INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Intrinsic positive end-expiratory pressure (PEEPi) due to dynamic hyperinflation is often present in patients with airway obstruction (1). Static PEEPi (PEEPi,st) is conventionally measured by an end-expiratory airway occlusion in which the end-expiratory plateau in airway pressure (Paw) reflects the elastic recoil pressure of the respiratory system (4). However, in spontaneously breathing, actively expiring patients, accurate assessment of the end-expiratory elastic recoil pressure is invariably confounded by increases in airway pressure related to expiratory muscle contraction (9, 10).

To estimate PEEPi,st under these circumstances, we recently proposed subtracting the average expiratory rise in gastric pressure (Pga), calculated from three breaths just prior to an airway occlusion, from the end-expiratory Paw of the first occluded inspiratory effort (11). An inherent assumption in this method is that expiratory muscle activity (expressed by the expiratory rise in Pga [Pga,exp rise] [5, 9]) prior to airway occlusion remains at approximately similar levels during the first breath postocclusion. However, we, as others (10), have observed that in some patients there is substantial variability in the intensity of expiratory muscle activity and hence in Pga,exp rise. In these patients, correction of PEEPi,st for expiratory muscle activity (11) may be inaccurate because the Pga,exp rise of breaths preceding airway occlusion may differ from that of the first postocclusion breath.

In this study, we introduce a new method of correcting PEEPi,st for expiratory muscle activity, which consists of synchronous subtraction of Pga,exp rise from Paw, both occurring during airway occlusion. The corrected PEEPi,st (referred to as PEEPi,st sub) is theoretically free of any confounding effect of expiratory muscle activity because the correction is based on simultaneous measurements of Paw and Pga. PEEPi,st sub and PEEPi,st measured by our previous method (11) (referred to as PEEPi,st avg) are each compared with the reference PEEPi,st obtained during muscular paralysis (i.e., complete absence of expiratory muscle activity) and simulation of the spontaneous breathing pattern by the ventilator (referred to as PEEPi,st ref).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Twenty-five critically ill patients (17 males; age, 63 ± 12 yr) with acute respiratory failure of different etiologies participated in the study. Ten patients had exacerbated chronic obstructive pulmonary disease, three had adult respiratory distress syndrome, four had cardiogenic pulmonary edema, two had postoperative respiratory failure, and six had pneumonia or other infectious disease. All were mechanically ventilated through a cuffed endotracheal (n = 19) or tracheostomy tube, and were selected on the basis of exhibiting abdominal muscle contraction during a brief trial of spontaneous breathing with a low level of pressure support (5 to 7 cm H₂O). Thirty-three patients were initially recruited based on clinical suspicion, but only 25 exhibited contraction of the abdominal muscles according to predefined criteria (i.e., rise of Pga during expiration associated with a decrease in abdominal cross-sectional area) and were included in the final analysis. The investigative protocol was approved by the institutional ethics committee and informed consent was obtained from the next of kin. On the day of the study, all patients were hemodynamically stable. All sedative and paralyzing medications were discontinued at least 4 h prior to the study. Mechanical ventilation was delivered by a Siemens 300 servo ventilator in the assist-control (A/C) mode with the settings prescribed by the primary physicians. Thirty minutes before the study, PEEP was removed and fraction of inspired oxygen (FI_O2) was increased to 1 in all patients for safety reasons.

