INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

STRATEGIES TO SUPPORT VENTILATION and oxygenation with normal and abnormal lungs have varied over the last 50 years. The most frequently used and studied mode of mechanical ventilatory support is intermittent positive pressure or volume, which, when applied to the airways via endotracheal tube or mask, inflates the lungs. In the presence of pulmonary parenchymal and airway disease, mechanical ventilation often requires high pressures and/or volumes to achieve adequate arterial oxygenation and carbon dioxide removal (15). High airway pressures and/or volumes may produce mechanical tearing of lung parenchyma and secondary inflammatory responses with fibrosis. Strategies that attempt to minimize these problems have used lower pressure and volume modes, including high-frequency ventilation, jet ventilation, permissive hypercapnia, and high-frequency vibration ventilation (10, 12, 13, 16, 17).

We observed that caretakers patting the buttocks of spontaneously breathing newborns to calm them also produced small volume changes at the airway that were superimposed on spontaneous tidal volume (VT). This clinical observation led us to design a device that accomplished this action automatically and noninvasively. A horizontal platform was constructed that moved periodically in a sinusoidal motion at frequencies of 1-15 Hz, with distances of 1-3 cm. Such motion imparts periodic sinusoidal inertial forces (due to acceleration) on the body in the spinal axis or z-plane (pG_z) and creates noninvasive motion ventilation (NIMV). The purpose of this investigation was to determine whether NIMV, which does not depend on positive pressure inflation of the lung, could effectively ventilate anesthetized and paralyzed animals with normal and diseased lungs, thereby potentially mitigating lung injury sometimes associated with conventional positive pressure mechanical ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Platform design. The motion platform was constructed around a linear displacement direct current motor (model 400, 12 volt; APS Dynamics, Carlsbad, CA). The motor is powered by a dual-mode power amplifier (model 144, APS Dynamics) connected to a sine-wave controller (model 140-072; NIMS, Miami Beach, FL). The controller allows for control of the frequency of the table oscillation, the amplitude of the voltage reaching the motor and subsequent acceleration of the stroke, and the duty cycle of the motor. The motor is secured in the bottom and center of a frame constructed from treated pine lumber. The table platform is directly driven by the underlying motor and articulates across the frame on stainless steel tracks and nylon wheels. The unit has a maximum weight capacity of ~30 kg.

Animal preparation. These studies were approved by the Institutional Animal Care and Use Committee and comply with the Animal Welfare Act. Juvenile piglets weighing between 3 and 12 kg were used for these studies. The animals were anesthetized with intramuscular ketamine (10 mg/kg) followed by intravenous pentobarbital sodium (20 mg/kg) titrated until a surgical plane of anesthesia was obtained. Paralysis was induced with pancuronium bromide at 0.1 mg/kg and supplemented throughout the experiment along with sedation as necessary. A tracheostomy was performed, and a 3.5-mm (ID) endotracheal tube was inserted. The proximal end of the endotracheal tube allowed for measurements of airway flow and pressure by interposing a pneumotachograph (Fleisch model 5751) and pressure transducer (Validyne, model MP45, full scale range ±50 cmH₂O), respectively. The airway pressure transducers were oriented so that the diaphragm of the transducer was 90° relative to the z-axis. The pneumotachograph was calibrated by moving known volumes of air through the unit by use of a 60-ml syringe. The femoral artery was cannulated for the measurements of arterial blood pressure using pressure transducers (Transpac, Abbott Critical Care Systems, North Chicago, IL) and for arterial blood sampling. All fluid-filled transducers were stabilized at the level of the heart and away from the motion platform. Fluid and drugs were administered via a cannula inserted into the femoral vein, and the animals were maintained at 38°C with a thermostatically controlled warming pad. The animals were placed on the motion platform in the supine posture, with the front and hind legs tied securely to the table so that they were closely coupled to the platform. The tracheostomy tube was connected to a conventional pressure-limited ventilator (Bear Cub BP-200, Inter Med) at frequency 14-20 breaths/min, peak inspiratory pressure 18-24 cmH₂O, and positive end-expiratory pressure (PEEP) 5 cmH₂O. Initial settings on the mechanical ventilator were adjusted to achieve arterial partial pressure of CO₂ (Pa_CO2) at ~38 Torr. To minimize any effect from hypoxia, all animals were placed on 100% oxygen during the entire protocol. Arterial blood gases were measured every 30 min or as needed during the experiment. An accelerometer (NIMS, model 140-060) was placed on the snout of the pig to determine the G forces applied to the z-plane (spinal axis) of the animal during NIMV. Another accelerometer oriented to the z-plane was placed on the platform. The values of acceleration on the snout of the animal and the table were identical during NIMV. An electrocardiogram was continuously monitored in a standard three-wire lead configuration. The analog signals from the transducers, accelerometer, and electrocardiogram were continuously recorded on a data-acquisition processor (Respitrace PT, NIMS). During the experiment and for playback afterward, the data were viewed on a personal computer using RespiEvents software (NIMS).

