INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Acute respiratory distress syndrome (ARDS) was once considered a problem of diffuse lung injury and a generalized increase in tissue recoil. However, studies using computed tomography have consistently shown that manifestations of acute lung injury are not distributed homogeneously throughout the lung (1); some areas of the lung retain normal morphologic features and density, whereas other areas are poorly aerated or nonaerated (1). Within better aerated portions of the lung, the compliance and fragility of lung tissue is likely to be more functionally normal than previously thought, especially in the earliest phase of ARDS (4). Consequently, the use of conventional mechanical ventilation with a high tidal volume (VT) and peak airway pressure (Paw_peak) can have detrimental effects in patients with ARDS. Because better aerated portions of the lung receive a larger portion of the VT, more normal alveoli may become overdistended, resulting in ventilator-induced volutrauma and barotrauma (5).

The concept of a lung-protective strategy in the management of patients with ARDS has been proposed as a method of overcoming the problems of volutrauma and barotrauma from alveolar overdistension during mechanical ventilation (5). When a lung-protective strategy is used, attempts are made to minimize iatrogenic worsening of lung injury through pressure- and volume-limited strategies such as permissive hypercapnia, pressure-control ventilation (PCV), and pressure-limited, volume-cycled ventilation (VCV) (7). As part of this ventilatory strategy, attempts are made to keep set inspiratory pressure at levels  35 cm H₂O (7, 9). However, the reduction in VT necessary to achieve this goal may result in an increase in Pa_CO2 (permissive hypercapnia), which itself has potentially adverse consequences (10). Further, there are a number of absolute and relative contraindications to the use of permissive hypercapnia, (e.g., cerebral edema, severe metabolic acidosis, refractory hypoxemia, and coronary artery disease) (10).

Tracheal gas insufflation (TGI), an adjunctive ventilatory technique, can be used to enhance CO₂ elimination efficiency (12). During expiration, fresh gas insufflated through the TGI catheter washes out the CO₂ that remains in the dead space proximal to the catheter tip. Consequently, less CO₂ is inhaled during inspiration (13, 20). In addition, turbulence generated by gas exiting from the catheter tip enhances gas mixing beyond the catheter tip, increasing CO₂ removal from more peripheral airways (20). Therefore, TGI has the potential to make ventilation with a lower VT more efficient (12- 25). The change in Pa_CO2 with TGI has been shown to be flow dependent, with higher flows causing a greater reduction in Pa_CO2 (13, 19, 23).

Although a number of studies have tested the impact of TGI in patients with ARDS, the primary focus of these studies was on the efficacy of TGI during PCV and VCV when minute ventilation (E) is held constant (12, 15, 19, 22). Only one study was identified that tested the ability of TGI to permit mechanical ventilation at a lower set inspiratory pressure during PCV, and this study was conducted in infants (25). We sought to determine whether TGI would allow a reduction in set inspiratory pressure without increasing Pa_CO2 in adults with ARDS who were managed with PCV and, if feasible, to quantify the extent of the reduction tolerated and the time interval required to evaluate the efficacy of the TGI. The findings of our study are clinically relevant because they help to define both the limits of efficacy of TGI when a decision is made to reduce set inspiratory pressure, and the time at which TGI-induced changes are likely to be complete.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

Eight consecutive patients with ARDS were enrolled (Table 1). All patients in intensive care units with ARDS who were managed with PCV (Model 7200; Nellcor-Puritan Bennett, Pleasanton, CA) were screened daily to determine whether they met the study inclusion criteria. The inclusion criteria were: (1) a fraction of inspired oxygen (FI_O2) < 0.9; (2) a positive end-expiratory pressure (PEEP) < 20 cm H₂O; (3) a peak airway pressure > 35 cm H₂O; and (4) hemodynamic stability (e.g., a stable vasopressor requirement during the 6 h preceding enrollment, no clinical or radiologic evidence of uncontrolled congestive heart failure, and no unstable arrhythmias. These criteria were intended to select patients at risk for mechanical ventilation-induced lung injury. Patients were considered to have ARDS if they met the criteria described in the American-European Consensus Conference, which include: (1) a predisposing clinical illness; (2) bilateral pulmonary infiltrates; (3) a pulmonary artery occlusion pressure of < 18 mm Hg (or no clinical evidence of left atrial hypertension); and (4) a Pa_O2-to-FI_O2 ratio < 200 mm Hg (26). The study received approval from the Institutional Review Board of the University of Pittsburgh. Once a patient had met all inclusion criteria, informed consent was obtained from the patient's next-of-kin. All subjects were managed clinically without changes in vasopressor requirements or intravascular fluid changes during data collection, and no invasive procedures were performed within 1 h of data collection.

