INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient-ventilator dyssynchrony is a major clinical problem (1) that commonly arises because the ventilator cycles out of phase with the patient's respiratory muscle activity. Such patient-ventilator dyssynchrony may originate at several points in the cycling of these two pumps. If the onset of machine assistance lags significantly behind the commencement of a patient's inspiratory muscle activity (i.e., trigger delay), the ventilator's ability to produce respiratory muscle rest is markedly impaired (2, 3). A second source of inadequate respiratory muscle rest is the failure of the gas delivery system to meet a patient's inspiratory flow demands (4). Most studies of patient-ventilator dyssynchrony have focused on these two inspiratory phenomena, and scant attention has been paid to expiratory events during ventilation. Depending on the algorithm used for "cycling off" of the ventilator, recruitment of the expiratory muscles can occur during mechanical inflation (7). A final source of dyssynchrony could be a delay in the termination of a patient's expiratory muscle activity before the onset of the next mechanical inflation. All of the foregoing considerations are further complicated by the possibility that the inspiratory and expiratory muscle groups may contract simultaneously (8).

Research into patient-ventilator interactions has been largely based on recordings of flow and pressures (airway, esophageal, and occasionally gastric), and on an estimate of the patient's chest wall characteristics under relaxed conditions (4, 11). With the flow and pressure recordings, however, it is difficult to precisely define the onset and termination of inspiratory and expiratory muscle activity because of the common occurrence of intrinsic positive end-expiratory pressure (PEEPi), which can be due to increased elastic recoil and/or expiratory muscle activity (7, 12, 13). Two different approaches have been advanced for distinguishing the contributions of elastic recoil from those of expiratory muscle activity to PEEPi, with the latter being estimated from measurement of either the increase in gastric pressure (Pga) over the course of expiration (14) or the decrease in Pga at the onset of the next inspiration (15). Consensus on the relative merit of these approaches has not been reached.

Electromyographic recordings make it possible to define more precisely the phase relationship of the "on-switch" and "off-switch" of pertinent muscles with cycling of the ventilator. We have obtained electromyographic recordings of the diaphragmatic crura and transversus abdominis muscles in healthy subjects receiving pressure-support ventilation. Since patient- ventilator synchronization is most difficult to achieve in patients with chronic obstructive pulmonary disease (COPD), we induced expiratory airflow limitation in healthy volunteers through the use of a Starling resistor. With this model we tested the hypothesis that the magnitude and timing of expiratory muscle activity modulate a subject's ability to trigger a mechanical ventilator in the presence of airflow limitation. A secondary aim of our study was to evaluate the relative accuracy of the two methods that have been reported (14, 15) for quantifying the expiratory muscle contribution to PEEPi.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Nine subjects (eight men, and one woman; age 31.7 ± 2 yr (mean ± SE), naïve to the purpose of the study protocol, participated in the study. All were nonsmokers and had normal pulmonary function. The study was approved by the ethics committee of Hines Veterans Administration Hospital, and informed consent was obtained from all subjects.

Experimental Setup

Airflow was measured with a pneumotachograph (No. 2; Fleisch, Lausanne, Switzerland) positioned between the Y of the ventilator circuit and the mouthpiece. Airway pressure (Paw) was measured close to the mouthpiece, using a pressure transducer (MP-45; Validyne, Northridge, CA). Esophageal (Pes) and gastric (Pga) pressures were measured with multilumen balloon catheter systems (16) attached to MP-45 pressure transducers. The esophageal balloon was filled with 0.5 ml of air and positioned in the midesophagus (17); the gastric balloon was filled with 1 ml of air and positioned 60 to 70 cm from the nares. Transdiaphragmatic pressure (Pdi) was obtained by electronic subtraction of Pes from Pga.