Measurements

All measurements were made in the semirecumbent position. Flow () was measured with a heated pneumotachograph (No. 2; Fleisch, Lausanne, Switzerland) inserted between the endotracheal tube and the Y piece of the ventilator, and a differential pressure transducer (Model MP-45, ± 2 cm H₂O; Validyne, Northridge, CA). Tidal volume (VT) was obtained by integrating the flow signal. Esophageal (Pes) and Pga were measured with two balloons placed in midesophagus and stomach, respectively, according to standard techniques. Each balloon was connected to a differential pressure transducer (Model MP-45, ± 100 cm H₂O; Validyne). Appropriate placement of the esophageal balloon was verified by an occlusion test (12). Transdiaphragmatic pressure (Pdi) was assessed as the difference between Pga and Pes. Paw was recorded at the distal end of the endotracheal tube with a differential pressure transducer (Model MP-45, ± 100 cm H₂O; Validyne). Rib cage (RC) and abdominal (AB) displacements were measured with a respiratory inductive plethysmograph (Respitrace Ambulatory Monitoring, Ardsley, NY). The bands were placed circumferentially around the RC and AB in such a way that they were at the level of the nipples and umbilicus, respectively. The electrical activity of the abdominal muscles (EMGab) was obtained with surface electrodes placed on the right anterior axillary line, midway between the costal margin and the iliac crest and conditioned with a Nihon-Kohden electromyograph amplifier (band-pass between 20 Hz and 1 kHz). All signals were simultaneously recorded on an eight-channel recorder (Gould ES 1000; Gould Instruments, Cleveland, OH) and stored on a personal computer (Wyse 486) via analog-to-digital conversion using Labdat Software (Labdat/Anadat; RHT-Info Dat, Montreal, PQ, Canada). Analysis was made using the Anadat Software.

Simulation of Spontaneous Breathing and Measurement of PEEPi,st ref

Patients were initially allowed to breathe spontaneously through the ventilator with a pressure support of 5 to 7 cm H₂O. When expiratory muscle activity became apparent, the airway was occluded at the end of a normal expiration for about 8 to 12 s, using the end-expiratory hold knob of the ventilator. Mechanical ventilation on the A/C mode was then reinstituted, the patients were sedated (propofol, 2.5 mg/kg intravenously) and hyperventilated in order to abolish respiratory muscle activity, as judged by the absence of both Pes swings and EMGab activity, and by an airway pressure wave contour representative of passive inflation. Flow and volume tracings recorded during spontaneous breathing were analyzed in terms of VT, frequency (f), and duty cycle (TI/Ttot) to obtain the breathing pattern preceding the airway occlusion. With the patient ventilated with control mechanical ventilation and constant inspiratory flow, we then simulated the pattern of spontaneous breathing by regulating the appropriate ventilator buttons (13). Values of VT, f, and TI/Ttot used for simulation by the ventilator were the average of the last three consecutive spontaneous breaths prior to airway occlusion. Simulated mechanical breaths were accepted as representative of the breathing pattern during spontaneous ventilation if they were within ± 0.02 L for VT, ± 0.1 breaths for f, and ± 0.02 for TI/Ttot.

The PEEPi,st ref, determined during the simulation of spontaneous breathing pattern by the ventilator (13) and corresponding to the period of spontaneous breathing just preceding airway occlusion, was considered as the gold standard used for validation. The airway was occluded at the end of a tidal expiration using the end-expiratory hold button of the ventilator (Figure 1). End-expiratory plateau of Paw directly reflected PEEPi,st ref (14). Respiratory mechanics were also assessed during the simulation of spontaneous breathing by the constant flow end-inspiratory occlusion method (15).

View larger version (24K): [in this window] [in a new window]   Figure 1.   Tracings of Pes, Pga, Pdi, , and Paw in a representative patient, spontaneously breathing (SB) through the ventilator (pressure support, 5 cm H2O) after discontinuation from full mechanical ventilation. (A) Corresponds to the airway occlusion achieved by using the end-expiratory hold knob of the ventilator. Note that the change of Pdi during expiration is zero, suggesting that the diaphragm remains relaxed between the occluded inspiratory efforts. (B) With the patient relaxed, the pattern of SB immediately preceding the airway occlusion was simulated by the ventilator. PEEPi,st ref was measured by the end-expiratory occlusion technique (14) as the value of end-expiratory plateau of airway pressure under static conditions. Pga,exp rise is the increase in gastric pressure during expiration from its minimal end-inspiratory level to the maximal level at end-expiration.