Effects of motion platform frequencies and accelerations on ventilation, gas exchange, and blood pressure. The first series of experiments were designed to determine the effects of various platform frequencies and accelerations. To maintain continued oxygenation and prevent alveolar collapse during NIMV, a bias flow of 100% oxygen with continuous positive airway pressure (CPAP) of 5 cmH₂O was applied. A diagram of the airway connections in this preparation is shown in Fig. 1. The effects of various platform frequencies on airway flow, blood gases, and blood pressure were determined while acceleration was held constant. Increased frequency slightly decreased acceleration, although the voltage applied to the motor remained constant. Next, the effects of various platform accelerations on ventilation, blood gases, and blood pressure were determined. These relationships were analyzed to determine optimal frequency and acceleration for ventilation of paralyzed, anesthetized animals.

View larger version (72K): [in this window] [in a new window]   Fig. 1.   Representation of the airway circuitry used in noninvasive motion ventilation (NIMV). Also shown are hypothesized force vectors acting on the thorax and abdominal compartments to produce pressure gradients and airflow during oscillatory periodic acceleration in the z-plane (pGz). PEEP, positive end-expiratory pressure.

Ventilatory efficiency of NIMV in normal lungs. The settings of NIMV obtained above were then used to test its effects on gas exchange and blood pressure in animals with normal lungs over a 2-h period. Eight juvenile piglets were anesthetized as previously described. The ventilator settings were intentionally adjusted to induce hypoventilation characterized by elevated Pa_CO2 and reduced pH, as well as increased heart rate and systemic central blood pressure. After hypoventilation on conventional mechanical ventilation (CMV) and PEEP of 5 cmH₂O, the animals were switched to NIMV at a frequency of 4.0 Hz and an amplitude of pG_z at about ±0.6 g, with CPAP of 5 cmH₂O. The animals remained at these settings for 120 min, with arterial blood sampling every 30 min, while physiological variables were recorded continuously.

Ventilatory efficiency of NIMV in diseased lungs. Eight piglets were placed on CMV with PEEP of 5 cmH₂O and the ventilator adjusted to obtain normal blood gases (PCO₂  38 Torr). After recording with CMV and PEEP for 30 min, 25% human meconium solution (6 ml/kg) was instilled into the endotracheal tube at the carinal level. The animals were maintained on the same CMV and PEEP settings for 30 min, after which respiratory and cardiovascular variables were again collected. This period defined the aspiration phase of the experiment. The animals were then placed on NIMV with settings identical to those used for the previous animals without aspiration, i.e., frequency = 4.0 Hz, ±0.6 g, and CPAP of 5 cmH₂O. The motion platform was continuously actuated for 120 min, and arterial blood samples and physiological variables were measured every 30 min.

Effects of NIMV on transpulmonary and transdiaphragmatic pressures. Four pigs were instrumented to obtain pleural and abdominal pressures during NIMV. In two of the animals, uncalibrated respiratory inductive plethysmography was utilized to measure phase angles between rib cage and abdominal movements (2). Phase angles were determined from Lissajous plots of the abdomen vs. rib cage compartments (with the abdomen leading). Pigs were anesthetized as previously described but without paralysis, and a balloon catheter was placed into the esophagus with its tip at the midthorax to estimate intrapleural pressures. Placement of the catheter was verified by the occlusion technique (4). The endotracheal tube was occluded during spontaneous ventilation, and changes in both esophageal and airway pressures were observed. Proper placement of the esophageal catheter was signified at the level at which the esophageal pressure swings during inspiratory efforts equaled airway pressure swings. Subsequently, a balloon cannula was placed into the stomach until the pressure during spontaneous inspiration became positive. Intragastric pressure measured at this level was assumed to equal intra-abdominal pressure. Both cannulas were sutured in place and secured. The difference between esophageal and gastric pressures represented transdiaphragmatic pressure (Pdi). Abdominal and pleural pressures were then measured during spontaneous breathing, CMV and PEEP of 5 cmH₂O, and NIMV. The motion platform settings were varied from frequencies of 1-8 Hz, and the amplitude settings were adjusted to produce forces of about ±0.05-0.8 g. Figure 2 shows a representative sample of data from an experiment in which all of these variables were recorded. At the end of the experiments, the animals are killed with pentobarbital sodium (120 mg/kg iv).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Data sample from a representative experiment where pleural and gastric pressures were measured and plotted against time. Pres, pressure.