The electromyogram of the transversus abdominus muscle (EMGTA) was obtained from six subjects with bipolar insulated wires (Medwire, New York, NY). In each subject, the individual layers of the abdominal wall muscles were visualized with a high-resolution 7.5-MHz ultrasound linear probe (Ultramark 9; Advanced Technology Laboratories, Boswell, WA). The three muscle layers were noted, and an inbuilt electronic caliper was used to measure the thickness of the transversus abdominis and its depth from the skin surface. Under ultrasound guidance, the needle housing the bipolar wires was advanced in the right anterior axillary line (midway between the costal margin and the iliac crest); gentle side-to-side movements helped to visualize and locate the tip of the needle. Once the tip was positioned within the thickness of the transversus abdominis, the needle was removed and the insulated bipolar wires were left behind. To further confirm placement of the wires and ensure that they were not displaced during the experiment, the following steps were taken: After placement, the wires were marked at the point of skin entry, and on completion of the experiment the wires were removed; the distance between the point of skin entry and the tip of the wire was measured and compared with the depth of the transversus abdominis beneath the skin surface as measured ultrasonically. After placement of the wires, a satisfactory signal-to-noise ratio for electromyographic amplitude (greater than 3:1) was verified while the subject counted aloud from one to five (18). Signals were also measured during a slow expiration from total lung capacity, and during a maximal expulsive effort against a closed mouthpiece at total lung capacity (TLC). These maneuvers were repeated at the end of the experiment, and the electromyographic signals were compared with the initially measured signals, thereby revealing whether the wires had been displaced during the course of the experiment. Esophageal electrodes were used to record the diaphragmatic electromyogram (EMG). These EMG signals were recorded and digitized at 3,000 Hz with an analog-to-digital converter (CODAS; DATAQ Instruments, Akron, OH).

Protocol

Subjects were tested while lying on a bed in a semirecumbent position and breathing through a mouthpiece attached to a unidirectional valve, which in turn was attached to the circuit of a Servo 900C ventilator (Siemens, Schaumburg, IL). To induce airflow limitation, a Starling resistor was attached to the expiratory limb of the ventilator circuit when required. Confirmation that expiratory flow limitation had been induced by the Starling resistor was achieved by applying negative expiratory pressure downstream to the resistor. Flow-volume loops of two consecutive breaths, one with and the other without 10 cm H₂O of negative expiratory pressure, were recorded while a subject was breathing on the ventilator. The curves were later superimposed, and failure to achieve a persistent increase of expiratory flow during negative expiratory pressure confirmed the presence of airflow limitation (19).

After placement of the balloon catheter system, subjects breathed through a mouthpiece attached to the ventilator. Pressure support (PS) levels of 0, 10, and 20 cm H₂O were applied in a random manner. After a subject had been allowed 2 to 5 min to become comfortable at a given level of PS, expiratory flow limitation was induced with the Starling resistor attached to the expiratory port of the ventilator. PEEP was set at 0 cm H₂O, and trigger sensitivity was set at 2 cm H₂O for all settings.

Computation of Physiologic Indices

From the 10-min recordings at each setting, the last minute free of swallowing artifacts on the Pes tracings was selected for analysis. The number of triggering and nontriggering attempts was recorded during these 1-min recordings. Nontriggering was defined as an inspiratory attempt (decrease in Pes > 1 cm H₂O with a simultaneous decrease in Paw and/ or change in flow) that failed to open the inspiratory valve of the ventilator (i.e., achieve a sudden increase in flow > 100 ml/s) (3). Gross dynamic intrinsic PEEP (PEEPi) during triggered breaths was calculated as the decrease in Pes from the onset of rapid decrease in Pes to the onset of inspiratory flow (3). Because triggering sensitivity was set at 2 cm H₂O, PEEPi was systematically overestimated by 2 cm H₂O.

Expiratory muscle activity was assessed to determine its effect on triggering and its contribution to PEEPi; two aspects of this were evaluated. First, the amplitude of the moving time average of the rectified EMGTA was compared with two methods that have been proposed to estimate the contribution of expiratory muscle activity to PEEPi: the increase in Pga over the course of expiration (14), and the decrease in Pga at the onset of the next inspiration (15) (Figure 1). This comparison was undertaken in six subjects. As described in the RESULTS section, the expiratory increase in Pga showed a close correlation with the amplitude of the moving time average of EMGTA; accordingly, the increase in Pga was used to estimate the magnitude of expiratory muscle activity before triggering and nontriggering attempts in all nine subjects. Second, the timing of expiratory muscle activity was analyzed with respect to both inspiratory muscle activity and the inflationary action of the ventilator. The onset and termination of inspiratory electromyographic activity was measured by a single observer who was blinded to the expiratory EMG signal, and the observer was blinded to the inspiratory electromyographic activity while determining the onset and offset of expiratory muscle activity. The effect of expiratory muscle activity on the subsequent inspiratory effort was estimated by using the time of overlap, defined as the period during which late expiratory (transversus abdominis) muscle activity overlapped with early inspiratory muscle (diaphragmatic) activity (Figure 1). Also, the synchronization of a subject's neural expiratory time with the ventilator's mechanical expiratory time was assessed by measuring the phase of mechanical expiratory time, expressed in terms of phase angle ( degrees), at which the neural expiratory effort began (defined as the onset of EMGTA activity) (20) (Figure 2).

