INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Achieving adequate oxygenation in mechanically ventilated patients with acute respiratory distress syndrome (ARDS) can pose a significant therapeutic challenge. Many times this can be accomplished only by the use of high alveolar pressure, which may potentiate lung damage and/or retard healing. Therefore, the treatment of ARDS has focused on maintaining adequate oxygen delivery and protecting lung from ventilator-induced injury until reversal of the underlying lung pathology occurs. Strategies that restrain the end-inspiratory pressure and maintain an "optimal" lung volume at the end of expiration have been proposed to achieve these goals (1). In addition to conventional support, other means to improve gas exchange such as extrapulmonary gas exchange, tracheal gas insufflation, partial liquid ventilation, nitric oxide or prostacyclin inhalation, and prone position have been employed (2). With regard to the latter, studies examining the impact of body position changes on oxygenation have been available for more than 20 yr (3). Specifically, several reports (4) have suggested a potential benefit in oxygenation by placing patients with ARDS in the prone position. However, not all patients respond favorably to this simple maneuver and the pathophysiology behind this variable response has not been elucidated yet.

Currently there are no clinical guidelines as to where, when, and how prone positioning should be applied. Further, with the exception of ARDS, there are no data in the literature regarding the clinical utility of the prone position in other clinical entities in the intensive care unit (ICU) setting.

In the present study, the effect of prone positioning was examined in three different disease states that seriously affect oxygenation. The first group of patients was composed of patients with hydrostatic pulmonary edema (HPE) who were mechanically ventilated in the prone position (Group 1). A group of patients with HPE, who were ventilated in the supine position, was studied as the control group (Group 2). Patients with ARDS (Group 3) and patients with pulmonary fibrosis (Group 4), both ventilated in supine and prone positions were studied as well. The common characteristic of all groups was persistent hypoxemia despite "optimal" ventilatory support.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The study was conducted in a 14-bed ICU setting, at the University Hospital of Ioannina (Ioannina, Greece). Patients with hydrostatic pulmonary edema requiring mechanical ventilation were studied on a consecutive and prospective basis. Patients with hemodynamic instability were excluded from the study. Patients were divided in the following groups.

Group 1 was composed of eight mechanically ventilated patients (seven men and one woman) with acute HPE, who were turned to the prone position. Patient demographics are shown in Table 1. Enrolled patients with HPE fulfilled the following criteria: Pa_O2  60 mm Hg and Sa_O2 < 90% with an inspired oxygen fraction equal to or more than 0.60 (FI_O2  0.6), regardless of positive end-expiratory pressure (PEEP), for at least 6 h after initiation of mechanical ventilation. Group 2 was composed of six intubated and mechanically ventilated patients with HPE, who fulfilled the criteria for prone positioning but were not turned to the prone position. Three of them had contraindications for prone positioning and the other three were not positioned prone owing to technical reasons. Three patients of this group fulfilled the criteria for prone positioning ~ 24 h after intubation. Two of them had no contraindication and they were turned to the prone position. This group of patients also served as control patients (Table 1).

Eight patients with HPE who fulfilled the criteria for intubation and mechanical ventilation but did not meet the criteria of prone positioning (Pa_O2 was more than 60 mm Hg with FI_O2  0.6, 6 h after initiation of mechanical ventilation) were excluded from the protocol.

Criteria for diagnosis of pulmonary edema. Sudden onset of severe dyspnea, bilateral rales, radiographic findings consistent with pulmonary edema in the absence of a history suggesting aspiration or infection (7). The diagnosis of HPE was confirmed by right heart catheterization.

Indications for intubation and mechanical ventilation. (1) Left ventricular failure refractory to medical therapy, (2) hypotension and/or severe hypoxemia, and (3) respiratory muscle fatigue (rapid and shallow breathing, exhaustion, rise in Pa_CO2).

Group 3 included 20 patients (16 men and 4 women) with ARDS (Murray lung injury score  2.5) (8). Patient demographics are shown in Table 2.

