INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The adult respiratory distress syndrome (ARDS) is characterized by a marked increase in respiratory elastance and resistance (1). Intrinsic positive end-expiratory pressure (PEEPi) and dynamic hyperinflation (DH) are also common in ARDS patients during mechanical ventilation with zero end-expiratory pressure (ZEEP) (2, 4), despite the overall reduction in FRC in ARDS (3, 8). DH and PEEPi were originally described in mechanically ventilated patients with chronic obstructive pulmonary disease (COPD) (9, 10), in whom they are almost invariably associated with tidal expiratory flow limitation (EFL) (11) and are caused primarily by loss of lung recoil and/or increased peripheral airway resistance (12). However, DH and PEEPi can also occur in the absence of EFL: a high expiratory resistance may by itself impede expiration to the extent that the next inspiration occurs before exhalation to the relaxation (elastic equilibrium, Vr) volume of the respiratory system (12). Indeed, in the absence of EFL, the rate of passive lung deflation is determined by the elastic recoil pressure stored during the preceding lung inflation and by the opposing flow resistance offered by the respiratory system (including an endotracheal tube, expiratory circuit of a ventilator, and additional equipment, if any). Accordingly, the higher the resistance, the slower the rate of lung emptying, with the consequence that at high resistance, the next inspiration may occur before complete exhalation to Vr, with ensuing PEEPi and DH. Since EFL has not been envisaged in ARDS, it has been suggested that in ARDS patients, PEEPi and DH are generated by a disproportionate increase in expiratory flow resistance (13). This hypothesis is supported by the marked increase in resistance found consistently in patients with ARDS (2). However, several mechanisms could induce EFL in ARDS. Because maximal flows depend on lung volume, the decrease in FRC exhibited by ARDS patients (3, 8) should decrease their expiratory flow reserve in the range of VT. This latter reduction is compounded by the marked reduction in the number of functional lung units in ARDS, a condition aptly termed "baby lung" (14). Breathing at low lung volume also promotes small-airway closure and gas trapping, particularly in the decubitus position (15). An increased closing volume and gas trapping have been reported with pulmonary congestion and edema (16), and florid pulmonary edema is typically present during the early stages of ARDS (17). By reducing the functional lung volume (i.e., the number of lung units with patent airways), enhanced small-airway closure should further decrease the expiratory flow reserve in ARDS.

Although tidal EFL may occur in ARDS and contribute to PEEPi, EFL has not to our knowledge been assessed in such patients. Therefore, in the present study, using the negative expiratory pressure (NEP) technique (18), we assessed whether EFL and PEEPi are present in ARDS patients ventilated mechanically at ZEEP in the supine and semirecumbent positions. Because bronchodilators may decrease airway resistance in ARDS patients (19), we also investigated the effect of bronchodilator administration on EFL and PEEPi in the semirecumbent position.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We studied 10 patients admitted to the intensive care unit of our hospital for management of acute respiratory failure resulting from ARDS. The diagnosis of ARDS was made according to the American-European Consensus Conference on ARDS (20). None of the patients had a history of chronic bronchitis or asthma. The anthropometric and clinical characteristics of the 10 ARDS patients are provided in Table 1, together with the patients' blood gas tensions measured in the semirecumbent (30 degrees) position immediately before our experiments at the baseline ventilatory settings and positive end-expiratory pressure (PEEP) prescribed by the patients' primary physicians. Six of the patients had never smoked (Patients 1, 2, 5, 6, 8, and 9), whereas the other four were smokers (8 to 20 pack-years). If patients were receiving ₂-agonists as part of their management, the administration of these drugs was discontinued at least 8 h before the study began. The investigative protocol was approved by the institutional ethics committee of our hospital, and informed consent was obtained from the next of kin of the patients.

