Closed-Loop Control of Respiratory Drive Using Pressure-Support Ventilation
http://www.100md.com
美国呼吸和危急护理医学 2005年第5期
Research Center, Respiratory Health Research Unit, Sacree-Coeur Hospital of Montreal
Pediatric Intensive Care Unit, Department of Pediatrics, Sainte-Justine Hospital Research Center
Department of Medicine, University of Montreal
Department of Adult Critical Care, Sir Mortimer B. Davis Jewish General Hospital, McGill University
Department of Kinanthropology, University of Quebec in Montreal, Montreal, Quebec
Department of Newborn and Developmental Pediatrics, Sunnybrook and Women's College Health Sciences Center
Interdepartmental Division of Critical Care, University of Toronto, St. Michael's Hospital, Toronto, Ontario, Canada
ABSTRACT
By using diaphragm electrical activity (multiple-array esophageal electrode) as an index of respiratory drive, and allowing such activity above or below a preset target range to indicate an increased or reduced demand for ventilatory assistance (target drive ventilation), we evaluated whether the level of pressure-support ventilation can be automatically adjusted in response to exercise-induced changes in ventilatory demand. Eleven healthy individuals breathed through a circuit (18 cm H2O/L/second inspiratory resistance at 1 L/second flow; 0.5eC1.0 L/second expiratory flow limitation) connected to a modified ventilator. Subjects breathed for 6-minute periods at rest and during 20 and 40 W of bicycle exercise, with and without target drive ventilation (the target was set to 60% of the increase in diaphragm electrical activity observed between rest and 20 W of unassisted exercise). With target drive ventilation during exercise, the level of pressure-support ventilation was automatically increased, reaching 13.3 ± 4.0 and 20.3 ± 2.8 cm H2O during 20- and 40-W exercise, respectively, whereas diaphragm electrical activity was reduced to a level within the target range. Both diaphragmatic pressure-time product and end-tidal CO2 were significantly reduced with target drive ventilation at the end of the 20- (p < 0.01) and 40-W (p < 0.001) exercise periods. Minute ventilation was not altered. These results demonstrate that target drive ventilation can automatically adjust pressure-support ventilation, maintaining a constant neural drive and compensating for changes in respiratory demand.
Key Words: airflow limitation diaphragm electromyography exercise mechanical ventilation
During acute respiratory failure, patients frequently require ventilatory assistance to unload the respiratory muscles. Traditionally, the adjustment of the ventilatory assist has been performed manually. Recently, however, there has been increased interest in "closed loop" mechanical ventilation, whereby the patient's own physiologic parameters are used to automatically adjust the ventilatory assist delivered (1, 2). In contrast to within-a-breath ventilator control, breath-to-breath adjustments of the ventilatory support are achieved by first establishing a target range or level within which the designated parameter should be maintained. This parameter, which likewise acts as the input signal, is then continuously measured and compared with the predetermined target, with differences prompting the adjustment of the ventilatory assist delivered on a breath-by-breath basis.
Diaphragm electrical activity (EAdi) can be used to infer respiratory motor output (3). Recent advances in the signal acquisition and processing of the crural EAdi (4eC7) make it possible to accurately measure EAdi online in mechanically ventilated subjects and to test whether EAdi can be used for the automatic adjustment of ventilatory assist when respiratory drive is altered (see Figure 1E in the online supplement). By establishing a target range of EAdi, and letting breath-by-breath changes in the EAdi above or below this range determine an increased or reduced requirement of pressure-support ventilation (PSV), we hypothesized that it may be possible to automatically adjust the level of ventilatory assist in response to changes in respiratory drive. The essential element underpinning this operation is that, in addition to unloading the respiratory muscles, PSV also deactivates them (i.e., it "downregulates" or suppresses the EAdi) (8eC10). Thus, when the level of EAdi for a given inspiration exceeds the upper limit of a preset target range, PSV is progressively increased until the inspiratory EAdi level is suppressed to within the target range. If the level of EAdi drops below the lower limit of the target range, PSV is gradually reduced until the EAdi level returns within the predetermined target range. This mode of ventilatory control is hereafter referred to as target drive ventilation (TDV).
The purpose of the present study was to determine whether EAdi can be maintained within a preset target range and whether the diaphragm can be simultaneously unloaded by the automatic closed-loop adjustment of the PSV level in response to changes in EAdi outside of the target range, when respiratory drive is altered by exercise.
Some of the results of the current study have been previously reported in the form of an abstract (11).
METHODS
See online supplement for additional details on the methods.
Subjects
Eleven healthy individuals (three women, eight men) with a mean age of 40 years (range, 35eC52) participated in the study. Written, informed consent was obtained from all subjects.
Experimental Protocol
While seated on a bicycle ergometer, each subject performed maximum sniff inhalations and inspiratory capacity maneuvers to determine maximum voluntary diaphragm activation (12). Thereafter, subjects breathed through a mouthpiece connected to a breathing circuit, in turn connected to a modified Siemens Servo 300 ventilator (Siemens-Elema AB, Solna, Sweden). The breathing circuit contained an inspiratory flow resistance (18 cm H2O/L/second at a flow of 1 L/second) and an expiratory Starling resistor (expiratory flow limitation of 0.5eC1.0 L/second). The experiment consisted of two experimental runs. Each run consisted of 6 minutes of resting breathing, 6 minutes of constant workload exercise performed at 20 W and then at 40 W, and 6 minutes of exercise recovery. No ventilator support was provided in the first run (control), whereas TDV was set to target EAdi at a level of approximately 60% of the difference in mean inspiratory EAdi observed between control resting breathing and 20-W exercise.
Instrumentation
Airflow was measured with a pneumotachograph and VT was obtained by integrating flow. The crural EAdi was obtained via a multiple-array esophageal electrode. Esophageal and gastric pressures were measured using two balloon-tipped catheters anchored within the lumen of the EAdi catheter. Transdiaphragmatic pressure (Pdi) was obtained by subtracting the esophageal from the gastric pressure. Mouth and ventilator pressures were measured via side ports in the mouthpiece and ventilator tubing (distal to the resistance circuit), respectively. In eight of the subjects, minute ventilation and pulmonary gas exchange were measured breath by breath with a SensorMedics Vmax computerized system (SensorMedics Corp., Yorba Linda, CA).
Online Automatic Processing of EAdi
Signals from each electrode pair were differentially amplified, digitized, and processed using previously described algorithms and techniques (4, 6, 13). The root-mean-square of the processed signal was used to quantify EAdi.