The standard diagnostic criteria of ARDS for inclusion in this study were as follows: (1) acute hypoxemic respiratory failure requiring mechanical ventilation, (2) diffuse bilateral alveolar infiltrates on the chest roentgenogram, (3) refractory hypoxemia (Pa_O2/FI_O2 < 200) regardless of PEEP level, (4) pulmonary artery wedge pressure < 18 mm Hg or no clinical evidence of left atrial hypertension, and (5) appropriate clinical setting or risk factor for the development of ARDS (9). All patients with ARDS exhibited a lung injury score greater than 2.5, as described by Murray and coworkers (8).

ARDS was characterized as early or late as follows:

Early ARDS. When  36 h elapsed from the onset of the clinical picture and diagnosis of ARDS

Late ARDS. When  36 h elapsed from the onset of the clinical picture and diagnosis of ARDS

Group 4 was composed of five mechanically ventilated patients (4 women and 1 man) with pulmonary fibrosis (PF) and severe hypoxemic respiratory failure. The demographics of patients with PF are shown in Table 2.

On the basis of the response to prone positioning, patients were characterized as responders (when their Pa_O2/FI_O2 increased by  20% during the first 30 min in the prone position) and nonresponders (when their Pa_O2/FI_O2 never increased by  20%). The patients who increased their Pa_O2/FI_O2 by  20% after the first 2 h were classified as late responders.

Prone Positioning

The prone positioning maneuver was executed in two steps: first side position and then prone. The whole procedure demanded the presence of four members of the nursing stuff and one physician. Before the maneuver, the patients were prepared by arranging the lines and the endotracheal tube. Thick pillows were put under the shoulders, the upper chest, and the pelvic area. These pillows prevent pressure sores and allow free abdominal wall motion and tension-free positioning of the head.

Measurements

The following parameters were recorded in every patient enrolled in the study: age, duration of mechanical ventilation, Pa_O2/FI_O2 ratio, arterial O₂ saturation (Sa_O2), mixed venous O₂ saturation (Sv_O2), Pa_CO2, pH, PEEP, tidal volume (VT), frequency (f), minute ventilation (VE), peak airway pressure (Ppeak), plateau pressure (Pplateau), compliance of respiratory system (Crs), pulmonary artery wedge pressure (Ppa,we), cardiac output (), cardiac index (CI), pulmonary artery pressure (Ppa), systemic artery pressure (Psa), and central venous pressure (Pcv). Systemic vascular resistance (Rsva), pulmonary vascular resistance (RL) and venous admixture (S/T) were calculated according to standard formulas (10).

Protocol

Patients were ventilated in a constant-flow, volume-control (CMV) or synchronized intermittent mandatory ventilation (SIMV) plus pressure support (PS) mode (900C; Siemens Servo, Solna, Sweden). Tidal volume was set at 6-8 ml/kg body weight, frequency was set at 15-25 per minute, PEEP was set just above the value of the lower inflation point (Pflex), and if the Pflex was invisible, PEEP was arbitrarily set at 10 cm H₂O. The pressure-volume curve was performed 20-60 min after intubation, when the patients were hemodynamically stable. Pplateau was kept lower than 35 cm H₂O. A pulmonary catheter (Opticath; Abbott, Abbott Park, IL), via an internal jugular or subclavian vein, and an arterial line were introduced. Patients who, after initiation of mechanical ventilation for at least 6 h, required an FI_O2  0.6 to achieve an Sa_O2 ~ 90%, independent of the level of the applied PEEP, and who did not respond to recruitment maneuvers (applied positive airway pressure 35-45 cm H₂O for 20-30 s), were turned to the prone position. Data were collected in the supine position just before patients were turned to the prone position. Data collection was repeated 30, 60, and 120 min and 6 h after the patients had been turned to the prone position. Data were collected again just before and 30 min after the patients were repositioned in the supine position. In Group 2 patients, who were not turned prone, the data, which were collected roughly 30 min after intubation (when the patients were hemodynamically stable), were used as baseline measurements. Data were also recorded at Hours 6, 7, 8, 12, and 24 after intubation in order to be comparable with the corresponding data of Group 1.

