INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dynamic hyperinflation and intrinsic positive end-expiratory pressure (PEEP_i) are frequently encountered in patients, especially those with severe airway obstruction (1). In mechanically ventilated patients PEEP_i is routinely measured under static conditions (PEEP_i,st) as the plateau in airway pressure during a prolonged end-expiratory airway occlusion (1). In actively breathing patients, either during spontaneous or assisted ventilation, PEEP_i has been assessed dynamically from records of esophageal pressure (Pes). The decrease in Pes needed to abruptly bring expiratory flow to zero during unoccluded breathing is taken as dynamic PEEP_i (PEEP_i,dyn) (2). The presence of PEEP_i has a number of clinical implications (1, 2) and its accurate measurement is important. However, expiratory muscle activity can increase the end-expiratory alveolar pressure independently of dynamic hyperinflation, leading to an overestimation of PEEP_i,dyn (3). Indeed, in this case part of the decrease in Pes preceding inspiration, which is measured as PEEP_i,dyn, is actually due to relaxation of the expiratory muscles rather than contraction of the inspiratory muscles to counterbalance PEEP_i.

The amount of pressure due to expiratory muscle activity that should be subtracted from the measured PEEP_i,dyn to obtain the actual ("true") PEEP_i,dyn elicited by dynamic hyperinflation remains unclear. Lessard and coworkers (5) proposed to subtract from the measured PEEP_i,dyn either the increase in gastric pressure over the course of expiration or the total decrease in gastric pressure observed at the beginning of inspiration attributed to expiratory muscle contraction and relaxation, respectively. In contrast, Appendini and coworkers (4) subtracted only a part of the decrease in gastric pressure measured at the beginning of inspiration, that is, that encompassed between the onset of inspiratory effort and the point of zero flow. In both studies (4, 5) the corrections proposed could not be validated because a gold standard measurement of PEEP_i,dyn was lacking, that is, the actual magnitude of PEEP_i,dyn could not be assessed. Yan and coworkers (6) and Parthasarathy and coworkers (7) have partly addressed this issue in normal subjects that became flow limited by expiring through Starling resistors, either directly by constructing Campbell diagrams (6) or indirectly by obtaining EMGs of abdominal muscles (7). However, these investigators have not examined all proposed corrections. Furthermore, the passive mechanics of the lung and chest wall are often much different in patients as compared with normal subjects, whereas the flow limitation provided by the Starling resistor is constant throughout expiration as contrasted to what happens in patients (6). Thus, the extent to which these observations in normal subjects are applicable to patients remains to be studied.

The purpose of our study was to extend these observations in patients and investigate which of the above methods of correcting PEEP_i,dyn for expiratory muscle contraction is more accurate. Validation was accomplished by comparing the results of these corrections to a reference PEEP_i,dyn (PEEP_i,dyn ref), obtained using the Campbell diagram (6). The study was performed in spontaneously breathing patients with acute respiratory failure, after temporary discontinuation and resumption of mechanical ventilation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Fifteen patients (eight males) with acute respiratory failure of different etiologies (Table 1) participated in the study. In nine patients with chronic obstructive pulmonary disease (COPD) diagnosis was based on previous clinical history and lung function tests (FEV₁ of ± 0.85 ± 0.24 L and FVC of 1.72 ± 0.40 L). Three patients with adult respiratory distress syndrome (ARDS) met conventional diagnostic criteria. The investigative protocol was approved by the institutional ethics committee and informed consent was obtained from the next of kin. Patients were mechanically ventilated through either a cuffed endotracheal (n = 11) or tracheotomy tube, and were clinically stable (systolic blood pressure above 100 mm Hg; heart rate less than 120 beats/ min; no significant fluctuations in blood pressure, heart rate, arterial blood gases, urine output, or mental status) on the day of the study. Patients were selected on the basis of exhibiting abdominal muscle contractions during a brief weaning trial with a low level of pressure support (5-7 cm H₂O). Twenty patients were initially recruited based on clinical assessment, but only 15 exhibited contraction of the abdominal muscles according to predefined criteria (see below) and were included in the study. All sedative and paralyzing medications were discontinued at least 6 h prior to the study. Mechanical ventilation was delivered by a Siemens 300 servo ventilator in the assist-control (A/C) mode with the ventilator settings prescribed by the primary physicians. Thirty minutes before the beginning of the study PEEP was removed and fraction of inspired oxygen (FI_O2) was increased to 1 in all patients for safety reasons. Baseline partial pressures of arterial O₂ and CO₂ were 151 ± 48 and 47 ± 9 mm Hg, respectively. During the study a physician not involved in the procedure was always present to provide for patient care. The electrocardiogram, the heart rate, the systemic arterial blood pressure, and the arterial O₂ saturation (Nellcor, CA) were continuously monitored.