Patients were studied while clinically and functionally stable. They were intubated (Portex cuffed endotracheal tube, I.D. = 7 to 9 mm) and mechanically ventilated in the semirecumbent position in the volume control mode with constant inspiratory flow, with an Evita 2 ventilator (Dräger, Lubeck, Germany) equipped with an NEP device attached to the distal end of the expiratory line of the ventilator. Patients were sedated (with an intravenous infusion of 0.03 to 0.2 mg/kg/h midazolam) and paralyzed (with an intravenous infusion of 0.03 to 0.06 mg/kg/h pancuronium bromide) as part of their management. Fractional inspired oxygen concentration (FI_O2) was fixed throughout the procedure (Table 1). The ventilator settings were those prescribed by the patient's primary physicians before our investigation (Table 2), and were kept constant throughout the study, with the following two exceptions: (1) external PEEP was discontinued during the study; and (2) respiratory frequency was decreased for a few seconds during end-expiratory and end-inspiratory occlusions, as described in the experimental procedure. When the Evita 2 ventilator used in the study was set at ZEEP, it actually applied an extrinsic PEEP of 0.8 cm H₂O. The patient's electrocardiogram, heart rate, blood pressure, and Sp_O2 (pulse oximetry) were continuously monitored (Life Scope 14; Nihon Kohden, Tokyo, Japan). A physician not involved in the research protocol was present to provide for patient care.

Airway pressure (Paw), flow (), and volume (V) were measured with the pressure transducers and pneumotachographs incorporated into the ventilator used in the study. In built software was used to monitor these variables on an Evita video screen attached to the ventilator, and to obtain on-line records of the time course of Paw, , and V, as well as plots of tidal flow-volume (-V) loops with and without NEP. The Paw, , and V signals were digitized at 125 Hz, using the analog-to-digital converter incorporated into the ventilator, and were stored in a personal computer for subsequent analysis. A device especially designed by Dräger was attached to the ventilator to allow application of the NEP technique. This device applies a negative pressure of 5 cm H₂O to the distal end of the expiratory circuit (i.e., distal to the expiratory valve). The device is activated by pressing a special button during inspiration, which causes NEP to be applied throughout the ensuing expiration, starting 8 ms after the onset of expiratory flow. To reduce the effects of compliance and resistance of the system connecting the subject to the ventilator on the measurements of respiratory mechanics, a standard low-compliance tube was used and the humidifier was omitted from the inspiratory line of the ventilator. Care was taken to avoid gas leaks in the equipment and around the tracheal cuff.

Experimental Procedure and Data Analysis

Assessment of respiratory mechanics and EFL was first done at ZEEP in seven of the ARDS patients (Patients 1 to 7) in the supine position. The patients were supine at ZEEP for about 20 min before data collection was begun. Only seven patients were studied in the supine position because the values of Sa_O2 in this position for the other three patients (Patients 8 to 10) fell below the acceptable limits. Assessment of respiratory mechanics and EFL was then done in 10 patients in the semirecumbent position, after they were ventilated at ZEEP in this position for about 20 min. Assessment of respiratory mechanics and EFL in this position was made before and 30 min after the administration of salbutamol via nebulization (5 mg salbutamol and 3 ml saline). Salbutamol was delivered in the form of a nebulized aerosol through the endotracheal tube, using a Dräger pneumatic driver connected to the inspiratory gas source. All measurements were made with the patient relaxed and with the same ventilator settings before and after bronchodilator administration. Patient relaxation was evidenced by regularity of sequential Paw, V, and records, an absence of inspiratory deflections in the Paw and signals during mechanical expiration, and an absence of visible signs of spontaneous respiratory efforts. The airways were suctioned before and (if needed) during the experiments.

For measurement of PEEPi, the airway was occluded at the end of a tidal expiration by use of the end-expiratory hold button of the ventilator (9). Pressing this button closes both the inspiratory and expiratory valves of the Evita 2 ventilator for a period that corresponds to the baseline inspiratory time. Since this time was less than the 2 to 3 s required for accurate measurement of PEEPi (4), the frequency of the ventilator was decreased to 6 breaths/min immediately after activation of the end-expiratory hold button. This produced end-expiratory pauses of sufficient duration (> 3 s) for the measurement of PEEPi. In all instances, duplicate measurements of PEEPi were made.