Patients were sedated (midazolam or propofol) and paralyzed (vecuronium) during data collection in the supine as well as the prone position. Paralysis was begun 15 min before data collection and withdrawn when it was no longer needed. During the first 2 h in the prone position, all ventilatory parameters were kept constant. A reassessment of PEEP and FI_O2 followed the Hour 2 measurements. The mode of mechanical ventilation was not changed until after Hour 6 of data collection. After that, and when Pa_O2/FI_O2 increased to more than 200, a pressure support (PS) or SIMV plus PS mode was initiated and sedation was discontinued or reduced. If the improvement in blood gases continued or no deterioration was observed during the next 12 h, patients were repositioned to the supine. The patients were extubated as soon as they were capable of unassisted breathing. The diuretics, inotrops, and vasodilators were given according to clinical indications and hemodynamic data. Fluid balance was guided by pulmonary artery wedge pressure.

The patients with HPE who were not turned prone (Group 2) were treated according to the preceding protocol, except for positioning. During the first 30 min after intubation PEEP was set arbitrarily and was usually less than 10 cm H₂O, because of hemodynamic instability due to intubation. When the hemodynamic instability was restored the pressure-volume curve was performed. PEEP was then set just above the value of the lower inflection point (Pflex), and if the Pflex was invisible, PEEP was arbitrarily set at 10 cm H₂O, as in Group 1.

The Ethics Committee of the University Hospital of Ioannina approved the protocol and the patients or their next of kin gave informed consent to the study.

Statistics

All data are expressed as means ± standard deviation. Comparisons between groups (multiple measurements over time and in different positions) were performed, using analysis of variance (ANOVA) for multiple measurements. Individual comparisons were performed with a nonparametric paired t test. The mortality of the patients was compared with the predicted mortality rate (PMR) from the APACHE II score (11). Data were analyzed using SPSS 8.0 software for Windows (SPSS, Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Data

Tables 1 and 2 depict age, sex, risk factor for developing HPE and ARDS, duration of mechanical ventilation in days (DMV), the predicted mortality rate (PMR), and outcome of our study population as well as the days of mechanical ventilation (MV) of patients with ARDS before these patients were turned to the prone position.

The application of prone positioning was not followed by any major side effects in any of the patient groups. Minor clinical side effects included facial edema as well as pressure sores and nasal bleeding in one patient.

Gas Exchange and Hemodynamics

Patients with HPE. Group 1, the prone position resulted in a significant increase in oxygenation and a decrease in S/T within 30 min compared with values just before prone positioning (p < 0.05). The improvement in oxygenation was continuing for the next 6-12 h. When the patients were repositioned to the supine, the oxygenation did not differ significantly from the last measurement of the prone position. PCO₂ decreased and pH increased significantly within 2 h compared with baseline, and PCO₂ decreased further when patients started to breathe spontaneously in the prone as well as in the supine position and sedation was reduced. Sv_O2 and CI increased significantly within 60 min (p < 0.05) and they did not change after repositioning of the patients in the supine. Heart rate (HR), mean systemic arterial pressure (, Pcv, and Rsva) did not change significantly. Mean pulmonary arterial pressure () and Ppa,we were significantly reduced after 6 and 12 h, respectively (p < 0.05). RL was reduced within 30 min (p < 0.01) at prone position (Table 3).