Measurements

All measurements were made with the patients in a semirecumbent position. Airflow () was measured with a heated pneumotachograph (Fleisch no. 2; Fleisch, Lausanne, Switzerland) inserted between the endotracheal tube and the Y piece of the ventilator, and a differential pressure transducer (Validyne MP-45, ± 2 cm H₂O; Validyne, Northridge, CA). Tidal volume (VT) was obtained by integrating the flow signal. Esophageal (Pes) and gastric (Pga) pressures were measured with conventional balloon-catheter systems placed in the midesophagus and the stomach, respectively; the esophageal balloon was filled with 0.5 ml of air, and the gastric balloon contained 1 ml of air. Both balloons were connected to separate differential pressure transducers (Validyne MP-45, ± 100 cm H₂O). Appropriate placement of the esophageal balloon was verified by the occlusion test (8). Airway pressure (Paw) was recorded at the distal end of the endotracheal tube with a differential pressure transducer (Validyne MP-45, ± 100 cm H₂O). 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 recorded with surface electrodes placed in 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, except EMGab, were continuously recorded on an eight-channel electrostatic recorder (Gould ES 1000; Gould Instruments, Cleveland, OH) at a paper speed of 10 or 25 mm/s and taped on a video recorder (including EMGab) via an analog-to-digital converter. The data were played back to a personal computer (Wyse 486) by the same analog-to-digital converter at a sampling rate of 1500 Hz (EMGab) or 200 Hz (other variables) for subsequent data analysis.

Simulation of Spontaneous Breathing and Measurement of PEEP_i,dyn and PEEP_i,st

Patients were initially allowed to breathe spontaneously through the ventilator with a small inspiratory assistance (pressure support of 5-7 cm H₂O) to compensate for the additional work due to the endotracheal tube and the inspiratory circuit (9). Fifteen to 60 min (33 ± 14) after the beginning of spontaneous breathing, Pa_CO2 increased  14 mm Hg in all patients (24 ± 9 mm Hg) and reinstitution of mechanical ventilation on A/C was required. In every patient arterial O₂ saturation remained higher than 90% throughout the spontaneous breathing trial. At the end of the trial Pa_O2 was 132 ± 43 mm Hg. As soon as mechanical ventilation on the A/C mode was reinstituted, the patients were sedated (midazolam) and some were also paralyzed (pancuronium bromide). After an initial bolus administration of these drugs (15 mg midazolam, 0.06 mg/kg pancuronium bromide), the patients continued receiving the same sedative and paralyzing medication in order to maintain respiratory muscle relaxation (judged by lack of inspiratory swings of Pes, airway pressure wave contour representative of passive inflation, and clinical assessment). Recordings of and volume during spontaneous breathing were analyzed in terms of VT, frequency (f), and duty cycle (TI/TT) to obtain the breathing pattern. Two periods of time were selected for analysis in every patient: (1) at the middle of the spontaneous breathing trial and (2) 1 min before reinstitution of mechanical ventilation, at the end of the spontaneous breathing trial. The data were the means of three consecutive breaths. With the patient ventilated on control mechanical ventilation and constant inspiratory flow, we subsequently simulated these two patterns of spontaneous breathing in each patient by regulating the appropriate buttons of the ventilator (10). Inclusion criteria for accepting simulated breaths as representative of the breathing pattern during spontaneous ventilation were deviations of less than ± 0.02 L for VT, ± 0.1 breaths for f, and ± 0.02 for TI/TT. To ensure that the mechanical characteristics of the lung did not change from the middle to the end of the trial and thus passive mechanics during the simulation immediately after the end of the trial are representative of passive mechanics at the middle of the trial we measured dynamic lung compliance (CdynL) and lung resistance (RL) (11) at these moments.

During spontaneous breathing both PEEP_i,dyn and PEEP_i,st were measured with previously described methods (2, 10). PEEP_i,dyn was measured as the negative deflection of Pes from the onset of inspiratory effort to the point of zero flow (2) (Figure 1). Values of PEEP_i, dyn were the average of three consecutive breaths. PEEP_i,dyn was measured twice in each patient: at the middle and at the end of the spontaneous breathing trial. The three consecutive breaths used to measure PEEP_i,dyn were analyzed to obtain the spontaneous breathing pattern, as previously described. PEEP_i,st was measured with the patient relaxed during the simulation of spontaneous breathing by the ventilator (10). The airway was occluded at the end of a tidal expiration (Figure 1), and the end-expiratory plateau of Paw was taken as PEEP_i,st (1). With the ventilator we simulated two spontaneous breathing patterns in each patient and hence we obtained two values of PEEP_i,st: one corresponding to the middle and the second to the end of the spontaneous breathing trial. During each simulation, the patient was disconnected from the ventilator after a normal inflation and was allowed to expire freely until expiratory flow became nil and the elastic equilibrium volume of the respiratory system (Vr) had been reached (10). The difference between the inspired and the expired volume represents the increase in FRC due to PEEP_i (FRCsim), which equals to the difference between the end-expiratory lung volume (EELV) during the simulation with controlled ventilation and the Vr. The value of FRCsim was calculated as the mean of two or three measurements.

