See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12889054

Pillow JJ, Neil H, Wilkinson MH, et al: Effect of I/E ratio on mean alveolar
pressure during high-frequency oscillatory ventilation. J Appl Physiol 87: 407-
414

Article in Journal of Applied Physiology: Respiratory, Environmental and Exercise Physiology · August 1999
DOI: 10.1152/jappl.1999.87.1.407 · Source: PubMed

CITATIONS READS

67 358

4 authors, including:

Jennifer Jane Pillow Malcolm Wilkinson

University of Western Australia Monash University (Australia)
223 PUBLICATIONS 7,683 CITATIONS 51 PUBLICATIONS 1,297 CITATIONS

SEE PROFILE SEE PROFILE

C. Andrew Ramsden
Monash University (Australia)
34 PUBLICATIONS 1,378 CITATIONS

SEE PROFILE

All content following this page was uploaded by Malcolm Wilkinson on 20 July 2018.

The user has requested enhancement of the downloaded file.

Effect of I/E ratio on mean alveolar pressure during
high-frequency oscillatory ventilation
J. J. PILLOW,1 H. NEIL,2 M. H. WILKINSON,3 AND C. A. RAMSDEN1
1Newborn Services, Monash Medical Centre; 2Department of Paediatrics, Monash University;
and 3Ritchie Centre for Baby Health Research, Institute of Reproduction and Development,
Monash University, Clayton, Victoria 3168, Australia

Pillow, J. J., H. Neil, M. H. Wilkinson, and C. A. during CMV at rapid rates, inspiratory-to-expiratory
Ramsden. Effect of I/E ratio on mean alveolar pressure time (I/E) ratios of at least 1:2 are commonly employed.
during high-frequency oscillatory ventilation. J. Appl. Physiol. Extrapolation of the principles underlying choice of
87(1): 407–414, 1999.—This study investigated factors contrib- I/E ratio for CMV has lead many authorities to also
uting to differences between mean alveolar pressure (PA) and recommend the use of I/E ratios of at least 1:2 for
mean pressure at the airway opening (Pao) during high- HFOV, with the same intent of avoiding gas trapping
frequency oscillatory ventilation (HFOV). The effect of the and the associated dangers of hyperinflation, air leak,
inspiratory-to-expiratory time (I/E) ratio and amplitude of and cardiovascular compromise. The extent to which
oscillation on the magnitude of PA ⫺ Pao (Pdiff) was exam-
gas trapping occurs during HFOV, elevating mean
ined by using the alveolar capsule technique in normal rabbit
lungs (n ⫽ 4) and an in vitro lung model. The effect of
alveolar pressure (PA) above mean airway opening
ventilator frequency and endotracheal tube (ETT) diameter pressure (Pao), remains controversial. Various investi-
on Pdiff was further examined in the in vitro lung model at an gators have reported that global or regional measures
I/E ratio of 1:2. In both lung models, PA fell below Pao during of PA may be the same (1, 5), higher (1, 2, 5, 7, 13, 19,
HFOV when inspiratory time was shorter than expiratory 21), or lower (13, 25) than Pao during HFOV. Differ-
time. Under these conditions, differences between inspiratory ences in ventilator settings including I/E ratio (13), Pao
and expiratory flows, combined with the nonlinear relation- (5), ventilator amplitude (⌬Pao) (1), and the effects of
ship between resistive pressure drop and flow in the ETT, are regional factors (1) and posture (1, 21) appear to
the principal determinants of Pdiff. In our experiments, the account, at least in part, for these disparate observa-
magnitude of Pdiff at each combination of I/E, frequency, lung tions.
compliance, and ETT resistance could be predicted from the To date, there has been no quantitative description of
difference between the mean squared inspiratory and expira- the effect of oscillatory ventilator settings and mechani-
tory velocities in the ETT. These observations provide an cal characteristics of the lung on the difference (Pdiff)
explanation for the measured differences in mean pressure between PA and Pao. This study was, therefore, under-
between the airway opening and the alveoli during HFOV taken to systematically examine the interaction of key
and will assist in the development of optimal strategies for ventilatory parameters on the magnitude of Pdiff in the
the clinical application of this technique.
lungs of young rabbits and in an in vitro lung model.
high-frequency ventilation; gas trapping; lung volume; mean
airway pressure MATERIALS AND METHODS