Group 2 fulfilled the criteria for prone positioning 6 h after initiation of mechanical ventilation, but patients were not initially turned to the prone position. As shown in Table 4, the mean values of the Pa_O2/FI_O2 ratio increased and S/T decreased after 12 h of mechanical ventilation in the supine position compared with baseline (p < 0.05). Twenty-four hours after initiation of mechanical ventilation, three patients exhibited a Pa_O2/FI_O2 ratio < 100. Prone positioning was contraindicated in only one of them, and the other two patients were turned prone. They immediately responded to prone positioning, increasing the Pa_O2/FI_O2 ratio > 30%. Twenty-four hours after prone positioning the mean value of the Pa_O2/FI_O2 ratio of these patients was significantly higher compared with that of patients of the same group who remained supine (Table 4). PCO₂ decreased and pH increased significantly at 12 and 24 h, respectively, compared with baseline. Ppa,we and were significantly reduced after 24 h of mechanical ventilation in the supine position. In contrast, HR, Pcv, and RL did not change significantly.

Comparison between Groups 1 and 2. The mean values of the Pa_O2/FI_O2 ratio in Group 2 were significantly lower than the corresponding values of patients in Group 1, who were turned prone (comparison of the values of both groups at Hours 7, 8, 12, and 24 after intubation and initiation of mechanical ventilation, p < 0.01). The S/T values exhibited a faster decrease in Group 1 after 1 h prone. During the first 2 h (8 h after intubation [AI]) of mechanical ventilation in the prone position (Group 1), PCO₂ values were higher compared with Group 2 values, but there was no significant difference between these groups after 12 h of mechanical ventilation (Tables 3 and 4). There was no significant difference in pH, Sv_O2, HR, Ppa,we, , Pcv, , and CI between Groups 1 and 2. The values of RL, after 8 h of mechanical ventilation, were found to be significantly lower in Group 1 (p < 0.05). Rsva was significantly higher in Group 1 compared with Group 2 during the entire protocol (p < 0.05).

Patients with ARDS (responders). Twelve of 15 patients exhibited a significant increase in oxygenation and a decrease in S/T, within 30 min in the prone position. Three of 15 patients had a late response, that is, after the first 2 h of prone positioning. The improvement in oxygenation continued for the next 12 to 48 h. When the patients were repositioned to the supine, a decrease in oxygenation was observed, but Pa_O2/ FI_O2 was significantly higher compared with baseline values. PCO₂ initially increased in the prone position, but it decreased continuously, reaching baseline values after 12 to 48 h. pH significantly increased after 12-48 h in the prone position. Sv_O2, HR, , Pcv, Ppa,we, CI, and Rsva did not change significantly. RL and were reduced within 6 h in the prone position (Table 5).

Patients with ARDS (nonresponders). Prone positioning did not affect oxygenation, early or even late, during 24 h. PCO₂ increased and pH decreased significantly within 2 h in the prone position. PCO₂ decreased and pH increased significantly when patients were repositioned to the supine, compared with the measurement that preceded repositioning. Sv_O2, HR, , Pcv, Ppa,we, CI, , Rsva, and RL did not change significantly between the prone and supine positions (Table 6).

Patients with fibrosis. Only PCO₂ increased significantly within 2 h in the prone position. No significant change was observed in any other parameter of gas exchange or hemodynamics, early or late, when patients were placed prone and then returned to the supine position (Table 7).

Comparison between Group 1 and other control groups. The significant differences between patients with HPE and responder and nonresponder ARDS as well as PF are marked in Tables 5, 6, and 7. The mean values of the Pa_O2/FI_O2 ratio in Group 1 were significantly lower during the first 2 h in the prone position than the corresponding values of patients in the responder-ARDS group (p < 0.01), but after 6 h in the prone position Pa_O2/FI_O2 ratios were similar in both groups of patients. The Pa_O2/FI_O2 ratio was higher in Group 1 in comparison with the nonresponder-ARDS and PF group throughout the duration of prone positioning. Although PCO₂ values exhibited a faster decrease in Group 1, there was no significant difference in PCO₂ values between Group 1 and the responder-ARDS group. After 2 h in the prone position, PCO₂ was found to be significantly lower in Group 1 than in the nonresponder-ARDS and PF groups. Ppa,we values were higher in Group 1 patients compared with other groups (ARDS responders and nonresponders, and patients with PF) during the first 7-8 h of mechanical ventilation. CI was lower in Group 1 patients compared with patients with ARDS (responders and nonresponders), but no difference was observed between Group 1 patients and patients with PF. The values of RL in Group 1 were significantly reduced during prone positioning compared with the corresponding values of the other groups. There was no difference in HR, , Pcv, and Sv_O2 between groups. CI values were lower in Group 1 patients compared with patients with ARDS (responders and nonresponders) and no difference was observed between Group 1 patients and patients with PF. There was no difference in between Group 1 patients and responder-ARDS patients. In contrast, the values of during prone positioning were significantly reduced in Group 1 patients compared with nonresponder-ARDS and PF patients.