View larger version (26K): [in this window] [in a new window]   Figure 1.   Measurement of the duration of overlap between inspiratory and expiratory muscle activity in a subject receiving PS of 20 cm H2O. The duration of overlap is the period of simultaneous activity of the transversus abdominis and the diaphragm, as measured from their respective EMG signals. Also shown is the rise and descent in Pga used to estimate elastic recoil from gross PEEPi.

View larger version (17K): [in this window] [in a new window]   Figure 2.   The relationship of neural expiratory time to mechanical expiratory time was assessed by measuring the phase angle, expressed in degrees. If neural activity began simultaneously with the machine, the phase angle () was zero. Neural activity beginning after the offset of mechanical inflation resulted in a positive phase angle (60 degrees for Subject 1). Neural activity beginning before the onset of mechanical inflation resulted in a negative phase angle (45 degrees for Subject 2).

Inspiratory muscle activity was assessed in the following manner. First, a modified pressure-time product (PTP) was calculated by measuring the time integral of respiratory pressure (i.e., the difference between Pes and the recoil pressure of the chest wall [P_{es cw}]). P_{es cw} is equal to (Ecw × VL) + P_{es cw, eexp}, where Ecw is the elastance of the chest wall, VL is the lung volume above end-expiration, and P_{es cw, eexp} is the recoil pressure of the chest wall at the transition point between end-expiratory effort and onset of inspiratory effort (7). To correct for the effect of expiratory muscle activity, we considered P_{es cw, eexp} to be equal to end-expiratory Pes minus the expiratory increase in Pga (14). With this line of reasoning, the decrease in Pes before P_{es cw, eexp} is achieved is attributed to expiratory muscle relaxation. The continued decrease of Pes until the onset of inspiratory flow is considered to equal the inspiratory muscle activity needed to overcome the threshold load of dynamic hyperinflation. This approach to partitioning of PEEPi, based on the method reported by Lessard and colleagues (14), appears to be the most robust of methods reported to date (21). Second, in addition to calculation of the modified PTP, the duration of neural inspiration was quantified as the time between the onset and end of phasic activity of the diaphragmatic EMG.

Data Analysis

During mechanical inspiration, we compared the following measurements: time of overlap for triggering versus nontriggering inspiratory attempts; inspiratory PTP of triggering versus nontriggering attempts; and neural TI of the inspiration before triggering versus nontriggering attempts. During mechanical expiration, comparisons were made between the magnitude of the expiratory effort (ascertained from either EMGTA or the expiratory increase in Pga) before triggering and nontriggering attempts, and of the phase shift between onset of expiratory activity and mechanical expiration. Comparisons were made with paired t tests.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Neural Inspiration

Inspiration muscle activity. All nine subjects had episodes of unsuccessful triggering attempts. At PS of 10 cm H₂O, inspiratory PTP/breath was less for nontriggering attempts than for triggering attempts, 5.2 ± 1.5 cm H₂O · s and 14.4 ± 8.3 cm H₂O · s, respectively (p = 0.04); a similar difference was seen at PS of 20 cm H₂O, 7.5 ± 3 and 13.1 ± 1.8 cm H₂O · s, respectively (p = 0.04). In contrast, at PS of 0 cm H₂O, inspiratory PTP was not significantly different for nontriggering and triggering attempts (p = 0.18).

Neural TI. At all PS levels, the duration of neural TI before nontriggering attempts exceeded that before triggering attempts, 1.1 ± 0.2 and 0.7 ± 0.1 s, respectively (p = 0.04).