Correction of PEEPi,st for Expiratory Muscle Activity

Correction of PEEPi,st for expiratory muscle activity was made by two methods, both using the airway occlusion. With our recently reported method (11), PEEPi,st was corrected by subtracting the average Pga,exp rise of the last three breaths preceding airway occlusion from the end-expiratory Paw of the first occluded inspiratory effort (PEEPi,st avg). With the present method, PEEPi,st was corrected by subtracting Pga from Paw during airway occlusion (PEEPi,st sub). Only the absolute increase in Pga from its lowest value (i.e., immediately after the end of inspiration where expiratory muscle activity was nil) was subtracted from the Paw, because this represents expiratory muscle activity. In other words, the amount of Pga at this point was given the value of zero irrespective of its true value relative to atmospheric pressure. The resulting airway pressure tracing was accepted as corrected for expiratory muscle activity (cPaw) only if it exhibited an end-expiratory plateau between the occluded inspiratory efforts. From this plateau we then measured the PEEPi,st sub (Figure 2). We accepted plateaus that remained almost constant and horizontal for 50% or more of the expiratory phase, whereas PEEPi,st sub was measured during the expiratory phase of the first and /or second occluded effort.

View larger version (34K): [in this window] [in a new window]   Figure 2.   Recordings of Pes, , Paw, Pga, "suppressed" gastric pressure (sPga), and on line "corrected" airway pressure (cPaw) in a representative actively expiring patient during airway occlusion. sPga is the Pga tracing suppressed to zero at its lowest value at the end-inspiratory level ( vertical line) where expiratory muscle activity is nil, i.e., at this level it was given the value of zero irrespective of its true value relative to atmospheric pressure. cPaw is obtained by subtracting sPga from Paw. Note a consistent end-expiratory plateau in cPaw despite marked variability in Pga swings. From this plateau PEEPi,st sub was measured.

Data Analysis

Abdominal muscle contraction was assessed by the rise of Pga from its minimum end-inspiratory level to the maximal level at end-expiration (Pga,exp rise) and its subsequent abrupt decrease at the beginning of the next inspiration (Figure 1). This pattern, associated with a decrease in abdominal cross-sectional area during expiration and a subsequent increase during the ensuing inspiration, yields a characteristic Pga-AB displacement loop (5, 9) and clearly indicates contraction of the abdominal muscles. The intensity of expiratory muscle contraction was assessed by averaging Pga,exp rise values of three consecutive breaths immediately preceding airway occlusion. The same three values of Pga,exp rise were also used to assess the variability of expiratory muscle contraction, which was expressed as the coefficient of variation.

All data are presented as mean ± SD. Statistical analysis was performed using Student's paired t test and linear regression analysis. A p value at 0.05 level was considered significant. The agreement between PEEPi,st sub and PEEPi,st ref, as well as between PEEPi,st avg and PEEPi,st ref, were evaluated by the method of Bland and Altman (16). The degree of agreement is summarized by calculating the mean difference and the SD of the differences.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For the entire patient group, static compliance of the respiratory system was 0.056 ± 0.016 L/cm H₂O, and the total and minimal (airway) respiratory system resistance (after subtraction of the endotracheal tube resistance [15]) was 19.8 ± 6.7 and 13.9 ± 6.1 cm H₂O/L/s, respectively. Breaths simulated by the ventilator were almost identical to those recorded during spontaneous breathing (0.38 ± 0.05 versus 0.38 ± 0.04 L for VT, 29 ± 5 versus 29 ± 5 breaths/min for f, and 0.32 ± 0.02 versus 0.32 ± 0.02 for TI/Ttot, respectively).

Figure 2 shows a representative cPaw tracing obtained after airway occlusion and subtraction of Pga from Paw. In all patients, cPaw had a contour quite similar to that found in patients with passive expiration, i.e., an end-expiratory plateau between the occluded inspiratory efforts was present (11). A clear plateau in cPaw could be identified in as many as 60% of respiratory efforts against the occluded airway. The mean value of PEEPi,st sub (5.3 ± 2.6 cm H₂O) was nearly identical to that of PEEPi,st ref (5.4 ± 2.4 cm H₂O, p > 0.1). PEEPi,st sub highly correlated with PEEPi,st ref (r = 0.986; p < 0.001, Figure 3A). Bland and Altman analysis showed that the mean difference of PEEPi,st sub and PEEPi,st ref was 0.06 cm H₂O and the limits of agreement (i.e., mean difference 2 SD to mean difference +2 SD) were 0.96 to 0.84 cm H₂O, indicating a strong agreement between these methods (Figure 4A).