Respiratory Mechanics

Patents with HPE. In Group 1, the prone position resulted in an initial nonsignificant decrease in Crs. This was followed by a gradual increase in Crs values, which were significantly increased 6 h after repositioning. Pplateau and total PEEP decreased significantly within 6 h of prone positioning, compared with baseline values (Table 3).

In Group 2, Crs started increasing after the initiation of mechanical ventilation, but only the Hour 24 measurements exhibited a significant increase compared with baseline values. Pplateau decreased, but not significantly. Total PEEP did not exhibit a significant change during 24 h of mechanical ventilation (Table 4).

Comparison between Groups 1 and 2. There was no significant difference in Pplateau between Groups 1 and 2 at comparable hours (Hour 7, 8, 12, and 24 measurements after intubation). With regard to PEEP values, 12 h after initiation of mechanical ventilation, PEEP values were set at a lower level in Group 1 than in Group 2 (p < 0.05). Although there was no significant difference in Crs between baseline measurements, 12 h after initiation of mechanical ventilation Crs was significantly higher in Group 1 compared with Group 2 (p < 0.05).

Patients with ARDS (responders). Crs was reduced and Pplateau increased significantly, when the patients were turned prone, in comparison with baseline (supine). After 6 h in the prone position, the values of Crs and Pplateau reverted to baseline levels. When patients were repositioned in the supine, the values of Crs were significantly higher and the values of Pplateau were significantly lower compared with baseline, as well as with the last prone position measurement. The values of total PEEP were gradually reduced, and the decrease became significant after 12 to 48 h (Table 5).

Patients with ARDS (nonresponders). Crs decreased and Pplateau increased within 30 min, when the patients were turned to the prone position, in comparison with baseline. When patients were repositioned in the supine, the values of Crs were higher and the values of Pplateau were lower compared with the last prone position measurement (Table 6).

Patients with fibrosis. Crs decreased and Pplateau increased when the patients were turned prone, in comparison with baseline. When patients were repositioned in the supine, the values of Crs were higher and the values of Pplateau were lower compared with the last prone position measurement (Table 7).

Comparison between Group 1 and other control groups. After 1 to 2 h of mechanical ventilation in the prone position, Crs was higher and Pplateau lower in Group 1 patients in comparison with patients with ARDS (responders and nonresponders) and PF. PEEP values in Group 1 patients were lower compared with other groups, after 12-48 h in the prone position (Tables 567).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

All patients with HPE and 75% of patients with ARDS exhibited improvement of oxygenation when positioned prone; the rise in Pa_O2 was persistent, without detrimental effect on hemodynamics. In contrast, none of the patients with PF responded favorably to prone positioning. Patients with HPE and early ARDS responded better to prone positioning than did patients with late ARDS and PF. Patients with HPE and ventilated in the supine position had lower Pa_O2/FI_O2 ratios and remained longer in mechanical ventilation in comparison with patients with HPE who were ventilated in the prone position.

The results of this study demonstrated an improvement in mortality rate for patients with HPE who were turned prone, as well as for patients with ARDS who were turned prone, compared with the predicted mortality. However, the small number of patients in the HPE group makes this comparison powerless to draw a definite conclusion about the influence of prone positioning on outcome.