Experiments were performed on the lungs of four adult

rabbits and an in vitro model of the intubated newborn
THE POSSIBILITY THAT GAS TRAPPING may occur during respiratory system. Animal experiments were performed in
high-frequency oscillatory ventilation (HFOV) has been accordance with the guidelines of the Australian National
Health and Medical Research Council and with the approval
a source of concern for many years. This concern has
of the Animal Ethics Committees of Monash Medical Centre
been heightened by perceived similarities with conven- and Monash University (approval no. A9452).
tional mechanical ventilation, where it is well recog-
nized that gas trapping can occur at high ventilator Lung Models
rates (17, 18, 20).
During conventional mechanical ventilation (CMV), Rabbit lung model. Four adult New Zealand White rabbits
(body weight, 4.0 ⫾ 1.8 kg) were administered a lethal dose of
the time required for pressure to equilibrate between
pentobarbital sodium (160 mg/kg) and intubated through a
the alveolar space and the patient’s airway, and thus for tracheostomy with an 11-cm-long, 3.0-mm-ID endotracheal
inspiration or expiration to be complete, is determined tube (ETT) (Portex, UK). Intermittent positive pressure
by the time constant, which is the product of compli- ventilation of the lungs (peak inspiratory pressure ⫽ 15
ance (C) and resistance (R) of the respiratory system. cmH2O; end-expiratory pressure ⫽ 5 cmH2O; rate ⫽ 30
Because expiratory resistance is commonly up to four- breaths/min; and inspiratory time ⫽ 0.7 s) was initiated with
fold greater than inspiratory resistance (17), applica- a Humming II ventilator (Senko Medical Instrument Manu-
tion of rapid ventilator rates during CMV may easily facturing, Japan) and maintained until the completion of
surgical procedures when HFOV was commenced with a
compromise expiration. Gas trapping results, with
SensorMedics 3100 high-frequency oscillator. To measure
elevation of alveolar pressure above airway pressure at alveolar pressure (PA) by the alveolar capsule technique (12,
end expiration, in a phenomenon commonly referred to 13), a right thoracotomy was performed and a small plastic
as inadvertent positive end-expiratory pressure (3, 17, capsule glued (Supaglue, Selleys Chemical, Australia) to the
20). To minimize the risk of this problem occurring exposed anterolateral surface of the right middle (n ⫽ 3) or

http://www.jap.org 8750-7587/99 $5.00 Copyright r 1999 the American Physiological Society 407

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
408 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV

right lower (n ⫽ 1) lobe of the lung. The pleural surface The time constant of the MP15 transducer response to a
enclosed within the capsule was punctured in five places with sudden change in pressure by balloon-burst testing was 0.7
a 23-gauge needle inserted to a uniform depth (2 mm). The ms for the MP15 transducer alone and 2.6 ms for the
capsule opening was then sealed by inserting the tip of a MP15-needle combination, indicating a frequency response
Micron MP15 pressure transducer (Micron Instruments), that was adequate up to at least 454 and 134 Hz, respectively,
which was carefully supported to exert minimal pressure on for Pao and PA measurements. The output of the MP15 was
the underlying lung. linear over the range ⫺70 to ⫹140 cmH2O. Transducer
The capsule was confirmed to be free of leaks by demonstrat- signals were amplified and low-pass filtered at 400 Hz
ing that PA remained constant after airway occlusion at end (Cyberamp 320, Axon Instruments), digitized at 1 kHz, and
inspiration. Immediately before HFOV was commenced, the stored on a personal computer by using data-acquisition
C and R of the respiratory system were determined by the software (Spike 2, Cambridge Electronic Design, UK).
airway-occlusion technique (16) by using a Hans Rudolph no. Mean pressures. Measurements of mean pressures were
1 pneumotachograph. Mean (⫾ SD) C was 2.67 ⫾ 0.64 determined from the average of at least 10 complete cycles of
ml/cmH2O, and mean R was 250 ⫾ 0.13 cmH2O · s · l⫺1. oscillation. The potential for errors arising from differences in
In vitro lung model. To simulate the mechanical properties transducer calibration when determining differences be-
of the respiratory system of the intubated human newborn
tween PA and Pao was eliminated in the in vitro lung model
infant suffering from hyaline membrane disease, we em-
by measuring PA and Pao with the same transducer con-
ployed a lung model comprising an 11-cm-long, 3.0-mm-ID
nected either to the airway opening or to the interior of the
ETT, sealed into the neck of a 590-ml glass flask (see Fig. 1).
The model had an adiabatic compliance C of 0.4 ml/cmH2O lung. In the rabbit lung, each transducer was carefully
and an R of 75 cmH2O · s · l⫺1 [measured at a flow (V8) ⫽ calibrated to the same gain. Potential errors arising from zero
0.1 l/s]. drift were minimized by frequently referencing each trans-
ducer to the same static pressure, achieved by briefly reduc-
Measurement of Pressure ing the amplitude of the pressure oscillation delivered by the
high-frequency ventilator to zero.
PA and pressure at the airway opening (Pao). Pao was Amplitude of pressure oscillation at the airway opening
measured with a Micron MP15 pressure transducer (Micron ⌬Pao. Reporting of the amplitude of pressure oscillation
Instruments) via a sideport positioned immediately above the produced by the SensorMedics 3100 at the airway opening is
connection to the ETT and perpendicular to the ventilator confounded by the waveform’s complex shape, which approxi-
tubing. Pressure in the interior of the in vitro lung model (PA) mates to a square wave, upon which are superimposed
was also measured with an MP15 transducer positioned at large-amplitude, damped oscillations at the onset of both
the end of an 11-cm-long, 12-gauge needle inserted into the inspiration and expiration. Where observations were made at
glass flask. Similar measurements of mean pressure were
various amplitudes, we have reported the amplitude dis-
obtained when Pao was measured at other sites close to the
played by the ventilator. The ventilator internal pressure
airway opening and when PA was determined at a variety of
measurement provides a damped estimate of the amplitude
sites within the glass flask (results not shown).
⌬Pao, since the pressure transducer is placed at the end of a
75-cm length of 3.5-mm-bore tubing attached to the inspira-
tory limb of the ventilator circuit, 1 cm before the patient’s
airway connection. Consequently, the ⌬Pao displayed by the
ventilator approximates the amplitude of the square-wave
component of the pressure waveform.
Calculation of tidal volume (VT) and V8. In the in vitro lung
model, VT was calculated by using the equation