Automatic Adjustment of PSV by Targeting EAdi
The processed EAdi signal (root-mean-square) was averaged for each inspiration, and a five-breath moving average (EAdiMEAN) was computed and used to control the ventilator. Once a target level of EAdi had been set (with a ± 10% range), a servo-control algorithm was used such that PSV was maintained at a constant level when the EAdiMEAN was within the targeted range; it increased or decreased by 0.5 cm H2O/breath when EAdiMEAN was above or below the targeted range, respectively. The ventilator was equipped with an external analog input for triggering of PSV using EAdi in combination with the built-in flow/pressure triggers operating on a first-come-first-serve basis. Ventilator cycling-off was also accomplished by using the EAdi and occurred when EAdi dropped to 80% of the peak activity for a given breath (Figure 1).
ANALYSIS
Offline Signal Analysis
The flow and pressures were acquired simultaneously with the EAdi data. Minute ventilation, VT, and timing parameters were determined breath by breath from the flow signal. Mean Pdi swings were calculated between the onset of EAdi and the end of inspiratory flow. The pressure-time product of the Pdi (PTPdi) was obtained for each breath by multiplying the area subtended by the Pdi signal by the respiratory frequency (60/total breath duration). The level of PSV delivered by the ventilator was identified as the plateau pressure from the ventilator pressure signal. For each subject, minute ventilation, VT, timing parameters, PTPdi, and the PSV level were averaged for each minute in each condition studied. Because it takes approximately 4 minutes to attain a steady-state condition during constant workload exercise, the stored gas exchange variables recorded in the last 2 minutes of each condition were averaged in the subsequent analysis.
Statistical Analysis
Variables were compared between control and TDV runs during resting breathing and the two levels of exercise using two-way repeated measures analysis of variance, and post hoc contrasts of significant effects were performed using the Student-Newman-Keuls test (SPSS, version 12.0; SPSS Inc., Chicago, IL). Values in text and figures are means ± SD unless otherwise indicated. The level of significance for all statistical tests was set to p < 0.05.
RESULTS
Averaged data from the 11 healthy subjects studied under conditions of control breathing and TDV at rest, during each minute of 20- and 40-W exercise, and recovery after exercise, are shown in Figure 2. In the absence of ventilatory assist, the EAdiMEAN, PTPdi, and minute ventilation progressively increased during exercise and then gradually declined after exercise was ceased. As indicated by the gray horizontal band in Figure 2, the EAdi that was used to target the TDV ranged from 14.9 ± 3.7 to 18.2 ± 4.5 (SD). During resting breathing with TDV, the EAdiMEAN remained below the target range and therefore no ventilatory assist was provided. During 20-W exercise, when EAdiMEAN rose above the target range, the PSV level was automatically increased by TDV, reducing the EAdiMEAN to within the target range and concomitantly reducing the PTPdi. During 40-W exercise, the level of the PSV delivered by TDV was further increased, and diaphragm activation was maintained within the target range. The level of PSV delivered with TDV during the last minute of 40-W exercise ranged between 7 and 35 cm H2O. After the cessation of exercise, EAdiMEAN initially dropped below the target range, which resulted in the progressive and automatic reduction and ultimate removal of the PSV. There was no difference in the minute ventilation observed between the two runs. Subjects demonstrated a slightly larger VT (p = 0.008), with no significant change in the respiratory rate during resting breathing with TDV, despite the fact that ventilatory assist was not provided. Although the VT was higher in the first minutes of 20- and 40-W exercise with TDV compared with control, it was not significantly different during the last minute of each exercise period once steady-state conditions were reached (Figures 2 and 3).
Ventilatory and gas exchange responses from the last 2 minutes of each condition obtained in 8 of the 11 subjects are presented in Figure 4. There was no significant difference in any of the measured parameters comparing control breathing and TDV at rest. Compared with control exercise, the end-tidal CO2 (PETCO2) was significantly lower with TDV during the last 2 minutes of 20- and 40-W exercise, when the level of PSV was automatically increased to 12.0 ± 9.6 and 18.0 ± 8.5 cm H2O, respectively. This was associated with a significant reduction in the O2 consumption (O2) and CO2 production (CO2) during TDV, but only at 40 W of exercise. Although there was a tendency for higher VTs with TDV at rest and during both levels of exercise, the differences were not statistically significant. A significant reduction in the EAdiMEAN with TDV was observed only during the last 2 minutes of 40-W exercise.
DISCUSSION
By using the EAdi as an index of the respiratory drive and letting changes in the EAdiMEAN over time control pressure support, we have shown that, during loaded breathing in healthy subjects, it is possible to automatically adjust the level of ventilatory assist from one breath to the next and to maintain diaphragm activation within a predetermined target range.
The ventilatory control system is integrative in nature; the central motor output is modulated by afferent feedback of sensory information from numerous sources (chemoreceptors, chest wall/muscle mechanoreceptors, lung/airway receptors, and others) and modified by inputs from other parts of the brain (voluntary control of breathing; Figure 1E). The electrical impulses originating centrally are transmitted via the spinal and peripheral nerves to ultimately activate the respiratory muscles. Thus, in individuals with a functional neuromuscular transmission, the EAdi can be used as an index of the respiratory drive (3). Moreover, as demonstrated in the current study (Figure 1E), the crural EAdi can be used in a control loop to automatically adjust the level of PSV delivered in response to changes in the respiratory drive. This is possible because, in addition to reflecting the ventilatory "demand" of the patient, the EAdi is likewise responsive to manipulation of the PSV (i.e., increasing the PSV level acts to downregulate or suppress the central motor output and the EAdi) (8eC10).
The current study used inspiratory loading and expiratory flow limitation to mechanically load the respiratory muscles and to increase the respiratory drive as has been observed to occur in patients with stable chronic obstructive pulmonary disease and with acute respiratory failure (12, 14, 15). Exercise was used to additionally stimulate the respiratory drive and further increase the EAdi. However, given the large inspiratory resistance and expiratory flow limitation that were used, a relatively low level of exercise was needed for this purpose. Moreover, because of a high respiratory workload, which was further increased during exercise, subjects were unable to eliminate the increased CO2 produced by their exercising muscles, causing the PETCO2 to rise to 52.0 ± 5.0 mm Hg during 40-W exercise. It could therefore be suggested that hypercapnic respiratory failure may have developed to a certain extent in these healthy subjects (16).
The determination of the target range to be used in the TDV run in the current study required that unassisted control breathing precede TDV. Although it could be suggested that this introduced a sequence bias into the study, we do not anticipate any effect of such on the ability of TDV to control and maintain the EAdiMEAN within the target range. However, we cannot exclude the possibility that the higher VT observed during quiet breathing in the TDV run might have occurred in anticipation of the run with ventilatory assistance that was to follow. Conceivably, a higher level of PSV may have been required to suppress the EAdiMEAN within a given target range.