View larger version (30K): [in this window] [in a new window]   View larger version (29K): [in this window] [in a new window]   Figure 1.   (Left panel ) Tracings of esophageal pressure (Pes), gastric pressure (Pga), transdiaphragmatic pressure (Pdi), flow (), airway pressure (Paw), and cross-sectional area of the rib cage (RC) and the abdomen (AB) in a representative patient (No. 1, Table 1) actively expiring during spontaneous breathing through the ventilator (pressure support of 6 cm H 2O). After discontinuation from full mechanical assistance, the patient was unable to sustain spontaneous breathing and resumption of mechanical ventilation was needed. (A) The beginning of spontaneous breathing (SB). (B) The end of spontaneous breathing, 1 min before the reinstitution of control mechanical ventilation. At this period of time the breathing pattern was analyzed and subsequently simulated by the ventilator (C ) with the patient relaxed. PEEPi,st was obtained during the simulation with control mechanical ventilation by the end-expiratory occlusion technique (1). The three vertical lines are passed through the onset of inspiratory muscle activity (i.e., beginning of Pes decay) and the beginning and the end of inspiratory flow, respectively. In contrast to the beginning of spontaneous breathing (A), note the large increase of Pga and Pes during expiration due to expiratory muscle recruitment at the end of spontaneous breathing trial (B). Also note the inward displacement of RC in phase with an outward displacement of AB during the initial part of inspiratory effort suggesting chest wall distortion. (Right panel ) Magnification of the left panel area encompassed by the parallelogram. Pes is the increase in end-expiratory Pes relative to its level at the onset of spontaneous breathing trial, when expiratory muscle activity was nil. Pga,total decay represents the abrupt decrease in Pga from its maximum end-expiratory value to its minimum value at the beginning of inspiration due to relaxation of the abdominal muscles. Pga,zf decay is the part of Pga decay during inspiration restricted between the onset of inspiratory effort and the point of zero flow. Pga,exp rise is the Pga rise from its minimal end-inspiratory level to the maximal level at end expiration. The decay in Pes between the first two vertical lines represents dynamic intrinsic positive end-expiratory pressure (PEEPi,dyn). For further explanation, see text.

Respiratory mechanics (Table 1) were assessed with the patient ventilated in control mode during simulation of the spontaneous breathing pattern by using the constant flow end-inspiratory occlusion method previously described in detail (12).

Correction of PEEP_i,dyn for Expiratory Muscle Activity and Measurement of PEEP_i,dyn ref

The measured PEEP_i,dyn was corrected for expiratory muscle activity by subtracting the increase in Pga during expiration (Pga,exp rise), the whole decay of Pga during inspiration (Pga,total decay) (methods of Lessard and coworkers [5]) or the part of Pga decay encompassed between the onset of inspiratory effort and the point of zero flow (Pga,zf decay) (method of Appendini and coworkers [4]).

Validation of these methods was performed by comparison with the reference PEEP_i,dyn (PEEP_i,dyn ref), which was measured by using the Campbell diagram (6). The latter was constructed from the passive compliance lines of the chest wall (Ccw) and the lung (CL,dyn) which were obtained by dynamically plotting of VT against Pes and transpulmonary pressure (PL = Pes  Paw), respectively, during simulation with control mechanical ventilation. The two points needed to construct the compliance lines of lung and chest wall corresponded to the zero flow values at the onset and end of inspiration of PL and Pes, respectively. To determine the actual passive chest wall compliance line below the EELV during controlled ventilation, volume was plotted against Pes during the maneuvers in which the patient was allowed to expire passively to Vr. CL,dyn was assumed to be linear and was extrapolated to Vr. The clockwise dynamic V-Pes loop was then plotted on the Campbell diagram. The loop was constructed by averaging three consecutive breaths at the middle or at the end of the spontaneous breathing trial. The same three breaths were also used to obtain the breathing pattern that was simulated by the ventilator. CL,dyn was used to position the dynamic V-Pes loop on the volume axis of the Campbell diagram, that is, the value of PL at the onset of inspiration (zero flow) was measured during spontaneous breathing and the loop was positioned so that this value fell on the CL,dyn. To validate the correct positioning of the dynamic V-Pes loop on the Campbell diagram, end-inspiratory PL (at zero flow) was measured during spontaneous breathing and it was compared with the corresponding value of the lung line, that is, that at the point where the dynamic loop intercepts the lung line of the Campbell diagram. An end-inspiratory PL value ± 1 cm H₂O of the lung line value was considered as acceptable indicating appropriate positioning of the loop. To assess the possible effect of expiratory muscle contraction on EELV during spontaneous breathing, we measured from the Campbell diagram the difference between the EELV and the Vr (FRCcamp) and compared it with the FRCsim. An important feature of the Campbell diagram is that it allows assessment of respiratory muscle recruitment at different times of a breathing cycle, which is of particular importance for measurement of PEEP_i,dyn (6). If the respiratory muscles are relaxed, Pes should lie along the Ccw. If the end-expiratory Pes lies to the right of Ccw, it indicates contraction of expiratory muscles during expiration. In contrast, if the end-expiratory Pes lies to the left of the Ccw, tonic inspiratory muscle activity during expiration is present (6, 13, 14). In all patients of the present study, PEEP_i,dyn ref was measured assuming that there was no inspiratory muscle activity during expiration. Therefore, PEEP_i,dyn ref was measured on the dynamic V-Pes loop as the horizontal distance between the point where inspiratory flow starts and the Ccw line.