Tidal volume (VT). At PS of 10 cm H₂O, breaths before a nontriggering attempt had a higher VT, 0.67 ± 0.05 L, than did breaths before a triggering attempt, 0.55 ± 0.02 L (p < 0.05); a similar difference was seen at PS 20 cm H₂O, 0.71 ± 0.06 and 0.56 ± 0.02 L, respectively (p = 0.03).

Inspiratory flow rate. At PS of 10 cm H₂O, peak inspiratory flow in breaths before nontriggering attempts was 26.5 ± 7.7% higher than in breaths before triggering attempts (p = 0.0008). Likewise, at PS of 20 cm H₂O, peak inspiratory flow of breaths before nontriggering attempts was 11.2 ± 3.6% higher than of breaths before triggering attempts (p = 0.002); no difference was observed at PS of 0 cm H₂O. At PS levels of 10 cm H₂O and 20 cm H₂O, breaths before nontriggering attempts had a slower rate of decrease in inspiratory flow over the last third of mechanical inflation, being 77.6 ± 11.2% and 69.1 ± 4.5% of the corresponding values before triggering attempts (p = 0.007 and p = 0.0001, respectively).

Timing of Neural Expiration

Onset of expiratory muscle activity before termination of mechanical inflation. At PS of 10 cm H₂O, the phase angle between neural and mechanical expiratory times was 32.6 ± 3.1 degrees before nontriggering attempts, versus 12.6 ± 1.8 degrees before triggering attempts (p = 0.0002) (Figure 3). Similarly, at PS of 20 cm H₂O, the phase angles between neural and mechanical expiratory times were 50 ± 9.4 degrees before nontriggering and 19.1 ± 3.4 degrees before triggering attempts (p = 0.01). That is, at PS levels of 10 cm H₂O and 20 cm H₂O, the period of neural expiratory time preceding the onset of mechanical expiratory time was longer before nontriggering attempts; no such difference was observed at PS of 0 cm H₂O (p = 0.122).

View larger version (12K): [in this window] [in a new window]   Figure 3.   Phase angle between neural and mechanical expiratory times before triggering (closed circles) and nontriggering (open circles) attempts. At PS of 10 cm H2O and 20 cm H2O, the phase angle before nontriggering attempts exceeded that before triggering attempts, indicating that neural expiratory time during late mechanical inflation was longer before nontriggering attempts than before triggering attempts (for PS 10, p = 0.0002; for PS 20, p = 0.01).

Expiratory and inspiratory muscle activity. In each of the six subjects for whom EMG recordings were made, EMGTA overlapped with the diaphragmatic EMG during early inspiration. During PS of 0 cm H₂O, the period of overlap of late transversus abdominus muscle activity and early diaphragmatic muscle activity, as inferred from their respective EMGs, was not different for triggering and nontriggering attempts, 61.9 ± 7.9 and 53.6 ± 4.5 ms, respectively (p = 0.67); the respective values at a PS level of 10 cm H₂O were 56.9 ± 3.7 and 55 ± 5.7 ms (p = 0.78); and the respective values at a PS level of 20 cm were 62 ± 4.4 and 64 ± 2.7 ms (p = 0.39).

Estimation of Magnitude of Expiratory Effort from Pga Recordings

In each of the six subjects for whom simultaneous recordings were made of Pga and EMGTA, the expiratory increase in Pga (14) was closely related to the amplitude of the moving average of EMGTA (Figure 4), with correlation coefficients ranging from 0.70 to 0.95 (Figure 5). In contrast, the relationship between the decrease in Pga at the onset of inspiration (15) and the amplitude of the moving average of EMGTA was weaker, ranging from 0.04 to 0.53 (Figure 6). In the nine subjects, the expiratory increase in Pga, an indirect measurement of the magnitude of expiratory muscle activity, was not different before triggering and nontriggering attempts (p > 0.3) (Figure 7).

View larger version (14K): [in this window] [in a new window]   Figure 4.   Tracings of gastric pressure (Pga) and the rectified transversus abdominis EMG in a subject breathing through a Starling resistor while receiving PS of 10 cm H2O. The configurations of the two signals closely resemble each other.

View larger version (17K): [in this window] [in a new window]   Figure 5.   The expiratory increase in Pga (14) plotted against the amplitude of a moving average of the transversus abdominis EMG signal in six subjects breathing through a Starling resistor. The correlation coefficients ranged from 0.7 to 0.95 among the six subjects.