The mechanism by which the prone position improves oxygenation is related to the reduction of shunting and the correction of ventilation-perfusion (/) heterogeneity (12). Since regional perfusion is relatively unaffected by a change from the supine to the prone position, the beneficial effect of the prone position on / matching should be due to a redistribution of ventilation. It has been postulated (12, 13) that oxygenation may be improved by the effect of prone position on the gravitational gradient of pleural pressure. Pleural pressure is more uniform in the prone compared with the supine position. When lung edema occurs, pleural pressure in the dependent lung regions becomes positive in the supine position, but it is less positive in the prone. These findings suggest that in the supine position patients with pulmonary edema may find their dorsal lung regions below closing volume; in the prone position the less positive pleural pressure in dependent regions could reduce the number of lung units that are below closing volume (12). Mutoh and coworkers (13) showed in pigs that the reduction in the gravitational pleural pressure gradient in the prone position compared with the supine was even more pronounced after volume infusion. Pappert and coworkers (14), however, found no correlation between the response to prone positioning and extravascular lung water, and they stated that extravascular lung water probably does not represent the excess tissue mass that correlates with the amount of superimposed pressure (14, 15).

Our observations are in agreement with the preceding mechanism by which prone positioning improves oxygenation. The majority of patients with pronounced pulmonary edema, such as those with HPE and early-phase ARDS, exhibited improved oxygenation when they were turned prone. In contrast, patients with a prominent alveolar capillary block (fibrosis), such as in late ARDS and PF, did not respond. There is no good explanation as to why some patients with late ARDS responded and some with early ARDS did not. This could be due to the classification of ARDS, which is quite arbitrary, and might not correspond accurately to pathophysiological alterations during the course of ARDS. It is noteworthy that the patients with late response had late-phase ARDS and the causative factor of ARDS was some direct insult to the lungs.

Another factor that could adversely affect ventilation in the dorsal region of the lung during supine positioning is the heart, especially when it is big. Wiener and colleagues (16) showed that patients with cardiomegaly exhibited reduced left mid- and lower zone ventilation in the supine position; that was probably due to the mechanical effect of the heart compressing the left lung. Prone positioning allows the heart to lay on the sternum and the mechanical effect on the left lower lobe is relieved. The compressing effect of the heart on dorsal lung regions has also been demonstrated in large animal models (17, 18). Our patients with congestive heart failure and cardiomegaly exhibited a significant, rapid, and persistent improvement in oxygenation. This improvement could be partly due to the decompression of the left lower lobe by the enlarged heart during prone positioning. The persisting improvement after turning the patient to the supine position might be due to a substantial decrease in pulmonary edema as well as to the relatively decreased heart size, because of the treatment.

The improvement in / matching due to prone positioning allows the clinician to apply milder transalveolar stretching forces to achieve the same gas exchange (19). Furthermore, prone positioning seems to contribute to a reduction of duration of mechanical ventilation in patients with HPE. Since ventilator-induced lung injury (VILI) remains one of the major problems of ventilating patients with low compliance, the maneuver could be employed as a means of prevention or inhibition of VILI. Mechanical ventilation in the prone position could itself prevent or reduce ventilator-induced lung injury. Broccard and colleagues observed that after oleic acid lung injury, animals ventilated with a high tidal volume and PEEP undergo less extensive histological change in the prone than in the supine position (19).

Stocker and colleagues (20) reported low mortality in patients with ARDS when they used a protocol similar to ours, including low peak inspiratory pressure and prone position. Apart from prone positioning several other factors could potentially contribute to these favorable results, such as low tidal volume and PEEP. Furthermore, Amato and colleagues (21) showed a decrease in mortality when using a low tidal volume, in comparison with using a large tidal volume, in patients with ARDS.