TGI Apparatus

To deliver continuous TGI, we passed a single-distal-lumen catheter (1.67 mm I.D.) through a jet ventilator adapter (Model 600101; Concord Portex, Keene, NH). Before inserting the TGI catheter into the patient, we assessed a chest radiograph to verify tracheal tube position. The TGI catheter was passed in vitro through a tracheal tube of the same size and model as that used for the patient, to a position that would place the tip of the TGI catheter 1 cm above the carina. The jet ventilator adapter was then adjusted to hold the TGI catheter in this position, and the catheter was inserted into the patient's tracheal tube. Blended oxygen and air (Air/O₂ Microblender; Bird Products Corp., Palm Springs, CA), at the same FI_O2 as that delivered by the ventilator, were passed through a calibrated flow generator (Bellofram, Newell, WV) and delivered to the TGI catheter via a small-bore tube under the TGI flow conditions. Because this was a short-term study, no TGI gas humidification was used.

Respiratory Inductive Plethysmography

Bands for respiratory inductive plethysmography (RIP) were positioned circumferentially around the subject's rib cage and abdomen, and were secured in place with adhesive tape. The respiratory inductive plethysmograph (RespiTrace Plus; NIMS, Miami Beach, FL) was calibrated through the semiquantitative single-position calibration method (27). Subjects were maintained in the same (supine) position throughout both calibration and the experimental protocol. To insure delivery of a known VT, we gave subjects VCV during calibration, and then returned them to PCV. RIP was used to directly measure gas entering the patient's lungs, since total inspired VT included ventilator and TGI-derived gas flow (15, 28). In addition, we used RIP to adjust ventilator PEEP to compensate for the development of TGI-induced auto-PEEP, and to thereby maintain total PEEP (ventilator PEEP + auto-PEEP) at its control (baseline) value (28). Accordingly, after TGI was initiated at 10 L/min, and before the 5 cm H₂O reduction in set inspiratory pressure, we progressively reduced ventilator PEEP in 1- to 2-cm H₂O increments until the VT matched the RIP-derived VT obtained during the baseline (zero TGI) condition. The mean reduction in ventilator PEEP was 2.9 ± 1.5 cm H₂O. Because RIP tracings have been shown to have significant signal drift over time (29), we compared RIP-derived values for VT obtained during the two bracketed (zero flow) conditions (see the experimental protocol). Mean VT increased by 2.3% from the first to the last zero-flow condition, an outcome we attributed to the relatively short experimental protocol.

Dead Space-to-VT Calculations

All exhaled gas was collected after passing through a nondiffusing gas collection bag (Model 6015; Hans Rudolph, Kansas City, MO), and the mixed expired CO₂ was measured by capnography (Model 7000; Novametrix, Wallingford, CT). A standard ventilator circuit with a known compliance (0.0014 L/cm H₂O) was used for each study. To calculate the dead space-to-VT ratio (VD/VT), we used the Enghoff modification of the Bohr equation, after accounting for the dilutional effects of the TGI delivered during expiration and the compressible volume of the ventilator tubing, calculated according to the method described by Nahum and colleagues (13).

Airway Pressure Measurements

Peak airway pressure was measured at the proximal end of the tracheal tube, using the digital output adapter on the mechanical ventilator.

Lung Injury Scores

Lung injury scores (LIS) were calculated through the method proposed by Murray and colleagues (30).

Hemodynamic Parameters

Mean arterial blood pressure (MAP) was measured by obtaining the continuous signal from an indwelling arterial catheter. The MAP was taken as the average of all data points from the tracing over a period of 5 min. Arterial blood gas samples were obtained from an indwelling catheter. Heart rate (HR) was obtained from the standard lead II electrocardiogram tracing.