Ten regular breaths after the end-expiratory occlusion test, the inspiratory pause time was prolonged to 3 s by once again decreasing the frequency of the ventilator to 6 breaths/min. During this end- inspiratory occlusion test, there was an immediate decrease in Paw from a maximal value (Pmax) to a lower value corresponding to zero flow (P1), followed by a gradual decrease in Paw to an apparent plateau (P₂), which was achieved after 2 to 3 s and which represented the static elastic recoil of the respiratory system at the end of the mechanical inflation (4, 10). In all instances, three end-inspiratory occlusion tests were performed and the values were averaged for subsequent analysis. The static inflation elastance of the total respiratory system (Est,rs) was computed as the ratio of the static end-inspiratory pressure (P₂) minus the total PEEP to V_T: Est,rs = (P₂  PEEP  PEEPi)/VT (4). In our experiments, PEEP amounted to 0.8 cm H₂O (see the previous discussion).

The resistive properties of the respiratory system were obtained as previously described in detail (21). Briefly, the minimal resistance (Rmin) was calculated as the difference between Pmax and P₁ divided by the inspiratory flow () preceding the end-inspiratory occlusion (i.e., [Pmax  P₁]/) and the maximal resistance (Rmax) was obtained as (Pmax  P₂)/. The latter includes Rmin and the "additional" resistance (Rrs = Rmax  Rmin) due to time-constant inequality and/or stress relaxation of the tissues of the lung and chest wall (4). Both Rmin and Rmax include the resistance of the endotracheal tube (Ret). The latter was computed for each endotracheal tube and each used, and was subtracted from the values given earlier for Rmax and Rmin in order to obtain the maximal and minimal resistance of the total respiratory system alone (Rmax,rs and Rmin,rs, respectively). Ret was computed with Rohrer's equation: Ret = k₁ + k₂, where k₁ and k₂ are constants. The values of these constants were obtained from Behrakis and coworkers (22). The total inspiratory work (WI) per breath was obtained by integration of Paw with respect to V during baseline ventilation. WI included the total work done on the respiratory system and the resistive work due to the endotracheal tube. The work due to PEEPi (WPEEP_i) was obtained as the product of PEEPi and VT.

Ten regular breaths after the end-inspiratory occlusion tests, a consecutive pair of tidal -V loops were recorded, one immediately before (control) and the other during the application of NEP (Figure 1). The two -V loops were automatically superimposed by the respirator's software. The presence of EFL was assessed by comparing the tidal expiratory -V curve with NEP against the reference curve (18). If the expiratory flow with NEP was higher than it was under control conditions, the subject was classified as being without EFL (Figure 1, right side). In contrast, if with NEP all or part of the expiratory -V curve was superimposed on the control curve, the subject was classified as having EFL (Figure 1, left side). The extent of EFL was quantified in terms of the portion of VT over which the expiratory flows with and without NEP were similar, and was expressed as a percentage of the control VT (EFL, %VT) (18). At least two NEP tests were performed on each subject, and the mean value of EFL %VT was used for further analysis.

View larger version (42K): [in this window] [in a new window]   Figure 1.   -V loops of NEP test breath and preceding control breath for two representative patients, one with EFL exceeding 35% of control VT (left side) and the other without EFL (right side). Thin lines = control breath; heavy lines = NEP. In the EFL patient, onset of EFL is heralded by an inflection point on the expiratory -V curve (arrow).

Since all measurements were made on relaxed subjects, the mechanics data and NEP tests were highly reproducible in repeated measurements.