Data analysis. Data were analyzed for frequency distribution and were found to have followed a Gaussian distribution. The data were then analyzed by ANOVA with Bonferroni's post test for repeated measures. Data comparisons between animals were performed by using the unpaired t-test. Frequency, amplitude, and pressure response curves were subjected to nonlinear regression analysis. Differences in sample means were considered statistically significant if P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Thoracoabdominal motion during NIMV. The ventilatory cycle during NIMV caused paradoxical motion of the rib cage and abdomen at all cycling frequencies (see Fig. 3), whereas, in spontaneous breathing, these compartments moved in phase. The phase angles between rib cage and abdomen were <10° during spontaneous breathing at 12-18 breaths/min. The phase angles measured during NIMV with the abdomen leading ranged from 164 to 176° at 4 Hz at all levels of pG_z. Phase angles approached 90° as cycling frequencies rose to 6 Hz. The studies were done in paralyzed animals, but even in anesthetized, nonparalyzed animals, there was paradoxical motion between the rib cage and abdomen. NIMV produced an inspiratory, inward displacement of rib cage volume along with outward displacement of abdominal volume and vice versa during expiration.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Effects of a pGz cycle during NIMV on changes in the abdomen (Ab) and rib cage (Rc) compartments. The paradoxical movements of the 2 compartments are depicted during inspiration and expiration.

Minute and tidal ventilation responses. The relationships of both motion platform acceleration (G_z) and frequency of movement on minute ventilation (E) were determined. The effect of the motion platform G_z on E when the frequency of oscillation was held constant is shown in Fig. 4A. As the force due to periodic acceleration was increased, E increased linearly until a plateau was reached. This plateau occurred at a force and acceleration dependent on animal size and lung volume. Increasing acceleration further did not increase E. The maximum acceleration observed in these studies to achieve maximum VT and, therefore, maximum E with constant frequencies was about ±1.0 G_z. Conversely, when the acceleration of the motion platform was held constant, increased frequency of the platform oscillation reduced E (Fig. 4B). The relationship shown in Fig. 4B did not strictly apply to constant acceleration but rather to accelerations that were held as constant as possible during altered frequency. This is because frequency inversely altered the length of the stroke displacement with the motion platform employed in this study and, in turn, slightly decreased the maximum attainable acceleration for any constant mass. At extremely high frequencies, i.e., above 12 Hz, E was dramatically reduced and approached zero. Families of these types of curves shown in Fig. 4 can be generated for any given value of either frequency or amplitude of acceleration when one is held constant and the other varies, although only one constant value is shown for each. Manipulating both frequency and force amplitude produced an overall algebraic effect on E (Fig. 3); this was typically done when the platform was used to achieve adequate ventilation. The effects of pG_z on VT, shown in Fig. 5, were similar to the observed effects of pG_z on E. Specifically, increased pG_z produced linearly increased VT but was modified by platform frequencies. Figure 5 shows that, for any given pG_z, the VT decreased as the frequency of oscillation increased.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Effects of periodic acceleration forces (A; Gz) and the frequency (f) of oscillation (B) on minute ventilation (E) in anesthetized and paralyzed piglets. When force was varied, f was held constant and vice versa. These are representative data from a group in which only 1 f and 1 force were held constant. However, a whole family of similar curves may be obtained when different frequencies or different forces are used as constants while the other is varied. The pGz values shown on the x-axis are displayed only for positive values for simplicity. Equal negative values are simultaneously generated during 1 complete motion cycle.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   The relationship between pGz and tidal volume (VT) observed during NIMV. Data were grouped to control for f and reveal that increases in f lower VT for any given inertial force (pGz). The data represent 137 observations from 4 animals. Only positive pGz values are shown.