A representative diagram obtained from a patient with strong expiratory muscle recruitment is shown in Figure 2. In this figure the V- Pga loop plotted on the relaxation compliance line of the abdomen (Cab) is also shown, which graphically illustrates the different corrections of PEEP_i,dyn for expiratory muscle contraction made in the present study. The patient had COPD and her tracings are those presented in Figure 1. This is an illustrative example of a case of dynamic hyperinflation on the Campbell diagram. Due to dynamic hyperinflation the EELV is 0.32 L above Vr. The values of Pes during expiration (from point D to C) lie to the right of Ccw, indicating expiratory muscle contraction during expiration. The pressure difference between points C and I represents the measured PEEP_i,dyn. The pressure decay between the points C and B represents the expiratory muscle relaxation during the onset of inspiration that should be subtracted from measured PEEP_i,dyn to correct for expiratory muscle contribution. Therefore, PEEP_i,dyn ref is equal to the pressure difference between I and B. The V-Pga loop is typical for expiratory muscle recruitment during expiration due to the increasing positive Pga at the same time that abdominal volume was decreasing during expiration. Cab was composed by plotting Pga against abdominal volume during controlled mechanical ventilation (Figure 1C), and was assumed to be linear. Abdominal volume was indirectly obtained from the abdominal band of the Respitrace. After the beginning of expiration (E) abdominal volume abruptly decreases and Pga also decreases to its "nadir" value (H). Afterward Pga progressively increases at the same time that abdominal volume decreases until the point at which inspiratory effort starts (C). From this point on, relaxation of the abdominal muscles begins and Pga abruptly decreases at the same time that abdominal volume increases until the start of inspiratory flow (I). Subsequently, after the onset of inspiratory flow Pga further decreases (F) and then increases until the end of inspiration (E). From I to E abdominal volume sharply increases. Pga,exp rise is the pressure difference between points C and H, and Pga,total decay is the pressure difference between points C and F. Both of them were proposed by Lessard and coworkers (5) as the amount of pressure that could be subtracted from measured PEEP_i,dyn to correct for expiratory muscle contribution. In turn, Pga,zf decay recommended by Appendini and coworkers (4) as the pressure that should be subtracted from measured PEEP_i,dyn is the pressure difference between points C and I. 

View larger version (16K): [in this window] [in a new window]   Figure 2.   (Left panel ) Average lung volume-esophageal pressure (V-Pes) loop of a representative patient with strong expiratory muscle activity at end of spontaneous breathing trial. Ccw = passive compliance line of the chest wall; CL,dyn = passive compliance line of the lung; Vr = relaxation volume of the respiratory system. B = Ccw at end-expiratory lung volume; E and D = CL,dyn and Ccw at end-inspiratory lung volume, respectively; I = Pes at onset of inspiratory flow; C = onset of inspiratory muscle activity and relaxation of expiratory muscles. (Right panel ) Average abdominal volume-gastric pressure (V-Pga) loop. It was constructed from the same breaths as the V-Pes loop. Cab = relaxation compliance line of the abdomen. I and E = Pga at onset and end of inspiratory flow, respectively; C = onset of inspiratory muscle activity and relaxation of abdominal muscles; F = lowest value of Pga during inspiration; H = lowest value of Pga during expiration. For further explanation, see text.

Data Analysis

Expiratory muscle activity was assessed three ways: (1) By the rise of Pga from its minimum end-inspiratory level to the maximum level at end expiration (Pga,exp rise) and its subsequent abrupt decrease at the beginning of next inspiration from its maximum to its minimum level (Pga,total decay) (Figure 1). This pattern, associated with a decrease in abdominal cross-sectional area during expiration and with a subsequent increase during the ensuing inspiration, gives characteristic Pga-AB displacement loops (3, 4) and clearly indicates contraction of the abdominal muscles during expiration. (2) By the increase in end-expiratory Pes (Pes) relative to its level at the onset of spontaneous breathing trial, when expiratory muscle activity was nil or almost nil (Figure 1). (3) By the abdominal muscle's EMG activity (EMGab).