The target EAdi level for TDV was set to 60% of the increase in EAdiMEAN observed between resting breathing and 20 W during the control exercise period. Such a level was chosen based on our assumption that setting the target EAdi too low would likely result in the delivery of excessive pressures or an inability to achieve suppression of EAdiMEAN within the target range. Similarly, if the target had been set to an EAdiMEAN level above that observed during 40-W exercise, there would have been no PSV delivered with TDV, because that range would never have been exceeded. For the group, the EAdi target corresponded to 15 to 20% of maximum EAdi. Although a method for determining the optimum target level of EAdi in mechanically ventilated patients has yet to be established, we can assume that the lower limit of the EAdi target range should be higher than 8%, which is the EAdi exhibited by healthy subjects during unloaded resting breathing (12), whereas the upper limit should not exceed 25 to 40% of maximum EAdi, which corresponds to the resting diaphragm activation observed in outpatients with stable chronic obstructive pulmonary disease (12, 17).
To reduce excessive breath-by-breath variation of the ventilatory assist, a five-breath moving average was applied to the processed EAdi signal and the target range around that level was set to ± 10%. The five-breath moving average used in the present study may have been on the high side when considering the work of Viale and coworkers (9), who demonstrated that it takes six to eight breaths for the EAdi to stabilize in response to changes in the PSV level. However, the combination of a ± 10% target range and the small PSV changes made (0.5 cm H2O/breath) appears to have successfully maintained EAdiMEAN close to target, with overshoot, undershoot, and oscillation minimized. However, the extent to which response time, the step changes in the PSV level, and the target range affect TDV responsiveness has yet to be determined.
The current study was essentially a first attempt at evaluating servo-control adjustment of the PSV level using EAdi, while endeavoring to ensure system stability. For this purpose, we applied a simple step function as the servo-control algorithm. However, given that the relationship between suppression of EAdi and increase in pressure support is likely not linear and that adjustment of the respiratory drive involves a time delay of several breaths, from a system control perspective, other methods may be superior to the one used. For example, proportional adjustment of the PSV level in direct response to the magnitude of overshoot and undershoot of the target signal may be feasible and should be evaluated in future studies.
When conventional cycling-off of the PSV was used during pilot experiments (the Siemens Servo 300 ventilator is cycled-off when flow drops to 5% of peak), and especially when high levels of PSV were applied, we experienced severe delays between the end of neural inspiration and breath termination of the ventilatory assist, likely from the high inspiratory resistance of our breathing circuit (18, 19). We therefore implemented an algorithm for neural cycling-off, which ultimately improved expiratory synchrony (Figure 1). However, the use of such neural cycling-off, in conjunction with neural triggering, may have contributed to the findings that inspiratory duration, expiratory duration, and breathing frequency were not significantly altered by the ventilatory assistance provided during exercise, once steady-state conditions were reached, despite the fact that large reductions in the EAdiMEAN, PTPdi, and PETCO2 were observed. Previous studies using conventional PSV have demonstrated that, as the level of ventilatory assistance is increased, there is a tendency for the VT to become larger and the respiratory rate to be reduced (8, 10, 18, 20eC24). However, other evidence suggests a lesser effect on breathing pattern when the ventilatory assist is delivered in synchrony with diaphragm efforts (25). Although the higher VTs during TDV may have contributed to the reduced PETCO2 at 20-W exercise, the absence of further ventilatory modifications at 40-W exercise suggests that the significant reduction in the PETCO2 occurred as a result of respiratory muscle unloading, as evidenced by a lower O2, CO2, and EAdiMEAN (Figure 4).
Other modes of mechanical ventilation have also endeavored to achieve servo-controlled adjustment of the PSV level in spontaneously breathing patients. Using the airway occlusion pressure at 0.1 seconds (P0.1), which is an indirect index of respiratory drive, as the controller signal, Iotti and colleagues (26) were able to automatically adjust the PSV in a group of stable patients recovering from acute respiratory failure. However, it is known that factors that contribute to neuromechanical uncoupling (i.e., weakness, fatigue, dynamic hyperinflation) (17) can also cause the P0.1 to underestimate the "true" neural inspiratory drive (Figure 1E) (27). Still other modes use ventilatory parameters, such as VT, minute ventilation, breathing frequency and PETCO2, as the controller signal (2, 28eC33). Although certain of these can, to some extent, compensate for changes that occur in the respiratory mechanics over time (33), in some cases increases in patient effort secondary to an increased respiratory demand can cause a reduction in the ventilatory assist that is delivered, when in fact more is required (31). Problems with patienteCventilator synchrony can also have an adverse effect on the use of such modes. Although patients often demonstrate reductions in the respiratory rate with increasing PSV, such reductions may not necessarily represent reductions in the neural respiratory rate but may instead be from ineffective inspiratory efforts (efforts failing to trigger the ventilator), which have been shown to occur more often with increasing PSV (18, 21, 34, 35). No wasted efforts were noted in the current study.
Our electrode array was 8 cm in length, from the centers of the most cephalad to the most caudal rings, and was sufficient to measure diaphragm movements during exercise corresponding to VTs of more than 1.5 L. Previous work demonstrated that the diaphragm can move up to 4 cm along the electrode array in healthy subjects performing maximum inspirations (13), whereas in mechanically ventilated patients with acute respiratory failure, diaphragm movement seldom exceeds one electrode pair along the array (i.e., 10 mm) (10). Therefore, if the catheter is properly secured in a position where the diaphragm is located centrally over the electrode array, using eight electrode pairs with a 10-mm interelectrode distance (ring-to-ring center) is sufficient to cover most possible movements of the diaphragm along the array.
Electrode motion artifacts and noise can have an important impact on the measurement of EAdi (5). The signal-processing methods used in the present study include detection of the electrode position with regards to the diaphragm (4, 12), filtering to eliminate motion artifacts, and reducing common noise and ECG leak-through, and replacement of residual artifacts by a previous value (6). Therefore, such signal-processing methods ensure an optimum signal quality for ventilator servo-control.
It should be noted that the current study was performed in healthy subjects breathing against external loads and it is not known if this technique will be suitable for long-term use in patients with abnormal respiratory system mechanics. Moreover, TDV might not be suitable for all patients, particularly those in whom increasing the ventilatory assist does not decrease the respiratory drive or EAdi. TDV could also hypothetically cause excessive delivery of assist in patients with impaired vagal feedback. Similar to other modes of mechanical ventilation, appropriate setting of upper pressure limits is necessary to protect the patient.
TDV could potentially be applied to neurally adjusted ventilatory assist (7), which provides intrabreath assist in proportion to the measured multiplied by a fixed-gain constant. With neurally adjusted ventilatory assist, an individual must increase his or her EAdi to receive more support within a given breath. However, TDV could be used to automatically adjust the neurally adjusted ventilatory assist gain factor from breath to breath, enabling the respiratory drive to be maintained within a given range over time.
Conclusions
This study demonstrates that, by using diaphragm electrical activity to indicate the need to increase or reduce ventilatory support, it is possible to maintain the electrical activity of the diaphragm within a predetermined target range and to automatically compensate for changes in respiratory demand.