Statistical Analysis

Data are presented as mean ± SD. Comparisons of data obtained before and after bronchodilator administration, as well as in the supine and semirecumbent positions, were made with Student's paired t test. Regression analysis was done with the least squares method. A value of p  0.05 was accepted as statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Semirecumbent Position

Respiratory mechanics. The respiratory mechanics data before and after salbutamol are given in Table 3. Salbutamol had no significant effect.

PEEPi and EFL. Before salbutamol, eight ARDS patients exhibited tidal EFL, which ranged from 5% to 37% VT. In all instances the onset of EFL was heralded by a distinct inflection point on the expiratory -V curves obtained with and without NEP (i.e., after the onset of EFL, the -V curves became convex toward the volume axis over the entire EFL range of VT [Figure 1, left side]). It is likely that the convexity, in the expiratory -V curves, which has been described as a hallmark of EFL by Gottfried and coworkers (11), reflects a progressive reduction in the number of functional lung units due to small-airway closure. Accordingly, the lung volume corresponding to the inflection point (i.e., the volume at which the slope, d-dV, starts to decrease progressively) can be regarded as the "closing volume." By contrast, in the absence of EFL, the -V curves exhibited a concavity toward the volume axis throughout expiration (Figure 1, right side).

After administration of salbutamol, EFL was abolished in two patients (Patients 3 and 7), whereas in the others there was little or no change (Figure 2, top panel ). The -V curves of NEP test breaths and preceding control breaths for Patient 3 before and after salbutamol are shown in Figure 3. Before salbutamol, the onset of EFL, which amounted to 12% VT, was heralded by a distinct inflection point on the expiratory -V curve. After salbutamol there was no EFL and the inflection point had disappeared. Similar results were obtained for Patient 7, except that this patient did not exhibit the transient increase in flow seen at the onset of expiration in Patient 3 (Figure 3). This initial increase in flow may reflect extension of the inflation volume into the upper flat (low compliance) part of the static volume-pressure curve of the respiratory system, with the result, that the initial flow during expiration was relatively high due to greater elastic recoil pressure. An alternative explanation is that the transient increase in initial flow reflects a sequential emptying of lung units with nonuniform mechanical properties, with units having shorter time constants emptying first and being responsible for the transient increase in flow (11). It should be noted, however, that distinct flow transients at the onset of expiration were seen only in Patient 3. 

View larger version (18K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 2.   Relationships of EFL (%VT) (top panel ) and PEEPi (bottom panel ) before and after salbutamol administration in 10 semirecumbent ARDS patients. Broken line = identity line.

View larger version (42K): [in this window] [in a new window]   Figure 3.   -V loops of NEP test breath and preceding control breath of Patient 3 before (left side) and after salbutamol (right side). Before salbutamol, EFL amounted to 12% VT, and the onset of EFL was heralded by an inflection point (arrow). After salbutamol the subject became free of EFL and the inflection point disappeared. Thin lines = control breath; heavy lines = NEP. For further information see text.

Although on average EFL (%VT) decreased after salbutamol administration, the change was not significant (Table 3). Despite the presence of EFL in only eight patients before salbutamol, PEEPi was present in all 10 subjects, ranging from 0.4 to 7.7 cm H₂O. After salbutamol, PEEPi was still present in all 10 subjects (Figure 2, bottom panel ), and the values did not change significantly relative to those before salbutamol (Table 3). Neither before nor after salbutamol were any significant correlations found between PEEPi and the ventilatory settings (Table 2) and respiratory mechanics data (Table 3). On the other hand, a significant correlation was found between PEEPi and EFL (%VT) both before and after salbutamol administration (Figure 4). Also, negative correlation was found between EFL (%VT) and Est,rs before salbutamol administration (r = 0.65, p = 0.04). There were no significant differences between the four smokers and the six nonsmokers in the study in any of the variables shown in Table 3. In fact, both patients in Figure 1 were nonsmokers. No significant correlation was found of either EFL (%VT) or PEEPi with days of mechanical ventilation.