Results are expressed as mean ± SD. Statistical analysis was performed using Student's paired t test and linear regression analysis. A p value at the 0.05 level was considered significant. The agreement between the difference PEEP_i,dyn  Pga,total decay and PEEP_i,dyn ref, as well as between the difference PEEP_i,dyn  Pga,exp rise or PEEP_i,dyn  Pga,zf decay and PEEP_i,dyn ref were evaluated by the method of Bland and Altman (15). The degree of agreement was quantified in terms of the mean difference and SD of the differences.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For the entire patient group, the breaths simulated by the ventilator were almost identical to those recorded during spontaneous breathing (VT of 0.41 ± 0.06 versus 0.41 ± 0.05 L, f of 30 ± 5 versus 30 ± 5, and TI/TT of 0.33 ± 0.03 versus 0.33 ± 0.04, respectively). CdynL values during the middle and the end of the spontaneous breathing trial and during the simulation with controlled ventilation were 0.048 ± 0.024, 0.045 ± 0.020, and 0.045 ± 0.021 L/cm H₂O, respectively (p > 0.1). RL values during the middle and the end of the trial were 12.4 ± 3.2 and 12.8 ± 3.0 cm H₂O/L/s, respectively (p > 0.1). The end-inspiratory PL during spontaneous breathing fell on the lung line of the Campbell diagram in all cases (mean difference between the end-inspiratory PL and the corresponding value of the lung line 0.6 ± 0.2 cm H₂O). The difference between the FRCsim and the FRCcamp was 0.026 ± 0.021 L (range 0-0.064 L).

The results obtained at the middle and at the end of the spontaneous breathing trial are presented in Table 2. PEEP_i, dyn  Pga,total decay was highly correlated with PEEP_i,dyn ref (r = 0.993; p < 0.001, Figure 3A). The Bland and Altman analysis showed that the mean difference of PEEP_i,dyn  Pga,total decay and PEEP_i,dyn ref was only 0.37 cm H₂O and the limits of agreement (i.e., mean difference 2 SD to mean difference +2 SD) were 0.09 to 0.83 cm H₂O, indicating a strong agreement between these methods (Figure 4A). PEEP_i, dyn  Pga,exp rise and PEEP_i,dyn ref were also well correlated (r = 0.877; p < 0.001), and the Bland and Altman analysis showed that their mean difference was 0.72 cm H₂O with limits of agreement 1.69 to 3.13 cm H₂O, indicating a good agreement between them. However, the agreement between PEEP_i,dyn  Pga,exp rise and PEEP_i,dyn ref was weaker compared with that of PEEP_i,dyn  Pga,total decay and PEEP_i,dyn ref. Although PEEP_i,dyn  Pga,zf and PEEP_i,dyn ref were correlated (r = 0.856; p < 0.001, Figure 3B), the Bland and Altman analysis showed that their mean difference was 3.14 cm H₂O with limits of agreement 0.46 to 6.74 cm H₂O, indicating lack of agreement between the two methods (Figure 4B).

View larger version (19K): [in this window] [in a new window]   Figure 3.   Identity plots between the difference PEEPi,dyn  Pga,total decay and PEEPi,dyn ref (A) and between the difference PEEPi,dyn  Pga,zf decay and PEEPi,dyn ref (B) in 15 actively expiring patients. PEEPi,dyn  Pga,total decay = the correction of PEEPi,dyn for expiratory muscle activity proposed by Lessard and coworkers (5); PEEPi,dyn  Pga, zf decay = the correction of PEEPi,dyn for expiratory muscle activity recommended by Appendini and coworkers (4); PEEPi,dyn ref = reference dynamic PEEPi computed by using the Campbell diagram (6); solid line = identity line; dotted line = regression line. In A, all values are on or very close to the line of identity, indicating that the difference PEEPi,dyn  Pga,total decay is quite similar to PEEPi,dyn ref. In contrast, all values are above 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,dyn  Pga,total decay and PEEPi,dyn ref (A) and PEEPi,dyn  Pga,zf decay and PEEPi,dyn ref (B) against their average (mean) values, respectively. d = mean difference; d + 2 SD = upper limit of agreement; d  2 SD = lower limit of agreement. The mean difference between PEEPi,dyn  Pga,total decay and PEEPi,dyn ref was 0.37 cm H2O and the limits of agreement (i.e., mean difference 2 SD to mean difference +2 SD) were 0.09 to 0.83 cm H2O, indicating a strong agreement between these methods. This signifies that the two methods can be used interchangeably, and consequently the difference PEEPi,dyn  Pga,total decay is a very good estimate of the actual dynamic PEEPi. In contrast, the mean difference between PEEPi,dyn  Pga,zf decay and PEEPi,dyn ref was 3.14 cm H2O with limits of agreement 0.46 to 6.74 cm H2O, indicating lack of agreement between the two methods. For definitions see legend to Figure 3.

The mean values of Pga,total decay (9.2 ± 4.9 cm H₂O) and Pga,exp rise (8.9 ± 4.3 cm H₂O) were higher than that of Pga,zf decay (6.4 ± 3.4 cm H₂O, p < 0.0001; n = 30). There was a good correlation between the difference (PEEP_i,dyn  Pga,zf decay)  PEEP_i, dyn ref and Pga,exp rise (r = 0.903; p < 0.001, Figure 5A). In contrast, there was no correlation between the difference (PEEP_i,dyn  Pga,total decay)  PEEP_i,dyn ref and Pga,exp rise (r = 0.22; p > 0.1, Figure 5B).