View larger version (16K): [in this window] [in a new window]   Figure 6.   The early inspiratory decrease in Pga (15) plotted against the amplitude of the moving average of the transversus abdominis EMG signal in six subjects breathing through a Starling resistor. The correlation coefficients ranged from 0.04 to 0.53 among the six subjects.

View larger version (14K): [in this window] [in a new window]   Figure 7.   In nine subjects the magnitude of expiratory effort (quantified as the expiratory increase in Pga) did not differ before triggering (closed bars) and nontriggering events (open bars) at each level of pressure support (p > 0.3). SEs are shown.

Elastic Recoil

At the onset of nontriggering attempts, gross PEEPi,dyn was greater than at the onset of triggering attempts, 34.8 ± 5.0% (p = 0.002) and 32.7 ± 11.1% (p = 0.03), at PS levels of 10 cm H₂O and 20 cm H₂O, respectively. Estimated elastic recoil, obtained by subtracting the increase in Pga from gross PEEPi (14), was greater for nontriggering attempts than for triggering attempts at all PS levels: 9.2 ± 1.0 cm H₂O and 2.8 ± 0.7 cm H₂O, respectively, at PS of 0 cm H₂O (p = 0.0001); 11.8 ± 1.4 cm H₂O and 4.2 ± 1.1 cm H₂O, respectively, at PS of 10 cm H₂O (p = 0.0004); and 14.1 ± 1.8 cm H₂O and 4.9 ± 1.3 cm H₂O, respectively, at PS of 20 cm H₂O (p = 0.0001).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The induction of airflow limitation in healthy subjects resulted in significant phase differences between the cycling of the subjects' expiratory muscles and that of the mechanical ventilator. With this model, the continuation of mechanical inflation into neural expiration was associated with failure of the subsequent inspiratory effort to trigger the ventilator. A delay in relaxation of the expiratory muscles did not directly interfere with the success of subsequent inspiratory efforts in triggering the ventilator. We also showed that the increase in Pga during expiration bears a close relationship to the magnitude of the increase in electrical activity of the expiratory muscles.

Timing and Magnitude of Expiratory Effort

The commencement of an expiratory effort before the off-switch of mechanical inflation (quantified in terms of phase angle) interfered with the ability of the next inspiratory attempt to trigger the ventilator (Figure 3). The continuation of mechanical inflation into neural expiration not only directly counters expiratory flow, but it also decreases the time available for unopposed expiratory flow. This leads to an increase in elastic recoil, which in turn necessitates a greater inspiratory effort to achieve effective triggering of the ventilator (22, 23).

The magnitude of the phase angle is a function of the ventilator's algorithm for the off-switch of inflation. With the Servo 900C ventilator, mechanical inflation ceases when a patient's inspiratory flow falls to 25% of the peak value. Accordingly, the time to reach the off-switch is determined by both the magnitude of the peak inspiratory flow and the rate of decrease in flow from its peak value. Since the magnitude of the decrease in inspiratory flow, in absolute terms, depends on the preceding peak value, more time should be required to reach the threshold off-switch when a breath has a high peak inspiratory flow. Our data confirm this reasoning, in that the peak inspiratory flow of breaths preceding nontriggering attempts was higher than the peak flow of breaths before triggering attempts at PS levels of 10 cm H₂O and 20 cm H₂O. This finding is in accord with our previous observation in critically ill patients receiving PS that breaths before nontriggering attempts had a higher VT (3). The time taken to reach the off-switch flow threshold is also determined by the rate of decrease in inspiratory flow, which in turn is determined by the net effect of the rate of increase in expiratory muscle activity and the braking action of postinspiratory inspiratory muscle activity. At PS levels of 10 cm H₂O and 20 cm H₂O, the slower rate of decrease in inspiratory flow for breaths before nontriggering attempts than for breaths before triggering attempts contributed to the greater phase angle in the former instance.

