant, edited by J. C. Sinclair,
and M. B. Brackens. New York: Oxford, 1992, p. 399-424.
13.
England, S. J.
Current techniques for assessing pulmonary function in the newborn and infant: advantages and limitations.
Pediatr. Pulmonol.
4:
48-53,
1988[Web of Science][Medline].
14.
Fletcher, M. E.,
M. Ewert,
C. Stack,
D. J. Hatch,
and
J. Stocks.
Influence of tidal volume on respiratory compliance in anesthetized infants and young children.
J. Appl. Physiol.
68:
1127-1133,
1990[Abstract/Free Full Text].
15.
Freezer, N. J.,
and
P. D. Sly.
Predictive value of measurements of respiratory mechanics in preterm infants with HMD.
Pediatr. Pulmonol.
16:
116-123,
1993[Web of Science][Medline].
16.
Frey, U.,
M. Silverman,
A. L. Barabási,
and
B. Suki.
Irregularities and power law distributions in the breathing pattern in preterm and term infants.
J. Appl. Physiol.
85:
789-97,
1998[Abstract/Free Full Text].
17.
Galal, M. W.,
R. H. Habib,
D. D. Jaeger,
and
G. Lister.
Effects of rate and amplitude of breathing on respiratory system elastance and resistance during growth of healthy children.
Pediatr. Pulmonol.
25:
270-277,
1998[Web of Science][Medline].
18.
Gerhardt, T.,
L. Reifenberg,
S. Duara,
and
E. Bancalari.
Comparison of dynamic and static measurements of respiratory mechanics in infants.
J. Pediatr.
114:
120-125,
1989[Web of Science][Medline].
19.
Gerhardt, T.,
L. Reifenberg,
R. Goldberg,
and
E. Bancalari.
Pulmonary function in preterm infants whose lungs were ventilated conventionally or by high frequency oscillation.
J. Pediatr.
115:
121-126,
1989[Web of Science][Medline].
20.
Greenspan, J. S.,
S. Abassi,
and
V. K. Bhutani.
Sequential changes in pulmonary mechanics in the very low birth weight (<1000 grams) infant.
J. Pediatr.
113:
732-737,
1988[Web of Science][Medline].
21.
Gupta, S. K.,
J. S. Wagener,
and
A. Erenberg.
Pulmonary mechanics in healthy term neonates: variability in measurements obtained with a computerized system.
J. Pediatr.
117:
603-606,
1990[Web of Science][Medline].
22.
Hanrahan, J. P.,
R. W. Brown,
V. J. Carey,
R. G. Castile,
F. E. Speizer,
and
I. B. Tager.
Passive respiratory mechanics in healthy infants. Effects of growth, gender, and smoking.
Am. J. Respir. Crit. Care Med.
154:
670-680,
1996[Abstract].
23.
Jackson, J. C.,
S. Palmer,
W. E. Truog,
T. A. Standaert,
J. H. Murphy,
and
W. A. Hodson.
Surfactant quantity and composition during recovery from hyaline membrane disease.
Pediatr. Res.
20:
1243-1247,
1986[Web of Science][Medline].
24.
Kaczka, D. W.,
G. M. Barnas,
B. Suki,
and
K. R. Lutchen.
Assessment of time domain analyses for estimation of low-frequency respiratory mechanical properties and impedance spectra.
Ann. Biomed. Eng.
23:
135-151,
1995[Web of Science][Medline].
25.
Kari, M. A.,
M. Hallman,
M. Eronen,
K. Teramo,
M. Virtanen,
M. Koivisto,
and
R. S. Ikonen.
Prenatal dexamethasone treatment in conjunction with rescue therapy of human surfactant: a randomized placebo-controlled multicenter study.
Pediatrics
93:
730-736,
1994[Abstract/Free Full Text].
26.
Lanteri, C. J.,
and
P. D. Sly.
Changes in respiratory mechanics with age.
J. Appl. Physiol.
74:
369-378,
1993[Abstract/Free Full Text].
27.
LeSouef, P. N.,
S. J. England,
and
A. C. Bryan.
Total resistance of the respiratory system in preterm infants with and without an endotracheal tube.
J. Pediatr.
104:
108-111,
1984[Web of Science][Medline].
28.
Lutchen, K. R.,
and
A. C. Jackson.
Effects of tidal volume and methacholine on low frequency total respiratory impedance in dogs.
J. Appl. Physiol.
68:
2128-2138,
1990[Abstract/Free Full Text].
29.
Maher, J. E.,
S. P. Cliver,
R. L. Goldenburg,
R. O. Davis,
and
R. L. Copper.
The effect of corticosteroid therapy in the very premature infant. March of Dimes Multicenter Study Group.
Am. J. Obstet. Gynecol.
170:
869-873,
1994[Web of Science][Medline].
30.
Mortola, J. P.,
J. T. Fisher,
B. Smith,
G. Fox,
and
S. Weeks.
Dynamics of breathing in infants.
J. Appl. Physiol.
52:
1209-1215,
1982[Abstract/Free Full Text].
31.
Nicolai, T.,
C. J. Lanteri,
and
P. D. Sly.
Frequency dependence of elastance and resistance in ventilated children with and without the chest opened.
Eur. Respir. J.
6:
1340-1346,
1993[Abstract].
32.
Petak, F.,
M. J. Hayden,
Z. Hantos,
and
P. D. Sly.
Volume dependence of respiratory impedance in infants.
Am. J. Respir. Crit. Care Med.
156:
1172-1177,
1997[Abstract/Free Full Text].
33.
Polgar, G.,
and
S. T. String.
The viscous resistance of the lung tissues in newborn infants.
J. Pediatr.
3:
787-792,
1969.
34.
Rousselot, J. M.,
R. Peslin,
and
C. Duvivier.
Evaluation of the multiple linear regression method to monitor respiratory mechanics in ventilated neonates and young children.
Pediatr. Pulmonol.
13:
161-168,
1992[Web of Science][Medline].
35.
Stocks, J.,
M. Henschen,
A. F. Hoo,
K. Costeloe,
and
C. Dezateux.
Influence of ethnicity and gender on airway function in preterm infants.
Am. J. Respir. Crit. Care Med.
156:
1855-1862,
1997[Abstract/Free Full Text].
36.
Suki, B.,
and
J. H. Bates.
A nonlinear viscoelastic model of lung tissue mechanics.
J. Appl. Physiol.
71:
826-833,
1991[Abstract/Free Full Text].
J APPL PHYSIOL 88(3):997-1005
8570-7587/00 $5.00
Copyright © 2000 the American Physiological Society
TOP
ABSTRACT
INTRODUCTION
![E<SC>l</SC><SUB>Z</SUB> = −2&pgr; ⋅ f<SUB><IT>k</IT> = 1</SUB> ⋅ Imaginary[Z(f<SUB><IT>k</IT> = 1</SUB>)]](/content/vol88/issue3/fulltext/997/img003.gif)