VL · ⌬PA
VT ⫽ (1)
␥ · P0

where VL is model lung volume, ⌬PA is amplitude of oscilla-

tory pressure waveform in the model lung, P0 is atmospheric
pressure, and ␥ is adiabatic gas constant (1.4 for air). V8 was
calculated throughout the oscillatory cycle by using numeri-
cal differentiation such that

dPA
V8 ⫽ C · (2)
dt

where C is model lung compliance and PA is the instanta-

neous pressure in the model lung.

Experimental Protocols
In all experiments, HFOV was delivered by a SensorMedics
3100 high-frequency oscillator, operated in accordance with
the manufacturer’s instructions. All observations were made
at a mean airway pressure of 10 cmH2O.
Fig. 1. In vitro lung model. Pao, pressure at airway opening; PA, Pdiff in the rabbit lung model. Experiments were per-
alveolar pressure; ETT, endotracheal tube. formed to test whether significant differences between PA and

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV 409

Table 1. Ventilator settings and in vitro lung model characteristics

Ventilator Settings Lung Model

Protocol Amplitude, cmH2O Frequency, Hz I/E Compliance, ml/cmH2O ETT ID, mm

Effect of amplitude 10–90* 15 1:1, 1:2 0.4 3.0

Effect of frequency 10–90* 7.5, 10, 15 1:2 0.4 2.5
Effect of I/E ratio 70† 15 1:1–1:2.3‡ 0.4 3.0
Effect of ETT diameter 10–90* 15 1:2 0.4 2.5, 3.0, 3.5
Effect of compliance 80 15 1:2 0.1–2.0§ 3.0
ETT ID, internal diameter of endotracheal tube. * Amplitude was adjusted in steps of 10–15 cmH2O. † Amplitude was adjusted at each
setting of inspiratory-to-expiratory time (I/E) ratio to deliver a calculated tidal volume of 7 ml. ‡ Inspiratory time, expressed as a percentage of
total cycle time, was adjusted in steps of 2–3% between 30% (I/E ⫽ 1:2.3) and 50% (I/E ⫽ 1:1). § Compliance was varied by filling a 3.1-liter flask
with measured volumes of water.

Pao could be identified in the rabbit lung model during HFOV different amplitudes (level 1) and between rabbits (level 2).
and to investigate the effect of the I/E ratio on these differ- Data at each I/E ratio were assessed separately.
ences. By using a ventilator frequency of 15 Hz, the magni-
tude of Pdiff was determined at I/E ratios of 1:1 and 1:2 across RESULTS
a range of ⌬Pao from a minimum of 10 cmH2O to a maximum
of 90 cmH2O. Two observations were made at each setting, Pdiff in the Rabbit Lung
and average values of Pdiff were determined for individual
rabbits at each ⌬Pao and I/E ratio. The measurements of Pdiff made during HFOV in
Pdiff in the in vitro lung model. Experiments were per- each of the rabbit lungs are illustrated in Fig. 2A. In the
formed to test whether differences between PA and Pao could rabbit lung, Pdiff was not significantly different from
be identified in the in vitro lung model and to examine the zero (maximum ⫺0.8 ⫾ 0.3 cmH2O) at an I/E ratio of
independent effects of ventilator amplitude, frequency, and 1:1. However, at an I/E ratio of 1:2, PA was substan-
I/E ratio as well as the model lung compliance and ETT tially less than Pao, and the magnitude of Pdiff in-
resistance on those differences. The range of ventilator set- creased with the square of the oscillatory pressure
tings examined and the characteristics of the lung model amplitude at the patient’s airway to a maximum value
employed in each protocol are summarized in Table 1. Ventila-
of ⫺5.0 ⫾ 0.2 cmH2O.
tor settings were changed in random sequence until triplicate
observations of PA and Pao had been made under each
experimental condition. Differences between PA and Pao were Pdiff in the In Vitro Lung Model
pooled to determine average values for Pdiff. As in the rabbit lung Pdiff in the in vitro lung was ⬍1
Statistical Analysis cmH2O at an I/E ratio of 1:1 (Fig. 2B). In contrast to the
rabbit lung, however, Pdiff was positive (PA ⬎ Pao),
Results are shown as means ⫾ SE. Statistical analysis of and a statistically significant difference in Pdiff was
the data derived from the rabbit experiments was compli- demonstrable between models for amplitudes in excess
cated by the unequal number of measurements of Pdiff
obtained across a range of ventilator amplitudes in each
of 30 cmH2O. At an I/E ratio of 1:2, PA was substan-
rabbit. A multilevel modeling approach using a nested hierar- tially less than Pao (maximum difference ⫺6.6 ⫾ 0.2
chical design was employed (statistical software MLn v 1.0a, cmH2O), and the difference increased with increasing
Institute of Education, London, UK). Constant variance was amplitude in a similar manner to the rabbit lung.
assumed. Two levels were employed in the analysis to account Figure 3 illustrates the effect of changing the ampli-
for differences between the observations of Pdiff made at tude, frequency, and I/E ratio of the ventilator and both