Experimental Protocol

The eight mechanically ventilated study subjects receiving PCV were investigated sequentially in four experimental steps: (1) TGI at 0 L/ min, with baseline ventilator settings; (2) TGI at 10 L/min, with set inspiratory pressure reduced by 5 cm H₂O below baseline; (3) TGI at 10 L/min, with set inspiratory pressure reduced by 10 cm H₂O below baseline; and (4) TGI at 0 L/min, with baseline ventilator settings (Table 2). Each experimental step lasted 30 or 60 min, depending on patient tolerance of the reduction in set inspiratory pressure. Data were recorded during the last 5 min of each step. During TGI, a pressure-relief valve (Model 04230; Bird Products Corp., Palm Springs, CA) was inserted into the expiratory circuit to maintain a constant set inspiratory pressure (23). The experimental protocol was terminated if any of the following occurred: (1) arterial oxygen saturation (Sa_O2) < 90% and unresponsive to an increase in FI_O2; (2) MAP < 60 mm Hg or > 120 mm Hg; (3) new cardiac dysrhythmia; or (4) attempts by the patient to breathe between ventilator breaths despite sedation.

Data Acquisition

Numerical data were displayed and stored on computer disks, using software (Notebook Pro, Labtech Inc., Wilmington, MA) installed in a desktop computer (P120; Dell Inc., Round Rock, TX). Numerical data were then converted to waveforms, analyzed, and stored, using data-acquisition software (Windaq/200; Dataq Inc., Akron, OH). Analogue data (Paw_peak, mixed-expired CO₂) were digitalized with an analogue-to-digital data-acquisition board (CIO-DAS 1600; Computer Boards, Mansfield, MA) to produce waveforms for analysis. All analogue signals were acquired at 100 Hz per channel. Signals from the RespiTrace were imported and stored in a computer (Colorbook DX2-50; Gateway 2000, North Sioux City, SD), utilizing RespiEvents software (version 4.1; NIMS) to determine VT/minute ventilation (E).

Data Analysis

Data are reported as mean ± SD. Statistical significance was considered at p < 0.05. All measured and calculated variables of the two bracketing (zero flow) conditions were tested for statistical significance. Because there were no statistically significant differences between the two bracketing (zero flow) conditions, we used the average value in the subsequent statistical analysis. Data were analyzed using multivariate analysis of variance for comparisons between TGI at 0 L/ min and 10 L/min with set inspiratory pressure reduced by 5 cm H₂O and 10 cm H₂O. Comparisons between 30-min and 60-min data were analyzed with Student's t test.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Ventilatory and Hemodynamic Effects of Reducing Paw_peak by 5 cm H₂O

When set inspiratory pressure was decreased by 5 cm H₂O, Paw_peak decreased from 40.2 ± 2.7 cm H₂O to 34.3 ± 2.7 cm H₂O (p = 0.0001) at 30 min, a 15% decrement (Table 3). In conjunction with the decrease in Paw_peak, there was a 16% decrease in VT, from 584 ± 154 ml to 490 ± 160 ml (p = 0.001). Despite the decrease in airway pressure and total inspired VT, Pa_CO2 was unchanged (p = NS) as compared with its baseline value (59 ± 10 mm Hg versus 57 ± 6 mm Hg). Consequently, there was no change in VD/VT (0.64 ± 0.05 versus 0.62 ± 0.12). This phase of the protocol was completed by all eight study subjects for 30 min and by seven of the subjects for 60 min. There was no change in Pa_O2, HR, or MAP for the seven subjects who completed 60 min of data collection. Data collection was terminated for the remaining subject because of an Sa_O2 < 90% that was unresponsive to an increase in FI_O2 and patient attempts to breathe between ventilator breaths despite sedation.

Ventilatory and Hemodynamic Effects of Decreasing Paw_peak by 10 cm H₂O

When set inspiratory pressure was decreased by 10 cm H₂O, Paw_peak decreased from 40.2 ± 2.7 cm H₂O to 29.7 ± 2.0 cm H₂O (p = 0.0001) at 30 min, a 26% decrement. In conjunction with the decrease in Paw_peak, there was a 26% decrease in VT, from 584 ± 154 ml to 430 ± 160 ml (p = 0.001). However, Pa_CO2 increased from 59 ± 10 mm Hg to 70 ± 10 mm Hg after 30 min (p = 0.034), and Pa_O2 decreased from 75 ± 9 to 65 ± 9 mm Hg (p = 0.033). There was no change in VD/VT, HR, or MAP. Six subjects tolerated the 10 cm H₂O reduction in Paw_peak for 30 min and one subject for 60 min. Reasons for terminating data collection were an Sa_O2 < 90% unresponsive to an increase in FI_O2 (three subjects) and hemodynamic instability (MAP < 60 mm Hg or > 120 mg Hg) accompanied by patient attempts to breathe between ventilator breaths despite sedation (three subjects).