Transpulmonary and transdiaphragmatic responses to NIMV. The effects of periodic acceleration on transpulmonary pressure (Ptp) and Pdi during NIMV revealed direct relationships between acceleration from the motion platform and both Ptp (Fig. 6A) and Pdi (Fig. 6B). Accelerations between ±0.1 and ±0.6 G_z proportionally increased pressures across both the diaphragm and the lungs. In all experiments, NIMV produced negative esophageal pressure, with no significant change in mean airway pressure during inspiration, similar to spontaneously breathing animals. The increases in Ptp also produced proportional increases in VT during both spontaneous breathing and NIMV (Fig. 7). However, ventilation with NIMV was less efficient than spontaneous breathing. Although the slope of the VT-Ptp relationship was the same, the paradoxical movement lowered the intercept.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Effects of pGz on transpleural pressure (Ptp; A) and transdiaphragmatic pressure (Pdi; B). Dotted lines represent 95% confidence intervals of the regression. Data represent pooled linear regression analysis from 4 experiments. Only positive pGz values are shown.

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Ptp effects on VT during spontaneous breathing and NIMV. Data are from 2 experiments.

Gas exchange in normal lungs during NIMV. The effect of NIMV on gas exchange in animals with normal lungs is shown in Table 1. The animals were deliberately hypoventilated on CMV to observe the responses to subsequent NIMV. The motion platform was set to the optimized values determined in the previous experiments (f = 4.0 and ±0.6 pG_z). NIMV rapidly reversed the hypercapnia produced by hypoventilation and corrected the acidosis. These responses occurred within 5 min, and arterial blood gas values remained normal throughout the 120-min observation period.

Gas exchange in diseased lungs during NIMV. To test the efficiency of the NIMV under adverse pulmonary conditions, meconium solution was instilled into the lung. The effects of NIMV on blood gases are summarized in Table 2. Thirty minutes after instillation of 25% meconium solution during CMV and PEEP of 5 cmH₂O, arterial blood pH and arterial partial pressure of oxygen (Pa_O2) declined, whereas Pa_CO2 significantly rose. Placing the animals on NIMV at frequencies of 4.0 Hz and ±0.6 pG_z significantly reduced Pa_CO2 and increased pH and Pa_O2 over the 150-min observation period. NIMV did not significantly affect heart rate or arterial blood pressure.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A platform was constructed that oscillates the whole body of animals in the horizontal G_z (spinal axis) plane with controlled frequency and acceleration. These movements impart periodic sinusoidal acceleration forces to the whole body that ventilate the lungs of paralyzed and anesthetized animals. Ventilation was computed from the product of respiratory frequency and in-line integrated pneumotachographic volume. Owing to the small VT attendant with NIMV, bias flow was necessary to remove expired CO₂ from the dead space. This same problem arises when measuring VT during high-frequency oscillatory ventilation (7). Ventilation was directly related to the acceleration and inversely related to the frequency of the platform oscillations. Ventilation that was achieved at airway pressures equivalent to a CPAP of 5 cmH₂O was capable of maintaining adequate blood gases in paralyzed, normal animals and those with experimental meconium aspiration.

The two controllable variables of the motion platform are the frequency and acceleration of its oscillations. The acceleration subsequently determines the force imparted to the subject. The acceleration forces are sinusoidal and, therefore, are positive and negative in one complete cycle. The footward deceleration and headward acceleration forces are responsible for inspiration, and the headward deceleration and footward acceleration forces are responsible for expiration. These periodic acceleration forces actively compress and decompress the diaphragm, resulting in inspiration and expiration, respectively. This is supported by the observation that increases in the inertial force generated by the platform produce directly proportional increases in the Pdi gradient (Fig. 6B). Although Pdi gradients with NIMV are established, the actual excursion of the diaphragm was not documented. The Pdi and Ptp in these studies were dependent on estimated pressures for the pleural and abdominal cavities using esophageal and gastric pressures, respectively. Although esophageal and gastric pressures have been commonly used to estimate pleural and abdominal pressures, the estimates may be problematic during NIMV due to the high frequencies of sinusoidal inertial forces generated on these structures. It seems possible that pressures measured in the esophagus and stomach may be independent of the actual pressures within the pleural and abdominal compartments, due to the structures' own inertial effects on the pressure balloons. Also, the pleural pressures during NIMV may not be as homogenous as the esophageal pressure tracings may, incorrectly, imply. Although the actual pressure measurements during NIMV may not be completely accurate or reflect all lung regions homogenously, the changes of pressure within these compartments probably are relative and proportional to the changes occurring across the lungs and diaphragm. These differences or pressure artifacts may not significantly influence the conclusions about the pressure changes and mechanics of airflow during NIMV.