Fig. 2. Comparison of difference between mean

PA and Pao (Pdiff) in both in vitro and in rabbit
lung models. Plots of Pdiff against ventilator
amplitude for rabbit lung model (A) and the in
vitro model (B) at inspiratory-to-expiratory time
(I/E) ratios of 1:1 (䊉) and 1:2 (䊊). Each data point
in A represents a single observation of Pdiff, with
different symbols representing different animals.
Each data point in B represents mean ⫾ SE at
each amplitude and I/E ratio. In both lung models
at 1:1, mean PA was not substantially different
from mean Pao, whereas at 1:2, mean PA was
considerably lower than mean Pao.

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
410 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV

Fig. 3. Effect of ventilator and model lung

characteristics on Pdiff. The effect of tidal
volume (A and C), % inspiratory time (B), and
model lung compliance (D) on the magnitude
of Pdiff. Ventilator and model lung character-
istics for each group of observations are sum-
marized in Table 1. At any one tidal volume,
Pdiff increased with increasing frequency (A),
decreasing %inspiratory time (B), and decreas-
ing ETT parameter (C). Pdiff was relatively
unaffected by changes in compliance, except
at very low values of compliance, where lower
magnitudes were observed.

the ETT diameter and compliance of the lung model, on (ⱖ1 cmH2O) were seen only at an I/E ratio of 1:2 and
the magnitude of Pdiff. An increase in ventilator fre- were opposite in sign (PA ⬍ Pao). Interestingly, differ-
quency (Fig. 3A) at an I/E ratio of 1:2 or reduction of the ences between PA and Pao (i.e., Pdiff) could be elicited
inspiratory time as a fraction of total cycle time (Fig. even in a very simple in vitro lung model, where their
3B) caused PA to fall progressively below Pao. Reduc- magnitude was very similar to that seen in whole
tion in the resistance of the lung model, by increasing rabbit lung oscillated at comparable settings, suggest-
ETT diameter, decreased Pdiff at a given VT (Fig. 3C), ing that the ETT itself is critical to the generation of
whereas a change in compliance had negligible effect Pdiff.
(Fig. 3D) until very low compliances were reached
(⬍0.5 ml/cmH2O), after which Pdiff fell rapidly with Critique of Methods
decreasing compliance. Our in vitro and animal lung models were chosen
DISCUSSION because their mechanical properties resemble the hu-
man newborn infant’s respiratory system. Thus the
In common with several previous publications (1, 6, compliance of the in vitro model (0.4 ml/cmH2O) was
19, 21), we found evidence in this study that PA may comparable to that seen in severe hyaline membrane
differ significantly from Pao during HFOV and that I/E disease (13), a condition for which HFOV is commonly
ratio is a crucial determinant of whether such a differ- employed. We also used a range of ETTs (ID 2.5–3.5
ence is present. The novel feature of our study is that it mm) that spanned the range of sizes normally used for
represents the first systematic evaluation of the effects human newborn infants. However, although the total
of individual HFOV settings, and the influence of resistance of the intubated neonatal respiratory system
mechanical properties of the lung, on the magnitude of has a significant contribution from the ETT (11), our in
differences between PA and Pao. We found no evidence vitro model lacked conducting airways beyond the ETT,
of significant gas trapping (PA ⬎ Pao) at an I/E ratio of and its resistance must therefore have been less than
1:1. Rather, differences between PA and Pao of suffi- that in the human newborn. Again, the rabbit was
cient magnitude to be of potential clinical significance chosen as our whole lung model for its similarity in size