Effect of Time Interval (30 versus 60 min) on Ventilatory and Hemodynamic Parameters

To determine the stability of changes in CO₂ elimination efficiency during TGI, comparisons were made between data obtained at the ends of the 30- and 60-min intervals during the 5 cm H₂O reduction in set inspiratory pressure. Pa_CO2 was unchanged during the interval from 30 to 60 min (57.6 mm Hg versus 59.9 mm Hg), as was VD/VT (0.62 ± 0.12 versus 0.59 ± 0.09). There was also no change in Pa_O2, HR, or MAP.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study of patients with ARDS who were managed with PCV showed that: (1) TGI can rapidly facilitate a 5 cm H₂O reduction in set inspiratory pressure without increasing Pa_CO2; (2) the efficacy of this strategy was evident within 30 min; and (3) although most patients easily tolerated a 5 cm H₂O reduction in set inspiratory pressure, attempts to further reduce set inspiratory pressure were associated with adverse physiologic effects that led to termination of the protocol.

To reduce exposure of the lung to high airway pressure, mechanical ventilation should be conducted with sufficient PEEP to prevent end-expiratory collapse and tidal recruitment, and with a small VT to avoid ventilator-induced lung injury (4). This goal can be achieved with VCV or PCV (31, 32). One advantage of PCV is the ability to easily designate a set inspiratory pressure, which represents a maximum pressure that alveolar pressure cannot exceed under conditions of passive ventilation (32). Consequently, PCV is commonly used in our institution when managing patients with acute lung injury. During PCV, attempts to limit set inspiratory pressure to  35 cm H₂O, in accord with a lung-protective strategy, may result in decreased alveolar ventilation and a consequent increase in Pa_CO2. In our patient sample at baseline, Paw_peak was 40.2 ± 2.7 cm H₂O and Pa_CO2 was 59 ± 10 mm Hg. We utilized continuous TGI at 10 L/min in an attempt to lower Paw_peak without a further increase in Pa_CO2. Our findings showed that a modest (5 cm H₂O) reduction in set inspiratory pressure could be achieved without adverse physiologic effects in the majority of patients (seven of eight). Ability to tolerate the 5 cm H₂O reduction in set inspiratory pressure was evident within 30 min, suggesting that the efficacy of this intervention can be rapidly determined. Similar findings were reported by Danan and colleagues (25) in a study of five premature newborns who were mechanically ventilated with PCV. In these infants, set inspiratory pressure was decreased by 5.4 ± 1.7 cm H₂O during TGI, with no change in transcutaneous carbon dioxide saturation (Ptc_CO2) after 30 min.

Conversely, a greater reduction in set inspiratory pressure (10 cm H₂O) led to a significant increase in Pa_CO2 and a significant decrease in Pa_O2. A retrospective analysis indicated that these changes probably resulted from a decrement in VT (27.7 ± 8.5%; range: 17% to 41%) associated with a lower mean airway pressure (17.1 ± 4.2 cm H₂O). Notably, the subject who tolerated 60 min at the 10 cm H₂O reduction was also the individual with the least (17%) change in VT. Whether or not such reductions in preset inspiratory airway pressure would have been tolerated if the reduction had occurred over a longer time interval or with the use of increased sedation was not addressed by this study.

In the clinical setting, a commonly used approach to managing hypercarbia is to increase E, and thereby alveolar ventilation, by increasing the respiratory rate (f) (13). Although this strategy is effective during VCV, inherent characteristics of respiratory dynamics preclude its use during PCV. Using an animal model, Nahum and colleagues (13) showed that when the inspiratory duty cycle (TI/Ttot) and set inspiratory pressure are held constant, E reaches a well-defined plateau when f is increased and there is a progressive fall in VT. There was an increase in Pa_CO2 with each increment in f because of a monotonically increasing VD/VT. Further, auto-PEEP tended to increase when f was increased (13). TGI provides an alternative strategy that can be used during PCV to maintain a constant Pa_CO2. Gas insufflated through the TGI catheter enhances CO₂ elimination efficiency, allowing Paw_peak to be decreased without a further increase in Pa_CO2.