View larger version (16K): [in this window] [in a new window]   Figure 3.   Relationships between PEEPi,st sub and PEEPi,st ref (A) and between PEEPi,st avg and PEEPi,st ref (B) in 25 actively expiring patients. PEEPi,st sub = static PEEPi assessed by the method introduced in the present study, as described in Figure 2 (see also text); PEEPi,st avg = static PEEPi measured by the recently reported method (11), i.e., by subtracting the average Pga,exp rise of the last three breaths preceding airway occlusion from the end-expiratory airway pressure of the first occluded inspiratory effort; PEEPi,st ref = reference static PEEPi computed as the value of end-expiratory plateau of airway pressure during simulation by the ventilator of the pattern of spontaneous breathing preceding airway occlusion, as described in Figure 1; solid line: identity line; dotted line: regression line. In A, all values are on or very close to the line of identity, indicating that PEEPi,st sub is identical to PEEPi,st ref. In contrast, several values are away from the line of identity in B.

View larger version (19K): [in this window] [in a new window]   Figure 4.   Bland and Altman plots of the differences between PEEPi,st sub and PEEPi,st ref (A) and PEEPi,st avg and PEEPi,st ref (B) against their average (mean) values, respectively. d = mean difference; d + 2SD = upper limit of agreement; d  2SD = lower limit of agreement. The mean difference between PEEPi,st sub and PEEPi,st ref was 0.06 cm H2O and the limits of agreement (i.e., mean difference 2 SD to mean difference +2 SD) were 0.96 to 0.84 cm H2O, indicating a strong agreement between these methods. This signifies that the two methods can be used interchangeably, and consequently PEEPi,st sub is a very accurate estimate of the true static PEEPi. In contrast, the mean difference between PEEPi,st avg and PEEPi,st ref was 0.73 cm H2O with limits of agreement 3.97 to 5.43 cm H2O, indicating lack of agreement between the two methods. For definitions, see legend to Figure 3.

PEEPi,st avg (6.1 ± 3.4 cm H₂O) was comparable to PEEPi,st ref (5.4 ± 2.4 cm H₂O, p > 0.1) and they were also correlated (r = 0.728; p < 0.001, Figure 3B); Bland and Altman analysis showed that their mean difference was 0.73 cm H₂O with limits of agreement 3.97 to 5.43 cm H₂O, indicating lack of agreement between the two methods (Figure 4B). The mean Pga,exp rise was 7.1 ± 3.3 cm H₂O (range, 2.6 to 13 cm H₂O). The mean coefficient of variation of Pga,exp rise was 14.3 ± 7.2% (range, 5.2 to 28.3%). There was a good correlation between the coefficient of variation of Pga,exp rise and the difference between PEEPi,st avg and PEEPi,st ref (r = 0.909; p < 0.001, Figure 5A). In contrast, there was no correlation between the coefficient of variation of Pga,exp rise and the difference of PEEPi,st sub and PEEPi,st ref (r = 0.129; p > 0.1, Figure 5B).