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV 411

and mechanical properties to the human newborn tively) outlined above each cause RE to exceed RI, and,
infant. in turn, have the capacity to cause PA to exceed Pao.
Measurements of PA in the rabbit lung were made None, however, explains the observations made in our
with the alveolar capsule technique, which has been study where we saw minimal elevation of PA above Pao
used previously by several investigators to measure PA at a 1:1 ratio in the in vitro lung and no elevation at all
during high-frequency ventilation (1, 11–13). This tech- in the rabbit lung, where such sources of asymmetry
nique has been demonstrated by Gerstmann and col- would have been most likely to be evident. The lack of
leagues (13) to potentially overestimate PA when cap- evidence of PA exceeding Pao may well be explained by
sule and transducer are relatively heavy. We attempted the observation made by Byran and Slutsky (5) that the
to minimize this problem in our experiments by care- level of Pao is a the crucial determinant of whether gas
fully supporting the pressure transducer to achieve a trapping occurred in normal or surfactant-depleted
near-weightless condition. Whereas we cannot exclude rabbits given HFOV at an I/E ratio of 1:1. Our studies
the presence of some residual effect, we note that by employed a Pao of 10 cmH2O, which is well above that
causing overestimation of PA such an effect would have employed in those earlier studies in which gas trapping
tended to lessen the Pdiff that we observed at an I/E was seen (1, 6, 17, 25, 28) but is no higher than the
ratio of 1:2. Employing a single alveolar capsule, we minimum Pao commonly employed in the clinical set-
were unable to examine whether any regional variation ting.
in PA was present in our experiments. However, when By contrast, the striking finding in our study was
regional variation is present, PA is usually lowest in the that, at an I/E ratio of 1:2, PA may fall substantially
upper lobe of the lung during HFOV (1, 13), and our below Pao. Several years ago, Gerstmann et al. (13)
choice of the middle or lower lobe for alveolar capsule made a similar observation that the average tracheal
placement would again have tended to give a minimum pressure may be less than Pao during HFOV in the
estimate of the magnitude of the Pdiff present at an I/E rabbit at an I/E ratio of 1:2, and more recently Hatcher
ratio of 1:2. et al. (14) have found that PA determined by airway
Although measurement of static pressure is rela- occlusion may be less than Pao at an I/E ratio of 1:2.
tively uncomplicated, several investigators have previ- The lower value of PA compared with Pao may again be
ously noted that pressure measurements in a moving accounted for by asymmetry between RE and RI, but
stream of gas may underestimate the driving pressure requires the unusual situation to arise, whereby RI
because of the Bernoulli effect (4, 8). In our measure- exceeds RE, which is the converse of the relationship
ment system, the most likely pressure to be affected by when PA is higher than Pao. Our data provide evidence
a Bernoulli effect was Pao; however, the worst-case that turbulence in the ETT, as first suggested by
calculated dynamic pressure at that site was ⬍0.1 Gerstmann et al. (13), can account for this phenom-
cmH2O. enon.
Difference Between PA and Pao Turbulence and Pdiff
In contrast to our finding of no more than small During HFOV, one may predict that the high inspira-
differences (ⱕ1 cmH2O) between PA and Pao in both the tory and expiratory flows are likely to be associated
rabbit and in vitro lung models at an I/E ratio of 1:1, with turbulence in the ETT. We may, therefore, expect
several previous studies in dogs (21), rabbits (13), and that both the resistance R and the resistive pressure
adult human subjects (19, 23) have each shown a drop (Pres) along the ETT can be predicted from
potential for PA to be significantly elevated above Pao Rohrer’s equation, such that R ⫽ k1 ⫹ k2V8 and Pres ⫽
when HFOV was employed at an I/E ratio of 1:1. A k1V8 ⫹ k2V82. If we assume that the values of k1 and k2
number of mechanisms have been postulated to ac- are similar in inspiration and expiration, we can then
count for this effect, and each depends on the presence predict the effect of changes in inspiratory or expiratory
of asymmetry between inspiratory and expiratory resis- flow V8 on the magnitude of Pres during inspiration and
tance. They include changes in airway caliber between expiration (see APPENDIX A).
inspiration and expiration (15) that may be particu- In the simple circumstance of an I/E ratio of 1:1, the
larly evident at low lung volumes (low Pao) and the average inspiratory flow V8I must equal the average
effects of flow separation at branch points in the airway expiratory flow V8E, and both the resistance and Pres
(6). Bryan and Slutsky (5) have suggested, however, during inspiration and expiration will be equal. In our
that the elevation of PA above Pao, observed in the in vitro lung model at least, with no airways beyond the
study of Saari and colleagues (19) and that of Simon ETT, we would therefore expect PA to equal Pao. In
and co-workers (21), might alternatively be explained contrast, at an I/E ratio of 1:2, the average inspiratory
by the development of ‘‘choke points’’ that cause expira- flow V8I must be twice the average expiratory flow V8E,
tory flow limitation at low Pao. In support of their since it is sustained for only half the time. As Pres
hypothesis, they found the level of Pao to be the crucial depends on the square of V8 under turbulent flow
determinant of whether gas trapping occurred in nor- conditions, the average V8I2 will be four times greater
mal or surfactant-depleted rabbits given HFOV at an than the average V8E2. The effect on Pres of doubling
I/E ratio of 1:1. inspiratory flow with respect to expiratory flow will
The various potential sources of asymmetry between therefore exceed the countering effect on Pres of expira-
expiratory and inspiratory resistance (RE and RI, respec- tory time doubling, relative to inspiratory time. Thus,