Acknowledgments
The authors thank Dr. Jaques Lacroix and Dr. Marissa Tucci for their support and Norman Comtois for his technical assistance.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
REFERENCES
Ranieri VM. Optimization of pressure-ventilator interactions: closed-loop technology to turn the century. Intensive Care Med 1997;23:936eC939.
Branson RD, Johannigman JA, Campbell RS, Davis K Jr. Closed-loop mechanical ventilation. Respir Care 2002;47:427eC435.
Lourenco RV, Cherniack NS, Malm JR, Fishman AP. Nervous output from the respiratory center during obstructed breathing. J Appl Physiol 1966;21:527eC533.
Beck J, Sinderby C, Lindstrm L, Grassino A. Influence of bipolar electrode positioning on measurements of human crural diaphragm EMG. J Appl Physiol 1996;81:1434eC1449.
Sinderby C, Lindstrm L, Grassino A. Automatic assessment of electromyogram quality. J Appl Physiol 1995;79:1803eC1815.
Sinderby C, Beck JC, Lindstrm L, Grassino A. Enhancement of signal quality in esophageal recordings of diaphragm EMG. J Appl Physiol 1997;82:1370eC1377.
Sinderby C, Navalesi P, Beck J, Skrobik Y, Comtois N, Friberg S, Gottfried SB, Lindstrm L. Neural control of mechanical ventilation. Nat Med 1999;5:1433eC1436.
Brochard L, Harf A, Lorino H, Lemaire F. Inspiratory pressure support prevents diaphragmatic fatigue during weaning from mechanical ventilation. Am Rev Respir Dis 1989;139:513eC521.
Viale JP, Duperret S, Mahul P, Delafosse B, Delpuech C, Weismann D, Annat G. Time course evolution of ventilatory responses to inspiratory unloading in patients. Am J Respir Crit Care Med 1998;157:428eC434.
Beck J, Gottfried SB, Navalesi P, Skrobik Y, Comtois N, Rossini M, Sinderby C. Electrical activity of the diaphragm during pressure support ventilation in acute respiratory failure. Am J Respir Crit Care Med 2001;164:419eC424.
Spahija J, Beck J, Gottfried S, Comtois N, Comtois A, Sinderby C. Target drive ventilation (TDV): autoregulation of ventilatory assist using diaphragm electrical activity . Am J Respir Crit Care Med 2001;163:A303.
Sinderby C, Beck J, Weinberg J, Spahija J, Grassino A. Voluntary activation of the human diaphragm in health and disease. J Appl Physiol 1998;85:2146eC2158.
Beck J, Sinderby C, Weinberg J, Grassino A. Effects of muscle-to-electrode distance on the human diaphragm electromyogram. J Appl Physiol 1995;79:975eC985.
Aubier M, Murciano D, Fournier M, Milic-Emili J, Pariente R, Derenne J-P. Central respiratory drive in acute respiratory failure of patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1980;122:191eC199.
Murciano D, Boczkowski J, Lecocguic Y, Emili JM, Pariente R, Aubier M. Tracheal occlusion pressure: a simple index to monitor respiratory muscle fatigue during acute respiratory failure in patients with chronic obstructive pulmonary disease. Ann Intern Med 1988;108:800eC805.
Roussos C, Koutsoukou A. Respiratory failure. Eur Respir J Suppl 2003;47:3seC14s.
Sinderby C, Spahija J, Beck J, Kaminski D, Yan S, Sliwinski P. Diaphragm activation during exercise in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;163:1637eC1641.
Giannouli E, Webster K, Roberts D, Younes M. Response of ventilator-dependent patients to different levels of pressure support and proportional assist. Am J Respir Crit Care Med 1999;159:1716eC1725.
Yamada Y, Du HL. Analysis of the mechanisms of expiratory asynchrony in pressure support ventilation: a mathematical approach. J Appl Physiol 2000;88:2143eC2150.
Jubran A, Van de Graff WB, Tobin M. Variability of patient-ventilator interaction with pressure support ventilation in patients with COPD. Am J Respir Crit Care Med 1995;152:129eC136.
Leung P, Jubran A, Tobin MJ. Comparison of assisted ventilator modes on triggering, patient effort, and dyspnea. Am J Respir Crit Care Med 1997;155:1940eC1948.
Fauroux B, Isabey D, Desmarais G, Brochard L, Harf A, Lofaso F. Nonchemical influence of inspiratory pressure support on inspiratory activity in humans. J Appl Physiol 1998;85:2169eC2175.
Perrigault PF, Pouzeratte YH, Jaber S, Capdevila XJ, Hayot M, Boccara G, Ramonatxo M, Colson P. Changes in occlusion pressure (P0.1) and breathing pattern during pressure support ventilation. Thorax 1999;54:119eC123.
Kondili E, Prinianakis G, Anastasaki M, Georgopoulos D. Acute effects of ventilator settings on respiratory motor output in patients with acute lung injury. Intensive Care Med 2001;27:1147eC1157.
Sharshar T, Desmarais G, Louis B, Macadou G, Porcher R, Harf A, Raphael JC, Isabey D, Lofaso F. Transdiaphragmatic pressure control of airway pressure support in healthy subjects. Am J Respir Crit Care Med 2003;168:760eC769.
Iotti GA, Brunner JX, Braschi A, Laubscher T, Olivei MC, Palo A, Galbusera C, Comelli A. Closed-loop control of airway occlusion pressure at 0.1 second (P0.1) applied to pressure-support ventilation: algorithm and application in intubated patients. Crit Care Med 1996;24:771eC779.
Whitelaw WA, Derenne JP, Milic-Emili J. Occlusion pressure as a measure of respiratory center output in conscious man. Respir Physiol 1975;23:181eC199.
Dojat M, Brochard L, Lemaire F, Harf A. A knowledge-based system for assisted ventilation of patients in intensive care units. Int J Clin Monit Comput 1992;9:239eC250.
Branson RD, MacIntyre NR. Dual-control modes of mechanical ventilation. Respir Care 1996;41:294eC302.
Dojat M, Harf A, Touchard D, Lemaire F, Brochard L. Clinical evaluation of a computer-controlled pressure support mode. Am J Respir Crit Care Med 2000;161:1161eC1166.
Branson RD, Davis K Jr. Dual control modes: combining volume and pressure breaths. Respir Care Clin N Am 2001;7:397eC408.
Linton DM. Adaptive Lung ventilation. Respir Care Clin N Am 2001;7:409eC423.
Campbell RS, Branson RD, Johannigman JA. Adaptive support ventilation. Respir Care Clin N Am 2001;7:425eC440.
Fabry B, Guttmann J, Eberhard L, Bauer T, Haberthur C, Wolff G. An analysis of desynchronization between the spontaneously breathing patient and ventilator during inspiratory pressure support. Chest 1995;107:1387eC1394.