The proposed diaphragmatic movement during NIMV is somewhat analogous to normal human ventilation, with the exception that expiration with NIMV is a completely active process using applied forces in addition to the potential force of elastic recoil used during normal ventilation (6). The upward and downward movement of the diaphragm associated with paradoxical contraction and expansion of the rib cage and abdominal compartments during pG_z has been observed in seated humans subjected to pG_z. The subjects were vibrated vertically on an electrohydraulic shaker table with the application of sinusoidal and square waves (±0.5 G_z from 2 to 10 Hz). The subjects breathed through a pneumotachograph, and Ptp was estimated by using esophageal pressure. The maximum volume of air driven in and out of the airway was 46 ml at 5 Hz. Ptp was 1.9 cmH₂O at 2 Hz and 5.4 at 5 Hz. Measurements of thoracic and abdominal displacement during vertical accelerations indicate that, as the abdominal viscera move upward, displacing the diaphragm cephalad, the abdominal wall moves inward and the thoracic wall outward. The opposite took place when the abdominal viscera move downward (22). Although electrohydraulic vibration in the previous study and motion ventilation in this study are artificial phenomena, there appears to be a precedent in nature for pG_z imposing respiratory forces within the abdominal and thoracic compartments. Bipedal hopping mammals, specifically wallabies and kangaroos, produce forces due to pG_z while hopping that entrain respiration to locomotion cycles in a 1:1 ratio. The hopping movements correlate to respiratory paradoxical motions of the abdomen and thorax and airflow patterns identical to those observed in this study using the motion platform (3).

The out-of-phase motion during NIMV may limit the magnitude of VT. In fact, Fig. 7 shows less VT for any given transpleural pressure gradient with NIMV, relative to spontaneous breathing. Studies also suggest that the presence of paradoxical respiratory movements, which often occur during rapid-eye-movement sleep in preterm infants, results in inefficient ventilation with increased work of breathing (11, 14). Despite occurrence of paradoxical motion, the NIMV-induced active compression and decompression of the diaphragm and, therefore, ventilation most likely occurs due to the transmitted inertial forces of the abdominal contents during NIMV (6). This is supported by observations in which it was noted that abdominal distension or extirpation significantly attenuates ventilation during NIMV (unpublished observations). The low-pressure ventilation characteristics of NIMV, akin to natural breathing (Figs. 6 and 7), may reduce injury due to overpressure or inflation that is sometimes observed with CMV.

The use of conventional positive-pressure ventilation can cause injury to the lung tissues when excess volume or pressure is applied. Because NIMV uses low volumes and low pressures (fluctuating around the imposed CPAP), then these injuries have to be mitigated with NIMV. However, the actual movements and strains imposed on the lung parenchyma and other tissues during NIMV may impose a different trauma to these structures that is completely independent from air pressure or volume. Shearing forces between and within structures with differing viscoelastic properties, such as alveoli and blood vessels, may cause injury ranging from the mechanical tearing of cells from basement membranes (21) to the stretch-induced release of proinflammatory mediators (20). Intuitively, the magnitude of these responses may be proportional to the magnitude of the applied inertial force, the frequency of the oscillations, or the time of exposure. Although no apparent injuries were observed at the settings tested in this study and in preliminary testing of a human prototype motion platform, higher acceleration forces or forces of longer duration may manifest clinically significant trauma.

E (Fig. 4B) and VT (Fig. 5) with NIMV are inversely related to the frequency of the motion platform oscillations. As the frequency increases, the VT decreases and E falls. This occurs although the number of breaths per minute increases. Similar responses in pressure-controlled conventional ventilation are observed when high ventilation rates inhibit E because the minimum time required for the movement of air in and out of the lungs exceeds the imposed time demands of the frequency. Although E and tidal ventilation volumes are low at the frequencies selected for the motion platform, ventilation may be promoted by nonconvective means, similar to gas exchange observed with high-frequency jet ventilation and high-frequency vibration. This is because the dead space gas, conducting airways, and lungs "vibrate" in response to the high-frequency sinusoidal motion. The motion platform may also promote mucociliary movement and clearance of airway debris through two-phase gas-liquid pumping (5, 18). Such effects have not yet been investigated during NIMV.