View larger version (17K): [in this window] [in a new window]   Figure 5.   Plots of the variability of Pga,exp rise against the differences between PEEPi,st avg and PEEPi,st ref (A) and PEEPi,st sub and PEEPi,st ref (B), respectively. When the variability of Pga,exp rise is small (< 13 to 15%) there are in essence no differences between PEEPi,st sub and PEEPi,st ref and PEEPi,st avg and PEEPi,st ref, respectively, which indicates that both methods of correcting static PEEPi for expiratory muscle activity are accurate. When the variability of Pga,exp rise is high, the difference between PEEPi,st sub and PEEPi,st ref remains close to zero. In contrast, the difference between PEEPi,st avg and PEEPi,st increases in direct proportion to the increase in the variability of Pga,exp rise. Consequently, only PEEPi,st sub is an accurate estimate of the static PEEPi in the presence of increased variability in the intensity of expiratory muscle contraction.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study shows that in spontaneously breathing, intubated patients with variable expiratory muscle contraction, PEEPi,st can be accurately measured by performing an airway occlusion at FRC and subtracting Pga from Paw. The pressure tracing obtained after subtraction of Pga from Paw has a consistent end-expiratory plateau between occluded inspiratory efforts and essentially reflects the true elastic recoil pressure of the respiratory system. This method is valid in patients with various diseases and a wide range of PEEPi,st, and independent of breath-to-breath variability of expiratory muscle recruitment.

Contraction of expiratory muscles during expiration interferes with accurate assessment of PEEPi,st (9, 10). This may be especially important in mechanically ventilated patients with dynamic hyperinflation, in whom accurate estimates of PEEPi,st are often helpful for managing hemodynamic or respiratory problems (1). For this particular group of patients, Lessard and colleagues (10) first suggested that measurements of Pga,exp rise could be used as a gross estimate of the pressure needed to correct PEEPi. In the absence of a validation method, however, the exact amount of Pga,exp rise transmitted into the airways could not be defined. In our previous study (11), we found that the difference of an average Pga,exp rise value from Paw during airway occlusion is very similar to end-expiratory elastic recoil pressure measured during muscular paralysis and simulation of the spontaneous breathing pattern. Our present study included patients with marked variability in the expiratory muscle contraction (mean, 14.3 ± 7.2%; range, 5.2 to 28%) and showed that, in this setting, concomitant subtractions of Pga,exp rise from Paw (i.e., PEEPi,st sub) give closer estimates of PEEPi,st ref than average Pga,exp rise values (i.e., PEEPi,st avg). Indeed, the agreement between PEEPi,st avg and PEEPi,st ref was not good and moreover there was a very strong correlation between the variability of Pga,exp rise and the difference between PEEPi,st avg and PEEPi,st ref, which indicates that as variability in the intensity of expiratory muscle contraction increases, so does the inaccuracy of estimation of PEEPi,st with the application of the previously described method (11). In contrast, when variability in the intensity of expiratory muscle contraction is small (< 13 to 15%), the difference of PEEPi,st avg  PEEPi,st ref approximates zero, and thus the previously described method is quite accurate (Figure 5A).

The physiologic background of the present method was based on our previous observation (11) that during the expiratory phase of respiratory efforts against an occluded airway, the diaphragm remains relaxed and consequently any increase in Pga caused by expiratory muscle contraction is freely transmitted to alveolar pressure and thence to Paw. Therefore, subtracting this transmitted Pga from Paw during airway occlusion would correct Paw for expiratory muscle activity, thus exhibiting an end-expiratory plateau between occluded inspiratory efforts reflecting the elastic recoil pressure of the respiratory system, i.e., PEEPi,st. This was confirmed in the present study where in every patient, after a very short (< 0.25 s) period of postinspiratory diaphragmatic activity, the diaphragm remained virtually relaxed throughout expiration (Figure 1A). Moreover, Paw exhibited an end-expiratory plateau (PEEPi,st sub) after subtraction of the increase of Pga caused by expiratory muscle contraction (Pga,exp rise) (Figure 2). The strong agreement observed between PEEPi,st sub and PEEPi,st ref (Figure 4A) indicates that the newly introduced method leads to a very accurate correction of PEEPi,st for expiratory muscle activity. Furthermore, the variability of Pga,exp rise does not affect at all the accuracy of the new method, as evidenced by the lack of correlation between the coefficient of variation of Pga,exp rise and the difference between PEEPi,st sub and PEEPi,st ref (Figure 5B), which was anyway very small. This was expected on theoretical grounds because the new method, contrary to the previous one, uses the same respiratory effort against an occluded airway to correct for expiratory muscle activity, and consequently no matter how intense or variable this activity is, it is always being subtracted from the same Paw tracing that this activity influences; thus, any influence of the expiratory muscle activity on the Paw is automatically being removed.