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
412 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV

at an I/E ratio of 1:2, the unusual circumstance whereby with differences in the inspiratory and expiratory ETT
RI is greater than RE arises, such that the inspiratory flow profiles observed by Chang and Mortola (9). Such a
Pres must exceed the expiratory Pres, with the inevitable small increase of expiratory resistance relative to inspi-
consequence that PA must fall below Pao in our in vitro ratory resistance could account for the small elevation
lung model. In more general terms, we can predict that, of PA above Pao in our in vitro lung model at an I/E ratio
at any I/E ratio other than 1:1, PA will deviate from Pao, of 1:1. A further point of interest is that the calculated
and the magnitude of that deviation will be directly slope of the relationship between Pdiff and U 2I ⫺ U E2 is
proportional to the difference between the mean squared ⫺0.012, which is in good agreement with the theoreti-
inspiratory (U 2I ) and expiratory velocities (U E2 ) in the cally predicted slope (⫺0.013) calculated from the ex-
ETT (see APPENDIX A). perimental data of Sly et al. (22) (see APPENDIX A).
To test the capacity of this model to explain the Clearly, these conclusions are predicated on the assump-
magnitude of Pdiff in our in vitro lung model, we have tion that parameters measured from steady-flow experi-
plotted each measurement of Pdiff illustrated in Fig. 3 ments are applicable in the high-frequency oscillation
against the corresponding difference between U 2I and setting. Only further experimental work directed at
U E2, which we determined (knowing the in vitro lung measuring k1 and k2 during oscillatory flow can resolve
compliance) by differentiating the pressure change this issue.
within the flask (see Fig. 4). The correlation between Clinical Implications
Pdiff and (U 2I ⫺ U E2) was very close (r 2 ⫽ 0.95), sug-
gesting that the effect of turbulence could largely Although a substantial Pdiff was only seen at rela-
account for the Pdiff seen in our in vitro lung model. tively high ventilator pressure amplitudes in these
Furthermore, the close similarity between the results experiments, it should be noted that the resistance of
obtained in the in vitro lung model and those obtained our in vitro lung model (75 cmH2O · s · l⫺1 ) was relatively
in the rabbit strongly suggests that, even in the whole low, compared with established values of resistance in
lung, events occurring in the ETT may be the principal the intubated newborn infant with respiratory distress
determinant of differences between PA and Pao. (120–380 cmH2O · s · l⫺1 ) (24). Given that our in vitro
In deriving the relationship between Pdiff and U 2I ⫺ studies show that Pdiff increases with increased air-
2
U E, we have assumed that the values of k1 and k2 way resistance, it is likely that Pdiff may reach sub-
during inspiration are equal to those during expiration. stantial levels at lower pressure amplitudes when an
Sly et al. (22) have published k1 and k2 values measured I/E of 1:2 is used in the sick intubated newborn baby.
during steady flow for a number of ETT values ranging Similarly, the resistance of the ETT itself may be
in size from 2.5 to 5.5 mm ID. Their observations higher in the clinical setting, where its interior is
suggest that k1 and k2 during expiration are slightly frequently coated with secretions. The development of
higher than during inspiration, by ⬃10%, in agreement high-frequency ventilators with a capacity to deliver
I/E ratios of 1:3 or more further increases the possibil-
ity of substantial Pdiff occurring even at relatively low
amplitude.
Importantly, the pressure difference that arises across
the ETT is a rapidly achieved, steady-state phenom-
enon. The main clinical consequence of this phenom-
enon, therefore, is that whenever the mechanical char-
acteristics of the respiratory system or ventilator
settings are altered, a new lung volume will result. The
magnitude of such changes in lung volume will increase
as I/E ratio moves away from 1:1. Our studies show
that the adoption of I/E ratios approaching 1:1 might
eliminate the fluctuations in lung volume, which might
otherwise result from the pressure drop across the ETT.
Where I/E ratios other than 1:1 are employed, the
accurate measurement of flow across the ETT would
provide a means of calculating the difference in mean
pressure between the airway opening and the lung.
The optimization of lung volume has been a focus of
clinical HFOV strategies during the last decade. Com-
mon clinical practice involves initial setting of Pao 1–2
cmH2O higher than that employed during CMV, with
increments in Pao being imposed until a substantial
Fig. 4. Relationship between Pdiff and inspiratory and expiratory reduction in the inspired O2 fraction, necessary to
flow velocities in ETT. Values for Pdiff from experimental observa-
tions illustrated in Fig. 3 plotted against difference between the
maintain normoxia, is achieved. To date, no data are
time-averaged squared inspiratory and expiratory velocities (U I2 ⫺ available to indicate the extent to which the Pao
U E2) for each observation. A linear relationship was observed with a required in clinical practice for optimal lung recruit-
slope of 0.012 and a correlation coefficient r ⫽ 0.97. ment may be related to the I/E ratio employed.