Patients with HPE exhibited a significant reduction in Pa_CO2 during the first 6 h in the prone position, compared with the supine. Provided that the delivered ventilation was kept constant, the patients were sedated and paralyzed, and the CO₂ production (VCO₂) remained more or less steady, the reduction in Pa_CO2 was due to the increase in alveolar ventilation in well-perfused areas. Similarly, in the responder-ARDS subgroup of patients, Pa_CO2 initially increased, but decreased 24- 48 h after they were turned prone. In contrast, in the nonresponder-ARDS subgroup of patients, as well as in patients with PF, increased values were demonstrated when they were turned prone, and this increase remained during prone positioning. Pelosi and co-workers (22) did not exhibit any change in Pa_CO2 values of ARDS patients, 2 h after they were turned to the prone compared to the supine position. Pappert and colleagues (14) found increased Pa_CO2 values when patients with ARDS were turned to the prone position, compared with the supine. This increase persisted for the whole time of prone positioning, and similar results were observed in our nonresponder-ARDS and PF groups. The observed discrepancy of results between the preceding studies and our groups of HPE and ARDS responders might be due to differences in study populations, in allocation of patients with ARDS into subgroups, as well as in protocol used. In our study, the patients were kept in the prone position much longer than in the preceding studies.

Of note is that in HPE (Group 1), alveolar ventilation exhibited a faster improvement compared with the responder-ARDS group. In contrast, oxygenation increased faster in the responder-ARDS group. Although there is no obvious explanation for this phenomenon, it could be related to the lower cardiac output in Group 1 patients. The decompression of the left lower lobe by the enlarged heart of Group 1 patients could initially result in rapid removal of CO₂, while owing to the low cardiac output the perfusion could not match the ventilation.

The hemodynamic variables during prone position remained unchanged with the exception of the HPE group, in which the cardiac index increased significantly. This increase could play a major role in the improvement of / matching that was observed in this group. The increase in CI could be related to a reduction in RL as well as to inotropic agents administration. Of note is that in the control group of patients with HPE, RL did not change significantly during ventilation in the supine position. The reduction in pulmonary vascular resistance by improving the performance of the right ventricle could have a beneficial effect on the diastolic function of the left ventricle. Furthermore, the improvement of oxygenation as well as inotropic agents could affect the systolic function of the left ventricle. The reduction in pulmonary vascular resistance in HPE and responder-ARDS patients probably is related to the decreased hypoxic pulmonary vasoconstriction, owing to recruitment of atelectatic areas (23). In patients with HPE the reduction in RL could partially be the result of the increase in cardiac output. As regards the responder-ARDS and HPE patients, the improvement in RL may also be related to the reduction in PEEP.

The initial decrease in Crs, when the patient is turned prone, is probably due to a decrease in chest wall compliance (22). In contrast, the improvement in Crs during prone positioning seems to relate to the increase in lung compliance. The absolute increase in lung compliance in patients with HPE could be explained by the rapid absorption of hydrostatic pulmonary edema fluid and the decompression of the lower left lobe. These results are consistent with those of Pelosi and colleagues (22). However, Blanch and coworkers (24) found that responder-ARDS patients exhibited increased Crs after 20 min in the prone position. These results are inconsistent with ours and those of Pelosi and colleagues (22). There is no good explanation for this discrepancy, which may be due to a different study population and protocol. Patients with HPE who were ventilated prone exhibited an increase in Crs compared with baseline measurements. This finding was not observed in responder-ARDS patients. One explanation for this increase is the decollapse of the left lower lobe in the prone position. The enlarged hearts of patients with HPE could compress the left lower lobe more than the normal-sized hearts of patients with ARDS. Furthermore, the excess lung water is more easily removed in patients with HPE rather than ARDS.

In summary, our results show that the prone position may be a useful maneuver in treating patients with severe hypoxemia due to pulmonary edema. The simplicity, safety, and effectiveness of the method suggest that it can be used routinely. Furthermore, our results increase the clinical utility of the maneuver, since they extend the indications of its use (i.e., to patients with HPE). The existence of pulmonary edema, as in ARDS and HPE, is a predictor of the beneficial effect of prone positioning on gas exchange. In contrast, the presence of fibrosis (as in late ARDS and pulmonary fibrosis) predisposes to nonresponsiveness to prone positioning.