Nava S, Bruschi C, Rubini F, Palo A, Iotti G, Braschi A. Respiratory response and inspiratory effort during pressure support ventilation in COPD patients. Intensive Care Med 1995;21:871eC879.(Jadranka Spahija, Jennife)
Pediatric Intensive Care Unit, Department of Pediatrics, Sainte-Justine Hospital Research Center
Department of Medicine, University of Montreal
Department of Adult Critical Care, Sir Mortimer B. Davis Jewish General Hospital, McGill University
Department of Kinanthropology, University of Quebec in Montreal, Montreal, Quebec
Department of Newborn and Developmental Pediatrics, Sunnybrook and Women's College Health Sciences Center
Interdepartmental Division of Critical Care, University of Toronto, St. Michael's Hospital, Toronto, Ontario, Canada
ABSTRACT
By using diaphragm electrical activity (multiple-array esophageal electrode) as an index of respiratory drive, and allowing such activity above or below a preset target range to indicate an increased or reduced demand for ventilatory assistance (target drive ventilation), we evaluated whether the level of pressure-support ventilation can be automatically adjusted in response to exercise-induced changes in ventilatory demand. Eleven healthy individuals breathed through a circuit (18 cm H2O/L/second inspiratory resistance at 1 L/second flow; 0.5eC1.0 L/second expiratory flow limitation) connected to a modified ventilator. Subjects breathed for 6-minute periods at rest and during 20 and 40 W of bicycle exercise, with and without target drive ventilation (the target was set to 60% of the increase in diaphragm electrical activity observed between rest and 20 W of unassisted exercise). With target drive ventilation during exercise, the level of pressure-support ventilation was automatically increased, reaching 13.3 ± 4.0 and 20.3 ± 2.8 cm H2O during 20- and 40-W exercise, respectively, whereas diaphragm electrical activity was reduced to a level within the target range. Both diaphragmatic pressure-time product and end-tidal CO2 were significantly reduced with target drive ventilation at the end of the 20- (p < 0.01) and 40-W (p < 0.001) exercise periods. Minute ventilation was not altered. These results demonstrate that target drive ventilation can automatically adjust pressure-support ventilation, maintaining a constant neural drive and compensating for changes in respiratory demand.
Key Words: airflow limitation diaphragm electromyography exercise mechanical ventilation
During acute respiratory failure, patients frequently require ventilatory assistance to unload the respiratory muscles. Traditionally, the adjustment of the ventilatory assist has been performed manually. Recently, however, there has been increased interest in "closed loop" mechanical ventilation, whereby the patient's own physiologic parameters are used to automatically adjust the ventilatory assist delivered (1, 2). In contrast to within-a-breath ventilator control, breath-to-breath adjustments of the ventilatory support are achieved by first establishing a target range or level within which the designated parameter should be maintained. This parameter, which likewise acts as the input signal, is then continuously measured and compared with the predetermined target, with differences prompting the adjustment of the ventilatory assist delivered on a breath-by-breath basis.
Diaphragm electrical activity (EAdi) can be used to infer respiratory motor output (3). Recent advances in the signal acquisition and processing of the crural EAdi (4eC7) make it possible to accurately measure EAdi online in mechanically ventilated subjects and to test whether EAdi can be used for the automatic adjustment of ventilatory assist when respiratory drive is altered (see Figure 1E in the online supplement). By establishing a target range of EAdi, and letting breath-by-breath changes in the EAdi above or below this range determine an increased or reduced requirement of pressure-support ventilation (PSV), we hypothesized that it may be possible to automatically adjust the level of ventilatory assist in response to changes in respiratory drive. The essential element underpinning this operation is that, in addition to unloading the respiratory muscles, PSV also deactivates them (i.e., it "downregulates" or suppresses the EAdi) (8eC10). Thus, when the level of EAdi for a given inspiration exceeds the upper limit of a preset target range, PSV is progressively increased until the inspiratory EAdi level is suppressed to within the target range. If the level of EAdi drops below the lower limit of the target range, PSV is gradually reduced until the EAdi level returns within the predetermined target range. This mode of ventilatory control is hereafter referred to as target drive ventilation (TDV).
The purpose of the present study was to determine whether EAdi can be maintained within a preset target range and whether the diaphragm can be simultaneously unloaded by the automatic closed-loop adjustment of the PSV level in response to changes in EAdi outside of the target range, when respiratory drive is altered by exercise.
Some of the results of the current study have been previously reported in the form of an abstract (11).
METHODS
See online supplement for additional details on the methods.
Subjects
Eleven healthy individuals (three women, eight men) with a mean age of 40 years (range, 35eC52) participated in the study. Written, informed consent was obtained from all subjects.
Experimental Protocol
While seated on a bicycle ergometer, each subject performed maximum sniff inhalations and inspiratory capacity maneuvers to determine maximum voluntary diaphragm activation (12). Thereafter, subjects breathed through a mouthpiece connected to a breathing circuit, in turn connected to a modified Siemens Servo 300 ventilator (Siemens-Elema AB, Solna, Sweden). The breathing circuit contained an inspiratory flow resistance (18 cm H2O/L/second at a flow of 1 L/second) and an expiratory Starling resistor (expiratory flow limitation of 0.5eC1.0 L/second). The experiment consisted of two experimental runs. Each run consisted of 6 minutes of resting breathing, 6 minutes of constant workload exercise performed at 20 W and then at 40 W, and 6 minutes of exercise recovery. No ventilator support was provided in the first run (control), whereas TDV was set to target EAdi at a level of approximately 60% of the difference in mean inspiratory EAdi observed between control resting breathing and 20-W exercise.
Instrumentation
Airflow was measured with a pneumotachograph and VT was obtained by integrating flow. The crural EAdi was obtained via a multiple-array esophageal electrode. Esophageal and gastric pressures were measured using two balloon-tipped catheters anchored within the lumen of the EAdi catheter. Transdiaphragmatic pressure (Pdi) was obtained by subtracting the esophageal from the gastric pressure. Mouth and ventilator pressures were measured via side ports in the mouthpiece and ventilator tubing (distal to the resistance circuit), respectively. In eight of the subjects, minute ventilation and pulmonary gas exchange were measured breath by breath with a SensorMedics Vmax computerized system (SensorMedics Corp., Yorba Linda, CA).
Online Automatic Processing of EAdi
Signals from each electrode pair were differentially amplified, digitized, and processed using previously described algorithms and techniques (4, 6, 13). The root-mean-square of the processed signal was used to quantify EAdi.
Automatic Adjustment of PSV by Targeting EAdi
The processed EAdi signal (root-mean-square) was averaged for each inspiration, and a five-breath moving average (EAdiMEAN) was computed and used to control the ventilator. Once a target level of EAdi had been set (with a ± 10% range), a servo-control algorithm was used such that PSV was maintained at a constant level when the EAdiMEAN was within the targeted range; it increased or decreased by 0.5 cm H2O/breath when EAdiMEAN was above or below the targeted range, respectively. The ventilator was equipped with an external analog input for triggering of PSV using EAdi in combination with the built-in flow/pressure triggers operating on a first-come-first-serve basis. Ventilator cycling-off was also accomplished by using the EAdi and occurred when EAdi dropped to 80% of the peak activity for a given breath (Figure 1).