Theoretical considerations that could potentially question the accuracy of this method should be pointed out. First, at the end of inspiration, as the diaphragm relaxes Pga drops rapidly and starts to rise again when the abdominal muscles become active during expiration. If the time available at end-inspiration for Pga to return to its relaxation value is insufficient, then the mechanical effect of the abdominal muscles on Paw might be underestimated by the increase in Pga observed between the end-inspiratory and peak end-expiratory values (17). Thus, in this instance, PEEPi,st sub would overestimate the elastic recoil pressure of the respiratory system (PEEPi,st). As in our previous study (11), this end-inspiratory increase in Pga above its relaxation value was most apparent in seven patients with strong expiratory muscle activation (Pga,exp rose 9.3 to 13 cm H₂O). Second, if the diaphragm is not fully relaxed during expiration then the increase in Pga may overestimate the expiratory muscle recruitment affecting Paw. In the abovementioned seven patients with the highest values of Pga,exp rise, expiratory Pdi baseline during the airway occlusion maneuver was higher (1.2 ± 0.5; range, 0.6 to 2.1 cm H₂O) than that during the beginning of spontaneous breathing trial where expiratory muscle activity was almost nil; this probably reflects a passive Pdi due to strong abdominal muscle recruitment (7). When expiratory Pdi is increased, the increase in Pga during expiration may not be transformed into an equal increase in Pes and, thereby, in Paw. Therefore, subtracting Pga from Paw would give values of PEEPi,st sub that would underestimate the elastic recoil pressure of the respiratory system. Finally, persistent inspiratory rib cage muscle activity during expiration would attenuate the expiratory muscle effect on Paw and would result in underestimation of PEEPi,st by PEEPi,st sub. Our design does not allow us to comment on the likelihood of persistent inspiratory rib cage muscle activity during expiration. Nevertheless, we found that in all patients PEEPi,st sub was nearly identical to PEEPi,st ref regardless of the intensity of expiratory muscle activity (Figures 3A and 4A). The finding that PEEPi,st sub accurately reflects PEEPi,st ref even when the expiratory muscles are strongly activated is probably due to the coexistence of the abovementioned conditions, which at the same time tend to over- and underestimate the PEEPi,st by the PEEPi,st sub measurement; by offsetting the two effects, PEEPi,st sub accurately reflects PEEPi,st ref.

Recently, Purro and coworkers (18) measured PEEPi,st by modifying the end-expiratory occlusion technique. However, their method necessitates the use of both gastric and esophageal balloons, which is technically demanding and is valid only if the occlusion test used to assess the placement of the esophageal balloon is satisfactory. The latter is not always the case, especially in critically ill intubated patients. Furthermore, this method requires the performance of a Mueller maneuver, which requires the patient's cooperation and coordination that are often absent in such patients. Finally, the method of Purro and coworkers was validated only in patients with severe chronic airway obstruction. In contrast, our method requires only the insertion of a gastric balloon, which is easy to do and does not require validation. In addition, patient coordination and cooperation are not needed, and the method has been validated in patients suffering from various diseases with a wide range of severity.

The potential clinical implication of our findings is mainly related to the accurate measurement of PEEPi,st in spontaneously breathing and actively expiring intubated patients with airflow limitation in order to apply the appropriate level of external PEEP or continuous positive airway pressure (CPAP) to decrease inspiratory muscle effort due to PEEPi (4) and improve patient-ventilator interaction (6). Because the application of a level of PEEP/CPAP that is less than PEEPi,st leads to inadequate reduction of the inspiratory muscle effort, and because the application of excessive PEEP/CPAP causes further hyperinflation, accurate measurement of PEEPi,st is clinically fundamental. Moreover, correct measurement of PEEPi,st is important in the hemodynamic status assessment (1) as well as in the estimation of the load imposed on the inspiratory muscles that may determine the ability to wean from mechanical ventilation (13).