To safeguard against leaks, the Servo 900C ventilator has a second algorithm that permits exhalation when airway pressure exceeds the set level of PS (plus PEEP) by 3 cm H₂O (24). During PS, however, a large inspiratory effort causes the driving pressure to fall below the preset level (25). As a subject's inspiratory effort increases, the discrepancy between the effective level of PS and the preset level will increase proportionally, as will the change in airway pressure needed to reach the off-switch threshold for this alternative algorithm. Consequently, the time required to reach this off-switch pressure threshold will be prolonged after a large inspiratory effort, resulting in a greater phase angle. This mechanism is likely to have contributed to the greater phase angle observed before nontriggering attempts in our subjects, in that breaths preceding nontriggering attempts had a larger VT, implying a greater inspiratory effort. A separate, although related, observation was that the phase angle increased as a function of the preset level of PS for both triggering and nontriggering attempts (Figure 3). This increase in phase angle can be explained by the fact that the pressure threshold for the termination of mechanical inflation increases pari passu with the preset level of PS.

In a study of critically ill patients receiving PS of 20 cm H₂O, we noted that five of 12 patients with COPD appeared to recruit their expiratory muscles during mechanical inflation (7). In that study, however, activity of the expiratory muscles was inferred on the basis of Pes recordings and an estimate of chest wall recoil, and direct recordings of expiratory muscle electrical activity were not obtained. In the present study, wire electrodes inserted into the transversus abdominis muscle under ultrasound guidance provided clear proof that the expiratory muscles were activated during the inflation phase of PS in healthy subjects in whom we induced airflow limitation. We have also confirmed the presence of expiratory muscle recruitment, using EMGTA recordings, during the inflation phase of PS in a critically ill patient with COPD (Figure 8).

View larger version (27K): [in this window] [in a new window]   Figure 8.   Recordings of flow, airway pressure (Paw), and transversus abdominis EMG in a critically ill patient with COPD receiving PS of 20 cm H2O. The onset of expiratory muscle activity (vertical dotted line) occurred when mechanical inflation was only partly completed.

With reduction of end-expiratory lung volume below relaxation volume, the expiratory muscles can share in the work of inspiration (26). Airflow limitation prevents this reduction of lung volume and thus hampers such sharing of inspiratory work (12), and also impairs any influence of expiratory muscle activity on ventilator triggering. A delay in relaxation of the expiratory muscles could cause them to remain active during the early phase of the next inspiration, and by opposing the downward motion of the diaphragm could hinder the efficacy of the subsequent inspiratory effort. The average duration of inspiratory and expiratory muscle overlap was on the order of 60 to 90 ms in our subjects, with a range of 0 to 200 ms. Although this period is comparable to the time taken to trigger a ventilator of 100 to 200 ms (2), it did not appear to have any consistent effect on the success of triggering attempts in our subjects.

In the presence of airflow limitation, it is not clear whether the expiratory muscles have a beneficial or detrimental effect on the subsequent inspiratory effort. Some investigators have reported that the decrease in pleural pressure associated with expiratory muscle relaxation does not generate inspiratory flow (12). Other investigators have suggested that activity of the expiratory muscles during early inspiration is likely to stretch the diaphgramatic fibers, putting them at a more favorable position on their length-tension relationship (8). Moreover, at the point of switching between activity of the expiratory and inspiratory muscle groups, the rib cage can undergo large fluctuations in distorting forces, and continuation of expiratory muscle activity into early inspiration could stabilize the rib cage during this transition (9). Conversely, if expiratory muscle activity persists for too long into early inspiration, it could, through its deflationary action on the rib cage (27), oppose the downward motion of the diaphragm and decrease the efficiency of inspiratory efforts. Moreover, in the presence of dynamic hyperinflation and the concomitant decrease in the zone of apposition, expiratory muscle activity may no longer exercise an inflationary action on the lower rib cage (27). In our experimental model, the magnitude of expiratory muscle recruitment, and the subsequent relaxation, did not influence the success of triggering attempts. Conceivably, the beneficial effects of expiratory muscle recruitment (an increase in diaphragmatic fiber length and rib-cage stabilization) may have been nullified by the detrimental effects of an increase in end-expiratory alveolar pressure and rib-cage distortion.