Several prior studies have evaluated the efficacy of TGI in adults with ARDS. In five of these studies, TGI was used in combination with VCV (12, 15, 16, 19, 22). Ravenscraft and colleagues (15) examined the effects of continuous TGI at various catheter flow rates and positions in eight sedated, paralyzed patients with acute respiratory failure caused by a variety of lung disorders. The highest catheter flow (6 L/min) and the most distal catheter position (1 cm above the carina) were the most effective combination, producing an average 15% reduction in Pa_CO2. Similar findings were reported by Nakos and colleagues (16) in a study of seven patients with acute respiratory failure. Kalfon and colleagues (22) studied the efficacy of permissive hypercapnia with and without the use of TGI (15 L/min) during expiration only (expiratory washout [EWO]). They found that EWO significantly decreased Pa_CO2 (30%; p < 0.0001). However, the reduction in Pa_CO2 was accompanied by a significant increase in end-inspiratory plateau pressure (p < 0.001) and a significant decrease in cardiac index (p < 0.01). In a subsequent study, Kalfon and colleagues (12) compared three strategies for reducing Pa_CO2 during VCV in patients with severe ARDS: (1) EWO at 15 L/min; (2) optimized mechanical ventilation, defined as an increase in f to the maximal rate possible without development of auto-PEEP, in combination with a reduction in instrumental dead space; and (3) the combination of both methods. Optimized mechanical ventilation was as efficient as EWO in reducing Pa_CO2. When used in combination, EWO and optimized mechanical ventilation had additive effects, which resulted in Pa_CO2 levels that were close to normal values. However, auto-PEEP had to be reduced by 5.3 ± 2.3 cm H₂O during EWO and by 7.3 ± 1.3 cm H₂O during the combination of the two modes, whereas it remained unchanged during optimized mechanical ventilation alone. Two studies examined the efficacy of TGI during PCV (17, 19). Belghith and colleagues (17) tested the effects of TGI at 4 L/min in six patients with ARDS who met criteria for extracorporeal membrane oxygenation and who were severely hypercapnic and hypoxemic despite optimal PCV. During PCV with TGI, Pa_CO2 decreased significantly (p < 0.05), with no change in hemodynamic variables. Kuo and colleagues (19) compared the efficacy of TGI (6 L/min) during VCV and PCV in 12 patients with ARDS and found it equivalent during the two ventilatory modes. Additionally, prior work with animals has compared inspiratory, expiratory, and continuous TGI to determine the CO₂ washout capabilities of each technique (18). Although all three modes of TGI decreased the Pa_CO2, continuous TGI offered the greatest decline in Pa_CO2, followed by expiratory TGI and lastly by inspiratory TGI (18).

When TGI is used, a number of technical and safety issues must be addressed. When TGI is used continuously throughout the respiratory cycle, several modifications are required to compensate for the additional gas delivered by the catheter. The combination of continuous TGI with VCV will increase VT and Paw_peak unless the ventilator-derived VT is decreased to compensate for the added inspiratory gas from the TGI flow (21). Under these circumstances the ventilator-derived VT must be reduced by an amount equivalent to the TGI- inspired flow to keep total E (ventilator + TGI) constant (15, 16, 19, 21). When continuous TGI is used with PCV, Paw_peak may increase above the set inspiratory pressure as the result of delivery of a higher than intended VT, especially when TGI is delivered at high flow rates ( 10 L/min) (21, 23). The increase in Paw_peak occurs because gas from the catheter continues to flow into the lung after the ventilator-derived flow has ceased. This phenomenon is caused by the ventilator's inability to recognize the excess flow derived from the tracheal catheter. Therefore, when continuous TGI is used in conjunction with PCV, it is necessary to insert a pressure-relief valve or to adjust set inspiratory pressure to maintain Paw_peak at the same level, and to keep total E (ventilator + TGI) constant (23, 24). Use of a pressure-relief valve, adjusted to activate at the set inspiratory pressure plus the set PEEP, eliminates problems resulting from overpressurization, since any additional gas from the TGI catheter, after the target pressure is reached, will be vented into the atmosphere. If the set inspiratory pressure or set ventilator PEEP are changed, it is necessary to make appropriate adjustments in order to maintain the target pressure (set pressure plus PEEP). The approach of using a pressure-relief valve is not appropriate when using expiratory TGI. If auto-PEEP develops during expiratory TGI, there will be a decrease in the pressure gradient between proximal airway and end-expiratory intrapulmonary pressure, which dictates delivered VT. Consequently, set inspiratory pressure must be adjusted to maintain the pressure gradient (24).