View larger version (12K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   Figure 4.   Relationship of PEEPi to EFL (%VT) before (top panel ) and after (bottom panel ) salbutamol administration in 10 semirecumbent ARDS patients.

Supine Position

Table 4 summarizes the respiratory mechanics data obtained for seven ARDS patients (three smokers and four nonsmokers) in the supine and semirecumbent positions. In the supine position, the values of EFL (%VT), PEEPi, and WPEEP_i were significantly higher than in the recumbent position. In the supine position there was a significant positive correlation of PEEPi with EFL (%VT) (r = 0.82, p = 0.03), as there was also in the semirecumbent position (Figure 4), and a negative correlation of both PEEPi (r = 0.79, p = 0.04) and EFL (%VT) (r = 0.94, p = 0.002) with Est,rs.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main finding of the present study was that with ZEEP, most ARDS patients exhibited tidal EFL with a concomitant PEEPi, the onset of which was heralded by a distinct inflection point on the expiratory -V curve. Since this point probably reflects the onset of significant small-airway closure, it can be termed the closing volume. In contrast, in anesthetized, paralyzed normal subjects there is no inflection point on the passive expiratory -V curve (23). This was the case in only two of our ARDS patients before administration of salbutamol.

In patients with COPD, the presence of EFL during tidal breathing has been shown to promote DH and PEEPi, with a concomitant increase in WI, impairment of inspiratory muscle function, and adverse effects on hemodynamics (9, 12, 24). This may contribute to dyspnea (24) and cause ventilatory failure in COPD patients (2, 12). Increased work of breathing as a result of PEEPi is also found in ARDS patients with tidal EFL (Table 3). Impairment of inspiratory muscle function, however, is less likely in ARDS patients because they breathe at low lung volume in spite of DH (3, 8). The adverse effects of hemodynamics should also be less marked in ARDS patients, who have stiffer lungs than COPD patients; consequently, a smaller component of PEEPi may be transmitted to the juxtacardiac space in ARDS than in COPD (25). The absence of severe parenchymal lung injury in COPD may allow PEEPi to fully compress the pulmonary capillaries, in which case the inflated Swan-Ganz catheter used to measure pulmonary artery wedge pressure will track alveolar rather than vascular pressure (25). This is less likely in ARDS because of the presence of severe parenchymal lung injury.

EFL. In both the semirecumbent and supine positions, most of the ARDS patients in our study exhibited EFL. Several mechanisms could induce EFL in ARDS. The expiratory flow reserve is diminished by the decreased FRC (3, 8) and by the reduction in number of functional lung units in ARDS (14). Breathing at low lung volume promotes airway closure and gas trapping, with further reduction in the number of functioning lung units (i.e., units with open airways) and hence in expiratory flow reserve. An increased closing volume has been observed with pulmonary congestion and edema (16). Surfactant deficiency in ARDS should also promote small-airway closure (13, 16). A close association between closing volume and EFL has been postulated by Rodarte and associates (26). Our results suggest that ARDS patients also experience a synchronous occurrence of EFL and small-airway closure during expiration (i.e., that there is a critical transpulmonary pressure [PC] that elicits both EFL and the onset of small-airway closure). In normal lungs, the values of PC for airways are somewhat higher than those for the alveoli (27). This is a useful functional feature because, as a result of small-airway closure, gas is trapped behind the closed airways, thus preventing atelectasis at low lung volumes. It is conceivable, however, that in ARDS, which is characterized by alveolar instability, the values of Pc for alveoli exceed those for small airways. Under these latter conditions, there should be continuous reexpansion and collapse of alveoli during cyclic breathing. In this case EFL and the onset of significant atelectasis could occur synchronously during expiration. Indeed, a sudden decrease in the number of functional lung units as a result of atelectasis should be associated with a sudden decrease in expiratory flow reserve, resulting in EFL. Although it is more likely that the inflection point on the -V curves (Figure 1) found in our study reflects the onset of small-airway closure rather than atelectasis, the latter possibility cannot be discarded.