Ventilation by NIMV can be compared with ventilation by other modes of high-frequency ventilation such as high-frequency oscillation (12, 19), jet ventilation (17), and high-frequency vibration (8, 10). The approximate ventilator frequencies for these modes are: CMV, 0.3 Hz; high-frequency oscillation ventilation, 12 Hz; high-frequency jet ventilation, 2 Hz; high-frequency vibration, 22 Hz; and NIMV, 4 Hz. Thus the operational frequency for the motion platform that was determined to be best for ventilation in this study using juvenile piglets (4 Hz) is intermediate among high-frequency modes of ventilation, although clearly higher than CMV. The optimal frequency in larger animals or humans may be different than 4 Hz. Although all high-frequency modes of ventilation used small volumes, only NIMV used negative pleural pressures to generate airflow, thereby reducing the possibility of barotrauma and volutrauma via excessive lung stretch. Mean airway pressures during both NIMV (Fig. 2) and high-frequency vibration (10), however, remained low and have been attributable only to the imposed CPAP (~5 cmH₂O) that was used during operation. The CO₂ removal or gas-exchange capacities of high-frequency modes of ventilation were usually equal to or better than CMV at lower airway pressures (8, 10, 12, 17, 19). Gavriely et al. (10) found that constant-flow or -pressure modes such as CPAP or intratracheal pulmonary ventilation produce most effective air transport in the trachea but not in the peripheral airways, whereas high-frequency chest vibration was most effective at the periphery. Combining tracheal gas insufflation with vibration produced the optimal gas exchange with minimal pulmonary pressure or mechanical changes. Because of the intermediate frequencies, NIMV probably produces good gas exchange by some chest wall movements but not to the extent observed with high-frequency oscillation or chest vibration ventilation. Also, NIMV does not produce pejorative changes in vascular pressures (1) commonly observed with high-frequency ventilation (3, 19).

This study has demonstrated that NIMV ventilates paralyzed piglets. This is a model most useful to describe ventilation on NIMV of human infants and small children but not larger animals. The use of paralysis may not be necessary because NIMV has been successfully performed in nonparalyzed anesthetized piglets (unpublished observations). However, activation of respiratory muscles during NIMV is often not synchronized with the rapid frequency of the motion platform and thus impedes airflow. The use of NIMV for animals or humans of larger mass is limited. Larger animals probably require different platform settings to achieve adequate ventilation because of both increased mass and differences in the configuration and geometry of the lungs and thoracic compartments. In fact, two conscious human adults subjected to NIMV at a pG_z of ±0.3 and frequencies up to 3 Hz produced VT of 50-70 ml (preliminary experiments). Clearly, these volumes are not sufficient for adequate ventilation, and it may be necessary to sedate or paralyze humans on NIMV, similar to the animals described in this study.

The motion platform ventilates anesthetized and paralyzed animals both with normal lungs and with severe lung injury induced by the aspiration of meconium (Tables 1 and 2).

Carbon dioxide removal, after either intentional hypoventilation in animals with normal lungs or in animals compromised by aspiration, occurred rapidly and was sustained over the 2-h experiment. The CO₂ removal is comparable to what may be achieved with conventional pressure-controlled mechanical ventilation (18). In fact, we found that Pa_CO2 levels in these animals could be driven down to values as low as 5 Torr by altering the motion platform acceleration and frequency settings. Similar normalization also occurs for Pa_O2 and pH. These changes in arterial blood gases all occur at mean airway pressures equivalent to atmospheric pressure raised by the CPAP pressure. In contrast, conventional ventilation of diseased lungs often requires high peak inspiratory pressures to achieve adequate gas exchange and may result in barotrauma with subsequent pulmonary inflammation and fibrosis. On the other hand, the concept of "protective CMV" using low VT (<6 ml/kg) has proven very valuable in patients with acute respiratory distress syndrome (16). In protective ventilation, the low volumes and pressures protect the diseased lungs from further stretch-induced injury, thus allowing repair to occur. The permissive hypercapnia and lower oxygen tensions attendant with low-volume protective ventilation (16) are tolerated in an overall trade-off of pejorative effects. NIMV, as used in this experimental setting, produces both low-volume ventilation and adequate gas exchange in meconium aspiration and may be a superior approach to the protective ventilation concept. The possibility of shearing trauma with NIMV, however, cannot be ruled out. The ability of NIMV to produce adequate ventilation and exchange gas has been established in injured lungs over short time periods (2 h). The effectiveness or complications of the motion platform in ventilating injured lungs over longer times, when lung compliance further declines, has not yet been established.