Currently, there is no commercially available system that provides the ventilator modifications needed to deliver expiratory TGI (e.g., use of the expiratory valve as a trigger for TGI). Because it does not require ventilator modification, we prefer to use continuous TGI, since it offers better CO₂ washout and is easier to implement (23, 24, 33).

Both expiratory and continuous TGI may produce auto-PEEP (12, 22). In our study, RIP was used to provide an independent measure of this variable (28). In the clinical setting, the amount of TGI-induced auto-PEEP can be estimated by determining the amount of reduction in ventilator PEEP that is necessary to restore the VT (in PCV) or inspiratory pressure (in VCV) to their baseline values (21). Clinically, we have found that a 3- to 5-cm H₂O reduction in ventilator PEEP is typically required to keep total PEEP constant during continuous TGI at 10 L/min. Alternately, ventilator PEEP can be adjusted to keep the inspiratory plateau airway pressure (Pplat) constant when using expiratory TGI (12).

Both continuous and expiratory TGI cause problems in monitoring exhaled volumes (33). Because gas flows during the expiratory phase of TGI, it adds to the exhaled volume and may obscure any leak in the system. One approach to solving this problem involves setting the low-exhaled VT alarm at a volume that takes into account both TGI-derived catheter flow and ventilator flow (34). A second problem relates to the potential effect of continued TGI flow in the setting of an occluded endotracheal tube (33). If the expiratory pathway of the ventilator circuit were to become occluded, the catheter could quickly deliver large volumes of gas, potentially causing serious barotrauma or hemodynamic compromise (15, 33). Danan and colleagues (25) solved this problem by incorporating a feedback system designed to activate a sound alarm and cut off power to the membrane pump if insufflation or tracheal pressure exceeded the maximal predefined values plus 10%. Alternately, TGI can be used only during the expiratory phase (18). There is also the potential for TGI catheter occlusion by a mucus plug. Only one study was identified that described this complication. Among a series of 12 adults who used TGI for  72 h, Kuo and colleagues (19) reported that two subjects developed catheter obstruction from a mucus plug, at 58 h and 67 h, respectively, despite use of humidified gas to deliver TGI. Both events were immediately detected by whistling sounds emitting from the pressure-release valve on the humidifier, and there were no adverse sequelae. Given the risk of complications should catheter obstruction develop, it seems advisable to briefly remove the TGI catheter at least every 24 h in order to inspect the tip for mucous ball formation.

Our study was subject to several limitations. Our experimental protocol was designed to evaluate the efficacy of a rapid reduction in set inspiratory pressure. Although a 5 cm H₂O decrement was easily tolerated, further reductions in set inspiratory pressure were associated with indicators of stress under conditions of the experimental protocol. It is possible that the 10 cm H₂O reduction in the set inspiratory pressure would have been better tolerated if effected in a more gradual manner (e.g., in small steps over several hours), and if additional strategies had been used (e.g., increased sedation, neuromuscular blockade, administration of bicarbonate). However, we doubt that these steps would have altered our findings, given the substantial change in VT and in mean airway pressure that occurred during the 10 cm H₂O decrement in set inspiratory pressure. Of further note was that we did not alter initial ventilator settings as defined by the attending physician before beginning our experimental protocol. All subjects were managed by university-based internists cognizant of the need to use a lung-protective strategy, supporting the concept that ventilatory management was appropriate at entry into the study. However, it is possible that the study findings would have differed had our experimental protocol involved changing initial ventilator settings according to a predetermined protocol for ARDS management prior to initiating TGI. Additionally, we followed subjects only for the duration of the protocol, and therefore did not assess long-term response in terms of the impact of TGI on morbidity and mortality.

In summary, the findings of this study of patients with ARDS who were managed with PCV showed that TGI can be used to rapidly facilitate a 5 cm H₂O reduction in set inspiratory pressure without an increase in Pa_CO2. The ability to achieve a 5 cm H₂O reduction in set inspiratory pressure without adverse physiologic effect was evident within 30 min. Attempts to further reduce set inspiratory pressure produced a significant increase in Pa_CO2 and a significant decrease in Pa_O2, resulting in the need to return settings to prior values.