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV 413

Although HFOV is frequently employed in the prema- and substituting Eq. A4a and A4b into Eq. A5 gives
ture infant with hyaline membrane disease, which is a
homogeneous lung disease, it is also used to ventilate Pres ⫽ k1IVI8 ⫺ k1EVE8 ⫹ k2IVI82 ⫺ k2EVE82 (A6)
neonates with other forms of respiratory disease. In Assuming that k1I ⫽ k1E and, further, that k2I ⫽ k2E ⫽ k2,
some cases, this might be associated with a degree of and recognizing that VI8 ⫽ VE8, Eqs. A3 and A6 yield
ventilation inhomogeneity. Although it is likely that
this scenario might create small regional differences in Pdiff ⫽ ⫺k2(VI82 ⫺ VE82) (A7)
PA, the pros and cons of a symmetric vs. asymmetric I/E Finally, rewriting Eq. A7 in terms of flow velocity U in the
ratio in the presence of conditions such as pulmonary ETT
interstitial emphysema remain to be tested.
In conclusion, our observations of minimal difference Pdiff ⫽ ⫺k2 A 2(U12 ⫺ UE2) (A8)
between PA and Pao at an I/E of 1:1, and the presence of
a substantial difference at an I/E ratio of 1:2, call into where A is the cross-sectional area of the ETT. Subject to the
assumptions above, Eq. A8 indicates that the relationship
question whether the common practice of employing an
between Pdiff and U I2 ⫺ U E2 is a straight line through the
I/E ratio of 1:2 is the optimal strategy for HFOV. origin with a negative slope of k2 A 2. Calculation of k2 A 2, using
Whereas it may be argued that the Pdiff generated at averaged k2 values derived from the inspiratory and expira-
an I/E ratio of 1:2 has the advantage of offsetting any tory k2 data of Sly et al. (22) for a 10-cm-long ETT, gives
tendency toward gas trapping, either due to expiratory ⫺0.014, ⫺0.013, and ⫺0.011 for 2.5-, 3.0-, and 4.0-mm ETTs,
flow limitation or reduced airway caliber in expiration, respectively. The average calculated slope for the three ETTs
this strategy is not without potential problems. Not is ⫺0.013, in excellent agreement with that deduced from Fig.
only does it create the potential for PA to be substan- 4, which is ⫺0.012.
tially different from the Pao displayed by the machine,
APPENDIX B
it also creates the opportunity for changes in ventilator
amplitude and frequency, or in the size of the infant’s The pressure drop Pin across the inertance (I) of the ETT is
ETT, to have quite unanticipated and undesirable
dV8
effects on PA. Pin ⫽ I (B1)
dt
APPENDIX A where V8 is the instantaneous flow in the ETT.
Assuming I is invariant, the average Pin is, therefore
The instantaneous pressure at the airway opening Pao is
equal to the sum of the resistive and inertive pressure drops I dV8
兰
T
associated with the ETT (Pres and Pin, respectively) and the Pin ⫽ dt (B2)
alveolar pressure PA T 0 dt

Pao ⫽ Pres ⫹ Pin ⫹ PA (A1) where the integration is taken over a full cycle with dura-
tion T.
Integrating term by term over a full cycle and regrouping, we Evaluation of Eq. B2 gives
can write
I
Pin ⫽ [V8(T ) ⫺ V8(0)] (B3)
PA ⫺ Pao ⫽ ⫺(Pres ⫹ Pin) (A2) T

(Note that in Eq. A2 and all that follow, the mean is taken If flow is periodic, so that V8(T ) ⫽ V8(0), then
over a complete cycle). Because Pin ⫽ 0 (see APPENDIX B), and
defining PA ⫺ Pao ⫽ Pdiff, Eq. A2 simplifies to Pin ⫽ 0 (B4)

Pdiff ⫽ ⫺Pres (A3)

We gratefully acknowledge the support of the SensorMedics
Corporation for the loan of a SensorMedics 3100 ventilator, which
When using Rohrer’s equation to represent the effects of
made these investigations possible.
turbulent flow in the ETT, the Pres for inspiration (I) and This study was supported by a Financial Services Foundation for
expiration (E) is, respectively Children Grant and the SW Shields Research Fellowship from the
Royal Australasian College of Physicians. The authors also gratefully
Pres,I ⫽ k1IVI8 ⫹ k2IVI82 (A4a) acknowledge the support of the Monash Medical Centre Newborn
Services Special Purpose Fund for assistance in providing the
and additional support required for completion of these studies.
Address for reprint requests and other correspondence: J. Pillow,
Pres,E ⫽ k1EVE8 ⫺ k2EVE82 (A4b) TVW Telethon Institute for Child Health Research, PO Box 855,
West Perth 6872, Western Australia, Australia (E-mail: janep@ichr.
where V8 is the flow in the ETT and k1I, k2I, k1E, and k2E are uwa.edu.au).
experimentally determined constants. Note that the negative Received 16 October 1998, accepted in final form 4 March 1999.
sign is necessary in Eq. A4b to take account of flow reversal
during expiration. Recognizing that the Pres over a complete REFERENCES
cycle is 1. Allen, J., I. Frantz III, and J. Fredberg. Heterogeneity of
mean alveolar pressure during high-frequency oscillations. J.
Pres ⫽ Pres I ⫹ Pres E (A5) Appl. Physiol. 62: 223–228, 1987.