Bronchoconstriction probably also played a role in eliciting tidal EFL, at least in two of our patients. In fact, two of our patients who exhibited EFL before salbutamol administration (Patients 3 and 7) lost their EFL after salbutamol.

In the semirecumbent position, EFL (%VT) and PEEPi were significantly lower than in the supine position (Table 4), probably reflecting the higher FRC in the semirecumbent position. A similar phenomenon was previously observed by Valta and colleagues (18) in four patients mechanically ventilated for acute ventilatory failure.

PEEPi. We found a significant correlation of PEEPi with EFL (%VT) in our ARDS patients in both the semirecumbent and supine positions. Although all patients who had EFL exhibited PEEPi, PEEPi was also present in the absence of EFL. In the absence of EFL the rate of passive lung deflation is modulated by the elastic recoil pressure stored during the preceding lung inflation, and by the opposing total flow resistance offered by the respiratory system (including an endotracheal tube and the expiratory line of a ventilator). Accordingly, the stiffer the respiratory system (i.e., as reflected by an increased Est,rs), the faster the rate of lung emptying. In fact, a negative correlation was found between EFL (%VT) and Est,rs in both the semirecumbent and supine positions. In the latter position there was also a significant negative correlation of PEEPi with Est,rs. However, no significant correlation was found of either EFL (%VT) or of PEEPi with Rmax,rs and its components (Rmin and R). This is not surprising, since most of our ARDS patients had EFL, and under such conditions the rate of lung emptying is independent of resistance because it is modulated by dynamic airway compression (28). On the basis of the foregoing considerations, it appears that PEEPi should reflect a disproportionate increase in expiratory flow resistance relative to Est,rs only in patients with ARDS who do not have EFL.

Administration of salbutamol had no significant effect on either PEEPi or EFL (%VT), although two patients became free of EFL after bronchodilator administration.

During spontaneous breathing and patient-triggered mechanical ventilation (e.g., assisted mechanical ventilation, pressure support), PEEPi imposes an inspiratory threshold load on the inspiratory muscles in both the presence and absence of EFL (23, 24). As a result, the mechanical work of breathing is increased because of the work done in overcoming PEEPi (WPEEP_i). In the presence of EFL, application of an external PEEP should reduce or abolish WPEEP_i. In fact, Eissa and colleagues (29) measured WI and WPEEP_i in 10 sedated, paralyzed, and mechanically ventilated ARDS patients with ZEEP and different levels of PEEP. With ZEEP, the patients'values of WI and WPEEP_i were essentially the same as those in the present study (Table 3). With a PEEP of 10 cm H₂O, WPEEP_i was markedly reduced, suggesting that in the ARDS patients studied by Eissa and colleagues (29), PEEPi was associated with EFL. In ARDS patients without EFL, the administration of external PEEP should not necessarily reduce WPEEP_i (12).

In most of our ARDS patients, PEEPi was associated with EFL, as in COPD patients. Since the putative role of PEEP in COPD is to reduce PEEPi without increasing the end-expiratory lung volume (EELV), PEEP levels lower than PEEPi have been advocated for COPD patients in order not to increase their EELV (30). In contrast, in ARDS the applied PEEP should increase EELV slightly above the inflection point (closing volume) on the expiratory -V curve, in order to prevent cyclic collapse and reopening of peripheral airways during mechanical ventilation, with concomitant inhomogeneous filling of adjacent air spaces and possible resultant lung injury (31). In this case, therapeutic levels of PEEP should be determined by on-line inspection of the configuration of the patient's expiratory -V curves: PEEP should be increased until the inflection point on the -V curve disappears. Figure 5 shows -V loops obtained for Patient 5 during ventilation with ZEEP (including the NEP test breath) and with a PEEP of 6.5 cm H₂O. With ZEEP this patient was flow-limited (EFL = 37% VT), the onset of EFL being heralded by a distinct inflection point in the -V loop (Figure 5, left side). With a PEEP of 6.5 cm H₂O the inflection point disappeared, indicating absence of EFL (Figure 5, right side). Similar results were obtained for Patient 6. Thus, inspection of tidal expiratory -V curves may guide the choice of PEEP in ARDS. In this connection it should be noted that the effect of bronchodilators on EFL can also be assessed by inspecting the configuration of expiratory -V curves. In fact, in two of our patients (Patients 3 and 7), the inflection point on the -V curves disappeared after salbutamol (Figure 3).