ANALYSIS
Offline Signal Analysis
The flow and pressures were acquired simultaneously with the EAdi data. Minute ventilation, VT, and timing parameters were determined breath by breath from the flow signal. Mean Pdi swings were calculated between the onset of EAdi and the end of inspiratory flow. The pressure-time product of the Pdi (PTPdi) was obtained for each breath by multiplying the area subtended by the Pdi signal by the respiratory frequency (60/total breath duration). The level of PSV delivered by the ventilator was identified as the plateau pressure from the ventilator pressure signal. For each subject, minute ventilation, VT, timing parameters, PTPdi, and the PSV level were averaged for each minute in each condition studied. Because it takes approximately 4 minutes to attain a steady-state condition during constant workload exercise, the stored gas exchange variables recorded in the last 2 minutes of each condition were averaged in the subsequent analysis.
Statistical Analysis
Variables were compared between control and TDV runs during resting breathing and the two levels of exercise using two-way repeated measures analysis of variance, and post hoc contrasts of significant effects were performed using the Student-Newman-Keuls test (SPSS, version 12.0; SPSS Inc., Chicago, IL). Values in text and figures are means ± SD unless otherwise indicated. The level of significance for all statistical tests was set to p < 0.05.
RESULTS
Averaged data from the 11 healthy subjects studied under conditions of control breathing and TDV at rest, during each minute of 20- and 40-W exercise, and recovery after exercise, are shown in Figure 2. In the absence of ventilatory assist, the EAdiMEAN, PTPdi, and minute ventilation progressively increased during exercise and then gradually declined after exercise was ceased. As indicated by the gray horizontal band in Figure 2, the EAdi that was used to target the TDV ranged from 14.9 ± 3.7 to 18.2 ± 4.5 (SD). During resting breathing with TDV, the EAdiMEAN remained below the target range and therefore no ventilatory assist was provided. During 20-W exercise, when EAdiMEAN rose above the target range, the PSV level was automatically increased by TDV, reducing the EAdiMEAN to within the target range and concomitantly reducing the PTPdi. During 40-W exercise, the level of the PSV delivered by TDV was further increased, and diaphragm activation was maintained within the target range. The level of PSV delivered with TDV during the last minute of 40-W exercise ranged between 7 and 35 cm H2O. After the cessation of exercise, EAdiMEAN initially dropped below the target range, which resulted in the progressive and automatic reduction and ultimate removal of the PSV. There was no difference in the minute ventilation observed between the two runs. Subjects demonstrated a slightly larger VT (p = 0.008), with no significant change in the respiratory rate during resting breathing with TDV, despite the fact that ventilatory assist was not provided. Although the VT was higher in the first minutes of 20- and 40-W exercise with TDV compared with control, it was not significantly different during the last minute of each exercise period once steady-state conditions were reached (Figures 2 and 3).
Ventilatory and gas exchange responses from the last 2 minutes of each condition obtained in 8 of the 11 subjects are presented in Figure 4. There was no significant difference in any of the measured parameters comparing control breathing and TDV at rest. Compared with control exercise, the end-tidal CO2 (PETCO2) was significantly lower with TDV during the last 2 minutes of 20- and 40-W exercise, when the level of PSV was automatically increased to 12.0 ± 9.6 and 18.0 ± 8.5 cm H2O, respectively. This was associated with a significant reduction in the O2 consumption (O2) and CO2 production (CO2) during TDV, but only at 40 W of exercise. Although there was a tendency for higher VTs with TDV at rest and during both levels of exercise, the differences were not statistically significant. A significant reduction in the EAdiMEAN with TDV was observed only during the last 2 minutes of 40-W exercise.
DISCUSSION
By using the EAdi as an index of the respiratory drive and letting changes in the EAdiMEAN over time control pressure support, we have shown that, during loaded breathing in healthy subjects, it is possible to automatically adjust the level of ventilatory assist from one breath to the next and to maintain diaphragm activation within a predetermined target range.
The ventilatory control system is integrative in nature; the central motor output is modulated by afferent feedback of sensory information from numerous sources (chemoreceptors, chest wall/muscle mechanoreceptors, lung/airway receptors, and others) and modified by inputs from other parts of the brain (voluntary control of breathing; Figure 1E). The electrical impulses originating centrally are transmitted via the spinal and peripheral nerves to ultimately activate the respiratory muscles. Thus, in individuals with a functional neuromuscular transmission, the EAdi can be used as an index of the respiratory drive (3). Moreover, as demonstrated in the current study (Figure 1E), the crural EAdi can be used in a control loop to automatically adjust the level of PSV delivered in response to changes in the respiratory drive. This is possible because, in addition to reflecting the ventilatory "demand" of the patient, the EAdi is likewise responsive to manipulation of the PSV (i.e., increasing the PSV level acts to downregulate or suppress the central motor output and the EAdi) (8eC10).
The current study used inspiratory loading and expiratory flow limitation to mechanically load the respiratory muscles and to increase the respiratory drive as has been observed to occur in patients with stable chronic obstructive pulmonary disease and with acute respiratory failure (12, 14, 15). Exercise was used to additionally stimulate the respiratory drive and further increase the EAdi. However, given the large inspiratory resistance and expiratory flow limitation that were used, a relatively low level of exercise was needed for this purpose. Moreover, because of a high respiratory workload, which was further increased during exercise, subjects were unable to eliminate the increased CO2 produced by their exercising muscles, causing the PETCO2 to rise to 52.0 ± 5.0 mm Hg during 40-W exercise. It could therefore be suggested that hypercapnic respiratory failure may have developed to a certain extent in these healthy subjects (16).
The determination of the target range to be used in the TDV run in the current study required that unassisted control breathing precede TDV. Although it could be suggested that this introduced a sequence bias into the study, we do not anticipate any effect of such on the ability of TDV to control and maintain the EAdiMEAN within the target range. However, we cannot exclude the possibility that the higher VT observed during quiet breathing in the TDV run might have occurred in anticipation of the run with ventilatory assistance that was to follow. Conceivably, a higher level of PSV may have been required to suppress the EAdiMEAN within a given target range.
The target EAdi level for TDV was set to 60% of the increase in EAdiMEAN observed between resting breathing and 20 W during the control exercise period. Such a level was chosen based on our assumption that setting the target EAdi too low would likely result in the delivery of excessive pressures or an inability to achieve suppression of EAdiMEAN within the target range. Similarly, if the target had been set to an EAdiMEAN level above that observed during 40-W exercise, there would have been no PSV delivered with TDV, because that range would never have been exceeded. For the group, the EAdi target corresponded to 15 to 20% of maximum EAdi. Although a method for determining the optimum target level of EAdi in mechanically ventilated patients has yet to be established, we can assume that the lower limit of the EAdi target range should be higher than 8%, which is the EAdi exhibited by healthy subjects during unloaded resting breathing (12), whereas the upper limit should not exceed 25 to 40% of maximum EAdi, which corresponds to the resting diaphragm activation observed in outpatients with stable chronic obstructive pulmonary disease (12, 17).