View larger version (17K): [in this window] [in a new window]   Figure 5.   Plots of the Pga,exp rise against the differences between PEEPi,dyn  Pga,zf decay and PEEPi,dyn ref (A) and PEEPi,dyn  Pga,total decay and PEEPi,dyn ref (B), respectively. Pga,exp rise = the rise of Pga from its minimum end-inspiratory level to the maximum level at end expiration, expressing the intensity of expiratory muscle contraction. (A) When the expiratory muscle activity is small or moderate, for example, Pga,exp rise is less than 6 cm H2O, the overestimation in measurement of the actual PEEPi,dyn due to subtracting Pga,zf decay from PEEPi,dyn (i.e., [PEEPi,dyn  Pga,zf decay]  PEEPi,dyn ref) does not exceed 2 cm H2O. When expiratory muscle contraction is powerful, for example, Pga,exp rise exceeds 6 cm H2O, subtracting Pga,zf decay from the measured PEEPi,dyn may overestimate the actual PEEPi,dyn from 2 cm H2O up to 7 cm H2O. This overestimation may have clinical importance and may also introduce a significant error in the measurement of inspiratory work of breathing. (B) The absence of correlation indicates that PEEPi,dyn  Pga,total decay represents the PEEPi,dyn ref regardless of the intensity of expiratory muscle contraction. For definitions see legend to Figure 3.

The Pes (9.4 ± 5.4 cm H₂O) was nearly identical to the Pga,exp rise (8.9 ± 4.3, p > 0.1; n = 30) and they were highly correlated (r = 0.971; p < 0.001, Figure 6); their mean difference was 0.43 cm H₂O with limits of agreement 2.19 to 3.05 cm H₂O. In 6 of 15 patients the decay in Pes and Pga at the beginning of inspiration occurred almost simultaneously (e.g., Figure 1). In the other 9 patients the onset of the fall of Pga preceded that of Pes by 14 to 30 ms (18 ± 5 ms); the Pga decay from its peak end-expiratory level to the point where Pes started to descend amounted to 0.8 ± 0.4 cm H₂O. The actual PEEP_i,dyn requires inspiratory muscle contraction to be counterbalanced. In fact, from the beginning of the fall of Pes to the point of zero flow there was an increase in transdiaphragmatic pressure (Pdi = Pga  Pes) in all patients (e.g., Figure 1) amounting to 5.3 ± 0.6 cm H₂O (range 2.8 to 9.2 cm H₂O).

View larger version (18K): [in this window] [in a new window]   Figure 6.   Identity plot for values of Pes and Pga,exp rise. Pes = the increase in end-expiratory Pes in the course of the spontaneous breathing trial due to expiratory muscle recruitment, relative to its level at the beginning of this trial, when expiratory muscle activity was nil or almost nil; Pga,exp rise = the rise of Pga from its minimum end-inspiratory level to the maximum level at end expiration, expressing the intensity of expiratory muscle contraction; solid line = identity line; dotted line = regression line. The finding that all values are on or very close to the line of identity indicates that during active expiration both the rib cage and the abdomen behave as a single compartment, at least at the end of expiration. This presupposes that the diaphragm is relaxed and plays the role of a passive membrane, thus permitting free transmission of pressure from one compartment to the other. It also indicates that inspiratory muscle activity, if present, is very small. Indeed, persistent inspiratory intercostal or accessory muscle action during expiration would eliminate the expiratory muscle effect on Pes and, hence, on Pes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This study yielded three main findings: 1) In spontaneously breathing and actively expiring patients, PEEP_i,dyn measured with the esophageal balloon technique overestimates the actual PEEP_i,dyn by an amount approximately equal to Pga,total decay. Thus, the subtraction of Pga,total decay from PEEP_i,dyn provides a very good estimate of the actual PEEP_i,dyn. 2) The subtraction of Pga,exp rise from PEEP_i,dyn gives an adequate yet inferior correction of PEEP_i,dyn for expiratory muscle contraction. 3) Subtracting Pga,zf decay from PEEP_i,dyn overestimates the actual PEEP_i,dyn. This overestimation is proportional to the intensity of expiratory muscle recruitment.