View larger version (15K): [in this window] [in a new window]   Figure 5.   Left side: -V loops of control and NEP test breath of Patient 5 with ZEEP in the semirecumbent position. Right side: control -V loop of same subject with PEEP of 6.5 cm H2O. With ZEEP, EFL amounted to 37% VT and was heralded by a distinct inflection point (arrow). With PEEP the inflection point disappeared and the entire -V curve became concave toward the volume axis, suggesting absence of EFL.

The criteria for selecting the "optimal" ventilatory settings for ARDS patients are still under debate. It seems reasonable, however, that six main goals should be achieved: (1) limited inspiratory pressures; (2) limited inspiratory volumes; (3) limited negative effects on hemodynamics; (4) recruitment of atelectatic alveoli; (5) absence of cyclic reopening and closure of small airways; and (6) good oxygenation. Our analysis, based on the inspection of expiratory -V curves, allows detection of small-airway closure during tidal breathing, and may guide in the selection of levels of PEEP appropriate for avoiding cyclic reopening and closure of small airways during mechanical ventilation. This should also improve the distribution of ventilation in the lung, and improve oxygenation (15). On the other hand, recruitment of atelectatic alveoli, which probably depends mainly on the end-inspiratory pressure and volume, should be assessed separately, along lines previously described (5, 14, 34).

In theory, our approach to selecting the therapeutic level of PEEP in ARDS conforms to the "open lung approach" recently advocated (32, 33). In the studies leading to this, however, the choice of PEEP to avoid tidal reopening and closure of small airways was based on assessment of the pressure corresponding to the lower inflection point (PLIP) on the static inflation volume-pressure (V-P) curve of the respiratory system (i.e., PEEP should slightly exceed PLIP). Assessment of PLIP is, however, technically complex and time consuming. Furthermore, titration of PEEP based on PLIP determined from static V-P curves of the total respiratory system may not be valid; rather, PLIP should be determined from static V-P curves of the lung, a procedure that may not be feasible in the decubitus position because of artifacts in esophageal pressure measurements. Moreover, although PLIP reflects the onset of significant recruitment of atelectatic alveoli or reopening of the small airways in lung units previously unventilated because of small-airway closure, or both, this phenomenon may continue at pressures beyond PLIP (5, 34). In this context it should be noted that recruitment of functional lung units does not depend only on PEEP, but more importantly on end-inspiratory volume and pressure (5). Further studies are required to determine the relationship of EFL to PLIP. These studies should include measurement of the static deflation curve of the lung, which is more relevant to the closing volume.

Respiratory mechanics. The values of our ARDS patients' respiratory mechanics (Table 3) were similar to those previously reported (2). In contrast to Pesenti and coworkers (19), we did not find a significant reduction in Rmin,rs after salbutamol administration. The reason for this is not clear. However, Pesenti and coworkers used a PEEP of 9 cm H₂O, whereas in our study the PEEP was only 0.8 cm H₂O. It is also possible that in some of our patients the dose of salbutamol administered to the distal airways was insufficient.

In conclusion, the present study shows that most patients with ARDS exhibit PEEPi in association with EFL. The degrees of EFL and PEEPi were lower in the semirecumbent than in the supine position, but in the semirecumbent position, neither factor was significantly affected by salbutamol.