To reduce excessive breath-by-breath variation of the ventilatory assist, a five-breath moving average was applied to the processed EAdi signal and the target range around that level was set to ± 10%. The five-breath moving average used in the present study may have been on the high side when considering the work of Viale and coworkers (9), who demonstrated that it takes six to eight breaths for the EAdi to stabilize in response to changes in the PSV level. However, the combination of a ± 10% target range and the small PSV changes made (0.5 cm H2O/breath) appears to have successfully maintained EAdiMEAN close to target, with overshoot, undershoot, and oscillation minimized. However, the extent to which response time, the step changes in the PSV level, and the target range affect TDV responsiveness has yet to be determined.
The current study was essentially a first attempt at evaluating servo-control adjustment of the PSV level using EAdi, while endeavoring to ensure system stability. For this purpose, we applied a simple step function as the servo-control algorithm. However, given that the relationship between suppression of EAdi and increase in pressure support is likely not linear and that adjustment of the respiratory drive involves a time delay of several breaths, from a system control perspective, other methods may be superior to the one used. For example, proportional adjustment of the PSV level in direct response to the magnitude of overshoot and undershoot of the target signal may be feasible and should be evaluated in future studies.
When conventional cycling-off of the PSV was used during pilot experiments (the Siemens Servo 300 ventilator is cycled-off when flow drops to 5% of peak), and especially when high levels of PSV were applied, we experienced severe delays between the end of neural inspiration and breath termination of the ventilatory assist, likely from the high inspiratory resistance of our breathing circuit (18, 19). We therefore implemented an algorithm for neural cycling-off, which ultimately improved expiratory synchrony (Figure 1). However, the use of such neural cycling-off, in conjunction with neural triggering, may have contributed to the findings that inspiratory duration, expiratory duration, and breathing frequency were not significantly altered by the ventilatory assistance provided during exercise, once steady-state conditions were reached, despite the fact that large reductions in the EAdiMEAN, PTPdi, and PETCO2 were observed. Previous studies using conventional PSV have demonstrated that, as the level of ventilatory assistance is increased, there is a tendency for the VT to become larger and the respiratory rate to be reduced (8, 10, 18, 20eC24). However, other evidence suggests a lesser effect on breathing pattern when the ventilatory assist is delivered in synchrony with diaphragm efforts (25). Although the higher VTs during TDV may have contributed to the reduced PETCO2 at 20-W exercise, the absence of further ventilatory modifications at 40-W exercise suggests that the significant reduction in the PETCO2 occurred as a result of respiratory muscle unloading, as evidenced by a lower O2, CO2, and EAdiMEAN (Figure 4).
Other modes of mechanical ventilation have also endeavored to achieve servo-controlled adjustment of the PSV level in spontaneously breathing patients. Using the airway occlusion pressure at 0.1 seconds (P0.1), which is an indirect index of respiratory drive, as the controller signal, Iotti and colleagues (26) were able to automatically adjust the PSV in a group of stable patients recovering from acute respiratory failure. However, it is known that factors that contribute to neuromechanical uncoupling (i.e., weakness, fatigue, dynamic hyperinflation) (17) can also cause the P0.1 to underestimate the "true" neural inspiratory drive (Figure 1E) (27). Still other modes use ventilatory parameters, such as VT, minute ventilation, breathing frequency and PETCO2, as the controller signal (2, 28eC33). Although certain of these can, to some extent, compensate for changes that occur in the respiratory mechanics over time (33), in some cases increases in patient effort secondary to an increased respiratory demand can cause a reduction in the ventilatory assist that is delivered, when in fact more is required (31). Problems with patienteCventilator synchrony can also have an adverse effect on the use of such modes. Although patients often demonstrate reductions in the respiratory rate with increasing PSV, such reductions may not necessarily represent reductions in the neural respiratory rate but may instead be from ineffective inspiratory efforts (efforts failing to trigger the ventilator), which have been shown to occur more often with increasing PSV (18, 21, 34, 35). No wasted efforts were noted in the current study.
Our electrode array was 8 cm in length, from the centers of the most cephalad to the most caudal rings, and was sufficient to measure diaphragm movements during exercise corresponding to VTs of more than 1.5 L. Previous work demonstrated that the diaphragm can move up to 4 cm along the electrode array in healthy subjects performing maximum inspirations (13), whereas in mechanically ventilated patients with acute respiratory failure, diaphragm movement seldom exceeds one electrode pair along the array (i.e., 10 mm) (10). Therefore, if the catheter is properly secured in a position where the diaphragm is located centrally over the electrode array, using eight electrode pairs with a 10-mm interelectrode distance (ring-to-ring center) is sufficient to cover most possible movements of the diaphragm along the array.
Electrode motion artifacts and noise can have an important impact on the measurement of EAdi (5). The signal-processing methods used in the present study include detection of the electrode position with regards to the diaphragm (4, 12), filtering to eliminate motion artifacts, and reducing common noise and ECG leak-through, and replacement of residual artifacts by a previous value (6). Therefore, such signal-processing methods ensure an optimum signal quality for ventilator servo-control.
It should be noted that the current study was performed in healthy subjects breathing against external loads and it is not known if this technique will be suitable for long-term use in patients with abnormal respiratory system mechanics. Moreover, TDV might not be suitable for all patients, particularly those in whom increasing the ventilatory assist does not decrease the respiratory drive or EAdi. TDV could also hypothetically cause excessive delivery of assist in patients with impaired vagal feedback. Similar to other modes of mechanical ventilation, appropriate setting of upper pressure limits is necessary to protect the patient.
TDV could potentially be applied to neurally adjusted ventilatory assist (7), which provides intrabreath assist in proportion to the measured multiplied by a fixed-gain constant. With neurally adjusted ventilatory assist, an individual must increase his or her EAdi to receive more support within a given breath. However, TDV could be used to automatically adjust the neurally adjusted ventilatory assist gain factor from breath to breath, enabling the respiratory drive to be maintained within a given range over time.
Conclusions
This study demonstrates that, by using diaphragm electrical activity to indicate the need to increase or reduce ventilatory support, it is possible to maintain the electrical activity of the diaphragm within a predetermined target range and to automatically compensate for changes in respiratory demand.
Acknowledgments
The authors thank Dr. Jaques Lacroix and Dr. Marissa Tucci for their support and Norman Comtois for his technical assistance.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
REFERENCES
Ranieri VM. Optimization of pressure-ventilator interactions: closed-loop technology to turn the century. Intensive Care Med 1997;23:936eC939.