Critique of Methods

First, we have used the Campbell diagram constructed during simulation of spontaneous breathing with patients ventilated in the control mode in the absence of any respiratory muscle activity. Such simulation presupposes that passive mechanics during control ventilation are the same as during spontaneous breathing. This was the case in our patients, as CdynL and RL were essentially the same at the middle and the end of the spontaneous breathing trial and during control mechanical ventilation afterwards. Second, we have positioned the dynamic V-Pes loop on the volume axis of the Campbell diagram by using the CL,dyn, that is, the value of PL at the onset of inspiration was measured during spontaneous breathing and the loop was positioned so that this value fell on the CL,dyn. To validate the correct positioning of this loop, the end-inspiratory PL (at zero flow) was measured during spontaneous breathing and it was compared with the corresponding value of the lung line of the Campbell diagram, accepting only values differing less than 1 cm H₂O. The end-inspiratory PL fell on the lung line of the Campbell diagram in all cases. Because forceful expiratory muscle contraction during spontaneous breathing could have reduced the EELV, which would have not been accounted for during the simulation, we measured from the Campbell diagram the difference between the EELV and the Vr (FRCcamp) and compared it with the corresponding difference during simulated controlled breathing (FRCsim). The difference between FRCsim and FRCcamp was quite small, probably due to the presence of flow limitation. Third, chest wall mechanics might of course have been slightly different between spontaneous and simulated breathing due to chest wall distortion in the former that cannot be accounted for during the latter. However, this is an acceptable limitation of using the Campbell diagram in cases in which chest wall distortion exists, such as during exercise.

How Should We Correct PEEP_i,dyn for Expiratory Muscle Activity?

In accordance with recent work (3), the present study clearly demonstrates that expiratory muscle activity may contaminate the measurements of PEEP_i,dyn. It was found that PEEP_i,dyn measurements overestimate the actual PEEP_i,dyn by an amount approximately equal to Pga,total decay. This finding indicates that Pga,total decay reflects the expiratory muscle relaxation that contributes to the fall in Pes measured as PEEP_i,dyn. This part of PEEP_i,dyn that is related to the relaxation of the expiratory muscles obviously does not represent an extra load for the inspiratory muscles. In contrast, the actual PEEP_i,dyn necessitates the contraction of the inspiratory muscles to be counterbalanced, as it is related to dynamic hyperinflation. It represents, therefore, an inspiratory threshold load that must be faced by the inspiratory muscles before initiating inspiratory airflow or triggering the demand valve of the ventilator (2, 5). Actual PEEP_i,dyn, as reflected by PEEP_i,dyn ref, was present in every patient of this study (range 2.3 to 9.1 cm H₂O, 5.3 ± 1.9 cm H₂O; Table 2). Inspiratory muscle contraction is necessary to offset this threshold load and initiate inspiration. Indeed, the time lag between the onset of the fall in Pes and the point of zero flow was associated with diaphragmatic contraction in all patients (e.g., Figure 1); Pdi amounted to 5.3 ± 0.6 cm H₂O (range 2.8 to 9.2 cm H₂O). Simultaneous contraction of the diaphragm and relaxation of the abdominal muscles minimize the anticipated reduction of abdominal pressure despite substantial changes in abdominal volume, at least during the first half of inspiration (16). Therefore, calculation of the contribution of expiratory muscle relaxation to PEEP_i,dyn from the measurement of Pga,zf decay, which represents the amount of Pga decay from the onset of inspiratory effort to the point of zero flow (i.e., the pressure difference between points C and I in Figure 2B), would underestimate this contribution as diaphragmatic contraction delays, partially offsetting the decrease in Pga. Indeed, the whole decay of Pga during inspiration, that is, the pressure difference between C and F in Figure 2B, represented by Pga,total decay is due to relaxation of the abdominal muscles and this amount of pressure should be considered as contributing to the PEEP_i,dyn (Figure 4A). In the present study, the Pga,zf decay was significantly lower than Pga,total decay (6.4 ± 3.4 versus 9.2 ± 4.9 cm H₂O, p < 0.0001). Moreover, the error in measurement of the actual PEEP_i,dyn due to subtraction of Pga,zf decay from PEEP_i,dyn (i.e., [PEEP_i,dyn  Pga,zf decay]  PEEP_i,dyn ref) was proportional to the intensity of the expiratory muscle contraction, expressed by Pga,exp rise (7) (Figure 5A). Thus, correcting PEEP_i,dyn for the contribution of expiratory muscle relaxation by subtracting Pga,zf decay should overestimate the actual PEEP_i,dyn in a manner proportional to the intensity of expiratory muscle contraction. When the expiratory muscle activity is small or moderate, for example, when the Pga,exp rise is less than 6 cm H₂O, this overestimation does not exceed 2 cm H₂O (Figure 5A). In contrast, when the expiratory muscle contraction is powerful, for example, Pga,exp rise > 6 cm H₂O as in the COPD patients of the present study, subtracting Pga,zf decay from the measured PEEP_i,dyn may overestimate the actual PEEP_i,dyn from 2 cm H₂O up to 7 cm H₂O (Figure 5A). This overestimation may have important clinical consequences in terms of augmenting hyperinflation if external PEEP or continuous positive airway pressure (CPAP) were used to reduce the inspiratory threshold load related to PEEP_i in patients with expiratory flow limitation (2, 4). In addition, this overestimation of the actual PEEP_i,dyn may significantly influence the measurement of inspiratory work of breathing or the pressure time product, which also require accurate estimation of PEEP_i (2, 5).