Inspiratory Events

Nontriggering attempts were preceded by breaths having a longer neural TI than was the case for breaths preceding triggering attempts. This confirms our previous observations in critically ill patients, in whom we inferred neural timing on the basis of pressure recordings (3). Nontriggering attempts were also characterized by inspiratory efforts that were weaker (smaller inspiratory PTP) than was the case for triggering attempts. The lower inspiratory PTP could have resulted from an increase in lung volume with associated inspiratory muscle shortening, which decreases the effectiveness of inspiratory pressure generation. An increase in lung volume also increases vagal afferent traffic and, through activation of the Hering- Breuer reflex, results in a decrease in respiratory motor output (28) with a consequent decrease in inspiratory PTP. Additionally, the natural variability in the magnitude of respiratory effort and timing on a breath-to-breath basis (29) could result in an inspiratory PTP that is insufficient to trigger some breaths.

PEEPi and dynamic hyperinflation were higher at the onset of inspiratory attempts that did not trigger the ventilator than at the onset of successful triggering attempts. We previously demonstrated that a high PEEPi is an important cause of failure to trigger the ventilator in critically ill patients (3). In that study, however, we did not quantify the relative contributions of elastic recoil and expiratory muscle activity to failure of triggering. The present study showed that it is the increase in elastic recoil that is responsible for failure to trigger the ventilator, and that the component of PEEPi consisting of expiratory muscle effort has no effect on the success of triggering.

Estimation of Relative Contributions of Expiratory Effort and Elastic Recoil to PEEPi

In early descriptions of PEEPi, authors attributed the increase in alveolar pressure to gas trapping associated with dynamic hyperinflation (30, 31). Subsequently, Ninane and associates (12) pointed out that the positive alveolar pressure at end- expiration in patients with COPD is frequently due to transmission of an increase in abdominal pressure, resulting from abdominal muscle contraction, through the relaxed diaphragm. Two different approaches have been proposed for distinguishing between the contributions of expiratory muscle activity and those of increased elastic recoil pressure to PEEPi.

Lessard and coworkers (14) proposed that the contribution of expiratory muscle contraction to PEEPi could be estimated by measuring the increase in Pga between its end-inspiratory level and its maximal level at end-expiration (i.e., immediately before its decay during the subsequent inspiration). The assumption behind this method is that the diaphragm acts as a passive membrane during expiration; consequently, the effect of contraction of all expiratory muscles is reflected by an increase in Pga, which in turn is transmitted to the alveolar space. A theoretical problem with this method is that it ignores postinspiratory activity of the diaphragm, which will hinder the transmission of abdominal pressure. Appendini and associates (15) proposed an alternative method in which the negative deflection in Pga is subtracted from the negative deflection in Pes during the interval between the onset of an increase in Pdi and the onset of inspiratory flow. The assumption with this method is that the decrease in Pga is due solely to abdominal muscle relaxation. This method may also be in error, since a decrease in Pga resulting from expiratory muscle relaxation during early inspiration could be accompanied by an increase in Pga due to diaphragmatic contraction. The net change in Pga in this interval will be determined by the relative magnitude of these two processes, which cannot be accurately apportioned.

In healthy subjects breathing through a Starling resistor, Yan and colleagues (21) estimated static PEEPi by using the Campbell diagram technique, and then assessed the accuracy of the method of Lessard and associates (14). They concluded that the correction factor in the latter method (i.e., the expiratory increase in Pga) could accurately estimate the expiratory muscle contribution to PEEPi in some but not all subjects. These investigators did not evaluate the correction factor proposed by Appendini and coworkers (15). It should be noted that Yan and colleagues did not make direct measurements of expiratory muscle activity (21).

We compared the two proposed methodologies against carefully obtained electromyographic recordings of transversus abdominis activity. We found that the correction factor of Lessard and associates (the expiratory increase in Pga) correlated well with the amplitude of the moving average of EMGTA (Figure 5). In contrast, the correction factor of Appendini and coworkers (the early inspiratory decrease in Pga) showed a weaker correlation with EMGTA (Figure 6). The superior performance of the expiratory increase in Pga may have been attributable to the much longer time available to the expiratory muscles for achieving equilibration (the entire expiratory period) than was the case with the early inspiratory decrease of Pga; also, Pga does not necessarily reach its nadir before the onset of inspiratory flow. Furthermore, if the diaphragm is actively contracting at the time at which the expiratory muscles are relaxing, the decrease in Pga will be delayed and the magnitude of the decrease in Pga will be reduced.