Branson RD, Johannigman JA, Campbell RS, Davis K Jr. Closed-loop mechanical ventilation. Respir Care 2002;47:427eC435.
Lourenco RV, Cherniack NS, Malm JR, Fishman AP. Nervous output from the respiratory center during obstructed breathing. J Appl Physiol 1966;21:527eC533.
Beck J, Sinderby C, Lindstrm L, Grassino A. Influence of bipolar electrode positioning on measurements of human crural diaphragm EMG. J Appl Physiol 1996;81:1434eC1449.
Sinderby C, Lindstrm L, Grassino A. Automatic assessment of electromyogram quality. J Appl Physiol 1995;79:1803eC1815.
Sinderby C, Beck JC, Lindstrm L, Grassino A. Enhancement of signal quality in esophageal recordings of diaphragm EMG. J Appl Physiol 1997;82:1370eC1377.
Sinderby C, Navalesi P, Beck J, Skrobik Y, Comtois N, Friberg S, Gottfried SB, Lindstrm L. Neural control of mechanical ventilation. Nat Med 1999;5:1433eC1436.
Brochard L, Harf A, Lorino H, Lemaire F. Inspiratory pressure support prevents diaphragmatic fatigue during weaning from mechanical ventilation. Am Rev Respir Dis 1989;139:513eC521.
Viale JP, Duperret S, Mahul P, Delafosse B, Delpuech C, Weismann D, Annat G. Time course evolution of ventilatory responses to inspiratory unloading in patients. Am J Respir Crit Care Med 1998;157:428eC434.
Beck J, Gottfried SB, Navalesi P, Skrobik Y, Comtois N, Rossini M, Sinderby C. Electrical activity of the diaphragm during pressure support ventilation in acute respiratory failure. Am J Respir Crit Care Med 2001;164:419eC424.
Spahija J, Beck J, Gottfried S, Comtois N, Comtois A, Sinderby C. Target drive ventilation (TDV): autoregulation of ventilatory assist using diaphragm electrical activity . Am J Respir Crit Care Med 2001;163:A303.
Sinderby C, Beck J, Weinberg J, Spahija J, Grassino A. Voluntary activation of the human diaphragm in health and disease. J Appl Physiol 1998;85:2146eC2158.
Beck J, Sinderby C, Weinberg J, Grassino A. Effects of muscle-to-electrode distance on the human diaphragm electromyogram. J Appl Physiol 1995;79:975eC985.
Aubier M, Murciano D, Fournier M, Milic-Emili J, Pariente R, Derenne J-P. Central respiratory drive in acute respiratory failure of patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1980;122:191eC199.
Murciano D, Boczkowski J, Lecocguic Y, Emili JM, Pariente R, Aubier M. Tracheal occlusion pressure: a simple index to monitor respiratory muscle fatigue during acute respiratory failure in patients with chronic obstructive pulmonary disease. Ann Intern Med 1988;108:800eC805.
Roussos C, Koutsoukou A. Respiratory failure. Eur Respir J Suppl 2003;47:3seC14s.
Sinderby C, Spahija J, Beck J, Kaminski D, Yan S, Sliwinski P. Diaphragm activation during exercise in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2001;163:1637eC1641.
Giannouli E, Webster K, Roberts D, Younes M. Response of ventilator-dependent patients to different levels of pressure support and proportional assist. Am J Respir Crit Care Med 1999;159:1716eC1725.
Yamada Y, Du HL. Analysis of the mechanisms of expiratory asynchrony in pressure support ventilation: a mathematical approach. J Appl Physiol 2000;88:2143eC2150.
Jubran A, Van de Graff WB, Tobin M. Variability of patient-ventilator interaction with pressure support ventilation in patients with COPD. Am J Respir Crit Care Med 1995;152:129eC136.
Leung P, Jubran A, Tobin MJ. Comparison of assisted ventilator modes on triggering, patient effort, and dyspnea. Am J Respir Crit Care Med 1997;155:1940eC1948.
Fauroux B, Isabey D, Desmarais G, Brochard L, Harf A, Lofaso F. Nonchemical influence of inspiratory pressure support on inspiratory activity in humans. J Appl Physiol 1998;85:2169eC2175.
Perrigault PF, Pouzeratte YH, Jaber S, Capdevila XJ, Hayot M, Boccara G, Ramonatxo M, Colson P. Changes in occlusion pressure (P0.1) and breathing pattern during pressure support ventilation. Thorax 1999;54:119eC123.
Kondili E, Prinianakis G, Anastasaki M, Georgopoulos D. Acute effects of ventilator settings on respiratory motor output in patients with acute lung injury. Intensive Care Med 2001;27:1147eC1157.
Sharshar T, Desmarais G, Louis B, Macadou G, Porcher R, Harf A, Raphael JC, Isabey D, Lofaso F. Transdiaphragmatic pressure control of airway pressure support in healthy subjects. Am J Respir Crit Care Med 2003;168:760eC769.
Iotti GA, Brunner JX, Braschi A, Laubscher T, Olivei MC, Palo A, Galbusera C, Comelli A. Closed-loop control of airway occlusion pressure at 0.1 second (P0.1) applied to pressure-support ventilation: algorithm and application in intubated patients. Crit Care Med 1996;24:771eC779.
Whitelaw WA, Derenne JP, Milic-Emili J. Occlusion pressure as a measure of respiratory center output in conscious man. Respir Physiol 1975;23:181eC199.
Dojat M, Brochard L, Lemaire F, Harf A. A knowledge-based system for assisted ventilation of patients in intensive care units. Int J Clin Monit Comput 1992;9:239eC250.
Branson RD, MacIntyre NR. Dual-control modes of mechanical ventilation. Respir Care 1996;41:294eC302.
Dojat M, Harf A, Touchard D, Lemaire F, Brochard L. Clinical evaluation of a computer-controlled pressure support mode. Am J Respir Crit Care Med 2000;161:1161eC1166.
Branson RD, Davis K Jr. Dual control modes: combining volume and pressure breaths. Respir Care Clin N Am 2001;7:397eC408.
Linton DM. Adaptive Lung ventilation. Respir Care Clin N Am 2001;7:409eC423.
Campbell RS, Branson RD, Johannigman JA. Adaptive support ventilation. Respir Care Clin N Am 2001;7:425eC440.
Fabry B, Guttmann J, Eberhard L, Bauer T, Haberthur C, Wolff G. An analysis of desynchronization between the spontaneously breathing patient and ventilator during inspiratory pressure support. Chest 1995;107:1387eC1394.
Nava S, Bruschi C, Rubini F, Palo A, Iotti G, Braschi A. Respiratory response and inspiratory effort during pressure support ventilation in COPD patients. Intensive Care Med 1995;21:871eC879.(Jadranka Spahija, Jennife)