The present study shows that although both Pga,total decay and Pga,exp rise are adequate corrections for PEEP_i,dyn, the former is better. The reason for this is not clear. It might be due to the fact that Pga,total decay is synchronous to PEEP_i, dyn, whereas Pga,exp rise precedes the PEEP_i,dyn measurement. More studies are needed to sort it out.

The results of our study are compatible with those of other investigators who have published studies on this issue (6, 7). Yan and associates (6) estimated PEEP_i,st by using the Campbell diagram, and then assessed the ability of Pga,exp rise to correct the measured PEEP_i,dyn in healthy subjects expiring through a Starling resistor. They found that Pga,exp rise could accurately estimate the expiratory muscle contribution to PEEP_i,dyn in most subjects. These investigators did not evaluate Pga,zf decay or Pga,total decay. Parthasarathy and coworkers (7), aiming to evaluate the relative accuracy of Pga,exp rise and Pga,zf decay for quantifying the expiratory muscle contribution to PEEP_i,dyn, related them to carefully obtained electromyographic recordings of transversus abdominis muscle activity (EMGta). In normal subjects expiring through a Starling resistor they found that Pga,exp rise correlated well with the moving average of EMGta (r = 0.7 to 0.95), whereas Pga,zf decay showed a weaker correlation with EMGta (r = 0.04 to 0.53). Our study extended these observations in patients including concurrent comparisons of Pga,zf decay, Pga,total decay, and Pga,exp rise. In accordance to the above-mentioned studies (6, 7) we found that Pga,exp rise is a good correction of PEEP_i,dyn for expiratory muscle activity. However, we have shown for the first time that Pga,total decay is in fact an even better correction.

Why Are Pga,total decay and Pga,exp rise Adequate Correction Factors of PEEP_i,dyn for Expiratory Muscle Activity?

The fact that in patients who exhibited expiratory muscle contraction during expiration, the end-expiratory increase in Pes (Pes) was nearly identical to that in Pga (Pga,exp rise) (Figure 6) indicates that both the rib cage and the abdomen behave as a single compartment, at least at the end of expiration. This presupposes that the diaphragm permits free transmission of pressure from one compartment to the other. Indeed, in all but five patients the expiratory baseline values of Pdi at the middle and at the end of spontaneous breathing trial were equal to those obtained at the beginning of the trial when expiratory muscle activity was nil. Expiratory baseline values of Pdi were zero when Pga had been adjusted to equal Pes at end inspiration during the beginning of the spontaneous breathing trial, indicating that the diaphragm was relaxed, playing the role of a passive membrane. In the other patients (No. 1, 2, 5- 7; Table 1) expiratory baseline values of Pdi at the middle and at the end of the spontaneous breathing trial were slightly higher (1.2 ± 0.3, range 0.8 to 1.5 cm H₂O) than at the beginning of the trial; because these patients had the highest values of Pga,exp rise (range 12.5 to 16.3 cm H₂O), this probably reflects a passive Pdi due to strong abdominal muscle contraction (6). Moreover, the fact that Pes was similar to the end- expiratory increase of Pga suggests that inspiratory muscle activity during expiration, if present, was very small. In fact, substantial persistent inspiratory action during expiration would reduce the expiratory muscle effect on Pes and, hence, Pes. Because the diaphragm was relaxed during expiration, any inspiratory activity would have been due to the inspiratory intercostal and accessory muscles (14). In the patients of the present study, diaphragmatic relaxation constantly started very early postinspiration (< 0.3 s) and lasted until the end of expiration (e.g., Figure 1). This early cessation of the postinspiratory activity of the diaphragm may be attributed to the increased inspiratory load (13) in our patients when forced to breathe spontaneously. Therefore, the rise in abdominal pressure produced by contraction of the abdominal muscles is transmitted through the relaxed diaphragm to the pleural space, which in the absence of significant inspiratory intercostal and accessory muscle activity during expiration increases the alveolar pressure, giving an expiratory contour to Pes almost identical to that of Pga (Figure 1). Although no other EMGs besides that of the abdominal muscles were recorded, it is possible that other expiratory muscles may have contributed to this increase in alveolar pressure, such as the triangularis sterni (17) or the internal interosseous intercostals (3, 18). Nevertheless, contraction of the rib cage expiratory muscles, which would augment the rise in alveolar pressure caused by abdominal muscle activity, would contribute at the same time to the increase in abdominal pressure through the relaxed diaphragm, and would thus be included in the correction of PEEP_i,dyn.

In conclusion, the present study demonstrates that an adequate correction of PEEP_i,dyn for the overestimation due to expiratory muscle activity can be made by subtracting either Pga,total decay or Pga,exp rise, the former achieving the best performance. Actual PEEP_i,dyn obtained this way can be used for the calculation of the inspiratory work of breathing or the pressure time product; it can also be used as an index of PEEP_i to adjust external PEEP or CPAP to decrease inspiratory effort without inducing further hyperinflation.