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

Copyright © 1999 American Physiological Society. All rights reserved.
414 I/E RATIO AND ALVEOLAR PRESSURE DURING HFOV

2. Bancalari, A., T. Gerhardt, E. Bancalari, C. Suguihara, D. neonatal, high-frequency, oscillatory-type ventilators. Crit. Care
Hehre, L. Reifenberg, and R. Goldberg. Gas trapping with Med. 26: 1081–1088, 1998.
high-frequency ventilation: jet versus oscillatory ventilation. J. 15. Lai Fook, S., R. Hyatt, and J. Rodarte. Effect of parenchymal
Pediatr. 110: 617–622, 1987. shear modulus and lung volume on bronchial pressure-diameter
3. Bancalari, E. Inadvertent positive end-expiratory pressure behavior. J. Appl. Physiol. 44: 859–868, 1978.
during mechanical ventilation. J. Pediatr. 108: 567–569, 1986. 16. LeSouëf, P., S. England, and A. Bryan. Passive respiratory
4. Bates, J. H., P. D. Sly, J. Sato, B. L. Davey, and B. Suki. mechanics in newborns and children. Am. Rev. Respir. Dis. 129:
Correcting for the Bernoulli effect in lateral pressure measure- 727–729, 1984.
ments. Pediatr. Pulmonol. 12: 251–256, 1992.
17. Pérez Fontán, J., G. Heldt, R. Targett, M. Willis, and G.
5. Bryan, A., and A. Slutsky. Long volume during high-frequency
Gregory. Dynamics of expiration and gas trapping in rabbits
oscillation. Am. Rev. Respir. Dis. 133: 928–930, 1986.
6. Bush, E., D. Spahn, P. Niederer, and E. Schmid. Flow during mechanical ventilation at rapid rates. Crit. Care Med. 14:
separation, an important mechanism in the formation of mean 39–46, 1986.
pulmonary pressure during high-frequency oscillation. J. Bio- 18. Ramsden, C., and E. Reynolds. Ventilator settings for new-
mech. Eng. 111: 17–23, 1989. born infants. Arch. Dis. Child. 62: 529–538, 1987.
7. Cha, E., E. Chow, K. Chang, and S. Yamashiro. Lung 19. Saari, A., T. Rossing, J. Solway, and J. Drazen. Lung
hyperinflation in isolated dog lungs during high-frequency oscil- inflation during high-frequency ventilation. Am. Rev. Respir. Dis.
lation. J. Appl. Physiol. 65: 1172–1179, 1988. 129: 333–336, 1984.
8. Chang, H. Flow dynamics in the respiratory tract. In: Respira- 20. Simbruner, G. Inadvertent positive end expiratory pressure in
tory Physiology—An Analytical Approach, edited by H. Chang mechanically ventilated newborn infants: detection and effect on
and M. Paiva. New York: Dekker, 1989, p. 57–138. lung mechanics and gas exchange. J. Pediatr. 108: 589–595,
9. Chang, H., and J. Mortola. Fluid dynamic factors in tracheal 1986.
pressure measurement. J. Appl. Physiol. 51: 218–225, 1981. 21. Simon, B., G. Weinmann, and W. Mitzner. Mean airway
10. Dorkin, H., A. Stark, J. Werthammer, D. Strieder, J. Fred- pressure and alveolar pressure during high-frequency ventila-
berg, and I. Frantz III. Respiratory system impedance from 4 tion. J. Appl. Physiol. 57: 1069–1078, 1984.
to 40 Hz in paralyzed intubated infants with respiratory disease. 22. Sly, P., K. Brown, J. Bates, S. Spier, J. Milic-Emili. Noninva-
J. Clin. Invest. 72: 903–910, 1983. sive determination of respiratory mechanics during mechanical
11. Frantz, I. D. III, and R. D. Close. Elevated lung volume and
ventilation of neonates. Pediatr. Pulmonol. 4: 39–47, 1987.
alveolar pressure during jet ventilation of rabbits. Am. Rev.
23. Solway, J., T. Rossing, A. Saari, and J. Drazen. Expiratory
Respir. Dis. 131: 134–138, 1985.
12. Fredberg, J., D. Keefe, G. Glass, R. Castile, and I. Frantz flow limitation and dynamic pulmonary hyperinflation during
III. Alveolar pressure nonhomogeneity during small-amplitude high-frequency ventilation. J. Appl. Physiol. 60: 2071–2078,
high-frequency oscillation. J. Appl. Physiol. 57: 788–800, 1984. 1986.
13. Gerstmann, D., J. Fouke, D. Winter, A. Taylor, and R. 24. Thomson, A., and M. Silverman. Single-breath measurement
deLemos. Proximal, tracheal and alveolar pressures during of lung mechanics in very low birth weight infants. Crit. Care
high-frequency oscillatory ventilation in a normal rabbit model. Med. 13: 4–8, 1985.
Pediatr. Res. 28: 367–373, 1990. 25. Zobel, G., D. Dacar, and S. Rödl. Proximal and tracheal airway
14. Hatcher, D., H. Watanabe, T. Ashbury, S. Vincent, J. Fisher, pressures during different modes of mechanical ventilation: an
and A. Froese. Mechanical performance of clinically available, animal model study. Pediatr. Pulmonol. 18: 239–243, 1994.

Downloaded from www.physiology.org/journal/jappl by ${individualUser.givenNames} ${individualUser.surname} (178.171.079.216) on July 19, 2018.

View publication stats Copyright © 1999 American Physiological Society. All rights reserved.